1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.43:
3 * enhancement: RUN-PROGRAM accepts :EXTERNAL-FORMAT argument to select the
4 external-format for its :INPUT, :OUTPUT, AND :ERROR :STREAMs.
5 * enhancement: ALLOCATION-INFORMATION also provides the page the object
7 * enhancement: default dynamic-space size can be configured at build-time
8 without touching source, using the --dynamic-space-size argument to make.sh.
9 * enhancement: DESCRIBE recognizes symbols naming optimization policy
10 qualities, and mentions ALWAYS-BOUND declarations.
11 * enhancement: ATOMIC-INCF now supports AREF of (SIMPLE-ARRAY SB-EXT:WORD (*))
13 * enhancement: ASDF has been updated to 2.009.
14 * enhancement: the system detects known type-erros in calls better,
15 signalling a full warning about violated proclaimed FTYPEs and violations
16 of derived FTYPEs within the same file, including self-calls.
17 * enhancement: new function: SB-EXT:DELETE-DIRECTORY is now provided.
18 * optimization: constant-folding exploits numeric and character types, in
19 addition member types.
20 * optimization: numeric, character and member types that are inhabited by
21 exactly one value are tested with EQL.
22 * optimization: more conditional branches are eliminated during IR1.
23 Branches are simplified before performing if/if-conversion, and simple
24 equivalent branches (that only read the same constant or variable) are
26 * improvements to the Windows port:
27 ** change: canonical unparsing form for pathname namestrings now uses / as
28 directory separator. NATIVE-NAMESTRING still uses \ as the separator.
29 ** bug fix: stackoverwriting due to incorrect usage of PeekConsoleInput
30 on Windows. (thanks to Kalyanov Dmitry)
31 ** bug fix: build now works on cygwin with GCC 4.x installed. (thanks to
33 ** bug fix: run-sbcl.sh now works on Cygwin. (thanks to Kalyanov Dmitry)
34 * bug fix: compiler failed to derive the result-type of MAKE-ARRAY as
35 (AND VECTOR (NOT SIMPLE-ARRAY)) when appropriate. (lp#309130)
36 * bug fix: (THE (VALUES ...)) in LOAD-TIME-VALUE caused a compiler-error.
38 * bug fix: interrupts arriving during CL:OPEN caused an error.
39 * bug fix: overeager character buffering could cause input to block
40 spuriously when reading from a pipe (lp#643686)
41 * bug fix: more efficient timer expiry should avoid starvation on systems
42 where number of SIGALRMs per second is restricted. (lp#375515)
43 * bug fix: non-unicode builds no longer fail (broken since 1.0.36.15).
44 * bug fix: compile-times no longer scale linearly with the size of
45 quoted lists in source-code. (lp#654289)
46 * bug fix: better errors for unknown external-formats in OPEN. (lp#561974)
47 * bug fix: DEBUG>SPEED no longer degrades ALIEN-FUNCALL performance quite
49 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCES-BY-NAME no longer signals
50 an error for eg. STRUCTURE. (lp#458015)
51 * bug fix: LOOP WITH NIL = ... signalled an unused variable style-warning.
52 (lp#613871, thanks to Roman Marynchak)
53 * bug fix: more reliable &REST list type derivation. (lp#655203)
54 * bug fix: PPRINT-LOGICAL-BLOCK multiply-evaluated :PER-LINE-PREFIX,
55 and issued pointles code-deletion notes for it, :PREFIX, and :SUFFIX.
56 * bug fix: the compiler didn't utilize the proclaimed ftype for functions
57 also declared NOTINLINE. (lp#655581)
58 * bug fix: the compiler could attempt to emit constant left shifts of
59 greater value than n-word-bits. (lp#309063)
60 * bug fix: (SETF SYMBOL-FUNCTION) and (SETF FDEFINITION) clear derived
61 type information for the function being set. (lp#659220)
62 * bug fix: the system used to signal bogus STYLE-WARNINGs when functions
63 containing self-calls were recompiled with a new signature, and failed
64 to warn when a self-call using the old signature was left in. (lp#655126)
65 * bug fix: incorrect FILE-POSITION on streams opened using :EXTERNAL-FORMAT
66 :DEFAULT when the default external had character size other than 8 bits.
68 * bug fix: derived types of functions in the same file were implicitly
69 trusted, leading to potential heap corruption when those function were
70 defined incompatibly. (lp#657499)
71 * bug fix: existing ASDF source registries are ignored when building
73 * bug fix: short-form DEFSETF checks that the second argument is a symbol
74 (lp#655824, thanks to Roman Marynchak)
75 * bug fix: (SETF MACRO-FUNCTION) clobbered macro-definitions before
76 package-lock violation was detected. (lp#660752)
77 * bug fix: the system can store different docstrings for a single function
78 under different names. (lp#661631, regression from 1.0.29.24)
79 * bug fix: source-locations of DEFGENERIC forms weren't getting recorded
81 * bug fix: (COERCE X '(SIMPLE-VECTOR 5)) and similar coercions to vectors
82 of specified length could confuse the type derivation in unsafe code.
84 * bug fix: SB-POSIX:FORK now signals an error if an attempt to
85 fork with multiple Lisp threads running is made on every platform,
86 not just Darwin. (lp#451111)
87 * bug fix: SB-BSD-SOCKETS and SB-SIMPLE-STREAMS could leave stale finalizers
88 around when core was saved, which could cause undefined behaviour when the
90 * bug fix: Allow UNTRACE of functions that are no longer defined
91 (lp#667657, thanks to Attila Lendvai)
93 changes in sbcl-1.0.43 relative to sbcl-1.0.42:
94 * incompatible change: FD-STREAMS no longer participate in the serve-event
95 event-loop by default. (lp#316072)
96 ** In addition to streams created by explicit calls to MAKE-FD-STREAM this
97 affects streams from CL:OPEN.
98 ** Streams from SOCKET-MAKE-STREAM still participate in serve-event by
99 default, but this is liable to change: applications needing serve-event
100 for socket streams should explicitly request it using :SERVE-EVENTS T
102 * enhancement: SB-EXT:WORD type is provided for use with SB-EXT:ATOMIC-INCF
104 * enhancement: CLOS effective method functions and defclass slot typechecking
105 function now have debug names for use in backtraces and profiles.
106 * enhancement: ASDF has been updated to version 2.004. (lp#605260, thanks to
108 * enhancement: symbols are printed using fully qualified names in several
109 error and warning messages which are often associated with package
110 conflicts or mixups (lp#622789, thanks to Attila Lendvai)
111 * bug fix: SB-BSD-SOCKETS:SOCKET-CONNECT was not thread safe. (lp#505497,
112 thanks to Andrew Golding)
113 * bug fix: reading /proc files on Linux works. (lp#425199)
114 * bug fix: DOTIMES accepted literal non-integer reals. (lp#619393, thanks to
116 * bug fix: WRITE-TO-STRING compiler macro binding special variable names,
117 breaking code that tried to write the value of a printer control variable.
118 (lp#581564, thanks to Stas Boukarev)
119 * bug fix: WRITE compiler macro did not handle output stream designators
120 correctly. (lp#598374, thanks to Stas Boukarev)
121 * bug fix: better availability of names of foreign functions in backtraces
122 on Linux. (lp#626962, thanks to Stas Boukarev)
123 * bug fix: scripting and build for Solaris and FreeBSD. (lp#615497,
124 lp#627581, thanks to Josh Elsasser and Jim Wise)
125 * bug fix: build fixes for OpenBSD -current and 4.8 (lp#615489, lp#615492,
126 thanks to Josh Elsasser)
127 * bug fix: using aliases for builtin classes as defmethod specializers
128 without adding DEFTYPEs for them works. (lp#618387)
129 * bug fix: timetravel by getrusage() no longer causes type-errors during GC.
131 * bug fix: legally dynamic-extent lists and vectors used as
132 initialization arguments to MAKE-ARRAY can be stack allocated. (lp#586105)
133 * bug fix: inline-expansion creating references to dead lambda-variables
134 (lp#454681, thanks to Alexey Dejneka)
135 * bug fix: better error message for bogus numerical arguments to RANDOM.
136 (lp#598986, thanks to Stas Boukarev)
137 * bug fix: the compiler occasionally inlined references from incompatible
138 environments occurs. (lp#308951)
139 * bug fix: the compiler threw an error when trying to compile a local
140 function (labels or flet) known to take a specialized complex argument.
141 (not in launchpad, reported by sykopomp in #lispgames)
142 * bug fix: package-locks failed to protect against compile-time effects of
143 DEFUN when the symbol previously had a macro definition. (lp#576637)
144 * bug fix: spurious ignore warnings even given (DECLARE IGNORE) in methods
145 when parameter bindings mutated. (reported by Faré Rideau; lp #611361)
146 * bug fix: workaround for compiler hang in ORDER-UVL-SETS (lp#308914)
147 * bug fix: evaluation in debugger REPL works using the global context when
148 in frames that do not have sufficient debug information.
149 * bug fix: exceeding FD_SETSIZE limit now results in an sensible error
151 * bug fix: (SETF DOCUMENTATION) of a macro works properly. (lp#643958, thanks
153 * bug fix: interrupt taking longer than the requested period caused SLEEP
154 to hang on Darwin. (lp#640516, thanks to Joe Lebroco for the analysis)
156 changes in sbcl-1.0.42 relative to sbcl-1.0.41
158 ** Cross-compilation host is now specified to make.sh using
159 command-line argument --xc-host=<command> instead of a positional
160 argument. (thanks to Daniel Herring)
161 ** Install location can be specified to make.sh using command-line
162 argument --prefix=<path>. (lp#550889s, thanks to Daniel Herring)
163 * optimization: The default implementation of
164 COMPUTE-DISCRIMINATING-FUNCTION does much less wasted work.
165 * enhancement: Explicit memory barrier operations are now available for use
166 by multithreaded code. See documentation for details.
167 * enhancement: Experimental support for threading on Linux/PPC.
168 * bug fix: RENAME-PACKAGE returns the package. (Thanks to Eric Marsden)
169 * bug fix: EXPT signals an error if first argument is a zero and second
170 argument is a floating point zero. (lp#571581, thanks to Roman Marynchak)
171 * bug fix: DEFTYPE signals an error for non-list lambda-lists.
172 (lp#576594, thanks to Roman Marynchak)
173 * bug fix: make ASDF-INSTALL compatible with the now-included ASDF2.
174 (lp#612998, reported by Phil Hargett; patch from Jim Wise)
176 changes in sbcl-1.0.41 relative to sbcl-1.0.40:
177 * optimization: validity of observed keyword initargs to MAKE-INSTANCE is
178 cached, leading to many fewer expensive calls to
179 COMPUTE-APPLICABLE-METHODS.
180 * optimization: in the (unoptimized) general method for MAKE-INSTANCE on a
181 CLASS argument, search for and call an appropriate optimized ctor function
183 * bug fix: WRITE always returns the correct value.
185 changes in sbcl-1.0.40 relative to sbcl-1.0.39:
186 * bug fix: readdir now works on :inode64 darwin builds (lp#592897)
187 * bug fix: Name conflicts between symbols passed as arguments to a single
188 call to IMPORT no longer add multiple symbols with the same name to the
189 package (detectable via DO-SYMBOLS).
190 * bug fix: support building without the dlshim on darwin x86 and x86-64
192 * bug fix: TRACE :ENCAPSULATE NIL now works on ppc/linux.
194 changes in sbcl-1.0.39 relative to sbcl-1.0.38:
195 * bug fix: Backtrace from undefined function on x86 and x86-64 now show
197 * bug fix: linkage-table entries on PPC now no longer overflow their
198 space allocation (potentially causing crashes if they are written out
200 * bug fix: Scrub control stack after scavenging in gencgc on non-x86oid
201 platforms, preventing the GC from seeing stale pointers on the control
202 stack in subsequent GCs (which would, and does, break invariants).
203 * bug fix: 32-bit unicode external formats now work on big-endian systems.
204 * bug fix: Literal characters with code points greater than about 32767
205 now work on PPC UNICODE builds.
206 * bug fix: Any noise left by SSE operations (sqrt and conversions) in the
207 high order bits are explicitly cleared out. In some contrived situations,
208 this could lead to wrong results in mixed real/complex float arithmetic.
209 * bug fix: Fix function/macro redefinition warnings when building with
210 clisp. (lp#576787, thanks to Josh Elsasser)
211 * new platform: experimental support for ppc/openbsd (thanks to Josh
213 * bug fix: Floating-point traps now work on ppc/linux.
215 changes in sbcl-1.0.38 relative to sbcl-1.0.37:
216 * incompatible change: Thread names are now restricted to SIMPLE-STRINGs
217 like for any other thread-related datastructure, MUTEX, etc. (lp#547095)
218 * deprecation: the SB-QUEUE contrib was merged into the SB-CONCURRENCY
219 contrib module. New code should depend on SB-CONCURRENCY, not SB-QUEUE.
220 * deprecation: SB-THEAD:GET-MUTEX was deprecated in favor of
221 SB-THREAD:GRAB-MUTEX.
222 * new contributed module: SB-CONCURRENCY is a new contrib; it's supposed to
223 contain additional data structures and tools for concurrent programming;
224 at the moment it contains a lock-free queue, and a lock-free mailbox
226 * new feature: added SB-THREAD:GRAB-MUTEX; it's like the now deprecated
227 GET-MUTEX but takes &key rather than &optional parameters. Also added
228 :TIMEOUT argument to GRAB-MUTEX on non-sb-lutex platforms like Linux and
230 * new feature: added SB-THREAD:TRY-SEMAPHORE, a non-blocking variant of
231 SB-THREAD:WAIT-ON-SEMAPHORE.
232 * new feature: SB-EXT:ATOMIC-DECF has been added as a companion to
234 * new feature: a CANCEL-DEADLINE is associated with DEADLINE-TIMEOUT
235 conditions to defer the deadline for forever.
236 * enhancement: *STANDARD-OUTPUT*, *STANDARD-INPUT*, and *ERROR-OUTPUT* are
238 * enhancement: errors from NO-APPLICABLE-METHOD and
239 NO-PRIMARY-METHOD now have a RETRY restart available to retry the
240 generic function call.
241 * enhancement: SB-BSD-SOCKET improvements
242 ** sockets and socket streams now have a more informative printed
243 representation based on the corresponding SOCKET-NAME and
245 ** SOCKET-MAKE-STREAM once more supports the :AUTO-CLOSE option.
247 ** SOCKET-CLOSE now accepts :ABORT argument, which is passed on to
248 CL:CLOSE when appropriate, and no longer disassociates the stream
249 from the socket if close failed. (lp#543951)
250 * improvements to the instrumenting profiler
251 ** new feature: report per-function GC overhead. (thanks to John Fremlin)
252 ** optimization: counters no longer use locks for the overflow mode.
253 ** bug fix: whenever a profiling counter wrapped into overflow mode, it
254 incurred an off-by-one miscount.
255 * enhancement: improved MAKE-HASH-TABLE documentation (lp#543473)
256 * enhancement: improved DEFMETHOD pretty-printing.
257 * enhancement: perform range reduction when arguments are too large for
258 x87's transcendentals (instead of returning 0). (lp#327192)
259 * enhancement: eliminate some spurious TYPE-WARNINGs. Should help with
260 some of CL-PPCRE's macros. (lp#570079)
261 * enhancement: our machine code is slightly less hostile to valgrind on
263 * enhancement: up-to-date versions of NetBSD-current are supported. (Thanks
264 to Robert Swindells and Aleksej Saushev)
265 * bug fix: correct restart text for the continuable error in MAKE-PACKAGE.
266 * bug fix: a rare case of startup-time page table corruption.
267 * bug fix: a semaphore with multiple waiters and some of them unwinding due
268 to timeouts could be left in an inconsistent state.
269 * bug fix: fix typo in "Reporting Bugs" section of the manual (lp#520366)
270 * bug fix: misoptimization of multiplication by one in
271 (SB-C::FLOAT-ACCURACY 0) policies.
272 * bug fix: miscounts in SB-PROFILE.
273 * bug fix: Fix lost wakeup bug between SB-THREAD:CONDITION-WAIT and
274 CONDITION-NOTIFY on Linux. See threads "lost wakeup in condition-wait /
275 condition-notify" (Feb 2010) and "Condition-Wait, Deadline handler, waking
276 up itself" (March 2010) for further details.
277 * bug fix: allow forward FIND and POSITION on lists to elide checking :END
278 against length of the list if the element is found before the specified
279 END is reached. (thanks to Alec Berryman, lp#554385)
280 * bug fix: errors signalled during package graph modification no longer
281 block FIND-SYMBOL and FIND-PACKAGE in other threads. (lp#511072)
282 * bug fix: SB-POSIX build was broken when SBCL was compiled without the
283 :SB-DOC feature. (lp#552564)
284 * bug fix: SB-INTROSPECT build issues on GENGC/PPC. (lp#490490)
285 * bug fix: more robust runtime executable path detection. (lp#375549)
286 * bug fix: GCD always returns positive values. (lp#413680)
287 * bug fix: Converting division to multiplication by reciprocal handles
289 * bug fix: We were too eager in eliding range reduction tests on x87.
290 The maximal magnitude is 2^63, not 2^64.
291 * bug fix: Transforms for TRUNCATE don't die when the result is completely
293 * bug fix: Maybe restore buildability on Alpha.
294 * bug fix: READ-BYTE isn't inline anymore, fixing weird streams failures.
296 * bug fix: RANDOM-STATE can be printed readably again.
297 * bug fix: Unreadable objects were sometimes printed like #<\nFoo>.
298 * bug fix: Using EQL with non-constant values of constant type (e.g. EQL
299 types) could result in type mismatches during compilation.
301 changes in sbcl-1.0.37 relative to sbcl-1.0.36:
302 * enhancement: Backtrace from THROW to uncaught tag on x86oids now shows
303 stack frame thrown from.
304 * enhancement: WITH-COMPILATION-UNIT :POLICY allows restricting changes to
305 compiler optimization qualities inside dynamic extent of its body.
306 * enhancement: LOAD-LOGICAL-PATHNAME-TRANSLATIONS can be used to load
307 translations from SYS:SITE;<HOST>.TRANSLATIONS.NEWEST (thanks to Michael
309 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) take advantage of
310 constraint propgation, allowing better compilation eg. when used to
311 access structures with WITH-SLOTS. (lp#520366)
312 * optimization: the compiler is now more aware of the type of the underlying
313 storage vector for multidimensional simple arrays resulting in better code
314 for accessing such arrays.
315 * optimization: passing NIL as the environment argument to TYPEP no longer
316 inhibits optimizing it. (lp#309788)
317 * optimization: more efficient register usage when handling single-float
318 arguments on x86-64. (thanks to Lutz Euler)
319 * optimization: ADJUST-ARRAY and STABLE-SORT on vectors no longer use
320 pre-allocated temporary vectors. (lp#496249)
321 * bug fix: Fix compiler error involving MAKE-ARRAY and IF forms
322 in :INITIAL-CONTENTS. (lp#523612)
323 * bug fix: FUNCTION-LAMBDA-EXPRESSION lost declarations from interpreted
324 functions. (lp#524707)
325 * bug fix: bogus style warnings from certain (SETF SLOT-VALUE) and
326 WITH-SLOTS usages during compilation.
327 * bug fix: SB-C::CLASS-INFO now prints correctly. (lp#514762)
328 * enhancement: Can now build with ud2 instead of int3 as trap instruction on
329 all x86oid platforms with :UD2-BREAKPOINTS target feature.
330 * bug fix: Breakpoints now work when using ud2 instead of int3 as trap
331 instruction (tested on x86oid linux with ud2-breakpoints).
332 * bug fix: slam.sh now works on win32.
333 * bug fix: better differences of numeric types. (lp#309124)
334 * bug fix: arrays declared intersection and union types can have their
335 upgraded element type derived. (lp#316078)
336 * bug fix: SB-SPROF allocation profiling for all threads failed to profile
337 threads started during profiling. (lp#472499)
338 * bug fix: SB-INTROSPECT test failure when building without SB-EVAL feature.
340 * bug fix: SB-CLTL2:DECLARATION-INFORMATION did not take
341 SB-EXT:RESTRICT-COMPILER-POLICY into account. (lp#313337)
342 * bug fix: Comma inside a backquoted array or structure resulted in nonsense
343 values instead of signaling an error. (lp#309093)
344 * bug fix: Spurious unused variable warning in a DEFSTRUCT edge case.
346 * bug fix: More consistent warnings and notes for ignored DYNAMIC-EXTENT
347 declarations (lp#497321)
348 * bug fix: FIND and POSITION on lists did not check sequence bounds properly
349 and failed to detect circular lists (lp#452008)
350 * bug fix: leakage from ~/.asdf-install into the ASDf-INSTALL contrib build
352 * bug fix: LOOP OF-TYPE VECTOR compile-time error. (lp#540186)
353 * bug fix: SIGNAL SB-SYS:INTERACTIVE-INTERRUPT before entering the debugger
354 due to it, so that handlers can run.
355 * bug fix: reparsing undefined types if they have become defined since
357 * bug fix: missing &REST type in a proclamation for a function with both
358 &REST and &KEY in lambda-list caused miscompilation (lp#458354)
359 * bug fix: WHO-CALLS information for source-transformed and compiler-macro
360 expanded calls (lp#542174)
361 * bug fix: more accurate WHO-MACROEXPANDS information; point into rather
362 than just at toplevel form.
364 changes in sbcl-1.0.36 relative to sbcl-1.0.35:
365 * new feature: SB-EXT:TYPEXPAND-1, SB-EXT:TYPEXPAND, and
366 SB-EXT:TYPEXPAND-ALL behave exactly like their MACROEXPAND counterparts
367 but work on type specifiers.
368 * new feature: SB-EXT:DEFINED-TYPE-NAME-P returns whether a symbol is known
369 to name a type specifier.
370 * new feature: SB-EXT:VALID-TYPE-SPECIFIER-P returns whether a given type
371 specifier is valid where "valid" basically means "would be accepted as
372 second argument of TYPEP".
373 * new feature: SB-INTROSPECT:FUNCTION-TYPE takes a function-designator and
374 returns the function's declared, or derived FTYPE.
375 * new feature: SB-POSIX now supports accessing the d_ino member of
376 dirent structures. (Thanks to Philipp Marek and Pierre THEIRRY)
377 * new feature: The function SB-EXT:SEED-RANDOM-STATE has been added to
378 provide for seeding a RANDOM-STATE object with user-provided data or
379 from the operating system's PRNG. Also, (MAKE-RANDOM-STATE T) will
380 attempt to initialize the returned state from the operating system's
381 PRNG where possible. (Thanks to Fare Rideau; launchpad bug lp#310116)
382 * bug fix: Fix SB-SIMPLE-STREAMS:READ-VECTOR to correctly set the
383 FILE-POSITION of the stream being read from. (launchpad bug lp#491087)
384 * bug fix: Fix grammar and style issues for the docstrings of
385 printer-related variables and functions. (Thanks to mon_key; launchpad
387 * bug fix: Fix compilation on chenygc platforms. Thanks to Larry Valkama and
389 * bug fix: SB-THREAD:CONDITION-WAIT sometimes signaled a deadline twice
390 in a row even though a handler defered the deadline long into the
392 * bug fix: A deadline handler was run without interrupts enabled for a
393 deadline signaled within SB-THREAD:CONDITION-WAIT. That could result
394 in infinitely spinning, non-killable threads.
395 * bug fix: Backtrace from internal-errors on x86-64 os x was truncated
396 before reaching the erring stack frame.
397 * bug fix: Fix type derivation for EXPT when raising a fixnum to a
398 real power. (launchpad bug lp#525949)
399 * bug fix: Fix SB-EXT:GENERATION-* accessors for generations > 0 on
400 GENCGC platforms. (launchpad bug lp#529014)
401 * bug fix: More robust checks for invalid DEFMETHOD argument specializers.
402 (launchpad bug lp#525916)
403 * bug fix: Fix building on Darwin when sysctl is not in the user's PATH.
404 (Thanks to Robert Goldman)
406 changes in sbcl-1.0.35 relative to sbcl-1.0.34:
407 * optimization: ROUND with a single single-float or double-float argument
408 is properly inlined when possible.
409 * optimization: Slightly better code is generated for integer<->float
410 conversions and for single-float<->double-float conversions on x86-64.
411 * optimization: SB-ROTATE-BYTE:ROTATE-BYTE now generates more efficient
412 code for 32-bit and 64-bit rotations on x86-64.
413 * bug fix: The install script changes the ownership of directories as well
414 as files for contrib modules using asdf. (thanks to Eugene Ossintsev;
415 launchpad bug lp#508485)
416 * bug fix: TRUNCATE with a single single-float or double-float argument is
417 properly inlined when possible. (launchpad bug lp#489388)
418 * bug fix: Passing a rotation count of zero to SB-ROTATE-BYTE:ROTATE-BYTE
419 no longer causes a compiler error on x86 and ppc.
420 * bug fix: GET-MACRO-CHARACTER bogusly computed its second return value
421 always relative to *READTABLE* rather than the passed argument.
423 changes in sbcl-1.0.34 relative to sbcl-1.0.33:
424 * minor incompatible change: threading support is now enabled by default
426 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
428 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
429 are now caught and reported just like errors during macroexpansion.
430 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
431 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
432 * enhancement: ASDF systems can now depends on SB-QUEUE.
433 * fixes and improvements related to Unicode and external formats:
434 ** bug fix: error handling and restart usage in the ucs-2 external format
436 ** there is now an implementation of the ucs-4 external format.
437 ** the utf-16 and utf-32 external formats are supported.
438 * bug fix: SB-POSIX wrapper for putenv no longer tries to put lisp strings
439 in the environment. setenv() and unsetenv() are also provided. (reported by
440 Fare Rideau; launchpad bug lp#460455)
441 * bug fix: LOAD of both .fasl and .FASL type files now forces fasl-style
442 loading. This ensures sensible errors for .FASL files from other
443 implementations on case-insensitive filesystems. (reported by Willem
444 Broekema; launchpad bug lp#489417)
445 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
446 Willem Broekema; launchpad bug lp#489698).
447 * bug fix: some minor code rearrangements to reenable warning-free building
448 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
449 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
451 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
454 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
455 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
456 * improvement: support O_LARGEFILE access to files larger than 2GB on
457 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
458 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
459 definitions for methods specializing on the passed class itself.
460 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
461 definitions for methods specializing on the passed class itself, or on
463 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
464 about itself during the build (e.g. for M-? in Slime), if this flag is
465 enabled in customize-target-features.lisp. This will increase the core
466 size by about 5-6mb, though, so it's mostly interesting to SBCL
468 * new feature: various GENCGC tuning parameters have been experimentally
469 documented and exported from SB-EXT. See documentation for details.
470 * fixes and improvements related to Unicode and external formats:
471 ** the Unicode character database has been upgraded to the
472 Unicode 5.2 standard, giving names and properties to a number of new
473 characters, and providing a few extra characters with case
475 ** improvement: restarts for providing replacement input/output on coding
476 errors for fd-stream external formats.
477 ** improvement: where :<encoding> is a keyword corresponding to an
478 external format the system supports, it is now possible to specify
479 (:<encoding> :replacement <character>) as an external format which will
480 automatically substitute <character> on encoding or decoding errors for
481 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
483 ** improvement: the file streams underlying the standard streams (such as
484 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
485 which uses the replacement mechanism to handle encoding errors,
486 preventing various infinite error chains and unrecoverable I/O
488 ** minor incompatible change: the utf-8 external format now correctly
489 refuses to encode Lisp characters in the surrogate range (char-codes
490 between #xd800 and #xdfff).
491 ** fix a typo preventing conversion of strings into octet vectors
492 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
494 ** fix a bug in the octet multibyte handling of decoding errors and the
495 USE-VALUE restart. (launchpad bug lp#314939)
496 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
497 restart on fd-stream decoding errors.
498 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
499 error is near the end of file.
500 ** fix a double-error case in unibyte octet conversions, when the first
501 use of USE-VALUE is ignored.
502 ** fix bugs in handling of undefined code points in unibyte encodings.
503 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
504 streams after an UNREAD-CHAR.
505 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
506 object is allocated in a boxed region of dynamic space.
507 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
508 fork with multiple Lisp threads running is made, instead of going
509 ahead with unpredictable consequences. (reported by Leslie Polzer)
510 * bug fix: uses of slot accessors on specialized method parameters within
511 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
512 error while finalizing the class. This fix may cause classes with slot
513 accessors to be finalized later than previously. (reported by Lars Rune
514 Nøstdal; launchpad bug lp#473699)
515 * bug fix: restore buildability on the MIPS platform. (regression from
516 1.0.30.38, reported by Samium Gromoff)
517 * bug fix: inspecting closures is less likely to fail with a type error.
518 * bug fix: no timer starvation when setting the system clock back.
519 (launchpad bug lp#460283)
520 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
521 standard pprint dispatch table as specified by CLHS.
522 * bug fix: give CLISP a hint about a type declaration to enable it to build
523 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
525 * bug fix: correctly dump literal objects in defaulting forms of arglists.
526 (reported by Attila Lendvai; launchpad bug lp#310132)
527 * bug fix: distinguish in type specifiers between arrays that might be
528 complex and arrays that are definitely complex. (launchpad bug lp#309129)
529 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
530 type. (reported by Levente Mészáros; launchpad bug lp#485972)
531 * bug fix: setting the value of a symbol-macro within a method in the
532 presence of type declarations works properly again. (reported by Iban
533 Hatchondo; launchpad bug lp#485019)
535 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
536 * optimization: faster FIND and POSITION on strings of unknown element type
537 in high SPEED policies. (thanks to Karol Swietlicki)
538 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
540 * improvement: better error signalling for bogus parameter specializer names
541 in DEFMETHOD forms (reported by Pluijzer)
542 * bug fix: DELETE-FILE once again works on logical pathnames (regression
544 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
546 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
547 removes previous default initargs (reported by Lars Rune Nøstdal and
549 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
550 lambda-lists. (reported by Peter Seibel)
551 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
552 correctly when starting from an executable core without saved runtime
553 options (reported by Faré Rideau, thanks to Zach Beane)
554 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
555 an optimization note instead. (reported by Martin Cracauer)
556 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
557 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
558 new process group if it doesn't need to share stdin with the sbcl
559 process. (thanks to Leslie Polzer)
560 * bug fix: SATISFIES could be misoptimized to refer to a local function.
561 (reported by Stanislaw Halik)
563 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
564 * improvement: stack allocation is should now be possible in all nested
565 inlining cases: failure to stack allocate when equivalent code is manually
566 open coded is now considered a bug.
567 * improvements related to Unicode:
568 ** the Unicode character database has been upgraded to the
569 Unicode 5.1 standard, giving names and properties to a number of new
570 characters, and providing a few extra characters with case
572 ** the system now recognizes and produces names for Unicode Hangul
574 ** the EBCDIC-US external-format is now supported for octet operations
575 (as well as for stream operations).
576 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
577 be used to output toplevel compile-time effects into a separate .CFASL
579 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
580 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
582 * optimization: TRUNCATE on known single- and double-floats is upto 25%
584 * optimization: division of floating point numbers by constants uses
585 multiplication by reciprocal when an exact reciprocal exists.
586 * optimization: multiplication of single- and double-floats floats by
587 constant two has been optimized.
588 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
589 sufficient type information is available. (thanks to Leslie Polzer)
590 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
591 known structure objects are as efficient as defstruct generated accessors.
592 * optimization: unused vector creation can now be optimized away.
593 * improvement: ASDF systems can now depends on SB-INTROSPECT.
594 * improvement: a STYLE-WARNING is signalled when a generic function
595 clobbers an earlier FTYPE proclamation.
596 * improvement: the compiler is able to track the effective type of
597 generic function across method addition and removal even in the
598 absence of an explicit DEFGENERIC.
599 * improvement: DESCRIBE now reports on symbols naming undefined
600 but assumed or declared function as well.
601 * improvement: recompilation of systems using SB-GROVEL now works
602 (thanks to Leslie Polzer)
603 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
604 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
606 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
607 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
608 well as user defined declaration names.
609 ** VARIABLE-INFORMATION is now aware of alien variables.
610 * improvement: improved address space layout on OpenBSD (thanks to Josh
612 * improvement: pretty-printing of various Lisp forms has been improved
613 (thanks to Tobias Rittweiler)
614 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
615 unused were deleted in safe code. (reported by John Fremlin)
616 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
618 * bug fix: some out-of-line array predicates were missing (reported by
620 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
622 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
623 (thanks to Larry D'Anna)
624 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
625 as well. (reported by Josh Elsasser)
626 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
627 leading to possible attempts to delete the same file twice. See docstring
628 on DELETE-FILE for details. (reported by John Fremlin)
629 * bug fix: DELETE-FILE once again deletes the file named by the pathname
630 designator argument, rather than its truename. (reported by Luis
632 * bug fix: the low-level debugger had 32-bit assumptions and was missing
633 information about some array types. (thanks to Luis Oliveira)
634 * bug fix: moderately complex combinations of inline expansions could
635 be miscompiled if the result was declared to be dynamic extent.
636 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
637 form (+ <variable> <integer>) were miscompiled under certain
639 * bug fix: in some cases no compiler note about failure to stack allocate
640 was emitted, even if the objects were in fact heap allocated.
641 * bug fix: minor violation of "otherwise inaccessible" rule for stack
642 allocation could cause objects users might reasonably expect to
643 be heap allocated to be stack allocated.
644 * bug fix: DESCRIBE signalled an error for generic functions under
645 certain circumstances. (thanks to Leslie Polzer)
646 * bug fix: Fixed spelling of an error message.
648 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
649 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
650 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
651 of SB-THREAD:THREAD-ERROR-THREAD.
652 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
653 * new feature: docstrings for local and anonymous functions are no longer
654 discarded. (thanks to Leslie Polzer)
655 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
656 values in other threads.
657 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
658 about object allocation.
659 * optimization: division of a real float by a complex float is implemented
660 with a specialised code sequence.
661 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
662 keywords is an order of magnitude faster.
663 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
664 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
665 methods, and similarly for non-standard metaclass classes as long as there
666 are no methods additional on MAKE-INSTANCE.
667 * optimization: more efficient type-checks for FIXNUMs when the value
668 is known to be a signed word on x86 and x86-64.
669 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
670 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
672 * optimization: compiler is smarter about delegating argument type checks to
674 * optimization: several character functions are now compiled somewhat more
675 efficiently. (reported by Lynn Quam)
676 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
677 * improvement: less unsafe constant folding in floating point arithmetic,
678 especially for mixed complex/real -float operations.
679 * optimization: constant double and single floats are stored in native
680 unboxed format on x86[-64].
681 * optimization: smarter code for arithmetic operations with constant floats,
682 complex floats, or integers on x86[-64].
683 * optimization: smarter code for conjugate/multiplication of float complexes
684 and abs/negate of floats on x86-64.
685 * optimization: more efficient complex float and real float operations on
687 * improvement: complex float division is slightly more stable.
688 * improvement: DESCRIBE output has been reworked to be easier to read and
689 contains more pertinent information.
690 * improvement: failure to provide requested stack allocation compiler notes
691 provided in all cases (requested stack allocation not happening without a
692 note being issued is now considered a bug.)
693 * bug fix: SB-POSIX exports the documented types and functions
694 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
695 types. (reported by "abhi")
696 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
697 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
698 (thanks to Paul Khuong)
699 * bug fix: looping around HANDLER-CASE could silently consume stack space
700 on each iteration. (reported by "foobar")
701 * bug fix: better error signalling when calls to functions seeking elements
702 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
703 (reported by Tobias Rittweiler)
704 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
705 directory components now works as it used to. (various prolems reported by
706 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
707 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
708 (reported by Elliot Slaughter, thanks to Stas Boukarev)
709 * bug fix: bogus undefined variable warnings from fopcompiled references to
710 global variables. (thanks to Lars Rune Nøstdal)
711 * bug fix: foreign function names should now appear in backtraces on
712 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
713 * bug fix: SETF compiler macro documentation strings are not discarded
715 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
716 (thanks to Alex Plotnick)
717 * bug fix: SLEEP accepts large integer arguments, truncating them to
718 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
719 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
720 * bug fix: compiler notes for expensive slot type checks could be emitted
721 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
723 changes in sbcl-1.0.29 relative to 1.0.28:
724 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
725 https://bugs.launchpad.net/sbcl
726 Bugs can be reported directly there, or by sending email to
727 sbcl-bugs@lists.sourceforge.net
728 (no subscription required.)
729 * minor incompatible change: under weak type checking policy integer
730 types are weakened less aggressively.
731 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
732 allowed to return, which causes SBCL to quit with exit status 0. Previously
733 if the function returned with a small integer return value, that value
734 was accidentally reused as the exit status.
735 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
736 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
737 :HASH-FUNCTION argument. Refer to user manual for details.
738 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
740 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
741 microseconds since the Unix epoch on all platforms.
742 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
743 allows the compiler to safely elide boundedness checks for special
745 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
746 the symbol, prohibits both lexical and dynamic binding. This is mainly an
747 efficiency measure for threaded platforms, but also valueable in
749 * new feature: UNC pathnames are now understood by the system on Windows.
750 * optimization: the compiler uses a specialized version of FILL when the
751 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
753 * optimization: accesses to potentially non-simple arrays where element type
754 is known are 50% faster.
755 * optimization: compiler now generates faster array typechecking code.
756 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
758 * optimization: multidimensional array accesses in the absence of type
759 information regarding array rank are approximately 10% faster due to
760 open coding of ARRAY-RANK.
761 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
762 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
764 * optimization: result of call to VECTOR can now be stack allocated.
765 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
766 as long as the resulting array is one-dimensional and has a known
767 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
768 length of the list matches the known length of the vector does not
769 allocate the list as an intermediate step. Ditto for VECTOR and simple
771 * optimization: MAKE-ARRAY can now stack allocate in the presence of
772 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
773 known element type, and is known to be simple and one dimensional.
774 * improvement: SBCL now emits a compiler note where stack allocation was
775 requested but could not be provided (not in all cases, unfortunately)
776 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
777 * improvement: pretty-printing loop has been implemented properly. (thanks
778 to Tobias Rittweiler)
779 * documentation: CLOS slot typechecing policy has been documented.
780 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
781 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
782 pointer, leading to memory faults. (thanks to Bart Botta)
783 * bug fix: better floating point exception handling on x86/OpenBSD.
784 (thanks to Josh Elsasser)
785 * bug fix: exit status from QUIT when called under --script was lost
786 (reported by Hubert Kauker)
787 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
788 same implementation of FILL to initialize the array, even if a more
789 efficient one was available (reported by Stas Boukarev, thanks to
791 * bug fix: potential miscompilation of array stack allocation on x86 and
792 x86-64. (reported by Time Tossavainen)
793 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
794 result in their subforms being treated as top level forms. (reported by
796 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
797 than a machine register could leave garbage in the high bits of the
798 result register (bug 316325).
799 * bug fix: disable address space randomization Linux/x86-64 as well,
800 not just x86-64. (reported by Ken Olum)
801 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
802 generate incorrect code.
803 * bug fix: #201; type inference for CONS and ARRAY types could derive
804 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
805 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
806 could not be assigned to or bound with PROGV. (reported by Lorenz
808 * bug fix: the value of CL:- in the inspector was the previous expression
809 evaluated rather than the expression being evaluated.
810 * bug fix: constants can no longer be locally declared special.
811 * bug fix: signals delivered to threads started from foreign land (read:
812 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
813 thread by blocking all signals and resignalling.
814 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
815 from :INITFORM, if any.
817 changes in sbcl-1.0.28 relative to 1.0.27:
818 * a number of bugs in cross-compilation have been fixed, with the ultimate
819 result that building under (at least) clisp should be much more reliable.
820 * minor incompatible changes: echo-streams now propagate unread-char to the
821 underlying input stream, and no longer permit unreading more than one
823 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
824 frames, allowing for instance more reliable backtraces.
825 * improvement: the debugger REPL can now reference lexical variables
826 by name directly for code compiled with (DEBUG 3).
827 * improvement: errors from malformed declarations now have better source
828 paths associated with them. (thanks to Tobias Rittweiler)
829 * optimization: faster local calls on x86/x86-64
830 * bug fix: some error messages for out-of-bound array indexes confused the
831 index and the bound. (thanks to Stas Boukarev)
832 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
834 * bug fix: running regressions tests in shells without OSTYPE set now works.
835 (reported by Harald Hanche-Olsen)
836 * bug fix: more robust static space exhaustion signalling from
837 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
838 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
839 docstring away instead of storing it under names such as (LAMBDA (X)).
840 (reported by Leslie Polzer)
841 * bug fix: timers could go off in the wrong order, be delayed indefinitely
842 (thanks to Ole Arndt for the patch)
843 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
844 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
845 potentially-invalid effective methods in its cache.
846 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
847 instances as well (thanks to Paul Khuong)
848 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
849 SB-POSIX:WAITPID and their C equivalents.
850 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
852 changes in sbcl-1.0.27 relative to 1.0.26:
853 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
854 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
855 * improvement: the system either recovers from stack exhaustion or dies
856 properly as opposed to leaving the user uncertain of whether the handler
857 trampled on some random memory next to the stack or having to rely on
858 --lose-on-corruption (which is still a good idea to use in production
859 because stack exhaustion can happen in signal handlers which will likely
861 * bug fix: fix gc related interrupt handling bug on ppc (regression from
862 1.0.25.37, reported by Harald Hanche-Olsen)
863 * bug fix: work around signal delivery bug in darwin (regression from
864 1.0.25.44, reported by Sidney Markowitz)
865 * bug fix: fix ERROR leaking memory (reported by David Thompson)
867 changes in sbcl-1.0.26 relative to 1.0.25:
868 * incompatible change: an interruption (be it a function passed to
869 INTERRUPT-THREAD or a timer function) runs in an environment where
870 interrupts can be enabled. The interruption can use
871 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
872 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
873 potentially running out of stack. Keep in mind that in the absence
874 of WITHOUT-INTERRUPTS some potentially blocking operation such as
875 acquiring a lock can enable interrupts.
876 * incompatible change: GC-OFF and GC-ON are removed, as they were
877 always unsafe. Use WITHOUT-GCING instead.
878 * new feature: runtime option --disable-ldb
879 * new feature: runtime option --lose-on-corruption to die at the
880 slightest hint of possibly non-recoverable errors: running out of
881 memory, stack, alien stack, binding stack, encountering a memory
882 fault, etc. In the absence of --lose-on-corruption a warning is
884 * enhancement: detect binding stack exhaustion
885 * enhancement: detect alien stack exhaustion on x86/x86-64
886 * improvement: generally more stable and reliable interrupt handling
887 * improvement: there is a per thread interruption queue,
888 interruptions are executed in order of arrival
889 * improvement: a repeating timer reschedules itself when the it has
890 finished, but expiration times are spaced equally. If an
891 expiration time is in the past it will trigger after a short grace
892 period that may give a chance to other things to run.
893 * optimization: slightly faster gc on multithreaded builds
894 * optimization: faster WITHOUT-GCING
895 * bug fix: when JOIN-THREAD signals an error, do it when not holding
896 important locks so that the debugger/handler doesn't produce
897 recursive errors or deadlock.
898 * bug fix: real-time signals are not used anymore, so no more
899 hanging when the system wide real-time signal queue gets full.
900 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
902 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
903 * bug fix: fix random memory faults related to interrupts on alpha
904 * bug fix: fix random memory faults related to interrupts on sparc
905 * bug fix: fix deadlocks related to starting threads
906 * bug fix: fix deadlines on locks on futex platforms
907 * bug fix: restore errno in signal handlers
908 * bug fix: fix deadlocks related to hash tables
909 * bug fix: fix deadlocks in pcl
911 changes in sbcl-1.0.25 relative to 1.0.24:
912 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
913 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
914 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
915 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
916 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
917 better name and does not return values so stale on multiprocessor systems.
918 Also, HOLDING-MUTEX-P was added for about the only sane usage of
920 * improvement: unithread builds keep track of MUTEX-VALUE.
921 * improvement: reading from a TWO-WAY-STREAM does not touch the output
922 stream anymore making it thread safe to have a concurrent reader and
923 a writer, for instance, in a pipe.
924 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
925 arguments, making compiler-macros for SETF-functions able to inspect
926 their constant arguments.
927 * improvement: COMPILE-FILE reports times with millisecond accuracy
928 (thanks to Luis Oliveira)
929 * optimization: CHAR-CODE type derivation has been improved, making
930 TYPEP elimination on subtypes of CHARACTER work better. (reported
931 by Tobias Rittweiler, patch by Paul Khuong)
932 * bug fix: setting alien structure fields of type struct by value now
933 computes the right offset for the memory copy.
934 * bug fix: compilation problem involving inlined calls to aliens with
935 result type VOID. (reported by Ken Olum)
936 * bug fix: #235a; sequential inline expasion in different policies no
937 longer reuses the functional from the previous expansion site.
938 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
939 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
940 * bug fix: named ENUMs in multiply-referenced alien record types no longer
941 cause an implied type redefinition. Regression from 1.0.21.29.
942 * bug fix: sign-extension of small signed return values now works with C
943 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
944 * improvements to the Windows port:
945 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
946 Westfold, thanks to Rudi Schlatte)
948 changes in sbcl-1.0.24 relative to 1.0.23:
949 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
950 vector of a multidimensional SIMPLE-ARRAY.
951 * new feature: the system now signals a continuable error if standard
952 readtable modification is attempted. (thanks to Tobias Rittweiler)
953 * new feature: DIRECTORY has been extended with a non-standard keyword
954 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
955 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
956 has been robustified and documented better. (thanks to Robert Goldman)
957 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
958 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
959 * tradeoff: constant FORMAT control strings are now compiled unless
960 SPACE > SPEED (previously only when SPEED > SPACE.)
961 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
963 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
965 * bug fix: lexical type declarations are now correctly reported by
966 SB-CLTL2. (reported by Larry D'Anna)
967 * bug fix: STRING-TO-OCTETS did not handle :START properly when
968 using UTF-8 as external format. (thanks to Luis Oliveira)
969 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
970 are signalled correctly. (thanks to Stas Boukarev)
971 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
972 designator. (thanks to Tobias Rittweiler)
973 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
974 designator, and returns T instead of the function. (thanks to
976 * bug fix: direct superclasses of STANDARD-CLASS and
977 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
978 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
979 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
980 when no class with the named slot yet exists no longer causes a
981 compile-time style-warning.
982 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
983 in safe code. (reported by Didier Verna)
984 * bug fix: #430; nested structure constructors can stack allocate.
985 * bug fix: on some 64-bit platforms dynamic space size was truncated
986 to #xffff0000 bytes. (reported by Benjamin Lambert)
987 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
988 REPL unusable. (reported by Daniel Herring)
989 * bug fix: return values of READ-SEQUENCE did not take :START into
990 account on file streams, regressions since 1.0.12.22. (reported by
991 Thomas Russ, patch by Paul Khuong)
992 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
993 method specializer used to confuse permuation vector optimization.
994 * bug fix: system inserted bogus implicit type declarations for local
995 special variables in DEFMETHOD bodies.
996 * bug fix: #354; duplicated frames in backtraces due to
997 non-tail-call-optimized XEPs to functions with return type NIL
999 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
1000 initializes structure object slots according to DEFSTRUCT initforms,
1001 and DEFSTRUCT forms :INCLUDEind structure classes defined using
1002 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
1003 (reported by Bruno Haible and Stephen Wilson)
1004 * bug fix: #395; fill-pointer output streams used now support
1005 element-type BASE-CHAR as well.
1006 * bug fix: compiler error when attempting to derive return value of
1007 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
1009 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
1011 * bug fix: Mac OS X binaries should now be portable between Leopard
1014 changes in sbcl-1.0.23 relative to 1.0.22:
1015 * enhancement: when disassembling method functions, disassembly
1016 for the associated fast function is also produced.
1017 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
1019 * optimization: printing with *PRINT-PRETTY* true is now more
1020 efficient as long as the object being printed doesn't require
1021 special handling by the pretty printer.
1022 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
1023 now interact correctly with type declarations.
1024 * partial bug fix: PCL detects infinite recursion during wrapper
1025 validation. (thanks to Attila Lendvai)
1026 * bug fix: #426; nested function calls are inlined properly.
1027 Previously if FOO was an inline function, in calls of the form
1028 (FOO (FOO ...)) the outer call was not inlined.
1029 * bug fix: long long arguments passed in registers to alien
1030 functions on PowerPC Linux and NetBSD are now handled correctly
1031 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
1033 changes in sbcl-1.0.22 relative to 1.0.21:
1034 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
1035 for the shared object in the current directory, but passes the native
1036 namestring of the designated pathname to the operation system's shared
1037 object loading function as-it.
1038 * minor incompatible change: --disable-debugger toplevel option now takes
1039 effect before processing of initialization files and --eval or --load
1041 * new feature: new commandline argument: --script, which supports
1042 shebang lines. See documentation for details. (based on work by
1044 * new feature: SAVE-LISP-AND-DIE can save current values of
1045 --dynamic-space-size and --control-stack-size in the executable core,
1046 causing it to skip normal runtime option processing. See documentation
1047 for details. (thanks to Zach Beane)
1048 * enhancement: inoccous calls to EVAL or generic functions dispatching
1049 on subclasses of eg. STREAM no longer cause compiler notes to appear.
1050 * enhancement: the system no longer resignals errors from --load and
1051 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
1052 associated with the original error to be lost. (thanks to Ariel
1054 * enhancement: :DONT-SAVE keyword argument has been added to
1055 LOAD-SHARED-OBJECT for controlling interaction with
1056 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
1057 the effects of an earlier LOAD-SHARED-OBJECT call.
1058 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
1059 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
1060 provided. (reported by Cedric St-Jean)
1061 * bug fix: circularity handling in the reader did not treat raw
1062 structure slots correctly. (reported by Cedric St-Jean)
1063 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
1064 file descriptors when there were none.
1065 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
1066 was broken. (reporteed by Willem Broekema)
1067 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
1068 pathnames without a directory.
1069 * bug fix: COMPILE-FILE did not warn about undefined variable
1070 references at toplevel, and LOAD of the resulting fasl did
1071 not signal an error.
1072 * bug fix: functions with non-required arguments used to end up with
1073 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
1074 * bug fix: redefining a function with non-required arguments didn't
1075 update the system's knowledge about its call signature properly.
1076 * bug fix: fixed #431; incompatible alien record type redefinitions
1077 are detected and handled. (reported by Neil Haven)
1078 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
1079 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
1081 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
1082 after alien stack frames.
1083 * bug fix: strings passed to mk{s,d}temp lacked null termination
1085 changes in sbcl-1.0.21 relative to 1.0.20:
1086 * new feature: the compiler is able to track the effective type of a
1087 generic function across method addition and removal.
1088 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
1089 appropriately typed structure slots without locking.
1090 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
1091 information like those gathered by TIME using a programming-friendly
1093 * new feature: TIME reports time taken even if the form performs a
1094 non-local transfer of control.
1095 * enhancement: reduced conservativism on GENCGC platforms: on
1096 average 45% less pages pinned (measured from SBCL self build).
1097 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
1098 mutate constant symbols or violate declaimed type of the symbol.
1099 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
1100 sleeping on the semaphore in heavily contested semaphores.
1101 * bug fix: semaphores and condition variables were not interrupt
1103 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
1104 owned by other threads anymore.
1105 * bug fix: FIND on lists called KEY outside the specified
1106 subsequence. (reported by budden)
1107 * bug fix: LOG doesn't use single-float intermediate results when
1108 given mixed integer and double-float arguments, leading to better
1109 precision. (reported by Bob Felts)
1110 * bug fix: LOG with base zero returned values of inconsistent type.
1112 changes in sbcl-1.0.20 relative to 1.0.19:
1113 * minor incompatible change: OPTIMIZE qualities
1114 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
1115 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
1116 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
1117 * documentation: some slot access efficiency guidelines have been
1118 added to the user manual.
1119 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
1120 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
1121 as ASSOC and MEMEBER.
1122 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
1123 to more efficient EQ-comparison versions more often.
1124 * optimization: enhanced derivation of DOLIST iteration variable type
1126 * optimization: constant folding of simple (LIST ...) forms as DOLIST
1128 * optimization: runtime lookup of function definitions can be
1129 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
1130 * optimization: compiler is able to derive the return type of
1131 (AREF (THE STRING X) Y) as being CHARACTER.
1132 * optimization: CLRHASH on empty hash-tables no longer does pointless
1133 work. (thanks to Alec Berryman)
1134 * optimization: REPLACE deftransforms don't punt when :START1 or
1136 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
1137 no longer emits a code deletion note. (reported by Andrew
1139 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
1140 functions with an odd number of &OPTIONAL arguments, a &REST
1141 argument, and one or more &KEY arguments at the call site.
1142 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
1143 arguments appeared at call sites as well.
1144 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
1145 READ-LINE &co can no longer read from them afterwards. (reported
1147 * bug fix: fixed #427: unused local aliens no longer cause compiler
1148 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
1150 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
1151 alien-stack leakage. (reported by Andy Hefner)
1152 * bug fix: PROGV signals an error when an attempt to violate declared
1153 type of a variable or bind a constant is made.
1154 * bug fix: SET signals an error when an attempt to violate declared
1155 type of a variable is made.
1156 * bug fix: restart computation during the execution of a restart
1157 test function no longer causes infinite recursion. (reported by
1159 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
1160 signals an error. (thanks to Pierre Mai)
1161 * bug fix: EXPORT left symbol unexported in conflict situations.
1162 (thanks to Michael Weber)
1163 * bug fix: correctly handle name conflicts involving CL:NIL.
1164 (thanks to Michael Weber)
1165 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
1166 conflicts arising from USEing package with conflicting symbols
1167 correctly. (thanks to Michael Weber)
1169 changes in sbcl-1.0.19 relative to 1.0.18:
1170 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
1171 warnings that go otherwise unhandled will be muffled if they are
1172 of the type that's the value of this variable.
1173 * optimization: stack allocation is slightly more efficient on x86
1175 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
1176 where a raw slot always is initialized using the initform whose
1177 type is not know sufficiently well a compile-time are now compiled
1178 correctly. (reported by John Morrison)
1179 * bug fix: compiler no longer makes erronous assumptions in the
1180 presence of non-foldable SATISFIES types.
1181 * bug fix: stack analysis missed cleanups of dynamic-extent
1182 arguments in non-let-converted calls to local functions.
1183 * improvements to the Windows port:
1184 ** adjusted address spaces for building on both Win32 and
1185 Win64. (thanks for John Connors)
1186 * fixed some bugs revealed by Paul Dietz' test suite:
1187 ** interval arithmetic during type derivation used inexact integer
1188 to single-float coercions.
1189 ** arithmetic operations involving large integers and single
1190 floats give the same results in compiled and interpreted code.
1191 ** deriving the result type of COERCE no longer signals an error
1192 if the derived type of the second argument is a MEMBER type
1193 containing invalid type specifiers.
1194 ** ADJOIN with constant NIL as second argument works correctly.
1195 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
1197 changes in sbcl-1.0.18 relative to 1.0.17:
1198 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
1199 profiles only the current thread.
1200 * minor incompatible change: changes to SYMBOL-VALUE of constants
1201 defined with DEFCONSTANT now signal an error.
1202 * enhancement: SB-SPROF now has support for wallclock profiling,
1203 and is also able to profile specific threads. REPORT output
1204 has also additional sorting options.
1205 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
1207 * optimization: structure allocation has been improved
1208 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
1209 ** out of line constructors are ~10% faster.
1210 ** inline constructors are ~15% faster.
1211 ** inline constructors are capable of dynamic extent allocation
1212 (generally on x86 and x86-64, in some cases on other platforms
1214 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
1216 * optimization: file compiler is now able to coalesce non-circular
1217 lists, non-base strings, and bit-vectors. Additionally, constants
1218 are never referenced through SYMBOL-VALUE at runtime.
1219 * optimization: code defining methods on PRINT-OBJECT (and other
1220 generic functions in the COMMON-LISP package) now loads faster.
1221 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
1223 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
1224 fasl is now deleted. (reported by Attila Lendvai)
1225 * bug fix: READ-LINE always returned NIL for the last line in files.
1226 (reported by Yoshinori Tahara)
1227 * bug fix: more accurate disassembly annotations of foreign function
1228 calls. (thanks to Andy Hefner)
1229 * bug fix: trimming non-simple strings and non-string string
1230 designators when the there is nothing to trim works properly.
1231 (thanks to James Knight)
1232 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
1233 (thanks to Travis Cross)
1234 * fixed some bugs revealed by Paul Dietz' test suite:
1235 ** NIL is a valid function name (regression at 1.0.13.38)
1236 ** FILL on lists was missing its return value (regression at 1.0.12.27)
1237 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1238 fill pointers properly (regression at 1.0.12.23)
1239 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1240 displacement indices properly (regression at 1.0.12.23)
1242 changes in sbcl-1.0.17 relative to 1.0.16:
1243 * temporary regression: user code can no longer allocate closure
1244 variable storage on stack, due to bug 419 without explicitly
1245 requesting it. Please consult sbcl-devel for advice if you need to
1246 use this feature in the meanwhile.
1247 * new feature: runtime argument --control-stack-size can be used to
1248 adjust thread default control stack size.
1249 * enhancement: improved TIME output
1250 ** all times are reported using the measured accuracy (milliseconds
1251 for real and GC times, microseconds for everything else.)
1252 ** processor cycle counts on x86 and x86-64.
1253 ** interpreted forms are counted for both evaluator modes.
1254 ** number of lambdas converted by the compiler is reported.
1255 ** CPU percentage report (computed from real and total run time.)
1256 ** more comprehensive run time reporting, using a condenced format
1257 ** interperted form, lambda, and page fault counts are omitted
1259 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
1261 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
1262 * optimization: two argument forms of LAST are upto ~10% faster
1263 in normal SPEED policies.
1264 * optimization: NCONC no longer needs to heap cons its &REST list
1265 in normal SPEED policies.
1266 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
1267 to Sidney Markowitz)
1268 * bug fix: UNION and NUNION work with :TEST-NOT once more,
1269 regression since 1.0.9.1. (thanks to Eric Marsden)
1270 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
1271 since 1.0.15.36. (thanks to Paul Khuong)
1272 * bug fix: LAST when always returned the whole list when given a bignum
1273 as the second argument.
1274 * bug fix: dynamic extent allocation of nested lists and vectors
1275 could leak to otherwise accessible parts.
1276 * bug fix: invalid optimization of heap-allocated alien variable
1278 * bug fix: fasl header checking is less vulnerable to different
1279 platform word lengths.
1280 * bug fix: more correct assembler syntax for GNU binutils
1281 2.18.50.0.4 support. (thanks to Marijn Schouten)
1282 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
1284 * bug fix: the fopcompiler can handle LOCALLY forms (with no
1285 declarations) successfully. (reported by Attila Lendvai)
1287 changes in sbcl-1.0.16 relative to 1.0.15:
1288 * minor incompatible change: revert the changes to sb-posix's error
1289 signaling added in 1.0.14.
1290 * minor incompatible change: change PROBE-FILE back to returning
1291 NIL whenever we can't get a truename, as was the case before 1.0.14.
1292 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
1293 inherits from ERROR instead of just CONDITION.
1294 * new feature: SB-INTROSPECT can provide source locations for instances
1295 as well. (thanks to Tobias Rittweiler)
1296 * optimization: binding special variables now generates smaller code
1297 on threaded platforms.
1298 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
1299 and cases where no :TEST is given but the compiler can infer that
1300 the element to search is of type (OR FIXNUM (NOT NUMBER)).
1301 * optimization: better LOGNOT on fixnums.
1302 * optimization: modular arithmetic for a particular requested width
1303 is implemented using a tagged representation unless a better
1304 representation is available.
1305 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
1306 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
1307 for wellformedness and returns a specialized vector. (reported by
1308 Francois-Rene Rideau)
1309 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
1310 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
1311 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
1312 unscheduling timer at the same time as another timer fires.
1313 * bug fix: don't reschedule timers for dead threads.
1314 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
1315 * bug fix: copying output from RUN-PROGRAM to a stream signalled
1316 bogus errors if select() was interrupted.
1317 * enhancement: add support for fcntl's struct flock to SB-POSIX.
1319 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
1320 * enhancement: cleaner backtraces for interactive interrupts, as
1321 well as other cases where the interesting frames used to be
1322 obscured by interrupt handling frames.
1323 * enhancement: untracing a whole package using (UNTRACE "FOO") is
1324 now supported, and tracing a whole package using (TRACE "FOO") now
1325 traces SETF-functions as well.
1326 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
1327 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
1328 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
1329 * unreadably printed representation of hash-tables now includes
1331 * bug fix: partially fixed #188: type propagation from assignments
1332 is now more efficient.
1333 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
1334 'DOUBLE-FLOAT) are not flushable.
1335 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
1336 full address of the object, and none of the tag bits.
1337 * bug fix: readably printing hash-tables now respects other printer
1338 control variables. (reported by Cedric St-Jean)
1339 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
1340 keyword with MAKE-HASH-TABLE.
1341 * bug fix: export SB-POSIX:MKSTEMP.
1342 * bug fix: SORT was not interrupt safe.
1343 * bug fix: XREF accounts for the last node of each basic-block as
1345 * bug fix: MAKE-INSTANCE optimizations interacted badly with
1346 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
1347 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
1348 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
1349 create obsolete instances in the case of redefinition or
1350 obsoletion of a superclass. (thanks to Andy Hefner)
1351 * bug fix: Support for the Alpha architecture has been revived; it had
1352 suffered somewhat from lack of maintenance since sbcl-1.0.
1353 * improvements to the Windows port:
1354 ** The system detects the codepage to use at startup, instead of
1355 using the value from the saved core. (thanks to Kei Suzuki)
1357 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
1358 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
1359 (see documentation for details.)
1360 * revived support for OpenBSD (contributed by Josh Elsasser)
1361 * partially fixed bug #108: ROOM no longer suffers from occasional
1362 (AVER (SAP= CURRENT END)) failures.
1363 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
1364 bodies no longer cause a WARNING to be signalled. (reported by
1366 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
1367 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
1368 the bug in TRUNCATE.) (reported by Michael Weber)
1369 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
1370 no samples. (reported by Andy Hefner)
1371 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
1372 no longer appear as (NIL ...) frames in backtraces.
1373 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
1374 is now actually exported from SB-EXT as documented. (reported by
1376 * bug fix: sb-aclrepl now correctly understands how to inspect
1377 single-floats on 64-bit platforms where single-floats are not boxed.
1378 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
1379 yet been finalized. (reported by Levente Meszaros)
1380 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
1381 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
1382 SBCL-specific optimize qualities.
1384 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
1385 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
1386 an executable in the search path, and does so in the child
1387 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
1388 been removed; it can be found in the CVS history, for somebody who
1389 needs that search behavior (see the manual).
1390 * minor incompatible change: compiler policy re. weakening type
1391 checks has changed: now type checks are weakened only if SAFETY < 2
1393 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
1394 unparsing of directory pathnames as files. Analogously,
1395 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
1396 filename to parse into a directory pathname.
1397 * enhancement: implicit generic function creation now signals a
1398 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
1399 users can bind handlers for and muffle around calls to LOAD.
1400 * enhancement: RUN-PROGRAM allows unicode arguments and environments
1401 to be used (using the default stream external format), and allows
1402 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
1403 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
1404 strings and vectors whose element-type or simplicity is not fully
1405 known at compile-time.
1406 * optimization: STRING-TRIM and related functions no longer allocate
1407 a new string when no trimming needs to be performed. These functions
1408 are also faster than before when the input string has been declared
1410 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
1411 * optimization: READ-LINE is significantly faster for files containing
1413 * optimization: non-open coded uses of character comparison operators
1414 (e.g. char=) no longer cons when called with more than one parameter
1415 on platforms supporting dynamic-extent allocation.
1416 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
1417 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
1418 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
1420 * bug fix: some sequence functions elided bounds checking when
1422 * bug fix: too liberal weakening of union-type checks when SPEED >
1424 * bug fix: more bogus fixnum declarations in ROOM implementation
1427 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
1428 * new feature: MAKE-HASH-TABLE now experimentally accepts a
1429 :SYNCHRONIZED argument, which makes the hash-table safe for
1430 concurrent accesses (but not iteration.) See also:
1431 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1432 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1433 * optimization: CONCATENATE on strings is an order of magnitude faster
1434 in code compiled with (> SPEED SPACE).
1435 * optimization: SUBSEQ is ~50% faster on lists.
1436 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1437 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1438 END is smaller then START.
1439 * bug fix: SB-PROFILE will no longer report extra consing for nested
1440 calls to profiled functions.
1441 * bug fix: ROOM implementation had bogus fixnum declarations which
1442 could cause type-errors when calling ROOM on large images.
1443 * bug fix: if file compilation is aborted, the partial fasl is now
1444 deleted, and COMPILE-FILE returns NIL as the primary value.
1445 * bug fix: number of thread safety issues relating to SBCL's internal
1446 hash-table usage have been fixed.
1447 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1448 be returned from its body when the values were being returned
1449 using unknown-values return convection and the W-P-O was wrapped
1450 inside an UNWIND-PROTECT.
1451 * bug fix: sb-posix should now compile again under Windows, enabling
1452 slime to work again.
1454 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1455 * incompatible change: hash-table accessor functions are no longer
1456 automatically protected by locks. Concurrent accesses on the same hash-table
1457 from multiple threads can give inconsistent results or even corrupt the
1458 hash-table completely. Multi-threaded applications should do their own
1459 locking at the correct granularity. In the current implementation it is
1460 still safe to have multiple readers access the same table, but it's not
1461 guaranteed that this property will be maintained in future releases.
1462 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1463 and will signal an error at runtime.
1464 * enhancement: SB-THREAD package now exports a semaphore interface.
1465 * enhancement: CONS can now stack-allocate on x86 and
1466 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1468 * enhancement: nested lists can now be stack allocated on
1469 platforms providing stack allocation support.
1470 * enhancement: dynamic-extent support has been extended to support
1471 cases where there are multiple possible sources for the stack
1473 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1474 if the mutex is uncontested on Linux.
1475 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1476 as specified by the CLHS. (thanks to Tobias Rittweiler)
1477 * bug fix: NaN comparison now works on x86-64.
1478 * bug fix: CLOSE :ABORT T on a stream with pending output now
1480 * bug fix: instances of non-standard metaclasses using standard
1481 instance structure protocol sometimes missed the slot type checks
1483 * bug fix: known functions can be inlined in high-debug code.
1485 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1486 * minor incompatible change: the MSI installer on Windows no longer
1487 associates .lisp and .fasl files with the installed SBCL.
1488 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1489 on Windows. (thanks to Luis Oliviera)
1490 * new feature: SB-CLTL2 contrib module now implements
1491 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1492 * optimization: scavenging weak pointers is now more efficient,
1493 requiring O(1) instead of O(N) per weak pointer to identify
1494 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1495 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1496 method bodies using constant slot-names when the first argument is
1497 a specializer parameter for the method.
1498 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1499 names, when the class of the instance is a direct instance
1500 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1501 as slow as the constant slot-name case.)
1502 * optimization: member type construction is now O(N) instead
1504 * optimization: UNION and NUNION are now O(N+M) for large
1505 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1507 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1508 now more readable in environments like Slime which display it.
1509 (thanks to Tobias C. Rittweiler)
1510 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1511 was able to infer the correct argument type for the object on which
1512 the CAS operation was being performed.
1513 * bug fix: copy propagation interfered with parallel assignment
1514 semantics in local calls. (reported by Paul Khuong)
1515 * bug fix: the signed modular fixnum shift compiled to wrong code on
1516 x86 and x86-64. (spotted by a slight modification to some of
1519 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1520 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1521 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1522 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1524 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1525 and generic function now print the object identity for anonymous
1526 instances. (thanks to Attila Lendvai)
1527 * enhancement: as an extension to MOP, SBCL now supports SETF of
1528 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1529 (thanks to Attila Lendvai)
1530 * enhancement: sb-cover has an alternate annotation mode which puts
1531 the form annotations at the car of the form, instead of annotating
1532 it completely. The benefit of this mode is that it shows better
1533 which forms were instrumented by the compiler.
1534 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1535 by printer control variables. (reported by Dan Corkill)
1536 * bug fix: system leaked memory when delayed output was performed by
1537 the OS in smaller chunks then expected. (thanks to David Smith)
1538 * bug fix: system leaked memory when file streams were not closed
1540 * bug fix: large objects written to slow streams that were modified
1541 after the write could end up with the modified state written to
1542 the underlying file descriptor.
1543 * bug fix: multiple threads operating in parallel on the same stream
1544 could cause buffer-overflows.
1545 * bug fix: source location information is stored correctly
1546 (broken since 1.0.6). This bug would generally show up as the
1547 Slime debugger highlighting the wrong form when the sldb-show-source
1549 * bug fix: Branch forms are again annotated as branches in the sb-cover
1551 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1552 on x86-64. (reported by Gregory Vanuxem)
1553 * bug fix: EQUALP could return wrong results for structures with raw
1554 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1555 word). (reported by Vjacheslav Fyodorov)
1556 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1559 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1560 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1561 atomic compare-and-swap operations on threaded platforms.
1562 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1563 allows assigning a global minimum value to optimization qualities
1564 (overriding proclamations and declarations).
1565 * enhancement: closed over variables can be stack-allocated on x86
1567 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1569 * optimization: slot definition lookup is now O(1). This speeds up
1570 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1571 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1572 * optimization: ASSOC and MEMBER can now be open-coded for all
1573 combinations of keyword arguments when second argument is constant
1574 and SPEED >= SPACE. In other cases a specialized version is
1576 * bug fix: using obsoleted structure instances with TYPEP and
1577 generic functions now signals a sensible error.
1578 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1579 (reported by Kristoffer Kvello)
1580 * bug fix: backtrace construction is now more careful when making
1581 lisp-objects from pointers on the stack, to avoid creating bogus
1582 objects that can be seen by the GC.
1583 * bug fix: defaulting of values in contexts expecting more than 7
1584 variables now works on x86-64. (reported by Christopher Laux)
1585 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1587 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1588 as the property-list of a symbol.
1589 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1590 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1591 situations requested, are once again file-compileable. (reported
1594 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1595 * MOP improvement: support for user-defined subclasses of
1596 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1597 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1598 the expansion of DEFMETHOD, is responsible for generating a form
1599 which creates a list of specializers when evaluated. Additional
1600 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1601 debugging and introspective support.
1602 * minor incompatible change: the (unsupported) spinlock interface
1603 has changed: free spinlock now has the value NIL, and a held spinlock
1604 has the owning thread as its value.
1605 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1606 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1608 * enhancement: name of a socket-stream is now "a socket" instead of
1609 "a constant string".
1610 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1611 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1612 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1613 for generic functions. (Thanks to Tobias C. Rittweiler)
1614 * optimization: bignum printing speed has been improved by 20-40%
1615 (depending on the bignum size.)
1616 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1618 * bug fix: the cache used by the CLOS to store precomputed effective
1619 methods, slot offsets, and constant return values is now thread and
1621 * bug fix: generic function dispatch function updating is now thread
1622 and interrupt safe (in the sense that the known issues have been
1624 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1625 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1626 * bug fix: asdf systems with dependencies to the SB-POSIX or
1627 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1628 * bug fix: interrupt safety of applicable method computation has been
1631 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1632 * new contrib: sb-cover, an experimental code coverage tool, is included
1633 as a contrib module.
1634 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1635 significantly faster.
1636 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1637 produce more efficient code on x86-64 if the compiler can determine
1638 that (+ INDEX <constant>) does not require a bounds check and FOO
1639 has an element type at least 8 bits wide.
1640 * enhancement: a new, experimental synchronous timeout facility is
1641 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1642 * enhancement: when a symbol name conflict error arises, the
1643 conflicting symbols are always printed with a package prefix.
1644 (thanks to Kevin Reid)
1645 * enhancement: stepping is now once again supported on the SPARC and
1646 MIPS platforms. (It is also now more likely to work on CheneyGC
1648 * enhancement: sb-sprof can now also track and report accurate call
1650 * bug fixes: the treatment of non-standard subclasses of
1651 SB-MOP:SPECIALIZER is more correct.
1652 * incompatible change: PURIFY no longer copies the data from the
1653 dynamic space into the static and read-only spaces on platforms
1654 that use the generational garbage collector
1655 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1657 * bug fix: GC race condition occasionally resulting in crashes with
1658 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1660 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1662 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1663 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1664 * bug fix: an asynchronous interrupt could previously leave the
1665 system running with GC inhibited.
1666 * bug fix: a DECLARE form evaluated at top-level now causes an error
1667 rather than silently (or verbosely) returning NIL.
1668 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1669 result in a error being signalled due to source form lookup errors.
1670 (reported by Peter Graves)
1672 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1673 * incompatible change: removed writer methods for host-ent-name,
1674 host-ent-addresses -- changing the values did not update the DNS
1676 * minor incompatible change: changed experimental JOIN-THREAD interface
1677 * documentation: the manual now lists reader and writer methods
1678 in class slot documentation sections. (thanks to Richard M Kreuter)
1679 * documentation: unwinding from asyncronous events has been
1680 documented as unsafe.
1681 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1682 in multithreaded application code.
1683 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1684 platforms. (thanks to James Anderson for the optimization hint)
1685 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1687 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1688 variants no longer cons.
1689 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1690 their NOT- variants no longer cons.
1691 * optimization: EQUAL hash tables no longer use SXHASH for objects
1692 of all data types, but instead use an EQL hash for types for which
1693 EQUAL is the same as EQL.
1694 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1695 are significantly faster.
1696 * optimization: new STRING-OUTPUT-STREAM implementation is much
1697 faster for large outputs and conses 30% less on average.
1698 * enhancement: XREF information is now collected for references made
1699 to global variables using SYMBOL-VALUE and a constant argument.
1700 * enhancement: SIGINT now causes a specific condition
1701 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1702 * bug fix: FILE-POSITION used to signal an error in some cases where
1703 ANSI requires it to return NIL.
1704 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1705 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1706 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1708 * bug fix: bad type declaration in the CLOS implementation has
1709 been fixed. (reported by James Anderson)
1710 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1711 (thanks to Lutz Euler)
1712 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1713 required by the CLHS. (reported by Samium Gromoff)
1714 * bug fix: dead unbound variable references now signal an error.
1715 (reported by Marco Monteiro)
1716 * bug fix: / with an unused value was being deleted in safe code.
1717 (thanks to Marco Monteiro and Kevin Reid)
1718 * bug fix: number of characters that can be written onto a single
1719 line in a file is unlimited.
1720 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1721 been fixed by inhibiting interrupts for when GC is disabled.
1722 * bug fix: some interrupt safety issues with GETHASH, (SETF
1723 GETHASH), CLRHASH and REMHASH have been fixed.
1724 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1725 specifier no longer causes infinite recursion.
1726 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1727 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1728 * bug fix: an &environment argument with the correct variable information
1729 is passed to macros that are expanded during byte compilation.
1730 (reported by Samium Gromoff)
1731 * bug fix: a logic error was causing the hash function for bignums to
1732 have excessive amounts of collisions. (reported by Faré Rideau)
1733 * bug fix: modifying the contents of an array could change the return
1734 value of SXHASH on that array, which is only allowed for strings
1735 and bit vectors. (bug introduced in 0.9.16)
1736 * improvement: the x86-64/darwin port now passes all tests (except
1737 for the debugger tests) but should still be considered
1738 experimental until this is fixed.
1739 * improvement: a style-warning is signaled for CASE (etc) clauses with
1740 duplicate keys. (thanks to Kevin Reid)
1741 * improvement: macos/x86 and macos/x86-64 now use mach exception
1742 handlers for dealing with illegal instructions (for trapping and
1743 error handling) and memory protection violations (for GC).
1745 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1746 * new platform: experimental support for x86-64/darwin (MacOS).
1747 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1748 and getnameinfo sockets functions are used instead of gethostbyaddr
1749 and gethostbyname, on platforms where the newer functions are available.
1750 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1752 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1753 don't print the startup banner, but behave as if --noinform was passed
1754 as a command line argument. (thanks to Kevin Reid)
1755 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1756 * optimization: code using alien values with undeclared types is much faster.
1757 * optimization: the compiler is now able to open code SEARCH in more cases.
1758 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1759 * bug fix: using standardized COMMON-LISP special variables as loop
1760 variables no longer signals bogus package lock violations. (reported
1762 * bug fix: declaring local loop variables to be of a range-limited type
1763 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1764 (reported by Andras Simon)
1765 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1766 bugs remain on x86-64.)
1767 * bug fix: the #= and ## reader macros now interact reasonably with
1768 funcallable instances.
1769 * bug fix: type-checks for function arguments were compiled using the
1770 compiler policy settings of the wrong lexical environment (in 1.0.2
1772 * bug fix: SHADOW accepts characters as string designators, as required
1773 by the spec (thanks to Eric Marsden)
1774 * bug fix: fixed GC safety issues when foreign functions are called with
1775 non-base strings as arguments
1776 * bug fix: more consistent error output for fatal-compiler-warnings, like
1778 * bug fix: functions with proper names were showing up as NIL in
1779 backtraces (reported by Edi Weitz, regression in 1.0.2)
1781 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1782 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1783 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1784 produce more efficient code on the x86 if the compiler can determine
1785 that (+ INDEX <constant>) does not require a bounds check and FOO
1786 has an element type at least 8 bits wide.
1787 * bug fix: references to undefined variables are handled the same way
1788 in toplevel forms as in the normal compiler
1789 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1791 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1792 (thanks to Jon Buller)
1793 * improvement: faster compilation times for complex functions
1794 * improvement: added readlink support to SB-POSIX (thanks to Richard
1797 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1798 * improvement: experimental support for mach exception handling on
1799 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1801 * improvement: support for GBK external format.
1802 (thanks to Chun Tian (binghe))
1803 * improvement: the debugger now displays variables that have been closed
1804 over, in code compiled with (DEBUG 2) or higher
1805 * improvement: support for executable cores on NetBSD (thanks to
1807 * new feature: added a RESTART-FRAME debugger command
1808 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1809 be used to provide an implementation for FILE-POSITION on Gray streams
1810 (thanks to Eric Marsden)
1811 * new feature: add syslog, openlog and closelog support to SB-POSIX
1812 (thanks to Richard Kreuter)
1813 * optimization: the function call overhead in code compiled with
1814 a high DEBUG optimization setting is significantly
1815 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1816 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1817 (thanks to Tony Martinez)
1818 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1819 evaluated once (reported by Kevin Reid)
1820 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1821 works again (thanks to Kilian Sprotte)
1822 * bug fix: an error is signaled for tagbodies with duplicate tags
1823 (thanks to Stephen Wilson)
1824 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1825 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1826 stack frames from alien callbacks.
1827 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1828 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1829 65k pages (thanks to David Woodhouse)
1830 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1832 changes in sbcl-1.0.1 relative to sbcl-1.0:
1833 * new platform: FreeBSD/x86-64, including support for threading.
1834 * new feature: the compiler stores cross-referencing information
1835 about function calls (who-calls), macroexpansion (who-macroexpands)
1836 and special variables (who-binds, who-sets, who-references) for code
1837 compiled with (< SPACE 3). This information is available through the
1838 sb-introspect contrib.
1839 * new feature: users may subclass SEQUENCE, and have instances of
1840 these classes interoperate with standard Common Lisp functions if
1841 a number of methods are defined. (This feature is experimental
1842 and the interface subject to change based on feedback from SBCL
1843 users and the general community)
1844 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1845 x86/x86-64, rather than the previous fixed depth of 8
1846 * improvement: another pthread back-end of mutex "pthread-futex".
1847 * bug fix: non-ascii command-line arguments are processed correctly
1848 (thanks to Yaroslav Kavenchuk)
1849 * bug fix: non-required arguments were not passed correctly when a method
1850 defined using DEFMETHOD was called from a mop-generated method using
1851 CALL-NEXT-METHOD (reported by Pascal Costanza)
1852 * bug fix: recursion is now permitted in accessors through
1853 SLOT-UNBOUND. (reported by Pascal Costanza)
1854 * bug fix: an error was signaled at startup if the HOME environment
1855 variable was defined, but had an empty value (reported by Peter Van Eynde)
1856 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1857 signal a PROGRAM-ERROR, not a BUG.
1858 * bug fix: SB-POSIX:READDIR works when built with large file support.
1859 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1861 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1862 for better automatic indentation support. (thanks to Matt Pillsbury)
1863 * optimization: loading generic functions no longer takes O(n^2) time,
1864 proportional to the amount of methods in the generic function
1865 (reported by Todd Sabin and Jeremy Brown)
1866 * optimization: the FIND and POSITION family of sequence functions
1867 are significantly faster on arrays whose element types have been
1869 * improvements to the Windows port:
1870 ** Intermittent heap corruption problems have been fixed. (thanks
1871 to Alastair Bridgewater)
1872 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1874 ** Lisp is able to unwind foreign exception frames from alien
1875 callbacks. (thanks to Alastair Bridgewater)
1877 changes in sbcl-1.0 relative to sbcl-0.9.18:
1878 * improvement: experimental support for threading on FreeBSD/x86.
1879 (thanks to NIIMI Satoshi)
1880 * improvement: runtime option --dynamic-space-size can be used
1881 to set the size of the dynamic space reserved on startup.
1882 * improvement: floating point modes in effect are now saved in
1883 core, and restored on startup.
1884 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1885 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1886 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1887 threads are not running after *SAVE-HOOKS* have run.
1888 * improvement: writes to CLOS instance slots are type-checked in code
1889 compiled with (SAFETY 3)
1890 * improvement: floating-point exception handling on FreeBSD (thanks to
1892 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1893 (thanks to Zach Beane)
1894 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1896 * improvement: added support for the Shift-JIS external format.
1897 (contributed by NIIMI Satoshi)
1898 * improvement: callbacks are supported on Linux/PPC. (thanks to
1900 * bug fix: compiler bug triggered by a (non-standard) VALUES
1901 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1902 * bug fix: file compiler no longer confuses validated and already
1903 dumped structurres. (reported by Kaersten Poeck)
1904 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1905 fill-pointer signals a type-error as required. (thanks to
1907 * bug fix: disassemly of funcallable instances works.
1908 * bug fix: single stepping on PPC.
1909 * bug fix: fix thread-safety problems in the type system (generally
1910 manifesting as nonsensical errors like "STRING is a bad type specifier
1911 for sequences" or "The value 1 is not of type FIXNUM")
1912 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1913 (reported by Josip Gracin)
1914 * bug fix: an error is signaled for attempts to displace arrays with
1915 incompatible element types (thanks to Mario Mommer)
1916 * bug fix: more correct handling of wide characters in debug info
1917 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1918 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1919 and don't cause extra consing
1920 * optimization: MAP and MAP-INTO are significantly faster on vectors
1921 whose elements types have been declared.
1922 * Improvements to SB-SPROF:
1923 ** Support for allocation profiling
1924 ** Reduced profiling overhead, especially for long profiling runs
1925 * Improvements to the Windows port:
1926 ** floating point exceptions are now reported correctly.
1927 ** stack exhaustion detection works partially.
1928 ** more accurate GET-INTERNAL-REAL-TIME.
1929 ** better RUN-PROGRAM behaviour in case of errors in the child
1931 ** PROBE-FILE now simplifies pathnames correctly.
1932 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1933 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1934 1st 1970 (although time timezone data will be unreliable).
1935 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1937 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1938 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1939 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1941 * bug fix: two potential GC deadlocks affecting threaded builds.
1942 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1943 returns T (reported by Anton Kazennikov)
1944 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1945 with non-variable places
1946 * bug fix: remove a race condition in the setting of
1947 funcallable-instance functions, this should make threaded CLOS
1948 code more stable against memory faults.
1949 * bug fix: corruption of specials when unbinding is interrupted by an
1950 asynchronous unwind (reported by Hannu Koivisto)
1951 * improvement: the debugger will now also display local variables that
1952 are only used once, for code compiled with a DEBUG optimization quality
1955 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1956 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1957 * incompatible change: External-format support for FFI calls. The
1958 SB-ALIEN:C-STRING no longer implies an ASCII
1959 external-format. Instead the string is subject to external-format
1960 conversion. Additionally return values of type SB-ALIEN:C-STRING
1961 are by default of type (SIMPLE-ARRAY CHARACTER), not
1962 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1963 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1964 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1965 * incompatible change: SB-EXT package no longer contains the
1966 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1967 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1969 * incompatible change: the single-stepper is no longer available
1970 on Alpha, Mips and Sparc platforms.
1971 * minor incompatible change: the direct superclasses of
1972 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1973 not (STANDARD-OBJECT FUNCTION). This makes the
1974 class-precedence-lists of GENERIC-FUNCTION and
1975 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1977 * minor incompatible change: the default stream external format on
1978 non-unicode SBCL is no longer detected from the locale, but is always
1980 * new feature: Add a version of evaluator that uses an interpreter instead
1981 of the compiler. EVAL still uses the compiler by default, to switch it
1982 to use the interpreter, set the value of the variable
1983 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1984 * minor incompatible change: the single-stepper REPL has been merged
1985 with the normal debugger (see the "Stepping" heading of the debugger help
1986 for more details). The debugger command STEP will no longer switch
1987 to the single-stepper REPL.
1988 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1989 for a type now works.
1990 * bug fix: loading of default sysinit file works. (thanks to Leonid
1992 * bug fix: better detection of circularities in the file-compiler.
1993 (reported by Marco Monteiro)
1994 * bug fix: the CL pathname functions now work with files that have
1995 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1996 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1997 systems with Unix98 pty semantics.
1998 * bug fix: ASDF-INSTALL will now work with bsd tar.
1999 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
2001 * bug fix: timers expiring in dead threads no longer cause a
2002 type-error (reported by Paul "Nonny Mouse").
2003 * bug fix: thanks to more lightweight single-stepper instrumentation,
2004 code compiled with (DEBUG 3) will compile and execute significantly faster,
2005 and will have more accurate type-inferencing than before
2006 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
2007 whose bindings are modified
2008 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
2009 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
2010 ** replace dummy implementations of CL:MACHINE-INSTANCE and
2011 CL:SOFTWARE-VERSION with proper version
2013 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
2014 * feature: implemented the READER-METHOD-CLASS and
2015 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
2016 as specified by AMOP.
2017 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
2019 * optimization: faster LOGCOUNT implementation on x86 and x86-64
2020 (thanks to Lutz Euler)
2021 * optimization: hashing of general arrays and vectors has been
2022 improved. (reported by Any Fingerhut)
2023 * enhancement: SB-INTROSPECT is now able to find definitions of
2024 profiled functions. (thanks to Troels Henriksen)
2025 * enhancement: compiler-macro expansion applies now to FUNCALL forms
2026 as well, allowing compiler-macros for SETF-functions to expand.
2027 * enhancement: step-instrumentation no longer wraps known
2028 single-value functions with multiple-value context, allowing
2029 better type inference.
2030 * fixed bug #337: use of MAKE-METHOD in method combination now works
2031 even in the presence of user-defined method classes. (reported by
2032 Bruno Haible and Pascal Costanza)
2033 * fixed bug #339(c): if there are applicable methods not part of any
2034 long-form method-combination group, call INVALID-METHOD-ERROR.
2035 (reported by Bruno Haible)
2036 * fixed bug #361: the :FUNCTION initarg in the protocol for
2037 initialization of methods can now be used to override
2038 internally-produced optimized functions. (reported by Bruno
2040 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
2041 system-provided lambda expression no longer cause warnings about
2042 unbound #:|pv-table| symbols.
2043 * bug fix: improved the handling of type declarations and the
2044 detection of violations for keyword arguments with non-constant
2046 * bug fix: potentially erronous calls to PATHNAME and
2047 MERGE-PATHNAMES were being flushed in some cases.
2048 (reported by Richard Kreuter)
2049 * bug fix: compiled calls to TYPEP were mishandling obsolete
2050 instances. (reported by James Bielman and Attila Lendvai)
2051 * bug fix: format strings with ~> without matching ~< no longer
2052 trigger an AVER, but signal an understandable error instead.
2053 (reported by Antonio Martinez)
2054 * bug fix: specifying an output-file without a pathname-type for
2055 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
2056 of input-file instead of "fasl". (reported by Robert Dodier)
2057 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
2058 (reported by James Y Knight).
2059 * bug fix: compiler-macros-function did not consider the environment
2060 argument for shadowing by local functions.
2061 * bug fix: compiler-macros expansion was inhibited by local INLINE
2063 * bug fix: inline expansions of known functions were subject to
2064 step-instrumentation in high debug policies, leading to problems
2065 with type-inference.
2066 * bug fix: compiler failed to differentiate between different CONS
2067 types in some cases.
2068 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
2069 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
2070 * bug fix: #368: incorrect use of expressed vs. upgraded array
2072 * bug fix: #306a: more precise unions of array types.
2073 * thread-safety improvements:
2074 ** CONDITION-WAIT could return early on Linux, if the thread was
2075 interrupted and subsequently continued with SIGCONT.
2076 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
2078 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
2079 * added support for the ucs-2 external format. (contributed by Ivan
2081 * minor incompatible change: pretty printing of objects of type
2082 (cons symbol) is, in the default pprint-dispatch-table, now
2083 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
2085 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
2086 called later in a class's lifetime, possibly as late as when the
2087 first instance of the class is created. Previously,
2088 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
2089 class became finalizeable.
2090 * fixed bug: FILE-POSITION sometimes returned inconsistent results
2091 for multibyte external-format streams. (thanks to "vbzoli")
2092 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
2093 with :ALLOCATION :CLASS inherited from superclasses of the
2095 * fixed bug: anonymous classes can now be created using the :NAME
2096 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
2097 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
2098 * fixed bug: core-files saved with :EXECUTABLE T can again be
2099 executed when SBCL_HOME isn't set. (reported by James Knight)
2100 * fixed bug: toplevel LOCALLY forms with declarations could
2101 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
2102 * fixed bug: printing from several different threads using different
2103 values of *print-case* could cause invalid output, due to
2104 some internal special variables of the printer not being bound
2105 thread-locally (reported by Max Mikhanosha)
2106 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
2107 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
2108 (reported by Pascal Costanza)
2109 * fixed bug: FINALIZE-INHERITANCE is called from
2110 REINITIALIZE-INSTANCE on classes when the class has previously
2111 been finalized, as required by AMOP.
2112 * minor code generation optimizations:
2113 ** better register allocation in CLOS dispatching functions
2114 ** overflow detection when coercing signed bytes to fixnums on x86-64
2115 ** is now implemented with one IMUL instruction instead of three shifts
2116 ** more efficient bit-vector access on x86 and x86-64
2117 ** more efficient access to raw structure slots on x86 and x86-64
2118 * fixed some bugs revealed by Paul Dietz' test suite:
2119 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
2120 return its argument.
2122 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
2123 * feature: thread support on Solaris/x86, and experimental thread support
2125 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
2127 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
2128 default method for SLOT-UNBOUND.
2129 * minor incompatible change: prevent the user from specializing the
2130 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
2131 somewhat counter to the intent of the protocol, I (CSR) think, and
2132 additionally it just doesn't work in SBCL as currently
2133 implemented, thanks to optimizations (that are always valid for
2134 the other three SLOT-VALUEish functions, but not for the setter).
2135 * minor incompatible change: the :SB-LDB feature is now enabled by
2136 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
2137 the low-level debugger.
2138 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
2139 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
2140 down to an SB-SYS:FD-STREAM, allowing direct communication in
2141 more cases. Temporary files and pipes are still used as a fallback
2143 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
2144 on the non-reentrant C functions localtime(3) and gmtime(3).
2145 * bug fix: global optimization policy was not visible in LOCALLY and
2147 * bug fix: class objects can be used as specializers in methods.
2148 (reported by Pascal Costanza)
2149 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
2150 with a type error. (reported by blitz_ on #lisp)
2151 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
2152 failed with a type error. (reported by Pascal Bourguignon)
2153 * bug fix: merging pathnames against defaults with :DIRECTORY
2154 starting with '(:RELATIVE :BACK) should preserve the :BACK.
2155 (reported by James Y Knight)
2156 * bug fix: saving large (>2GB) cores on x86-64 now works
2157 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
2158 constant index and a (simple-array (signed-byte 32)) array
2159 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
2160 error (patch by Robert J. Macomber)
2161 * feature: TIME now displays the amount of run-time spent in GC
2162 * bug fix: The debugger now does a better job of respecting
2163 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
2164 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
2165 workaround for bug 403.)
2166 * bug fix: floating point exception handling now works on FreeBSD
2167 (thanks to NIIMI Satoshi)
2168 * fixed some bugs revealed by Paul Dietz' test suite:
2169 ** MISC.641: LET-conversion were not supposed to work in late
2171 * improvements to the Win32/x86 port:
2172 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
2173 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
2175 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
2176 * new feature: source path information is generated for macro-expansion
2177 errors for use in IDE's like Slime (thanks to Helmut Eller)
2178 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
2179 * bug fix: compiler does not loop forever on an invalid type in
2181 * improvement: compilation of most CLOS applications is significantly
2183 * optimization: added a limited bytecode compiler for simple toplevel
2184 forms, speeding up compilation and FASL loading
2185 * bug fix: the statistical profiler now properly distinguishes anonymous
2188 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
2189 * minor incompatible change: in sbcl-0.9.11 (but not earlier
2190 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
2191 into subclasses of the finalized class. Now user calls to
2192 FINALIZE-INHERITANCE finalize just the one class, and calls by the
2193 system return before any subclasses are finalized.
2194 * minor incompatible change: The reader no longer ignores errors
2195 regarding non-existent packages in #+ and #- feature tests.
2196 * new feature: command line options --no-sysinit, --no-userinit to
2197 inhibit loading the corresponding init files
2198 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
2199 for sending data through UDP sockets (thanks to François-René Rideau)
2200 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
2201 error is signalled instead (thanks to François-René Rideau)
2202 * minor incompatible change: Error signalling behaviour of lexical
2203 operations violating package locks has changed slightly. Refer to
2204 documentation on package locks for details.
2205 * enhancement: EVAL can process IF-expressions without resorting to the
2207 * bug fix: Some operations involving SIN, COS, or TAN weren't being
2208 constant-folded properly due to missing out-of-line functions.
2209 (reported by Mika Pihlajamäki)
2210 * bug fix: LISTEN sometimes returned T even in cases where no data was
2211 immediately available from the stream
2212 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
2213 were reversed. (reported by Levente Mészáros)
2214 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
2215 when given a vector displaced to an adjustable array. (reported
2217 * bug fix: derivation of float boundaries from numbers outside the
2218 appropriate float range (reported by John Wiseman)
2219 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
2221 * bug fix: precomputing cache entries for generic functions with
2222 some subclasses of specializers as yet invalid does not attempt to
2223 fill a cache line with a negative offset. (reported by Levente
2225 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
2226 allow condition class objects as arguments to DOCUMENTATION and
2227 (SETF DOCUMENTATION); only find and set documentation for
2228 structure names for the STRUCTURE doc-type. (suggested by Gary
2230 * improvements to the Win32/x86 port:
2231 ** (user-homedir-pathname) and default initialization file
2232 locations now know about the user's "Documents and Settings"
2233 directory (thanks to Yaroslav Kavenchuk)
2234 ** run-program is implemented (thanks to Mike Thomas)
2235 ** sockets support (thanks to Timothy Ritchey)
2236 ** better backtrace support (thanks to Alastair Bridgewater)
2237 ** sb-grovel supported
2238 ** asdf-install and sb-posix work somewhat
2239 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
2240 * improvements to the Solaris/x86 port:
2241 ** works on Solaris 11/Solaris Express
2242 ** floating-point exception handling support
2243 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
2244 * fixed some bugs revealed by Paul Dietz' test suite:
2245 ** REMOVE-METHOD returns its generic function argument even when
2246 the method is not one of the generic functions' methods.
2247 ** objects declared as MEMBER types can be admissible arguments to
2248 structure accessors.
2249 ** printing characters should simply be printed by the FORMAT ~:C
2251 ** compiler failure when compiling functions with hairy constant
2252 defaults for optional parameters.
2253 ** compiler produces wrong code when MAYBE-INLINE-expanding a
2254 function, which is already optimized.
2256 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
2257 * new platform: experimental support for SBCL x86/Darwin, including
2258 MacOS X 10.4.5 on Intel.
2259 * new feature: Unicode character names are now known to the system
2260 (through CHAR-NAME and NAME-CHAR).
2261 * new feature: the filesystem location of SBCL's core file is
2262 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
2263 * minor incompatible change: the contrib modules SB-POSIX and
2264 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
2265 this change is to make it easier to distribute
2266 location-independent binaries.
2267 * bug fix: as implied by AMOP, standardized classes no longer have
2268 slots named by external symbols of public packages. (reported by
2270 * optimization: calling functions via constant symbols -- as in
2271 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
2272 via the function object as in (FUNCALL #'FOO).
2273 * optimization: on x86 and x86-64, the calling convention for the
2274 case of dealing with an unknown number of values has been altered
2275 to be friendlier to the prediction heuristics implemented,
2276 particularly with reference to CALL and RET pairing. (thanks to
2277 Alastair Bridgewater)
2278 * optimization: on x86, the code for access to thread-local symbol
2279 values and binding/unbinding of thread-local symbols is smaller.
2280 (thanks to Alastair Bridgewater)
2281 * enhancement: CONSTANTP is now able to determine constantness of
2282 more complex forms, including calls to constant-foldable standardized
2283 functions and some special forms beyond QUOTE.
2284 * fixed bug: occasional GC crashes on Solaris/x86
2285 * optimization: x86-64 supports stack allocation of results of simple
2286 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
2287 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
2288 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
2289 (thanks to James Knight)
2290 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
2291 by the standard (thanks to Stephen Compall)
2293 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
2294 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
2295 be used for bundling the runtime and the core file into one
2296 executable binary. This feature is not currently supported on all SBCL
2297 platforms. (thanks to James Bielman and NIIMI Satoshi)
2298 * new feature: a generational or ephemeral garbage collector is now
2299 the default on the PowerPC platform (both Linux and Darwin). The
2300 old Cheney (stop and copy) collector is a build-time option.
2301 (thanks to Cyrus Harmon, after Raymond Toy)
2302 * minor incompatible change: the method by which SBCL finds its
2303 contributed modules has changed; it no longer relies on symbolic
2304 links from an $SBCL_HOME/systems directory, but searches directly
2305 in the subdirectories of $SBCL_HOME.
2306 * enhancement: the dynamic heap size on the Linux/PPC platform is
2307 markedly larger, even using the older Cheney garbage collector.
2308 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
2309 certain complicated string types. (reported by Gary King)
2310 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
2311 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
2312 * fixed bug: DOCUMENTATION on structures no longer signals an
2313 error if no structure type of the right name exists. (reported by
2315 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
2316 applicable non-standard primary or :AROUND method on
2317 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
2318 no longer cause unbound variable errors. (reported by Kevin Reid)
2319 * fixed bug: metacircle resolution in cases where methods have slots
2320 added before the slots from STANDARD-METHOD. (reported by Jean
2322 * fixed bug: the Power PC assembly code for calling into lisp
2323 sometimes computed its return address wrongly (depending on the
2324 15th bit of the address, fixed at link time). This used to
2325 manifest itself in a segmentation violation while building PCL.
2326 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
2327 many others over the years)
2328 * fixed bug: ROOM no longer reports silly numbers for stack usage if
2329 the stack pointer (treated as a signed integer) is negative.
2330 (thanks to Peter van Eynde)
2332 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
2333 * new platform: experimental support for the Windows operating
2334 system has been added. (thanks to Alastair Bridgewater)
2335 * fixed several bugs in and robustified the PPC FFI (including
2336 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
2337 * bug fix: allow non-simple string symbol names (reported by Paul
2339 * bug fix: interrupt handling on NetBSD (thanks to Richard M
2341 * bug fix: saving a core corrupted callbacks on x86/x86-64
2342 * bug fix: closed a loophole in metacircularity detection and
2343 grounding in the PCL implementation of CLOS.
2344 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
2345 (reported by tomppa on #lisp)
2346 * bug fix: clear the direction flag on Lisp -> C transitions, as
2347 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
2348 (reported by Andrej Grozin and Hendrik Maryns)
2349 * optimization: major improvements to GC efficiency on GENCGC platforms
2350 * optimization: faster implementation of EQUAL
2351 * optimization: emit more efficient opcodes for some common
2352 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
2353 * optimization: several other minor code-generation improvements on x86-64
2354 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
2356 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
2357 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2358 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
2359 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
2360 by ANSI as a generic function, and for consistency (SETF
2361 GENERIC-FUNCTION-NAME) is treated likewise.
2362 * fixed bug #233.b: make constraint propagation notice when a variable
2363 value is changed after it is referenced but before it is used
2364 * fixed bug #296: no more arbitrary behaviour from filenames with
2365 odd characters as --load arguments. (reported by Adam Warner)
2366 * fixed bug #390: :CHARACTER-SET pathname components now work as
2367 expected. (reported by Tim Daly Jr)
2368 * fixed bug #391: complicated :TYPE intersections in slot
2369 definitions no longer cause an error in PCL internals.
2370 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
2371 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
2372 returning the number of octets which would be written to the
2373 file-stream. (thanks to Robert J. Macomber)
2374 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
2375 arguments to the internal stream functions. (thanks to David
2377 * bug fix: stack exhaustion is now handled on the x86/SunOS
2378 platform. (thanks to Daisuke Homma)
2379 * bug fix: more accurate ROOM results on GENCGC platforms
2380 * optimization: improved type inference for arithmetic-for
2381 index variables in LOOP
2382 * optimization: faster floating-point SQRT on x86-64
2383 * fixed some bugs revealed by Paul Dietz' test suite:
2384 ** DOCUMENTATION returns NIL instead of "" for method combinations
2385 that don't have a docstring
2387 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
2388 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2389 GENERIC-FUNCTION-NAME) are no longer generic functions, and
2390 therefore are not customizeable by user code (as seems to be at
2391 least permitted and maybe required by AMOP). As a consolation,
2392 however, the SBCL implementation of these functions now calls
2393 REINITIALIZE-INSTANCE as specified by AMOP.
2394 * bug fix: slot-definition documentation is propagated as per ANSI
2395 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
2396 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
2397 argument, as do eslotd initialization methods. (from Pascal
2398 Costanza's "Closer" project)
2399 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
2400 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
2402 * bug fix: it is now possible to have more than one subclass of
2403 STANDARD-GENERIC-FUNCTION without causing stack overflow.
2404 (reported by Bruno Haible, Pascal Costanza and others)
2405 * bug fix: the dependent update protocol now works for generic
2406 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
2407 and Pascal Costanza)
2408 * bug fix: condition-class instances corresponding to
2409 DEFINE-CONDITION forms are now created eagerly. (reported by
2410 Kalle Olavi Niemitalo on comp.lang.lisp)
2411 * bug fix: floating point printing is more accurate in some
2412 circumstances. (thanks to Simon Alexander)
2413 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
2414 merged with *DEFAULT-PATHNAME-DEFAULTS*.
2415 * bug fix: callbacks on OS X now preserve stack-alignment.
2416 (thanks to Cyrus Harmon)
2417 * enhancement: the x86-64 disassembler is much better at
2418 disassembling SSE instructions. (thanks to Lutz Euler)
2419 * enhancement: CHECK-TYPE on a variable now allows type-inference
2420 (thanks to Vincent Arkesteijn)
2421 * optimization: improved performance of EUC-JP external format.
2422 (thanks to NIIMI Satoshi)
2423 * optimization: performance improvements to IO on file streams of
2424 :ELEMENT-TYPE CHARACTER
2425 * optimization: much faster memory allocation on x86-64
2426 * optimization: faster garbage collections (latency and throughput)
2428 * optimization: faster bignum subtraction and division on x86 and x86-64
2429 (thanks to Lutz Euler)
2430 * optimization: more accurate type-inference for loops that use a
2431 floating point index variable or a negative step.
2433 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2434 * bug fix: add a workaround to SBCL looping infinitely at startup on
2435 Linux kernels with apparently buggy implementations of personality().
2436 (thanks to Svein Ove Aas)
2437 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2438 (thanks to David Lichteblau)
2439 * bug fix: local bindings shadow global symbol macros for
2440 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2441 * optimization: non-open coded uses of numeric comparison operators
2442 (e.g. >) no longer cons when called with more than one parameter
2443 on platforms supporting dynamic-extent allocation.
2444 * enhancement: saving cores with foreign code loaded is now supported
2445 on MIPS/Linux in addition to the previously supported platforms.
2446 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2447 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2448 * bug fix: division by zero in sb-sprof when no samples were collected
2449 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2450 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2451 (thanks to Alexander Kjeldaas)
2452 * new feature: ignoring the return values of destructive functions
2453 like NREVERSE signals a compile-time style-warning.
2454 (thanks to Kevin Reid)
2455 * new feature: the alignment of alien structure fields can be
2456 explicitly specified. (thanks to Cyrus Harmon)
2457 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2458 and dump core on SIGQUIT
2460 ** incompatible change: threads do not inherit values of specials
2461 from their parents (see manual)
2462 ** bug fix: threads stacks belonging to dead threads are freed by the
2463 next exiting thread, no need to gc to collect thread stacks anymore
2464 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2465 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2466 ** bug fix: dynamic variable and thread start related gc lossage
2467 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2469 * fixed some bugs revealed by Paul Dietz' test suite:
2470 ** SUBTYPEP is slightly more accurate on heinously complicated
2471 CONS types where some of the members have uncertain (in the
2472 NIL, NIL sense) type relationships to each other.
2474 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2475 no longer purified unless :PURIFY T is explicitly specified.
2476 ** Non-purified cores are significantly smaller than before
2478 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2479 * new feature: timers based on Zach Beane's excellent timer package
2480 * added support for the following external formats: koi8-u,
2481 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2482 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2483 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2484 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2485 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2486 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2487 * incompatible change: a threaded SBCL will no longer revert to
2488 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2489 * bug fix: interrupts are disabled until startup is complete; no
2490 more sigsegvs when receiving a signal to soon
2491 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2493 * bug fix: add a workaround for the memory randomization features in
2494 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2495 workaround will only be in effect on systems with the proc filesystem
2497 * bug fix: printing objects of type HASH-TABLE signals a
2498 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2500 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2501 have been running for over 50 days. (reported by Gilbert Baumann)
2502 * bug fix: the logic for getting names of functions gets less
2503 confused when confronded with alternate-metaclass
2504 funcallable-instances. (reported by Cyrus Harmon)
2505 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2506 consistent, even on internal alternate-metaclass objects.
2507 * bug fix: flush closure information collected by physical
2508 environment analysis prepass before the main pass. (bug reported
2510 * bug fix: compiler pack phase does not modify a hashtable, which is
2511 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2512 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2513 may be expanded. (reported by Alexander <malishev> on c.l.l)
2514 * bug fix: no more style-warnings when compiling code using LAST
2515 under high-SPEED low-SPACE settings. (reported by David Wallin
2517 * fixed some incompatibilities between SBCL's MOP and the MOP
2519 ** the METAOBJECT class is now implemented;
2520 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2521 STANDARD-OBJECT, as required;
2522 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2523 now compatible (as required by VALIDATE-SUPERCLASS); there
2524 remains a consistency requirement over the presence or absence
2525 of the FUNCTION class in the superclass list of finalized
2526 classes; see the manual for more details;
2527 ** the :AROUND method for COMPUTE-SLOTS on
2528 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2529 requested slot ordering.
2531 ** bug fix: parent thread now can be gc'ed even with a live
2533 ** bug fix: binding a special with PROGV to no value is not
2534 the same as that symbol not having been bound (thanks to
2536 ** bug fix: binding specials is thread safe (thanks to
2538 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2539 and SB-SPROF do not die with 'no handler for signal XX in
2540 interrupt_handle_now(..)' anymore
2541 ** bug fix: WITH-TIMEOUT works with multiple running threads
2542 * fixed some bugs revealed by Paul Dietz' test suite:
2543 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2544 the :method-class keyword argument.
2546 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2547 * new port: the Solaris operating system on x86 processors is now
2548 mostly supported, though some rough edges in the environment
2549 remain. (thanks to Daisuke Homma)
2550 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2551 space for its heap. (thanks to Thiemo Seufer)
2552 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2553 closures. (thanks to Thiemo Seufer)
2554 * minor incompatible change: eof selects abort in the debugger.
2555 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2556 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2558 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2559 deprecated, and will go away in a future revision of SBCL.
2560 * minor incompatible change: GC-ON and GC-OFF are no longer
2561 implemented with a counter, it does not matter how many times gc
2562 is switched on or off
2563 * bug fix: discriminating functions for generic function classes
2564 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2565 make invalid assumptions about method precedence order. (reported
2567 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2568 platforms. (thanks to Thiemo Seufer)
2569 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2570 * bug fix: correct pathname printing with printer escaping is on.
2571 (thanks to Kevin Reid)
2572 * bug fix: complex VOP definitions in "user-space" no longer trigger
2573 package locks. (reported by Zach Beane)
2574 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2575 for funcallable-instance functions, and is no different from
2577 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2578 characters in the current readtable. (reported by Nicholas Neuss)
2579 * bug fix: SUBTYPEP on various CONS types returns more a more
2580 accurate acknowledgment of its certainty.
2581 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2582 lists in some cases. This partially fixes bug 384.
2583 * flush all standard streams before prompting in the REPL and the
2585 * bug fix: signal handling and triggering gc do not conflict
2586 directly anymore, in particular a high frequency sb-sprof does
2587 not prevent gc from running
2588 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2589 approximation for timezone and DST information between the
2590 universal time epoch and the smallest negative 32-bit time_t.
2591 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2592 year 1899 so long as the encoded time is non-negative.
2593 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2594 the correct answer for non-constant rotate amounts when performing
2595 an inline 32-bit rotation.
2597 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2598 there is only one thread in the session
2599 ** bug fix: memory leak for streams created in one thread and
2600 written to in another
2601 ** bug fix: lockup when compiled with gcc4
2602 ** bug fix: race that allows the gc to be triggered when gc is
2604 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2605 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2607 ** bug fix: finished threads are gc'ed properly
2608 * fixed some bugs revealed by Paul Dietz' test suite:
2609 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2610 has arguments with a different set of applicable methods from
2611 the orignal arguments.
2612 ** The type error thrown by MAP now has a correct expected-type
2614 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2615 name a compiled function.
2616 ** Three MISC tests where a large bignum was improperly coerced to
2617 a float that couldn't represent that bignum during type
2618 derivation were fixed.
2619 ** SUBTYPEP can now handle types involving the negation of a
2620 list-form FUNCTION type.
2621 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2622 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2623 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2625 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2626 * New feature: Experimental support for bivalent streams: streams
2627 opened with :element-type :default now allow character and binary
2628 (unsigned-byte 8) I/O
2629 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2630 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2632 * bug fix: no more highly sporadic "couldn't check whether ~S is
2633 readable" when reading a stream and an interrupt hits in the middle
2634 of a select system call
2635 * compiler better recognizes complex arrays (reported by Eduardo
2637 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2638 for variables on SBCL built with linkage-tables. (reported by Luis
2640 * various error reporting improvements.
2641 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2642 (Thanks to Thiemo Seufer)
2643 * bug fix: nested reader invokations work correctly
2644 * bug fix: it is possible to have simultaneous references to foreign
2645 code and foreign data with the same name.
2647 ** added x86-64 support
2648 ** incompatible change: the threading api now works with thread
2649 objects instead of thread ids
2650 ** bug fix: threads are protected from signals and interruption when
2651 starting up or going down
2652 ** bug fix: a race where an exiting thread could lose its stack to gc
2653 ** bug fix: don't halt on infinite error in threads if possible
2654 ** fixed numerous gc deadlocks introduced in the pthread merge
2655 ** bug fix: fixed thread safety issues in read and print
2656 ** bug fix: debugger doesn't hang on session lock if interrupted at
2657 an inappropriate moment
2658 ** bug fix: run-program is now thread safe(r)
2659 ** bug fix: inner with-recursive-lock no longer releases the mutex
2660 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2661 * fixed some bugs revealed by Paul Dietz' test suite:
2662 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2663 EXPECTED-TYPE slots filled.
2664 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2666 ** MAKE-CONDITION accepts classes as type-designators.
2667 ** COMPILE may never return NIL.
2668 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2669 range before calling Unix time functions
2671 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2672 * numerous signal handling fixes to increase stability
2673 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2674 * minor incompatible change: we now correctly canonize default
2675 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2676 as per the MOP, rather than the historical (INITARG INITFUNCTION
2677 INITFORM). (reported by Bruno Haible)
2678 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2679 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2680 for more information.
2681 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2682 pathname is a directory pathname.
2683 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2684 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2686 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2687 :METACLASS argument in addition to a class name. (reported by
2688 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2689 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2690 (reported by Bruno Haible; patch from Gerd Moellmann)
2691 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2693 * bug fix: more cleanups to the floating point exception handling on
2694 x86-64 (thanks to James Knight)
2695 * bug fix: the compiler does not try to constant fold calls of
2696 COERCE and BIT-* functions when they return freshly constructed
2697 objects. (reported by Nikodemus Siivola and Paul Dietz)
2698 * optimization: improved the MIPS versions of generic (in the
2699 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2700 * optimization: direct conversion of (unsigned-byte 32) to floats on
2701 the PowerPC platform.
2702 * optimization: structure instances with raw slots now use less
2703 memory, and probably show better memory locality. (thanks to
2705 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2706 are treated as requests for stack allocation on the x86-64,
2707 Alpha-32, PPC and SPARC platforms.
2708 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2709 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2710 (thanks to Tony Martinez)
2712 ** gcing a dead thread can no longer lead to lockups
2713 ** threads block signals until they are set up properly
2714 ** errno is no longer shared by threads
2715 ** interrupt-thread restores the eflags register on x86
2716 ** fixed some lockups due to gc/thread interaction
2717 * fixed some bugs revealed by Paul Dietz' test suite:
2718 ** invalid dotted lists no longer raise a read error when
2719 *READ-SUPPRESS* is T
2720 ** signal an error if a symbol that names a declaration is used
2721 as the name of a type, or vice versa
2722 ** allow using the (declare (typespec var*)) abbreviation for
2723 (declare (type typespec var*)) with all type specifiers
2724 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2725 ** always return NIL from PPRINT-POP when OBJECT is NIL
2726 ** don't signal errors when pretty-printing malformed LABELS,
2727 FLET or MACROLET forms
2728 ** declarations in a DOLIST body are also in scope for the
2730 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2732 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2735 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2736 * fixed cross-compiler leakages that prevented building a 32-bit
2737 target with a 64-bit host compiler.
2738 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2739 opened with :IF-EXISTS :OVERWRITE.
2740 * fixed bug 281: error for an invalid qualifier in a short-form method
2741 combination method is not signalled until the faulty method is called.
2742 * bug fix: iteration variable type inferrer failed to deal with open
2743 intervals. (reported by Alan Shields)
2744 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2745 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2746 or more runtime options were provided to the sbcl binary.
2747 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2748 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2749 * optimization: there's now a fast-path for fixnum arguments in the
2750 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2751 * optimization: the code generated on x86-64 is more compact thanks
2752 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2753 * bug fix: floating-point exception handling now partly works on
2754 x86-64. (Thanks to James Knight)
2755 * improvement to the MIPS backend from Thiemo Seufer: C-style
2756 64-bit long long arguments and return values to alien functions
2758 * fixed some bugs revealed by Paul Dietz' test suite:
2759 ** the type-error signalled from WARN has a filled-in DATUM slot.
2760 ** the type-error required when a stream is not associated with
2761 a file has the stream as its datum.
2762 ** type-errors on single-floats on x86-64 no longer have
2763 :INVALID-OBJECT as the datum
2764 ** the type-errors signalled for invalid function names now have
2765 a correct expected type
2766 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2767 for structures defined with a :TYPE. Documentation strings for
2768 typed structures are no longer immediately discarded
2769 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2770 broadcast streams and synonym streams. FILE-LENGTH now also works
2771 on broadcast streams.
2773 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2774 * incompatible change: the --noprogrammer option, deprecated since
2775 version 0.7.5, has been removed. Please use the equivalent
2776 --disable-debugger option instead.
2777 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2779 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2780 inoperational for a while now) has been completely removed.
2781 * null lexical environments are now printed as #<NULL-LEXENV>,
2782 significantly reducing the amount of clutter in typical backtraces.
2783 * documentation on weak pointers, finalization, and after GC hooks
2784 has been added to the manual.
2785 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2786 as other specialized array types, is much faster. SUBSEQ and
2787 COPY-SEQ on such arrays have also been sped up.
2788 * optimization: EQL is now more efficient when at least other argument
2789 is known to be of type (OR FIXNUM (NOT NUMBER)).
2790 * fixed bug: the runtime is now less vulnerable to changes in the
2791 size of the SBCL object on OS X, and virtual memory is reserved for
2792 all spaces that need to be at a fixed address.
2793 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2794 * fixed bug: finalizers and after GC hooks that cause consing are now
2796 * fixed bug: compiler error messages and summaries are now printed to
2797 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2798 * fixed inference of the upper bound of an iteration variable.
2799 (reported by Rajat Datta).
2800 * fixed bug 211e: calling local functions with duplicated constant
2801 keyword argument no longer causes a bogus style warning about an
2803 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2804 ftype declaration to be disregarded. (reported by Dave Roberts)
2805 * fixed bug 373: caused by erronous compilation of references to alien
2806 variables in the runtime on ppc/darwin.
2807 * fixed bug 376: CONJUGATE type deriver.
2808 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2809 (reported by Baughn on #lisp)
2810 * a cross-compiler bug on non-x86 platforms has been identified and
2811 fixed. (thanks to Bruno Haible)
2812 * improvements to the MIPS runtime code for increased stability.
2813 (thanks to Thiemo Seufer)
2814 * increased the maximimum compact environment size to allow
2815 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2816 * partial workaround for bug 135: don't name the function we're
2817 calling for hairy cases of EVAL, so as not to accumulate one environment
2818 entry per call to EVAL. (reported by Kevin Reid)
2819 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2820 * optimization: type testing for non-vector arrays should be faster.
2821 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2823 * bug fix: setting 31st element of a bit vector to zero did not work
2825 * bug fix: redefining a class definition which failed due to a
2826 previous accessor / function clash now works (but see BUGS entry
2827 #380 for more problems in this area). (thanks to Zach Beane)
2828 * the long form of DEFINE-METHOD-COMBINATION disables method group
2829 checking when given a single method group with pattern *. (thanks
2831 * on x86 compiler supports stack allocation of results of simple
2832 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2833 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2834 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2835 * fixed some bugs related to Unicode integration:
2836 ** the restarts for recovering from input and output encoding
2837 errors only appear when there is in fact such an error to
2839 * fixed some bugs revealed by Paul Dietz' test suite:
2840 ** MISC.549 and similar: late transformation of unsafe type
2841 assertions into derived types caused unexpected code
2843 ** SCALE-FLOAT type deriver is less wrong.
2844 ** type derivers for EXP, LOG and similar functions compute result
2845 types for complex arguments better.
2846 ** (MISC.563) CONJUGATE type deriver works for very restricted
2848 ** out-of-line type testers for character strings are available.
2849 ** EQUAL compiler transform understands specialness of objects
2851 ** accessing double-floats stored on the stack now works on x86-64.
2852 ** debugger internals could sometimes create invalid lispobjs,
2853 resulting in GC crashes.
2854 ** MISC.548: type check weakening can convert required type into
2856 ** initialization forms for bindings are not in scope of free special
2859 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2860 * incompatible change: thread support for non-NPTL systems has
2861 been removed - locking is buggy and unreliable. A threaded
2862 SBCL build will now warn at startup and refuse to create
2863 new threads, unless futex support is detected in the kernel
2864 * incompatible change: the top level REPL now has only an ABORT
2865 restart associated with it, not TOPLEVEL and ABORT as it used to.
2866 TOP and TOPLEVEL are now available as debugger commands for
2867 returning to the top level.
2868 * incompatible change: forms evaluated in the REPL now use the
2869 global optimization policy.
2870 * incompatible change: user- and system-initialization files are
2871 no longer processed with LOAD, but by READ and EVAL; hence the
2872 global optimization policy, startup package, readtable, etc,
2874 * The .fasl file format number has been incremented because of
2875 various incompatible changes.
2876 * internal entry point details and argument counts no longer appear
2877 in backtraces unless explicitly requested by setting
2878 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2879 * built-in and standard functions no longer have names like "top
2880 level local call to FOO".
2881 * fixed bug 32: functions defined in non-null lexical environments
2882 now have more legible printed representation
2883 * fixed bug 33: functions defined in non-null lexical environemnts
2884 are now more amenable to inspection by INSPECT.
2885 * workaround for bug 354: XEPs no longer appear in backtraces unless
2886 explicitly requested.
2887 * fixed bug: receiving the signal which results from attempting to
2888 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2889 SIGBUS on other *BSDs) on architectures where the C stack is also
2890 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2891 to debugger confusion or wild instability. (reported by Cheuksan
2893 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2894 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2895 (reported by Lutz Euler)
2896 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2897 compiled code. (reported by Paul Dietz)
2898 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2899 specialized parameter to a method to something that is not TYPEP
2900 the specializer is now possible.
2901 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2902 face of package deletion.
2903 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2904 pathname. (reported by David Tolpin on c.l.l)
2905 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2906 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2907 STRING-STREAM and FILE-STREAM, respectively.
2908 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2909 than previously; in addition, its test suite is now run on build.
2910 (thanks to Luke Gorrie)
2911 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2912 * optimization: added a immediate representation for single-floats
2914 * fixed some bugs revealed by Paul Dietz' test suite:
2915 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2916 ** attempting to create a package with a colliding nickname causes
2917 correctable errors to be signalled.
2918 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2919 ** several x86-64 backend bugs related to sign-extension of immediate
2922 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2923 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2924 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2926 * fixed loading of multiply forward-referenced layouts.
2927 (thanks to Cheuksan Wang)
2928 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2929 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2930 (see COMPILE-FILE documentation for details of :PRINT option)
2931 * fixed bugs 19 and 317: fixed-format floating point printing is
2932 more accurate. This also fixes a bug reported by Adam Warner
2933 related to the ~@F format directive.
2934 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2936 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2937 dispatch table if the from-char is a dispatch macro character.
2938 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2939 by the wordsize no longer ignore the last word. (reported by Lutz
2941 * fixed bug in type checking of dynamic-extent variables. (reported
2943 * optimization: sequence traversal functions use their freedom to
2944 coerce function designators to functions.
2945 * optimization: code with many calls to CLOS methods specialized on
2946 CLOS classes has had redundant type checks removed.
2947 * optimization: added declarations to speed up operations that access
2948 the internal character database (for example STRING-UPCASE)
2949 * optimization: comparison operations between floats and sufficiently small
2950 fixnums no longer create extra rationals
2951 * fixed some bugs related to Unicode integration:
2952 ** portions of multibyte characters at the end of buffers for
2953 character-based file input are correctly transferred to the
2954 start of the buffer at the next read.
2955 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2956 passing it through to OPEN.
2957 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2958 argument given to internal calls to OPEN.
2959 ** the built-in comment readers (introduced by character sequences
2960 ";" and "#|") are more forgiving to encoding errors; they will
2961 STYLE-WARN and then attempt to resync the stream at a character
2962 boundary. (thanks to Teemu Kalvas)
2963 * fixed some bugs in the x86-64 port:
2964 ** Negative short int return values from c-calls are sign-extended
2966 ** The stack is aligned to 16-bytes for c-calls, as required by
2967 the ABI. (thanks to Cheuksan Wang)
2968 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2969 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2970 for enums. (thanks to Vincent Arkesteijn)
2971 ** Multiple small optimizations and bugfixes for floating point
2973 * fixed some bugs revealed by Paul Dietz' test suite:
2974 ** Space, Tab, Linefeed, Return and Page have the invalid
2975 secondary constituent character trait.
2976 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2978 ** WITH-INPUT-FROM-STRING should only update the index place on
2980 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2981 works more reliably.
2982 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2983 with one-dimensional arrays with fill pointers.
2984 ** TYPE-OF failed on a complex with an integer realpart and a
2986 ** compiler failure during type inference for the code of form
2987 (IF (EQL X (THE ...)) ...) (MISC.535).
2989 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2990 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2991 platform. The port supports 61-bit fixnums, large memory spaces
2992 and reloading shared object files.
2993 * enhancement: saving cores with foreign code loaded is now
2994 supported on ppc/Darwin in addition to the previously supported
2996 * enhancement: the statistical profiler now walks deeper into the
2997 call stack for more meaningful call-graphs and accrued time
2998 reports (x86/x86-64 only). It also now reports time spent in
3000 * enhancement: it is now possible to trace most individual methods
3001 of a generic function in addition to tracing the generic function
3003 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
3005 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
3006 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
3007 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
3008 produces an error. (thanks to Vincent Arkesteijn)
3009 * bug fix: NAMESTRING on pathnames with :WILD components in their
3010 directories works correctly. (thanks to Artem V. Andreev)
3011 * fixed bug 125: compiler preserves identity of closures. (reported
3013 * bug fixed: functions with &REST arguments sometimes failed with
3014 "Undefined function" when compiled with (DEBUG 3). (reported by
3016 * bug fix: overflow during compiling of setting element of a bit
3017 vector with constant index and value. (reported by Timmy Douglas)
3018 * build fix: fixed the dependence on *LOAD-PATHNAME* and
3019 *COMPILE-FILE-PATHNAME* being absolute pathnames.
3020 * on x86 compiler partially supports stack allocation of dynamic-extent
3022 * GO and RETURN-FROM do not check the extent of their exit points
3023 when compiled with SAFETY 0.
3024 * fixed some bugs related to Unicode integration:
3025 ** encoding and decoding errors are now much more robustly
3026 handled; it should now be possible to recover even from invalid
3027 input or output to the terminal. (thanks to Teemu Kalvas)
3028 ** provided a first cut at implementing STRING-TO-OCTETS and
3029 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
3030 ** altered the SB-MD5 contributed module slightly, changing the
3031 interface just enough for it to be supportable for builds where
3032 lisp characters are not eight bits.
3033 * fixed some bugs revealed by Paul Dietz' test suite:
3034 ** the FORMATTER-generated functions for ~V[ conditionals require
3035 the correct number of arguments.
3036 ** READ-FROM-STRING returns the mandated second value when applied
3037 to displaced strings.
3038 ** the #\Rubout and #\Backspace characters are treated as invalid
3039 constituent characters by the tokenizer.
3041 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
3042 * new feature: reloading changed shared object files with
3043 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
3044 * new feature: references to foreign variables and functions
3045 can now be compiled and loaded before the corresponding shared
3046 object file is loaded, as long as the foreign definitions are
3047 available at runtime.
3048 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
3049 supported. (thanks to Dan Debertin)
3050 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
3051 just keywords. (thanks to Vincent Arkesteijn)
3052 * fixed bug #331: structure-class instances corresponding to
3053 DEFSTRUCT forms are now created eagerly.
3054 * fixed bug #345: backtraces from calls to undefined functions work
3055 on x86 as well. Related bug #61 is now also partially fixed on x86
3056 (backtraces from throws to unknown catch tags.)
3057 * bug fix: lambda-list parsing is now stricter vrt. order and number
3058 of lambda-list keywords.
3059 * bug fix: as specified by AMOP, an error is signalled if a
3060 class-option appears multiple times in a DEFCLASS form. (reported
3062 * bug fix: RANDOM can be compiled when the compiler derives the type
3063 of its numeric argument as a disjoint set of small integers.
3064 (reported by Paul Dietz)
3065 * bug fix: starting a core saved with shared objects loaded when
3066 those objects are not available no longer causes threaded SBCL to
3067 hang. (reported by Sean Ross)
3068 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
3070 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
3071 argument, which must always be NIL. (reported by Kalle Niemitalo)
3072 * bug fix: printing 1.0d+23 no longer results in an error.
3073 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
3074 * bug fix: structure slot setters preserve evaluation order. (thanks
3076 * fixed some bugs related to Unicode integration:
3077 ** RUN-PROGRAM can allow its child to take input from a Lisp
3078 stream. (reported by Stefan Scholl)
3079 ** ASDF-INSTALL successfully downloads .tar.gz files in all
3080 locales. (reported by Ken Causey)
3081 * fixed some bugs revealed by Paul Dietz' test suite:
3082 ** INCF, DECF and REMF evaluate their place form as specified in
3084 ** FORMATTER expands ~{ ~} iteration directives with V or #
3085 parameters correctly.
3086 ** FORMATTER deals with the ~@[ ~] conditional directive where the
3087 consequent uses no arguments correctly.
3088 ** the system has a partial understanding of the (COMPLEX RATIO)
3090 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
3093 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
3094 * new feature: a build-time option (controlled by the :SB-UNICODE
3095 keyword feature, enabled by default) for building the system with
3096 support for the entire 21-bit character space defined by the
3098 * new feature: the system now has rudimentary external-format
3099 support; the primary user-visible change at this time is that
3100 characters with the high bit set (such as lower-case-e-acute) will
3101 print correctly to a terminal in a UTF-8 environment.
3102 * minor incompatible change: BASE-CHAR no longer names a class;
3103 however, CHARACTER continues to do so, as required by ANSI.
3104 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
3105 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
3106 has been moved to the SB-EXT package (temporarily re-exported from
3108 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
3109 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
3111 * fixed bug #351: better error-handlind and reporting for malformed
3113 * fixed bug #350: bignum-printing is now more memory-efficient,
3114 allowing printing of very large bignums, eg. (expt 2 10000000).
3115 (reported by Bruno Haible)
3116 * fixed bug #302: better primitive-type selection for intersection
3118 * fixed bug #308: non-graphic characters now all have names, as
3119 required. (reported by Bruno Haible)
3120 * bug fix: redefining a class with different superclasses now correctly
3121 removes it from the direct-subclasses of its previous superclasses.
3122 (reported by David Morse)
3123 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
3124 new value now works. (reported by Bruno Haible)
3125 * bug fix: correct canonicalization of multiple non-standard slot
3126 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
3127 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
3128 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
3129 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
3130 now exists, an signals an error.
3131 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
3132 during macroexpansion for non-iteration variables as well. (reported
3133 by Bruno Haible for CMUCL)
3134 * bug fix: Cyclic structures and unprintable objects in compiler
3135 messages no longer cause errors. (reported by Bruno Haible)
3136 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
3137 and READ-FROM-STRING all now return a primary value of NIL if
3138 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
3139 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
3140 (reported by Bruno Haible for CMUCL)
3141 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
3142 do not designate a condition. (reported by Bruno Haible for
3144 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
3145 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
3146 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
3147 permitting resolution in favour of any of the conflicting symbols.
3148 (reported by Bruno Haible for CMUCL)
3149 * bug fix: EQUAL compiler optimizations is less aggressive on
3150 strings which can potentially compare true despite having distinct
3151 specialized array element types.
3152 * bug fix: unit enumerations can be defined without dividing by
3153 zero. (thanks to Vincent Arkesteijn)
3154 * FORMAT compile-time argument count checking has been enhanced.
3155 (report from Bruno Haible for CMUCL)
3156 * a partial workaround for the bug 262: the compiler does not try to
3157 inline-expand a local function doing RETURN-FROM from a deleted
3158 BLOCK. (thanks to Peter Denno for the bug report and to David
3159 Wragg for the simple test case)
3160 * fixed some bugs revealed by Paul Dietz' test suite:
3161 ** RENAME-PACKAGE allows all package designators as new package
3163 ** constraint propagation and lambda variable substitution are
3164 more cautious in dealing with partially deleted code.
3165 ** compiler sometimes lost reoptimization passes.
3166 ** CERROR, given a condition as condition designator, passes its
3167 remaining arguments to the continue format control without
3169 ** Case-altering FORMAT directives work correctly on non-ASCII
3171 ** The REAL type specifier handles bounds outside the floating
3172 point ranges without signalling FLOATING-POINT-OVERFLOW.
3173 ** Functions with IR1-transformations can create intercomponent
3174 references to global functions.
3175 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
3177 ** FORMAT ~R treats a nil value for its first parameter correctly.
3179 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
3180 * enhancement: saving cores with foreign code loaded is now
3181 supported on x86/NetBSD and sparc/Linux in addition to the previously
3182 supported platforms.
3183 * bug fix: on some platforms repeated installations caused multiple
3184 copies of HTML documentation to be installed -- should not happen
3185 any more. (reported by Stefan Scholl)
3186 * bug fix: parsing self-recursive alien record types multiple times
3187 no longer causes infinite recursion. (reported by Thomas F. Burdick,
3188 original patch by Helmut Eller for CMUCL)
3189 * bug fix: stack-exhaustion detection works now on NetBSD as well.
3190 (thanks to Richard Kreuter)
3191 * bug fix: defining classes whose accessors are methods on existing
3192 generic functions in other (locked) packages no longer signals
3193 bogus package lock violations. (reported by François-René Rideau)
3194 * bug fix: special variables as DEFMETHOD parameters no longer have
3195 associated bogus type declarations. (reported by David Wragg and
3197 * bug fix: read-write consistency on streams of element-type
3198 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
3199 * bug fix: redefiniton of the only method of a generic function with
3200 no DEFGENERIC no longer emits a full WARNING. In addition,
3201 redefinition of generic functions with no DEFGENERIC to an
3202 incompatible lambda list now signals an error. (thanks to Zach
3204 * bug fix: DEFGENERIC now works even when there's a function of the
3205 same name in an enclosing lexical environment. (thanks to Zach
3207 * fixed compiler failure, caused by instrumenting code during
3208 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
3209 * optimization: added loop analysis and improved register allocation
3210 * fixed some bugs revealed by Paul Dietz' test suite:
3211 ** POSITION on displaced vectors with non-zero displacement
3212 returns the right answer.
3213 ** (SIMPLE-STRING) is a valid type specifier for sequence
3215 ** *PRINT-LEVEL* handling for slotless structures is pedantically
3217 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
3218 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
3220 ** The justification version of the FORMAT ~< directive treats
3221 non-zero minpad parameter correctly.
3223 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
3224 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
3225 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
3226 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
3227 the supported interface.
3228 * new feature: Single-stepping of code compiled with DEBUG 2 or
3229 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
3230 * new feature: saving cores with foreign code loaded is now
3231 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
3232 Timothy Moore's work for CMUCL)
3233 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
3234 parameters to * instead of NIL if no initform is supplied.
3235 (reported by Johan Bockgaard)
3236 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
3237 correctly when FUNCALL appears as the car of the form. Note:
3238 despite this FUNCALL forms are not currently subject to
3239 compiler-macro expansion. (port of Raymond Toy's fix for the
3240 same from CMUCL, reported by Johan Bockgaard)
3241 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
3242 (thanks for Teemu Kalvas)
3243 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
3244 case instead of terminating the iteration (thanks for Julian
3245 Squires, Sean Champ and Raymond Toy)
3246 * bug fix: incorrect expansion of defgeneric that caused a style
3247 warning. (thanks for Zach Beane)
3248 * bug fix: slot accessor effective method computation works properly
3249 for classes with multiple non-standard applicable methods on
3250 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
3251 * on x86 compiler supports stack allocation of results of LIST and
3252 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
3253 CMUCL implementation by Gerd Moellmann)
3254 * fixed some bugs revealed by Paul Dietz' test suite:
3255 ** FORMAT strings with both the ~<~:;~> form of the justification
3256 directive and pretty-printing directives cause an error.
3258 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
3259 * incompatible change: the internal functions
3260 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
3261 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
3262 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
3263 instead of the old functions.
3264 * new feature: on platforms where "dladdr" is available foreign
3265 function names now appear in backtraces. (based on Helmut Eller's
3267 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
3268 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
3270 * documentation: ASDF manual is now distributed with SBCL.
3271 * bug fix: SBCL can now load its contributed modules with REQUIRE
3272 even if the system-provided entries have been removed from
3273 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
3275 * bug fix: backtraces involving undefined functions or assembly
3276 routines are more informative. (thanks to Brian Downing)
3277 * bug fix: mutually referent alien structures now work correctly.
3278 (reported by Rick Taube)
3279 * bug fix: structures defined by WITH-ALIEN can be referred to
3280 within other definitions in the same WITH-ALIEN.
3281 * bug fix: division operators (MOD, TRUNCATE and the like) with
3282 constant zero divisors and integer dividends no longer generate
3284 * bug fix: provide default methods for INPUT-STREAM-P and
3285 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
3286 * bug fix: improve the detection and resolution of MOP metacycles
3287 (where effective method computation depends on the generic function
3288 having its effective method computed). (reported by Bruno Haible)
3289 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
3290 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
3291 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
3292 * bug fix: pathnames with relative directory components can be
3293 represented relative to default pathnames.
3294 * optimization: in taking the GCD of bignums, reduce the two bignums
3295 to approximately the same size (using Euclid's algorithm) before
3296 applying the more sophisticated binary GCD. (thanks to Juho
3298 * optimization: COUNT on bitvectors now operates word-at-a-time.
3299 * optimization: ASH with a positive, but not necessarily constant,
3300 (leftwards) shift, when in a modular context, is compiled to a
3302 * fixed some bugs revealed by Paul Dietz' test suite:
3303 ** FORMAT variable parameters ("~V<char>") are defaulted properly
3304 if the corresponding argument is NIL.
3305 ** FORMAT directives accepting a minpad parameter treat negative
3307 ** literal commas as character parameters to FORMAT directives are
3309 ** literal spaces directly after ~<Newline> directives within a
3310 format-logical-block (~:< ~@:>) do not induce :FILL-style
3311 conditional newlines.
3312 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
3313 as specified: it no longer includes conditional newlines).
3314 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
3316 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
3317 :PER-LINE-PREFIX argument does not evaluate to a string.
3319 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
3320 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3321 the manual for details; for now, package locks can be disabled by
3322 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
3323 if no major problems are found then it is likely that they will be
3324 compiled in unconditionally.
3325 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
3326 unsupported operators on all platforms. To load a shared library
3327 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
3328 object file, link it into a shared library outside of SBCL and
3329 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
3331 * bug workaround: ROOM T and the GC were not working together
3332 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
3333 reporting facility aren't preserved. That reporting has been
3334 disabled, so now until and unless someone figures out how to make it
3335 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
3336 * minor incompatible change: as threatened around sbcl-0.8.0, the
3337 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
3338 The symbols are also exported from SB-PCL for backwards
3339 compatibility, but more so than before SB-PCL should be treated as
3340 an implementation-internal package.
3341 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
3343 * the SB-SPROF contrib now works on (most) non-x86 architectures.
3344 It is known as of this release not to work on the Alpha, however.
3345 * fixed bug #167: errors signalled due to illegal syntax in method
3346 bodies are now more legible.
3347 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
3348 designators and can hence be used with TYPEP.
3349 * fixed bug #333: CHECK-TYPE now ensures that the type error
3350 signalled, if any, has the right object to be accessed by
3351 TYPE-ERROR-DATUM. (reported by Tony Martinez)
3352 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
3353 5.1.2.3 for multiple-value place subforms. (reported by Kalle
3355 * fixed bug #334: programmatic addition of slots using specialized
3356 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
3357 and :ALLOCATION :CLASS slots.
3358 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
3359 float exponents. (rereported by Peter Seibel)
3360 * fixed a bug: #\Space (and other whitespace characters) are no
3361 longer considered to be macro characters in standard syntax by
3362 GET-MACRO-CHARACTER.
3363 * fixed bug: initialization of condition class metaobjects no longer
3364 causes an instance of the condition to be created. (reported by
3366 * fixed bug: it is now possible to have slots such that
3367 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
3368 description is neither :INSTANCE nor :CLASS.
3369 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
3370 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
3371 SLOT-VALUE-USING-CLASS), and no longer causes errors with
3372 non-standard SLOT-DEFINITION-ALLOCATION values.
3373 * fixed bugs: various race conditions handling exiting threads.
3374 CL-PPCRE's thread torture test now passes.
3375 * fixed arguably-a-bug: GC time proportional to number of threads in
3376 system even when most of them are idle
3377 * optimization: improved performance of BIT and SBIT on bit-vectors.
3378 * .fasl file incompatibility: The fasl file version number has
3379 been incremented because of changes associated with package locks.
3381 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
3382 * minor incompatible change: the system no longer provides
3383 optimization hints (conditions which are TYPEP
3384 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
3385 mode; these hints are still emitted when higher SPEED optimization
3387 * new contrib module: a sampling profiler (profiling by statistical
3388 sampling, rather than by instrumenting functions) is available as
3389 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
3390 * the behaviour of the standard function ED is now customizeable by
3391 third parties through a hook variable: see ED's documentation
3392 string for information on the protocol.
3393 * the compiler no longer emits efficiency notes for (FUNCALL X)
3394 when the type of X is uncertain under default optimization
3396 * fixed bug 276: mutating a binding of a specialized parameter to a
3397 method to something that is not TYPEP the specializer is now
3399 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
3400 exhibit the required behaviour on the x86 platform. (thanks to
3401 Peter van Eynde, Eric Marsden and Bruno Haible)
3402 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
3403 even for difficult arguments. (reported by Peter Graves)
3404 * fixed bug 141a: the backquote printer now descends quoted
3406 * fixed another bug in backquote printing: no more destructive
3407 modification of the form's list structure. (reported by Brian
3409 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
3410 move between its address being taken and the call to
3411 interrupt_thread, fixing a crashing race condition.
3412 * the SB-POSIX contrib implementation has been adjusted so that it
3413 no longer exhibits ridiculously poor performance when constructing
3414 instances corresponding to C structs.
3416 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
3417 * minor incompatible change: the sb-grovel contrib now treats C
3418 structures as alien (in the sense of SB-ALIEN) objects rather than
3419 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
3420 has implications for memory management of client code
3421 (sb-grovel-returned objects must now be manually managed) and for
3422 type safety (alien objects now have full types).
3423 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
3424 used to control emission of compiler diagnostics, rather than the
3425 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
3426 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
3427 quality should be considered deprecated.
3428 * install.sh now installs the user manual as well
3429 * (not quite a new documentable feature, but worth considering in
3430 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3431 beginnings of a semantically meaningful condition hierarchy is
3432 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3434 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3435 designator as the defaults argument.
3436 * fixed bug: Displaced arrays whose displaced-to array has become
3437 too small now cause ARRAY-DIMENSION to signal an error, providing
3438 for safer bounds-checking. (reported by Bruno Haible)
3439 * fixed bug: DEFCLASS slot definitions with identical :READER and
3440 :WRITER names now signal a reasonable error. (reported by Thomas
3442 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3444 * fixed bug: Invalid slot specification errors now print correctly.
3445 (thanks to Zach Beane)
3446 * fixed bug 320: Shared to local slot value transfers in class
3447 redefinitions now happen corrently from superclasses as
3448 well. (reported by Bruno Haible)
3449 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3451 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3452 improper lists correctly. (reported by Bruno Haible)
3453 * fixed bug 313: source-transform for <fun-name> was erroneously
3454 applied to a call of a value of a variable with name <fun-name>.
3455 (reported by Antonio Menezes Leitao)
3456 * fixed bug 307: The obsolete instance protocol ensures that
3457 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3458 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3459 remove dead unknown-values globs from the middle of the stack.
3460 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3461 easier to resume long computations after using *BREAK-ON-SIGNALS*
3462 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3463 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3464 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3465 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3466 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3467 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3469 * fixed a (fairly theoretical) bug in string printing: if
3470 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3471 does not have array-element-type equal to the most general string
3473 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3474 function name. (thanks to Nikodemus Siivola)
3475 * fixed bug: compile-time format string checker failed on
3476 non-closed ~{. (reported by Thomas F Burdick)
3477 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3478 don't warn on legal loop code involving more than one aggregate
3479 boolean. (thanks to Nikodemus Siivola)
3480 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3481 in keyword position to MAKE-PACKAGE was not being reported as
3483 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3484 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3486 * fixed bugs 280 and 312: the checking for multiple definitions in a
3487 file is less likely to become confused by uses of inline
3489 * fixed bug: the #S reader macro performs the keyword coercion
3490 specified for slot names. (reported by Kalle Niemitalo)
3491 * fixed bug: lambda lists may contain symbols whose names start with
3492 & but are not lambda-list-keywords; their occurrence triggers a
3494 * fixed bug 321: define-method-combination argument lambda lists do
3495 not cause invalid code to be generated when &optional or &aux
3496 variables have default values. (reported by Bruno Haible)
3497 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3498 have CLOS classes; this appears also to have fixed reported
3499 instability in stack exhaustion detection.
3500 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3501 MUFFLE-WARNING when no associated restart is present is now
3503 * optimization: rearranged the expansion of various defining macros
3504 so that each expands into only one top-level form in a
3505 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3507 * optimization: used a previously little-used slot in symbols to
3508 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3510 * fixed some bugs revealed by Paul Dietz' test suite:
3511 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3513 ** Fixed an optimization bug related to inheritance of initforms
3514 from local to shared slots.
3515 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3516 ** CAST optimizer forgot to flush argument derived type.
3517 ** print/read consistency on floats is now orders of magnitude
3518 more likely. (thanks also to Bruno Haible for a similar report
3520 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3521 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3522 for objects of type REAL. Make it so.
3523 ** SXHASH is specified (infelicitously) to respect similarity,
3524 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3525 it so. (thanks to Markus Ziegler)
3526 ** on the Alpha, the compiler succeeds in compiling functions
3527 returning a known number of arguments greater than 63.
3528 ** fixed handling of invalid NIL arguments in keyword position
3530 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3532 ** provided a readably-printable representation for RANDOM-STATE
3534 ** ensured that pathnames, where they have a namestring, always
3535 print using #P"..." syntax.
3537 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3538 * Support for the forthcoming 2.0 version of the NetBSD kernel
3539 running on x86 hardware has been added. (thanks to Perry
3540 E. Metzger most immediately, and others for their past work)
3541 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3542 current release version; SBCL's OpenBSD support had been broken
3543 since about the time of OpenBSD's switch to ELF binary format.)
3544 * [placeholder for DX summary]
3545 ** user code with &REST lists declared dynamic-extent, under high
3546 speed or space and low safety and debug optimization policy.
3547 * The manual has been converted to Texinfo format and the debugger
3548 chapter from the cmucl manual has been added.
3549 * A facility has been added to extract documentation strings from
3550 sbcl and store them as Texinfo-formatted snippets for inclusion in
3551 the manual (via Texinfo's @include directive)
3552 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3553 the test case to Dave Roberts)
3554 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3555 pointers. (reported by Sean Ross)
3556 * bug fix: PROFILE output is printed nicely even for large numerical
3557 values. (thanks to Zach Beane)
3558 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3559 greater than 32 handle EOF correctly.
3560 * bug fix: on X86 an immediate argument of the IMUL instruction is
3561 correctly printed in disassembly. (thanks to Lutz Euler)
3562 * bug fix: class slots in redefined classes preserve their old
3563 values. (thanks to Bruno Haible and Nikodemus Siivola)
3564 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3565 succeeds. (reported by Marco Baringer)
3566 * fixed some bugs revealed by Paul Dietz' test suite:
3567 ** READ-SEQUENCE now works on ECHO-STREAMs.
3568 ** RATIONALIZE works more according to its specification. (thanks
3571 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3572 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3573 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3574 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3575 should matter to you only if you rebind the printer control
3576 variables and then find you want different bindings in the
3577 debugger than in the ordinary execution of your program.)
3578 * The runtime build system has been tweaked to support building
3579 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3580 assembler and linker. (thanks to Nikodemus Siivola)
3581 * Unbound, undefined, undeclared variables now trigger full
3582 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3583 more often programmer error than deliberate exploitation of undefined
3585 * optimization: the hash algorithm for strings has changed to one
3586 that is less vulnerable to spurious collisions. (thanks to Juho
3588 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3589 needless bounds checking. (thanks to Juho Snellman)
3590 * optimization: implemented multiplication as a modular
3591 (UNSIGNED-BYTE 32) operation on the PPC backend.
3592 * fixed some bugs revealed by Paul Dietz' test suite:
3593 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3595 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3596 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3597 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3599 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3601 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3603 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3604 their output stream on EOF from read.
3605 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3606 have been read to end-of-file.
3607 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3609 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3610 description of determination of which consecutive characters
3612 ** Printing the "Space" character with escaping on now yields "#\\ ",
3613 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3614 ** Reading floating-point numbers with *READ-BASE* set to a number
3615 less than 10 works correctly.
3616 ** Reading floating-point numbers with *READ-BASE* set to a number
3617 more than 10 works correctly.
3618 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3619 the readtable currently in effect.
3621 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3622 * minor incompatible change: parsing of namestrings on a physical
3623 (Unix) host has changed; numbers after the final #\. in a
3624 namestring are no longer interpreted as a version field. This is
3625 intented to be largely invisible to the user, except that the
3626 meaning of the namestring "*.*.*" has changed: it now refers to a
3627 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3628 should usually be replaced by
3629 (make-pathname :name :wild :type :wild :version :wild)
3630 with the added benefit that this is more likely to be portable.
3631 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3632 option to OPEN now signals an error if the file being opened
3633 exists; this may have an impact on existing code.
3634 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3635 signals received, on the PowerPC platforms (both Linux and
3636 Darwin). (thanks to Pierre Mai for pointing out the location of
3638 * several fixes on OS X: The system now builds and runs cleanly on
3639 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3640 (thanks to Brian Mastenbrook)
3641 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3642 second-resolution: integer multiples of 1/3600 between -24 and 24.
3643 (thanks to Vincent Arkesteijn)
3644 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3645 type when called with 1 argument; PEEK-CHAR checked type of
3646 PEEK-TYPE only after having read first character from a
3647 stream. (reported by Peter Graves)
3648 * bug fix: the garbage collector now has much better locality
3649 behaviour, and in particular no longer treats all memory as being
3650 exhausted when many small objects point to each other in a deeply
3652 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3653 recognized as being TYPEP their class.
3654 * bug fix: the PUSHNEW documentation string has been corrected.
3655 (thanks to Vincent Arkesteijn)
3656 * bug fix: defaulting of the value for the last of an atypically
3657 large number of multiple values being bound was not being
3658 performed correctly on the Alpha or PPC platforms
3659 * optimization: implemented multiplication as a modular
3660 (UNSIGNED-BYTE 32) operation on the x86 backend.
3661 * optimization: SEARCH on simple-base-strings can now be open-coded.
3662 (see also contrib/compiler-extras.lisp for inspiration for
3663 teaching the compiler about the Boyer-Moore algorithm).
3664 * value, returned by MAX (and MIN) called with several EQUALP, but
3665 not EQL, arguments now does not depend on compiler settings.
3666 * fixed some bugs revealed by Paul Dietz' test suite:
3667 ** in stack analysis liveness information is propagated from
3668 non-local entry points.
3669 ** pathwise CAST removing failed when the CAST node did not start
3671 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3672 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3674 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3675 host is already defined.
3676 ** RENAME-FILE works on streams instead of signalling an internal
3678 ** PEEK-CHAR uses the current readtable when determining whether
3679 or not a character is whitespace.
3680 ** MERGE-PATHNAMES handles the case when the pathname does not
3681 specify a name while the default-pathname specifies a version.
3682 ** Pathnames now stand a better chance of respecting print/read
3684 ** Attempting to use standardized file system operators with a
3685 pathname with invalid :DIRECTORY components signals a
3687 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3689 ** DIRECTORY on logical pathnames is more correct.
3690 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3691 signal a TYPE-ERROR if their argument is not a stream.
3692 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3693 designator argument does not designate a stream.
3694 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3695 examining the synonym.
3696 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3698 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3699 element-type larger than ([UN]SIGNED-BYTE 32).
3701 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3702 * When built with the :SB-FUTEX feature, threaded builds now take
3703 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3704 for faster/more reliable mutex and condition variable support.
3705 * Incompatible change (but one you probably shouldn't have been using
3706 anyway): the interface and code for arbitrating between multiple
3707 threads in the same user session has been redesigned.
3708 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3709 error when called without an explicit environment argument.
3710 (thanks to Nikodemus Siivola)
3711 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3712 confused about their position. (thanks to Adam Warner and Gerd
3714 * bug fix: Pretty printing backquoted forms with unquotations in the
3715 argument list position of various code constructs such as LAMBDA
3716 now works correctly. (reported by Paul Dietz)
3717 * bug fix: Pretty printing unquotations no longer loses all
3718 stream position information.
3719 * optimization: performance of string output streams is now less
3720 poor for multiple small sequence writes.
3721 * optimization: performance of CSUBTYPEP in the presence of complex
3722 expressions involving CONS and NOT many times has been improved.
3723 (reported by Paul Dietz)
3724 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3726 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3727 streams for the REPL, for the debugger, and so on. (thanks to
3729 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3730 * fixed some bugs revealed by Paul Dietz' test suite:
3731 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3732 is ignored if it is too small, rather than propagating through
3733 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3734 ** extremely complex negations of CONS types were not being
3735 sufficiently canonicalized, leading to inconsistencies in
3737 ** VALUES tranformer lost derived type.
3739 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3740 * fixed a bootstrapping bug: the build process no longer assumes
3741 that the various BOOLE-related constants have the same value in
3742 host and target lisps. (noted by Paul Dietz' test suite on an
3743 SBCL binary built from CLISP)
3744 * The system can now be dynamically linked on the MIPS platform,
3745 which enables dynamic loading of foreign code from Lisp. (thanks
3746 to Ralf Baechle for discussions on the MIPS ABI)
3747 * The system now records debugging information for its own source
3748 files in a filesystem-position-independent manner, relative to
3749 the "SYS" logical host.
3750 * fixed a compiler bug: MV-LET convertion did not check references
3751 to the "max args" entry point. (reported by Brian Downing)
3752 * tweaked disassembly notes to be less confident about proclaiming
3753 some instruction as an LRA. (thanks to Brian Downing)
3754 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3755 now each have their own history, command character, and other
3756 characteristics. (thanks to David Lichteblau)
3757 * fixed some bugs revealed by Paul Dietz' test suite:
3758 ** compiler failure in compiling LOGAND expressions including a
3760 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3761 shift greater than 32.
3762 ** FLUSH-DEST did not mark blocks for type check regeneration.
3763 ** HANDLER-CASE failed to accept declarations in handler clauses
3764 in some circumstances.
3766 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3767 * New code in contrib/sb-introspect (still probably not entirely
3768 stable yet) provides some support for smart Lisp development
3769 environments like SLIME.
3770 * The conditions signalled for errors occurring when loading .fasl
3771 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3772 in a way which should help ASDF recover gracefully.
3773 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3774 stuff has been cleaned up. If you code contrib/ stuff, this might
3775 affect you, and you can look at contrib/README, contrib/STANDARDS,
3776 and/or the 0.8.4.27 diff to check.
3777 * In full calls the compiler now does not generate checks for declared
3778 argument types for all arguments.
3779 * various threading fixes
3780 ** and some experimental patches which didn't make it into
3781 the main tree for this release, but which are shipped in
3782 contrib/experimental-thread.patch as a possible fix for some
3783 failures (deadlock, spinning...) in GC-intensive multithreaded
3785 * fixed PPC build problem (source code incompatibility of different
3786 library versions): added offsetof() hackery which attempts to divine
3787 where glibc maintainers put uc_mcontext today
3788 * fixed bug 282: compiler does not trust type assertions while passing
3789 arguments to a full call.
3790 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3791 &OPTIONAL VALUES type parameter.
3792 * fix bug 214: algorithm for noting rejected templates is now more
3793 similar to that of template seletion. (also reported by rydis on
3795 * fixed bug 141b: printing backquoted information readably and prettily
3796 inserts a space where necessary.
3797 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3798 (CHARACTER 10) are now reported as errors, rather than propagated
3799 as unknown types. (reported by piso on #lisp)
3800 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3801 with values NIL and :ERROR. (thanks to Milan Zamazal)
3802 * fixed bug 191c: CLOS now does proper keyword argument checking as
3803 described in CLHS 7.6.5 and 7.6.5.1.
3804 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3805 longer raise an error; further, using a list as a for-as-arithmetic
3806 counter now raises a meaningful error.
3807 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3808 now understood by sequence creation functions such as MAKE-SEQUENCE
3810 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3811 a STRING-INPUT-STREAM.
3812 * compiler enhancement: SIGNUM is now better able to derive the type
3814 * type declarations inside WITH-SLOTS are checked. (reported by
3816 * fixed some bugs revealed by Paul Dietz' test suite:
3817 ** incorrect optimization of TRUNCATE for a positive first
3818 argument and negative second.
3819 ** compiler failure in let-convertion during flushing dead code.
3820 ** compiler failure while deriving type of TRUNCATE on an
3821 interval, containing 0.
3822 ** ASH of a negative bignum by a negative bignum count now returns
3824 ** intersection of CONS types now canonicalizes properly, fixing
3825 inconsistencies in SUBTYPEP.
3827 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3828 * incompatible change: The --disable-debugger command line
3829 option now clobbers the debugger at a more fundamental
3830 level, by redefining #'INVOKE-DEBUGGER instead of by
3831 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3832 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3833 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3834 enter the debugger REPL and then suffer recursive errors
3835 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3836 while under the new system, BREAK in --disable-debugger mode
3837 terminates the system just as an unhandled error would.
3838 * fixed compiler performance when processing loops with a step >1;
3839 * bug fix: DOCUMENTATION now retrieves generic function
3840 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3841 support has been systematized, and now supports the methods
3842 specified by ANSI, along with a default method and a method for
3843 slot documentation. (reported by Nathan Froyd)
3844 * bug fix: effective methods associated with a generic function are
3845 no longer cached over a change of that generic function's method
3846 combination. (reported by Andreas Fuchs)
3847 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3848 names a non-existent directory. (thanks to Andreas Fuchs)
3849 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3850 the PPC platform, lead to stack corruption; this has been fixed.
3851 (reported by Rainer Joswig)
3852 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3853 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3854 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3855 arithmetic in general, is now much more reliable on the PPC
3857 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3858 platform now returns the right answer.
3859 * optimization: restored some effective method precomputation in
3860 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3861 precomputation is now tunable.
3862 * optimization: compiler-internal data structure use has been
3863 reviewed, and changes have been made that should improve the
3864 performance of the compiler by about 20%.
3865 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3866 simple-base-strings and simple-bit-vectors is improved.
3867 * optimization: the optimization of 32-bit logical and arithmetic
3868 functions introduced in version 0.8.3 on the x86 has been
3869 implemented on the mips, ppc and sparc platforms; an
3870 implementation of the same facility, but for 64-bit arithmetic,
3871 has been added for the alpha.
3872 * microoptimization: the compiler is better able to make use of the
3873 x86 LEA instruction for multiplication by constants.
3874 * bug fix: in some situations compiler did not report usage of
3875 generic arithmetic in (SPEED 3) policy.
3876 * bug 145b fix: compiler used wrong type specifier while converting
3877 MEMBER-types to numeric.
3878 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3880 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3881 index argument now works properly on the Alpha platform.
3882 * bug fix: floating point exception treatment on the Alpha platform
3884 * bug fix: FILE-POSITION works much better on string input and
3885 output streams. (thanks to Nikodemus Siivola)
3886 * bug fix: many threading/garbage collection symptoms sorted.
3887 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3888 might be pseudo-atomic.
3889 * internal change: Stopping for GC is now done with signals not
3890 ptrace. GC is now done in whichever thread wanted it, instead of
3892 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3894 * bug fix: result form in DO is not contained in the implicit
3896 * incompatible change: ICR structure is changed; the value part of
3897 CONTINUATION is now called LVAR; corresponding functions are
3898 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3899 * added type deriver for ISQRT (thanks to Robert E. Brown).
3900 * bug fix: better support for loading from the command line when an
3901 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3903 * fixed some bugs revealed by Paul Dietz' test suite:
3904 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3905 ** ROUND and FROUND now give the right answer when given very
3906 small float arguments.
3907 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3909 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3910 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3911 ** LCM should return a non-negative integer.
3912 ** PARSE-INTEGER returned the index of a terminator instead of the
3913 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3914 ** PARSE-INTEGER returned an incorrect index being applied to a
3916 ** LCM with two arguments of 0 returns 0 rather than signalling
3918 ** unsigned addition of a 32-bit constant with the high bit set no
3919 longer causes an internal compiler error on the x86.
3920 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3921 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3922 with negative last argument.
3923 ** byte specifiers with zero size and position no longer cause
3924 an error during type derivation.
3925 ** bignum multiplication on the Alpha platform now returns the
3927 * porting: The system now builds on SuSE AMD64, although it still
3928 generates a 32-bit binary.
3929 * .fasl file incompatibility: The fasl file version number has
3930 been incremented (because of the changes to internal compiler
3931 data structures referred to above).
3933 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3934 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3935 more accurately, on the Darwin kernel running on PowerPC hardware.
3936 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3937 * Compiler code deletion notes now signal a condition of type
3938 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3939 an associated MUFFLE-WARNING restart.
3940 * The compiler now performs limited argument count validation of
3941 constant format strings in FORMAT, and where appropriate in ERROR,
3942 CERROR and WARN. (thanks to Gerd Moellmann)
3943 * New ASDF-INSTALL contrib can be used for automatic download and
3944 installation of third-party Lisp code from CCLAN or other sites
3946 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3947 which forces another thread to execute a function supplied by the
3949 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3950 now accept and act upon their :ELEMENT-TYPE keyword argument.
3951 (reported by Martin Atzmueller, Edi Weitz)
3952 * bug fix: FILE-POSITION now accepts position designators up to
3953 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3954 is the greater. (thanks to Patrik Nordebo)
3955 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3956 Kalle Olavi Niemitalo)
3957 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3958 printable. (reported by Eric Marsden)
3959 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3960 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3961 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3962 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3963 (thanks to Stig Sandoe)
3964 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3965 UNREAD-CHAR now correctly includes the unread character in the
3966 target sequence. (thanks to Gerd Moellmann)
3967 * bug fix in threaded builds: the system can now be suspended and
3968 resumed by shell job control with minimal disruption.
3969 * bug fixes in times and timezones >2038AD
3970 * better handling of "where is GNU make?" problem in build scripts
3971 (thanks to Nikodemus Siivola)
3972 * new optimization: inside a named function any reference to a
3973 function with the same name is considered to be a self-reference;
3974 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3975 optimization quality.
3976 * new optimization on x86: logical functions and + now have
3977 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3978 used when the result is truncated to 32 bits.
3979 * VALUES declaration is partially enabled.
3980 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3981 * fixed some bugs revealed by Paul Dietz' test suite:
3982 ** The system now obeys the constraint imposed by
3983 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3984 types form a lattice under type intersection.
3985 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3986 ** ASSOC now ignores NIL elements in an alist.
3987 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3988 and (1+ MOST-POSITIVE-FIXNUM) answers.
3989 ** The addition of a method with invalid qualifiers to a generic
3990 function does not cause an error to be signalled immediately;
3991 a warning is signalled, and the error is generated only on
3992 calling the generic function.
3993 * changed .fasl file version number, in part to add type codes for
3994 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3995 obscure ANSI requirements
3997 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3998 * fixed bug 148: failure to inline-expand a local function left
3999 garbage, confusing the compiler.
4000 * fixed bugs 3cd: structure slot readers perform type check if the
4001 slot can have an invalid value (i.e. it is either not initialized
4002 or can be written with a less specific slot writer).
4003 * bug fix: the compiler now traps array references to elements off
4004 the end of an array; previously, the bounds checking in some
4005 circumstances could go off-by-one.
4006 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
4008 * type declarations for array element types now obey the description
4009 on the CLHS page "Declaration TYPE", as per discussions on
4010 sbcl-help around 2003-05-08. This means that a declaration
4011 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
4012 declaration, all references to BAR will be asserted or assumed
4013 (with THE, so dependent on compiler policy) to involve objects of
4014 type FOO. Note that no such declaration is implied in
4015 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
4016 * declared types of functions from the "Conditions"
4017 chapter. (reported by Paul Dietz)
4018 * bug fix: CERROR accepts a function as its first argument.
4019 * bug fix: NTH an NTHCDR accept a bignum as index
4020 arguments. (reported by Adam Warner)
4021 * optimization: character compare routines now optimize comparing
4022 against a constant character. (reported by Gilbert Baumann)
4023 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
4024 argument now works properly on the MIPS platform.
4025 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
4026 * repeated evaluation of the same DEFSTRUCT, a slot of which is
4027 declared to have a functional type, does not cause an error
4029 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
4030 many :DEFAULT-INITARGS. (reported by Istvan Marko)
4031 * fixed bug: if last continuation of a deleted block has a
4032 destination, this destination should be deleted too. (reported by
4034 * fixed a bug in the bootstrap process: the host compiler's values
4035 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
4036 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
4037 test case from Patrik Nordebo)
4038 * improved the ability of the disassembler on the PPC platform to
4039 provide helpful disassembly notes.
4040 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
4041 the class in more cases than previously.
4042 * bug fix: FILE-POSITION now understands :START and :END for
4043 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
4044 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
4045 8) data. (thanks to David Lichteblau)
4046 * bug fix: it is possible to add a method to a generic function
4047 without lambda list.
4048 * bug fix: reader failed to signal END-OF-FILE inside an
4049 object representation. (reported by Nikodemus Siivola)
4050 * fixed some bugs revealed by Paul Dietz' test suite:
4051 ** LAST and [N]BUTLAST should accept a bignum.
4052 ** condition slot accessors are methods.
4053 ** (VECTOR NIL) is a subtype of STRING.
4055 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
4056 * minor incompatible change: some nonsensical specialized lambda
4057 lists (used in DEFMETHOD) which were previously ignored now signal
4059 * minor incompatible change: the system is now aware of the types of
4060 variables in the COMMON-LISP package, and will signal errors for
4061 most violations of these type constraints (where previously they
4062 were silently accepted).
4063 * minor incompatible change: COMPILE-FILE now uses the freedom
4064 afforded (ANSI 3.2.2.3) to use derived function types for
4065 functions defined in the same file. This also permits the system
4066 to warn on static type mismatches and function
4067 redefinition. (Currently it does not work with high DEBUG level.)
4068 * minor incompatible change: VALUES declaration is disabled.
4069 * When issuing notes, the compiler now signals a condition of type
4070 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
4071 restart for use in user handlers. It is expected that the
4072 COMPILER-NOTE condition will eventually become a condition
4073 supertype to a hierarchy of note types, which will then be
4074 handleable in a similar fashion. However, at the moment, no such
4075 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
4076 but it's an implementation detail, not a classification for the
4078 * Changes in type checking closed the following bugs:
4079 ** type checking of unused values (192b, 194d, 203);
4080 ** template selection based on unsafe type assertions (192c, 236);
4081 ** type checking in branches (194bc).
4082 * A short form of VALUES type specifier has ANSI meaning (it has
4083 increased the number of situations when SBCL cannot perform type
4085 * fixed bug in DEFSTRUCT: once again, naming structure slots with
4086 keywords or constants is permissible.
4087 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
4088 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
4089 classes. (thanks to Antonio Martinez)
4090 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
4091 outputs FRESH-LINE or TERPRI, and no longer converts its stream
4092 argument to a pretty-print stream. Instead, it leaves any such
4093 operations to DESCRIBE-OBJECT methods.
4094 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
4096 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
4097 lambda lists. (thanks to David Lichteblau)
4098 * bug fix: Defining a generic function with a :METHOD-CLASS being a
4099 subclass of STANDARD-METHOD no longer causes stack exhaustion.
4100 (thanks to Gerd Moellmann)
4101 * fixed bug 246: increased compilation speed of long
4102 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
4104 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
4106 * DEFCONSTANT now throws a condition of type
4107 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
4108 constant to a non-EQL value; CONTINUE and ABORT restarts
4109 respectively change and preserve the value.
4110 * fixed bug 63: The code walker, part of the implementation of CLOS,
4111 is now better at handling symbol macros.
4112 * bug fix: There is no longer an internal implementation type named
4113 CL:LENGTH. (reported by Raymond Toy)
4114 * bug fix: In macro-like defining macros/special operators the
4115 implicit block does not enclose the lambda list.
4116 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
4117 enclosed in parentheses) are not suitable as type specifiers, and
4118 their use properly signals an error now.
4119 * bug fix: An argument count mismatch for a type specifier in code
4120 being compiled no longer causes an unhandled error at compile
4121 time, but signals a compile-time warning.
4122 * fixed simple vector readable printing
4123 * bug fix: DESCRIBE takes more care over whether the class
4124 precedence list slot of a class is bound before accessing it.
4125 (reported by Markus Krummenacker)
4126 * bug fix: FORMATTER can successfully compile pretty-printer format
4127 strings which use variants of the ~* directive inside.
4128 * bug fix: SEARCH now applies its TEST predicate to the elements of
4129 the arguments in the correct order. (thanks to Wolfhard Buss)
4130 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
4131 in type inference. (thanks to Robert E. Brown)
4132 * bug fix: Reading in symbols with an explicit package name of ""
4133 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
4134 * fixed some bugs revealed by Paul Dietz' test suite:
4135 ** NIL is now allowed as a structure slot name.
4136 ** Arbitrary numbers, not just REALs, are allowed in certain
4137 circumstances in LOOP for-as-arithmetic clauses.
4138 ** Multiple class redefinitions before slot access no longer
4139 causes a type error.
4140 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
4141 association between the name and a class.
4142 ** Generic functions with non-standard method-combination and over
4143 five methods all of which return constants no longer return NIL
4144 after the first few invocations. (thanks to Gerd Moellmann)
4145 ** CALL-NEXT-METHOD with no arguments now passes the original
4146 values of the arguments, even in the presence of assignment.
4147 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
4149 ** DISASSEMBLE works with closures and funcallable instances.
4150 ** ADD-METHOD now returns the generic function, not the new method.
4151 ** FIND-METHOD signals an error if the lengths of the specializers
4152 is incompatible with the generic function, even if the ERRORP
4154 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
4155 which its argument is a member.
4156 ** DEFCLASS only redefines the class named by its class-name
4157 argument if that name is the proper name of the class;
4158 otherwise, it creates a new class.
4159 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
4160 of the UNBOUND-SLOT condition to the name of the slot.
4161 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
4162 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
4163 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
4164 SLOT-MAKUNBOUND in the specified fashion.
4166 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
4167 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
4168 cross-compilation host. As a consequence, we can now bootstrap our
4169 way up to SBCL starting with a bare gcc toolchain and human-readable
4170 source code (first the source to CLISP, then the source to SBCL).
4171 * A contributed module containing a partial implementation of the
4172 simple-streams interface has been included. (thanks to Rudi
4174 * A contributed module implementing the RFC1321 Message Digest
4175 Algorithm, known as MD5, has been included.
4176 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
4177 no longer has any effect, as the code controlled by this feature
4178 has been deleted. (As far as we know, no-one has ever built using
4179 this feature, and its semantics were confused in any case).
4180 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
4181 work on conditions (as required by the ANSI specification),
4182 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
4183 expected behaviour on conditions. Users should note, however,
4184 that such behaviour is not required by the ANSI specification,
4185 and so use of this behaviour may render their code unportable.
4186 * fixed some bugs revealed by Paul Dietz' test suite:
4187 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
4189 ** &ENVIRONMENT parameter in macro lambda list is bound first.
4190 ** SXHASH on condition objects no longer returns NIL.
4191 ** :ALLOCATION :CLASS slots are better treated; their values are
4192 updated on class redefinition, and initforms inherited from
4193 superclasses are applied.
4194 ** REMOVE-METHOD returns its generic function argument even when
4195 no method was removed.
4196 ** SHARED-INITIALIZE now initializes the values of the requested
4197 slots, including those with :ALLOCATION :CLASS.
4198 ** ALLOCATE-INSTANCE now works on structure classes defined via
4199 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
4201 ** SLOT-EXISTS-P now works on conditions, as well as structures
4203 ** MAKE-LOAD-FORM now has the required methods on
4204 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
4205 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
4206 keyword, and now implements the SLOT-NAMES argument.
4207 ** methods with &OPTIONAL arguments no longer allow too many
4208 arguments to be passed in the call without error.
4209 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
4210 option is consistent with the required arguments of the generic
4211 function lambda list.
4212 * bug fix: REQUIRE accepts a string designator. (Thanks to
4214 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
4215 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
4216 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
4217 * bug fix: The compiler now checks for duplicated variables in macro
4219 * bug fix: SETQ on globals returns the correct value.
4220 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
4221 not a COMPILER-ERROR (followed by some other strange error on
4222 choosing the CONTINUE restart).
4223 * bug fix: make.sh and friends are now more consistent in the way that
4224 they look for GNU "make".
4226 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
4227 * experimental native threads support (on x86 Linux >=2.4 only).
4228 This is not compiled in by default: you need to add :SB-THREAD to
4229 the target features. See the "Beyond ANSI" chapter of the manual
4231 * fix for longstanding nonANSIism: The old distinction between
4232 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
4233 The return value from CL:FIND-CLASS is now a CLOS class, and
4234 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
4235 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
4236 * An interface to the MetaObject Protocol, as described in Kiczales,
4237 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
4238 MIT Press, 1991, is available from the SB-MOP package.
4239 * incompatible change: the SB-PCL package should now be considered
4240 a private implementation detail, and no longer a semi-private MOP
4242 * minor incompatible change: due to rearrangement for threads, the
4243 control stack and binding stack are now allocated at arbitrary
4244 addresses instead of being hardcoded per-port. Users affected by
4245 this probably have to be doing advanced things with shared
4246 libraries, and will know who they are.
4247 * minor incompatible change: Previously, all --eval forms used were
4248 processed with READ before any of them were processed with EVAL.
4249 Now each --eval form is processed with both READ and EVAL before
4250 the next --eval form is processed. (Thus package operations like
4251 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
4252 work as the user might reasonably expect.)
4253 * minor incompatible change: *STANDARD-INPUT* is now only an
4254 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
4256 * minor incompatible change: Y-OR-N-P is now character-oriented,
4257 not line oriented. Also, YES-OR-NO-P now works without errors.
4258 (thanks to Antonio Martinez)
4259 * sb-aclrepl module improvements: an integrated inspector, added
4260 repl features, and a bug fix to :trace command.
4261 * Known functions, which cannot be open coded by the backend, are
4262 considered to be able to check types of their arguments. (fixing
4263 a bug report by Nathan J. Froyd)
4264 * fixed a bug in computing method discriminating functions: It is
4265 now possible to define methods specialized on classes which have
4266 forward-referenced superclasses. (thanks to Gerd Moellmann)
4267 * fixed evaluation order in optional entries (reported by Gilbert
4269 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
4270 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
4271 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
4272 required-by-AMOP NAME argument, as well as CLASS and
4273 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
4274 * fixed bug 20: DEFMETHOD can define methods using names that are
4275 not the proper names of classes to designate class specializers.
4276 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
4277 Unix file descriptors, instead of blowing up. (thanks to Antonio
4279 * Garbage collection refactoring: user-visible change is that a
4280 call to the GC function during WITHOUT-GCING will not do garbage
4281 collection until the end of the WITHOUT-GCING. If you were doing
4282 this you were probably losing anyway.
4283 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
4284 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
4285 TYPEP the latter but not the former.
4286 * The compiler issues a full WARNING for calls to undefined functions
4287 with names from the CL package.
4288 * MAP-INTO for a vector destination is open coded. (reported by
4289 Brian Downing on c.l.l)
4290 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
4291 documentation string.
4292 * fixed some bugs revealed by Paul Dietz' test suite:
4293 ** COPY-ALIST now signals an error if its argument is a dotted
4295 ** Condition slots are now accessed more correctly in the presence
4296 of multiple initargs for a given slot.
4297 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
4298 exclude restarts of the same name associated with a different
4300 ** DEFCLASS of forward-referenced classes with another
4301 forward-referenced class in the superclasses list no longer
4303 ** Condition slots are now initialized once each, not multiple
4304 times. (thanks to Gerd Moellmann)
4305 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
4307 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
4308 ** NSET-EXCLUSIVE-OR does not return extra elements when its
4309 arguments contain duplicated elements.
4310 ** RESTART-CASE understands local macros.
4311 ** RESTART-CASE associates exactly its own restarts with a condition.
4312 ** ENDP in safe mode checks its argument to be of type LIST.
4313 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
4314 in question is unbound.
4315 ** Optimized MAKE-INSTANCE functions no longer cause internal
4316 assertion failures in the presence of duplicate initargs.
4317 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
4318 ** Side-effectful :DEFAULT-INITARGS have their side-effects
4319 propagated even in the ctor optimized implementation of
4321 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
4323 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
4324 * a better implementation of SXHASH on (simple) bit vectors,
4325 measured both in execution speed and in distribution of results
4326 over the positive fixnums, has been installed. Likewise, a better
4327 implementation of EQUAL for simple bit vectors is now available.
4328 * fixed CEILING optimization for a divisor of form 2^k.
4329 * fixed bug 240 (emitting extra style warnings "using the lexical
4330 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
4331 by Antonio Martinez)
4332 * fixed SXHASH, giving different results for NIL depending on type
4333 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
4334 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
4335 variable. (found by Rolf Wester)
4336 * SBCL does not ignore type declarations for special
4337 variables. (reported by rif on c.l.l 2003-03-05)
4338 * some bug fixes in contrib/sb-aclrepl/
4339 * fixed some bugs revealed by Paul Dietz' test suite:
4340 ** a bug in the CONS type specifier, whereby the CAR and CDR
4341 types got intertwined, has been fixed;
4342 ** the type system is now able to reason about the interaction
4343 between INTEGER and RATIO types more completely;
4344 ** APPEND, [N]REVERSE and NRECONC check that those their
4345 arguments, which must be proper lists, are really so;
4346 ** An array specialized to be unable to hold elements has been
4347 implemented, as required -- yes, really -- by ANSI;
4348 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
4349 on malformed property lists;
4351 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
4352 * incompatible packaging change: in line with Unix convention,
4353 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
4354 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
4355 * REQUIRE and PROVIDE are now optionally capable of doing something
4356 useful. See the documentation string for REQUIRE.
4357 * infrastructure for a managed SBCL contrib system: contributed
4358 modules in this release include:
4359 ** the ASDF system definition facility;
4360 ** an interface to the BSD Sockets API;
4361 ** an ACL-like convenience interface to the repl;
4362 (thanks to Kevin Rosenberg)
4363 ** an implementation of ROTATE-BYTE, with efficient implementation
4365 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
4366 gives a closer approximation to the right answer than previously.
4367 (thanks to Raymond Toy)
4368 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
4369 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
4370 optional environment arguments, as required by ANSI.
4371 * fixed bugs in other functions taking environment objects, allowing
4372 calls with an explicit NIL environment argument to be compiled
4374 * fixed bug 228: primary return values from
4375 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
4376 COMPILE or FUNCTION.
4377 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
4378 :TYPE will no longer signal errors on innocuous objects.
4379 * fixed bug 231b: SETQ is better at respecting type declarations in
4380 the lexical environment.
4381 * fixed a bug in DEFCLASS: classes named by symbols with no or
4382 unprintable packages can now be defined.
4383 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
4384 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
4385 * fixed compiler failure related to checking types of functions.
4386 (reported by Robert E. Brown)
4387 * the compiler is now much more consistent in its error-checking
4388 treatment of bounding index arguments to sequence functions: in
4389 (SAFETY 3) code, errors will be signalled in almost all cases if
4390 invalid sequence bounding indices are passed to functions defined
4391 by ANSI to operate on sequences.
4392 * fixed a bug in the build procedure: documentation of SBCL-specific
4393 packages is now preserved and available in the final Lisp image.
4394 * lifted FDEFINITION lookup out of loops in the implementation of
4395 many list operations. (thanks to Robert E. Brown)
4396 * fixed a bug in the reader: the #n# reader macro now works for
4397 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
4398 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
4399 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
4400 E. Brown and Tony Martinez respectively)
4401 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
4402 SLOT-BOUNDP: in particular, it is now able to optimize them much
4403 better, and is now not vulnerable to having packages renamed.
4404 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
4405 always called when a slot is not present in an instance. (thanks
4407 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
4408 longer signal an error on unreading a character following EOF on
4409 the previous constituent stream. (thanks to Tony Martinez)
4410 * fixed some bugs revealed by Paul Dietz' test suite:
4411 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
4412 not just nonnegative fixnums;
4413 ** the logical bit-array operators such as BIT-AND now accept an
4414 explicit NIL for their "opt-arg" argument (to indicate a
4415 freshly-consed result bit-array);
4416 ** ELT now signals an error on an invalid sequence index in safe
4418 ** the type system is now cleverer about negations of numeric
4419 types, and consequently understands the BIGNUM and RATIO types
4421 ** the type system is now cleverer about the interaction between
4422 INTEGER and RATIO types: while bugs still remain, many more
4423 cases are accurately computed;
4424 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
4425 if it is in the last clause;
4426 ** CONSTANTLY now correctly returns a side-effect-free function in
4428 ** DECLARE is no longer treated as a special-operator; in
4429 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
4430 * incremented fasl file version number due to the change in the
4431 DEFSTRUCT-SLOT-DESCRIPTION structure.
4433 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4434 * minor incompatible change: code processed by the "interpreter" or
4435 EVAL now has a compilation optimization policy of (DEBUG 2)
4436 (changed from (DEBUG 1)) to improve debuggability of interactive
4437 development, and to allow the use of the debug RETURN command in
4439 * an experimental implementation of the RETURN command for the
4440 debugger has been included. (thanks to Frederik Kuivinen)
4441 * fixed bug 62: constraints were not propagated into a loop.
4442 * fixed bug in embedded calls of SORT (reported and investigated by
4444 * fixed some bugs revealed by Paul Dietz' test suite:
4445 ** printing and reading of arrays with some dimensions having
4446 length 0 (thanks to Gerd Moellmann);
4447 ** BOA constructor with &AUX argument without a default value does
4448 not cause a type error;
4449 ** CONSTANTP now returns true for all self-evaluating objects.
4451 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4452 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4453 accessors that are related by inheritance, as specified in the
4454 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4455 Valtteri Vuorikoski)
4456 * The compiler is now able to inline functions that were defined in
4457 a complex lexical environment (e.g. inside a MACROLET).
4458 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4460 * Support for the upcoming FreeBSD-5.0 release has been included.
4461 (thanks to Dag-Erling Smorgrav)
4462 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4463 effect when it is not in a toplevel context.
4464 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4465 stand a better chance of being correct. (thanks to Gerd
4467 * fixed bug in COERCE, which now signals an error on coercing a
4468 rational to a bounded real type which excludes the expected
4470 * The compiler is now able to derive types more accurately from the
4471 COERCE and COMPILE functions.
4472 * fixed bug 223: functional binding is considered to be constant
4473 only for symbols in the CL package.
4474 * fixed bug 231: SETQ did not check the type of a variable being set
4475 (reported by Robert E. Brown)
4476 * A new optimization for MAKE-INSTANCE has been included, fixing
4477 various bugs (including relating to :ALLOCATION :CLASS slots and
4478 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4480 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4481 ** As required by ANSI, LOOP now disallows anonymous collection
4482 clauses such as COLLECT I in conjunction with aggregate boolean
4483 clauses such as THEREIS (= I 1);
4484 ** LOOP now signals an error when any variable is reused in the
4485 same loop (including the potentially useful construct analogous
4486 to WITH A = 1 WITH A = (1+ A);
4487 ** IT is only a special loop symbol within the first clause of a
4488 conditional loop clause;
4489 ** LOOP with a typed iteration variable over a hashtable now
4490 signals a type error iff it should.
4491 * fixed some other bugs revealed by Paul Dietz' test suite:
4492 ** FILE-STREAM now names the class previously known as FD-STREAM;
4493 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4494 argument) no longer signals an error;
4495 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4496 of the slot symbol, rather than using the current package
4497 ((:CONC-NAME "") continues to intern the slot's name in the
4499 * incremented fasl file version number, because of the incompatible
4500 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4501 of the new implementation of DEFINE-COMPILER-MACRO.
4503 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4504 * Support for building SBCL for MIPS platforms running in
4505 little-endian mode has now been checked in, and basic
4506 functionality on said platforms verified.
4507 * minor incompatible change: PCL now records the pathname of a file
4508 in which methods and the like are defined, rather than its
4510 * minor incompatible change: TRUENAME now considers the truename of
4511 a file naming a directory to be the pathname with :DIRECTORY
4512 component indicating that directory.
4513 * minor incompatible change: a NAMED clause in the extended form of
4514 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4515 reason for the previous behaviour is unclear.
4516 * more systematization and improvement of CLOS and MOP conformance
4517 in PCL (thanks to Gerd Moellman and Pierre Mai):
4518 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4520 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4521 primary methods with no specializers;
4522 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4524 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4525 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4526 FORWARD-REFERENCED-CLASSes; error reporting on
4527 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4529 ** SXHASH on CLOS instances now uses a slot internal to the
4530 instance to return different numbers on distinct instances,
4531 while preserving the same return value through invocations of
4533 ** DEFMETHOD signals errors when methods with longer incongruent
4534 lambda lists are added to generic functions;
4535 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4536 CLASS, as specified in AMOP;
4537 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4538 on the order returned by the primary method for classes of
4539 class STANDARD-CLASS;
4540 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4541 * fixed some bugs shown by Paul Dietz' test suite:
4542 ** DOLIST puts its body in TAGBODY;
4543 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4545 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4546 value producing form;
4547 ** if more variables are given to PROGV than values, extra
4548 variables are bound and made to have no value;
4549 ** NSUBSTITUTE on list arguments gets the right answer with
4551 ** ELT signals an error of type TYPE-ERROR when the index argument
4552 is not a valid sequence index;
4553 ** LOOP signals (at macroexpansion time) an error of type
4554 PROGRAM-ERROR when duplicate variable names are found;
4555 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4556 ** FUNCALL of special-operators now cause an error of type
4558 ** PSETQ now works as required in the presence of side-effecting
4559 symbol-macro places;
4560 ** NCONC accepts any object as its last argument;
4561 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4563 ** loop-for-as-package does not require a package to be explicitely
4565 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4567 * fixed bug 166: compiler preserves "there is a way to go"
4568 invariant when deleting code.
4569 * fixed bug 172: macro lambda lists with required arguments after
4570 &REST arguments now cause an error to be signalled. (thanks to
4572 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4573 bug. (thanks to Matthew Danish)
4574 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4576 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4577 arguments are lexically rebound. (thanks to Gerd Moellmann and
4579 * fixed bug 194: error messages are now more informative when there
4580 is no primary method applicable in a call to a generic
4581 function. (thanks to Gerd Moellmann)
4582 * fixed bug in command line argument checking (thanks to Julian
4584 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4585 (thanks to Matthew Danish)
4586 * incremented fasl file version number, because of the
4587 SXHASH-related changes in the layout of CLOS data structures
4589 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4590 * minor incompatible change: The runtime (the Unix executable named
4591 "sbcl") is now much pickier about the .core files it will load.
4592 Essentially it now requires .core files to descend from the same
4593 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4594 as the runtime does. (The intent is to prevent the crashes which
4595 can occur, and which can even be reported as mysterious failures,
4596 when people patch the sources or change the build parameters
4597 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4598 sbcl and .core files.)
4599 * fixed bug: VALUES-LIST is no longer optimized away.
4600 * fixed bug 142: The FFI conversion of C string values to Lisp
4601 string values no longer conses excessively. (thanks to Nathan
4602 Froyd porting Raymond Toy's fix to CMU CL)
4603 * began to systematize and improve MOP conformance in PCL (thanks to
4604 Nathan Froyd, Gerd Moellman and Pierre Mai):
4605 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4607 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4608 ** FINALIZE-INHERITANCE is now called on class finalization;
4609 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4610 argument precedence order.
4611 * fixed bug 202: The compiler no longer fails on functions whose
4612 derived types contradict their declared type.
4613 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4614 so it can be non-toplevel.
4615 * The fasl file version number has changed (because of the new
4616 implementation of DEFMACRO).
4617 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4618 safe code, that any length requirement by their type-specifier
4619 argument is valid. The exceptions to this are described in bug
4621 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4622 functions now check the types of their inputs as required by ANSI.
4623 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4624 introduced symbol is DECLAREd to be SPECIAL.
4625 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4626 * fixed bug: :COUNT argument to sequence functions may be negative
4627 * fixed bug: body of DO-SYMBOLS may contain declarations
4628 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4629 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4630 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4631 symbol macro only once
4632 * fixed printing of call frame when argument list is unavailable
4633 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4634 * compiler no longer signals WARNING on unknown keyword
4637 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4638 * A beta-quality port to the mips architecture running Linux,
4639 based on the old CMUCL backend, has been made. It has been tested
4640 on a big-endian kernel, and works sufficiently well to be able to
4641 rebuild itself; it has not been tested in little-endian mode.
4642 * fixed an inconsistency between gencgc.c and purify.c which made
4643 dumping/loading .core files unreliable
4644 * fixed bug 120a: The compiler now deals correctly with IFs where
4645 the consequent is the same as the alternative, instead of
4646 misderiving the return type. (thanks to Alexey Dejneka)
4647 * fixed bug 113: Logical pathnames are now dumpable (the logical
4648 host is resolved at load-time, throwing an error if it is not
4650 * fixed bug 174: FORMAT's error message is slightly clearer when a
4651 non-printing character is used in a format directive.
4652 * fixed several bugs in compiler checking of type declarations, i.e.
4653 violations of the Python "declarations are assertions" principle
4654 (thanks to Alexey Dejneka)
4655 * fixed several bugs in PCL's error checking (thanks to Gerd
4657 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4659 * fixed bug in compilation of functions as first class values
4660 (thanks to Antonio Martinez-Shotton)
4661 * The compiler's handling TYPE-ERRORs which it can prove will
4662 inevitably happen at runtime has been cleaned up and corrected
4663 in several ways. (thanks to Alexey Dejneka)
4664 * improved argument type checking for various basic arithmetic
4665 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4666 lavished on them in the past that they can be compiled in many
4667 ways in different special cases
4668 * fixed bug 181: compiler checks validity of user supplied type
4670 * cleaned up code flushing in optimization: Function calls which
4671 should signal errors for safety purposes (e.g. which ANSI says
4672 should signal errors when their arguments are of incorrect type)
4673 are no longer optimized away.
4674 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4675 * incremented fasl file version number, because changes in the
4676 implementation of sequence functions like COERCE caused
4677 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4678 in old inline expansions) to become undefined. (Actually these
4679 changes were later undone, so we might very well be binary
4680 compatible with 0.7.7 after all, but leaving the version number
4681 incremented seemed like the simplest and most conservative
4684 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4685 * An alpha-quality port to the parisc architecture running Linux,
4686 based on the old CMUCL backend, has been made. This, even more so
4687 than the other backends, should be considered still a work in
4688 progress; known problems include that the Linux kernel in 64-bit
4689 mode does not propagate the correct sigcontext structure to
4690 userspace, and consequently SBCL on a parisc64 kernel will not
4692 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4693 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4694 suppressed.) Also now that the compiler is looking at declarations
4695 in the environment, it checks optimization declarations as well,
4696 and suppresses inlining when (> DEBUG SPEED).
4697 * More fixes have been made to treatment of floating point exception
4698 treatment and other Unix signals. In particular, floating point
4699 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4700 * The detection and handling of control stack exhaustion (infinite
4701 or very deeply nested recursion) has changed. Stack exhaustion
4702 detection is now done by write-protecting pages at the OS level
4703 and applies at all optimization settings; when found, a
4704 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4705 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4706 be caught using IGNORE-ERRORS.
4707 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4708 correctly when transforming recursive LABELS functions to LETs.
4709 (thanks to Alexey Dejneka)
4710 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4711 that are names of constants or global variables.
4712 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4713 alien routines with docstrings.
4714 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4715 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4717 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4718 (thanks to Eric Marsden)
4719 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4720 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4721 * Bug fix: The SPARC backend can now compile functions involving
4722 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4723 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4724 to a functions expecting a PATHNAME-DESIGNATOR.
4725 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4726 lambda lists. (thanks to Alexey Dejneka)
4727 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4728 functions so that EXPORTing the name of the class doesn't cause
4729 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4730 to Antonio Martinez for reporting this)
4731 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4732 constant-folded without failing an assertion. (thanks to Einar
4733 Floystad Dorum for reporting this)
4734 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4735 types are now dealt with more correctly by the compiler.
4736 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4737 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4738 behaviour analogous to RENAME-FILE. This puts its behaviour more
4739 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4741 * The fasl file version number has changed again. (because of the
4742 bug fix involving the names of PCL MAKE-INSTANCE functions)
4744 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4745 * bug fix: Floating point exceptions are treated much more
4746 consistently on the x86/Linux and PPC/Linux platforms.
4747 * Array initialization with :INITIAL-ELEMENT is now much faster for
4748 cases when the compiler cannot open code the array creation, but
4749 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4750 array accesses have also seen a speed increase.
4751 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4752 specified. (This was at the root of some bad interactions between
4753 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4755 * bug fix: Internal error arguments for undefined functions are now
4756 computed correctly on the PPC/Linux platform.
4757 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4758 Raymond Toy's patch for CMU CL)
4759 * Support for the Solaris 9 operating environment has been included
4760 (thanks to Daniel Merritt)
4761 * A very ugly but hopefully complete draft of the missing FFI chapter
4762 of the manual has been created by reformatting the corresponding
4763 CMU CL manual chapter into (currently very ugly and incoherent)
4764 DocBook and bringing it up to date for SBCL behavior. Thus, the
4765 manual is now essentially complete, at least by my extreme
4766 once-and-only-once standards, whereby it's acceptable to refer to
4767 the doc strings of SB-EXT functions as the primary documentation.
4768 * The fasl file version number has changed again, due to cleanup of
4769 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4770 is no longer a static symbol.)
4772 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4773 * SBCL now builds with OpenMCL (version 0.12) as the
4774 cross-compilation host; also, more progress has been made toward
4775 bootstrapping under CLISP.
4776 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4778 * bug 158 fixed: The compiler can now deal with integer loop
4779 increments different from 1; fixing this turned out also to fix
4781 * bug 169 fixed: no more bogus warnings about using lexical bindings
4782 despite the presence of perfectly good SPECIAL declarations (thanks
4783 to David Lichteblau)
4784 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4785 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4786 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4788 * bug 180 fixed: Method combination specifications no longer ignore
4789 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4790 * bug fix: Structure type predicate functions now check their argument
4791 count as they should.
4792 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4793 correctly. (thanks to Pierre Mai)
4794 * minor incompatible change: The --noprogrammer option is deprecated
4795 in favor of the new --disable-debugger option, which is very similar.
4796 (The major difference is that it takes effect at a slightly different
4797 time at startup, causing handling of errors in --sysinit and
4798 --userinit files will be affected differently.) The
4799 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4800 been added to allow this functionality to be controlled from ordinary
4801 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4802 maintainers, who might want to run non-interactive scripts to
4803 build SBCL cores which will later be used interactively.)
4804 * minor incompatible change: The LOAD function no longer, when given
4805 a wild pathname to load, loads all files matching that pathname.
4806 Instead, an error of type FILE-ERROR is signalled.
4808 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4809 * bug 147 fixed: The compiler preserves its block link/count
4810 invariants more correctly now so that it doesn't crash. (thanks
4812 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4814 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4815 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4816 no-value-for-this-character as NIL (as specified by ANSI).
4817 * HOST-NAMESTRING on physical pathnames now returns a string that is
4818 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4819 (thanks to Christophe Rhodes)
4820 * The Alpha port handles icache flushing more correctly. (thanks to
4822 * More progress has been made toward bootstrapping under CLISP. (thanks
4823 to Christophe Rhodes)
4824 * The fasl file format has changed again, because dynamic loading
4825 on OpenBSD (which has non-ELF object files) motivated some cleanups
4826 in the way that foreign symbols are transformed and passed around.
4827 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4828 is no longer treated as whitespace by the reader, but instead as
4829 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4830 |A
\7fB|, instead of A as it used to.
4832 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4833 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4834 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4835 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4836 this as of 0.7.1.45, but was left out of the previous news section
4837 (thanks to Dan Barlow)
4838 * SBCL now runs on the Solaris operating system on SPARC architectures
4839 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4840 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4841 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4842 * SPARC backend cleanups, allowing builds of cores optimized for V8
4843 and V9 SPARCS, and also emission of code targeted to a particular
4844 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4846 * SBCL is closer to bootstrapping under CLISP, thanks to various
4847 fixes by Christophe Rhodes.
4848 * The fasl file format has changed again, to allow the compiler's
4849 INFO database to support symbol macros.
4850 * The user manual (in doc/) is formatted into HTML more nicely.
4851 (thanks to coreythomas)
4852 * The system is smarter about SUBTYPEP relationships, especially
4853 those involving NOT types (including types such as ATOM which are
4854 represented internally using NOT types). Thus SUBTYPEP is less
4855 likely to return (VALUES NIL NIL) in general, and in particular
4856 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4858 * The fasl file format has changed again, because the internal
4859 representation of types now includes a new slot to support the new
4860 SUBTYPEP-of-NOT-types logic.
4861 * (not a change in the main branch of SBCL, but a related prototype
4862 which can hopefully be merged into the main branch of SBCL in the
4863 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4864 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4865 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4866 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4867 dispatch macro character combinations. (thanks to Alexey Dejneka)
4868 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4869 that superficially look like logical namestrings correctly.
4870 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4872 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4873 default supplied-p) for &optional and &key arguments. (thanks to
4876 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4877 * incompatible change: The compiler is now less aggressive about
4878 tail call optimization, doing it only when (> SPACE DEBUG) or
4879 (> SPEED DEBUG). (This is an incompatible change because there are
4880 programs which relied on the old CMU-CL-style behavior to optimize
4881 away their unbounded recursion which will now die of stack overflow.)
4882 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4883 for non-GENCGC systems has been increased to 20M (since that
4884 seems much closer to the likely performance optimum for modern
4885 systems than the old 4M value was)
4886 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4887 and *DEBUG-PRINT-LEVEL*
4888 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4889 of CMU CL's support for SPARC, and various endianness and other
4890 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4891 * new syntactic sugar for the Unix command line: --load foo.bar is now
4892 an alternate notation for --eval '(load "foo.bar")'.
4894 ** The system now detects stack overflow and handles it gracefully,
4895 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4896 optimization settings. (This is a good thing in general, and
4897 its introduction in this version should be particularly timely
4898 for anyone whose code fails because of suppression of tail
4900 ** The system now hunts for the C variable "environ" in a more
4901 devious way, to avoid segfaults when the C library version differs
4902 between compile time and run time. (thanks to Christophe Rhodes)
4903 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4904 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4905 ** The compiler no longer issues bogus style warnings for undefined
4906 classes in the same source file as the DEFCLASSes which defined
4907 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4909 ** fixes in CONDITION class precedence list for undefined function
4910 errors (thanks to Alexei Dejneka)
4911 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4912 correctly. (thanks to Dan Barlow)
4913 ** portability fixes aiming at bootstrapping under CLISP (thanks
4914 to Dave McDonald and Christophe Rhodes)
4915 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4916 ** fixes in type translation and and type inference (thanks to
4918 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4919 ** various fixes in the new ports (thanks to Dan Barlow)
4920 * several changes related to debugging:
4921 ** suppression of tail recursion, as noted above
4922 ** stack overflow detection, as noted above
4923 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4924 is now the default. (For some time encapsulation has been more
4925 reliable than the breakpoint-based :ENCAPSULATE NIL
4926 implementation, at least on X86 systems; and I just noticed that
4927 encapsulation also seems closer to the spirit of the ANSI
4930 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4932 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4933 up properly again. (There was a packaging bug in 0.7.0 which
4934 left their definitions in SB-SYS::LOAD-FOREIGN and
4935 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4936 vital for most things which interface to C-level interfaces,
4937 like extensions working with sockets or databases or
4938 Perl-compatible regexes or whatever, and the need to fix
4939 this bug is the main reason that 0.7.1 was released so
4941 ** DEFGENERIC is now choosier about the methods it redefines, so that
4942 reLOADing a previously-LOADed file containing DEFGENERICs does
4943 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4944 cycle now works as it should. (thanks to Alexey Dejneka)
4945 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4946 fixed by Christophe Rhodes. (He actually submitted this patch
4947 months ago, and I delayed until after 0.7.0.)
4948 ** Bug 111 (internal compiler confusion about runtime checks on
4949 FUNCTION types) was fixed by Alexey Dejneka.
4950 * Some internal cleanups (getting rid of variables which aren't
4951 needed now that the byte interpreter is gone) caused the fasl
4952 file format number to change again.
4954 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4955 * major incompatible change: The default fasl file extension, i.e. the
4956 default extension for files produced by COMPILE-FILE, has changed
4957 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4959 ** There are many changes in the implementation of the compiler.
4960 SBCL is now essentially a compiler-only implementation of ANSI
4961 Common Lisp. EVAL still "interprets" a few special cases, but
4962 almost all the interesting cases are handled by creating
4963 a LAMBDA expression, calling COMPILE on it, then calling
4964 FUNCALL on the result.
4965 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4966 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4967 Since the code is newer, there might still be some new bugs
4968 (though not as many as before Martin Atzmueller's fixes:-). But
4969 the new code is substantially simpler and clearer, and hopefully
4970 any remaining bugs will be simpler, less fundamental, and more
4971 fixable then the bugs in the old code.
4972 ** The revised compiler is still a little unsteady on its feet.
4974 *** The debugging information it produces (particularly the names
4975 of FUNCTION objects) is sometimes much less useful than what
4976 the old compiler produced.
4977 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4978 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4979 within a MACROLET) has been temporarily weakened.
4980 ** There are new compiler optimizations for various functions:
4981 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4982 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4983 *** the math functions TRUNCATE, FLOOR, and CEILING
4984 *** the function-of-all-trades COERCE
4985 Mostly these should be transparent, but there's one
4986 potentially-annoying problem (bug 117): when the compiler
4987 inline-expands a function and does type analysis on the result,
4988 it can create control paths which have type mismatches, and
4989 when it can't prove that those control paths aren't taken,
4990 it will issue WARNINGs about the type mismatches. This is
4991 a particular problem in practice for the new sequence functions.
4992 It's not clear how this should be fixed, and for now, a
4993 workaround is given in the entry for 117 in the BUGS file.
4994 ** (Because of the interaction between the two previous items --
4995 occasional inlining problems and new inline expansions -- some
4996 of the new sequence function optimizations won't really kick in
4997 completely until debugging information, and then inlining, are
4998 straightened out in some future version.)
4999 * minor incompatible changes:
5000 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
5001 DIRECTORY no longer implicitly promotes NIL slots of its
5002 pathname argument to :WILD. In particular, when you ask for the
5003 contents of a directory (which you used to be able to do without
5004 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
5005 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
5006 ** changes in behavior that ANSI explicitly defines to be
5007 implementation dependent:
5008 *** The new compiler-only implementation still conforms with ANSI,
5009 but acts a little different than before. Besides the obvious
5010 changes in performance tradeoffs (that the cost per form passed
5011 to EVAL has gone up, and the cost per form executed by EVAL
5012 has gone down), the behavior of the system changes a little
5013 because there are no longer any interpreted function objects.
5014 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
5015 e.g. doing COMPILE on the output of interactive DEFUN is
5017 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
5019 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
5020 has changed from (:CL) to NIL.
5021 *** The CHAR-NAME of unprintable ASCII characters which, unlike
5022 e.g. #\Newline and #\Tab, don't have names specified in the
5023 ANSI Common Lisp standard, is now based on their ASCII symbolic
5024 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
5025 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
5026 are no longer used for output.
5027 ** changes in internal implementation constants:
5028 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
5029 4 million. (If your application spends a lot of time GCing and
5030 you have a lot of RAM, you might want to experiment with
5031 increasing it even more.)
5032 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
5033 However, almost all old code should still continue to work without
5034 immediate update, as SB-C-CALL is now a (deprecated) nickname
5036 ** Old operator names in the style DEF-FOO are now deprecated in
5037 favor of new corresponding names DEFINE-FOO, for consistency with
5038 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
5039 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
5040 internal symbols, but a few supported extensions like
5041 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
5042 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
5043 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
5044 etc. as you get deeper into recursive calls to the debugger
5045 command loop, instead of the old "5]", "5]]", "5]]]"
5046 sequence. (I was motivated to do this when squabbles between
5047 ILISP and SBCL left me very deeply nested in the debugger. In the
5048 short term, this change will probably provoke more ILISP/SBCL
5049 squabbles, but hopefully it will be an improvement in the long run.)
5050 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
5051 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
5052 function, since declaiming return types involving aliens is
5053 (1) annoyingly messy to do by hand and (2) vital to efficient
5054 compilation of code which calls such functions.
5055 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
5056 longer reexported by the SB-EXT package. They're solely useful
5057 for alien code, so it seems more logical that you should get
5058 them from the SB-ALIEN package, not in SB-EXT.
5059 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
5060 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
5061 features. Instead, the code that they used to control is always
5062 built into the system.
5063 * many other bug fixes
5064 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
5065 advantage of the new EVAL-WHEN stuff and to clean them up in
5066 general, and they are now more ANSI-compliant in a number of
5067 ways. Martin Atzmueller is responsible for a lot of this.
5068 ** Besides the cleanups discussed above, Martin Atzmueller fixed
5070 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
5071 *** correct ERROR type for various file operations
5072 *** some fixes for Lisp streams
5073 *** DEFMETHOD syntax checking
5074 *** changing old weird representation of debug information as
5075 strings (which, among their other deficiencies, don't transform
5076 correctly when you rename packages, and don't change their
5077 print representation when you change things like *PACKAGE*
5078 and *PRINT-LENGTH*) to symbols and lists of symbols
5079 He also made several improvements and fixed several bugs in DESCRIBE.
5080 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
5082 *** misbehavior of WRITE-STRING/WRITE-LINE
5083 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
5084 and several other LOOP problems as well
5085 *** DIRECTORY when similar filenames are present
5086 *** DEFGENERIC with :METHOD options
5087 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
5088 *** bug in the optimization of ARRAY-ELEMENT-TYPE
5089 *** argument ordering in FIND with :TEST option
5090 *** mishandled package designator argument in APROPOS-LIST
5091 *** various problems in the backquote readmacro
5092 *** a bug in APROPOS
5093 *** probably some others that I'm not describing very well here,
5094 since the CVS log documents them by reference to sbcl-devel
5095 messages, and the SourceForge archives aren't working well.:-(
5096 ** Dan Barlow improved the Alpha port (and is making progress on the
5097 PPC port, for those of you who think different).
5098 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
5099 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
5100 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
5101 continues to work on the SPARC port (for those of you in a position
5102 to look down upon our little PC-compatible boxes from a great height).
5103 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
5104 and uses that copy, rather than the current dynamic value, when
5105 it's trying to decide whether to truncate output. Thus e.g.
5106 (let ((*print-lines* 50))
5107 (pprint-logical-block (stream nil)
5109 (let ((*print-lines* 8))
5110 (print (aref possiblybigthings i) stream)))))
5111 should now truncate the logical block only at 50 lines, instead of
5112 often truncating it at 8 lines, as it did before.
5113 * The doc/cmucl/ directory, containing old CMU CL documentation
5114 from the time of the fork, is no longer part of the base system.
5115 SourceForge has shut down its anonymous FTP service, and with it
5116 my original plan for distributing the old CMU CL documentation
5117 there. For now, if you need these files you can download an old
5118 SBCL source release and extract them from it.
5119 * The fasl file version number changed again, for dozens of reasons,
5120 some of which are apparent above.
5122 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
5123 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
5124 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
5125 patch, so that DISASSEMBLE output is much nicer.
5126 * The code in the SB-PROFILE package now seems reasonably stable.
5127 I still haven't decided what the final interface should look like
5128 (I'd like PROFILE to interact cleanly with TRACE, since both
5129 facilities use function encapsulation) but if you have a need
5130 for profiling now, you can probably use it successfully with
5131 the current CMU-CL-style interface.
5132 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
5133 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
5134 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
5135 link now returns the dangling link itself, and for similar
5136 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
5137 link itself. (In these cases the old code signalled an error and
5138 looped endlessly, respectively.) Thus, DIRECTORY now works even
5139 in the presence of dangling and cyclic symbolic links.
5140 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
5141 is now a supported extension again, since the consensus on
5142 sbcl-devel was that it can be useful for ordinary development
5143 work, not just for debugging SBCL itself.
5144 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
5145 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
5146 that currently-defined functions might be redefined later with
5147 different return types.
5148 * Hash tables can be printed readably, as inspired by CMU CL code
5149 of Eric Marsden and SBCL code of Martin Atzmueller.
5150 * better error handling in CLOS method combination, thanks to
5151 Martin Atzmueller porting Pierre Mai's CMU CL patches
5152 * more overflow fixes for >16Mbyte I/O buffers
5153 * A bug in READ has been fixed, so that now a single Ctrl-D
5154 character suffices to cause end-of-file on character streams.
5155 In particular, now you only need one Ctrl-D at the command
5156 line (not two) to exit SBCL.
5157 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
5159 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
5160 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
5161 does the right thing.
5162 * The implementation of some type tests, especially for CONDITION
5163 types, is now tidier and maybe faster, due to CMU CL code
5164 originally by Douglas Crosher, ported by Martin Atzmueller.
5165 * Some math functions have been fixed, and there are new
5166 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
5167 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
5168 * (There are also some new optimizers in contrib/*-extras.lisp. Those
5169 aren't built into sbcl-0.6.13, but are a sneak preview of what's
5170 likely to be built into sbcl-0.7.0.)
5171 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
5172 CL, ported to SBCL by Martin Atzmueller)
5173 * DESCRIBE now gives more information in some cases. (Pierre Mai's
5174 patch to CMU CL, ported to SBCL by Martin Atzmueller)
5175 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
5176 * There's a new slam.sh hack to shorten the edit/compile/debug
5177 cycle for low-level changes to SBCL itself, and a new
5178 :SB-AFTER-XC-CORE target feature to control the generation of
5179 the after-xc.core file needed by slam.sh.
5180 * minor incompatible change: The ENTRY-POINTS &KEY argument to
5181 COMPILE-FILE is no longer supported, so that now every function
5182 gets an entry point, so that block compilation looks a little
5183 more like the plain vanilla ANSI section 3.2.2.3 scheme.
5184 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
5185 returns the number of bytes consed since the system started,
5186 rather than the number consed since the first time the function
5187 was called. (The new definition parallels ANSI functions like
5188 CL:GET-INTERNAL-RUN-TIME.)
5189 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
5190 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
5191 supported. Now DIRECTORY always does the abstract Common-Lisp-y
5192 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
5193 * Fasl file version numbers are now independent of the target CPU,
5194 since historically most system changes which required version
5195 number changes have affected all CPUs equally. Similarly,
5196 the byte fasl file version is now equal to the ordinary
5199 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
5200 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
5201 is no longer recognized. I apologize for this, because it was
5202 listed in SB-EXT as a supported extension, but I found that
5203 its existing behavior was poorly specified, as well as incorrectly
5204 specified, and it looked like too much of a mess to straighten it
5205 out. I have enough on my hands trying to get ANSI stuff to work..
5206 * many patches ported from CMU CL by Martin Atzmueller, with
5207 half a dozen bug fixes in pretty-printing and the debugger, and
5208 half a dozen others elsewhere
5209 * fixed bug 13: Floating point infinities are now supported again.
5210 They might still be a little bit flaky, but thanks to bug reports
5211 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
5212 as flaky as they were.
5213 * The --noprogrammer command line option is now supported. (Its
5214 behavior is slightly different in detail from what the old man
5215 page claimed it would do, but it's still appropriate under the
5216 same circumstances that the man page talks about.)
5217 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
5218 are now supported, and enabled by default. Thus, the compiler can
5219 handle many floating point and complex operations much less
5220 inefficiently. (Thus e.g. you can implement a complex FFT
5222 * The compiler now detects type mismatches between DECLAIM FTYPE
5223 and DEFUN better, and implements CHECK-TYPE more correctly, and
5224 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
5225 thanks to patches from Martin Atzmueller.
5226 * various fixes to make the cross-compiler more portable to
5227 ANSI-conforming-but-different cross-compilation hosts (notably
5228 Lispworks for Windows, following bug reports from Arthur Lemmens)
5229 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
5230 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
5231 fixed thanks to Pierre Mai's CMU CL patches.
5232 * fixes to keep the system from overflowing internal counters when
5233 it tries to use i/o buffers larger than 16M bytes
5234 * fixed bug 45a: Various internal functions required to support
5235 complex special functions have been merged from CMU CL sources.
5236 (When I was first setting up SBCL, I misunderstood a compile-time
5237 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
5238 * improved support for type intersection and union, fixing bug 12
5239 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
5240 more obscure bugs as well
5241 * some steps toward byte-compiling non-performance-critical
5242 parts of the system, courtesy of patches from Martin Atzmueller
5243 * Christophe Rhodes has made some debian packages of sbcl at
5244 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
5245 From his sbcl-devel e-mail of 2001-04-08 they're not completely
5246 stable, but are nonetheless usable. When he's ready, I'd be happy
5247 to add them to the SourceForge "File Releases" section. (And if
5248 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
5249 * new fasl file format version number (because of changes in
5250 internal representation of (OR ..) types to accommodate the new
5251 support for (AND ..) types, among other things)
5253 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
5254 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
5256 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
5258 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
5259 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
5260 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
5261 * DESCRIBE now works on structure objects again.
5262 * Most function call argument type mismatches are now handled as
5263 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
5264 know whether the function will be redefined before the call is
5265 executed. (The compiler could flag local calls with full WARNINGs,
5266 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
5267 it doesn't keep track of enough information to know whether calls
5268 are local in this sense.)
5269 * Compiler output is now more verbose, with messages truncated
5270 later than before. (There should be some supported way for users
5271 to override the default verbosity, but I haven't decided how to
5272 provide it yet, so this behavior is still controlled by the internal
5273 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
5274 src/compiler/ir1util.lisp.)
5275 * Fasl file format version numbers have increased again, because
5276 support for the Gray streams extension changes the layout of the
5277 system's STREAM objects.
5278 * The Gray subclassable streams extension now works, thanks to a
5279 patch from Martin Atzmueller.
5280 * The full LOAD-FOREIGN extension (not just the primitive
5281 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
5282 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
5283 but like most other programs, it defaults to copying the Unix
5284 environment from the original process instead of starting the
5285 new process in an empty environment.
5286 * Extensions which manipulate the Unix environment now support
5287 an :ENVIRONMENT keyword option which doesn't smash case or
5288 do other bad things. The CMU-CL-style :ENV option is retained
5289 for porting convenience.
5290 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
5291 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
5293 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
5295 * A patch from Martin Atzmueller seems to have solved the SIGINT
5296 problem, and as far as we know, signal-handling now works cleanly.
5297 (If you find any new bugs, please report them!)
5298 * The system no longer defaults Lisp source file names to types
5299 ".l", ".cl", or ".lsp", but only to ".lisp".
5300 * The compiler no longer uses special default file extensions for
5301 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
5302 seems to expect a single default extension for all compiled code,
5303 and there's no compelling reason to try to stretch the standard
5304 to allow two different extensions.) Instead, byte-compiled files
5305 default to the same extension as native-compiled files.
5306 * Fasl file format version numbers have increased again, because
5307 a rearrangement of internal implementation packages made some
5308 dumped symbols in old fasl files unreadable in new cores.
5309 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
5310 many fewer weird special cases.
5311 * Bug #17 (differing COMPILE-FILE behavior between logical and
5312 physical pathnames) has been fixed, and some related misbehavior too,
5313 thanks to a patch from Martin Atzmueller.
5314 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
5315 by Tim Moore, ported to SBCL by Martin Atzmueller.
5316 * Martin Atzmueller fixed several filesystem-related problems,
5317 including bug #36, in part by porting CMU CL patches, which were
5318 written in part by Paul Werkowski.
5319 * More compiler warnings in src/runtime/ are gone, thanks to
5320 more patches from Martin Atzmueller.
5321 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
5324 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
5326 * DESCRIBE now works on CONDITION objects.
5327 * The debugger now handles errors which arise when trying to print
5328 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
5330 * The build system now uses an additional file, customize-target-features.lisp,
5331 to allow local modifications to the target *FEATURES* list. (The point of
5332 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
5333 debugging features enabled, without having to worry about propagating it
5334 into everyone's system when I do a "cvs update".) When no
5335 customize-target-features.lisp file exists, the target *FEATURES* list
5336 should be constructed the same way as before.
5337 * fixed bugs in DEFCONSTANT ANSI-compatibility:
5338 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
5339 warn about behavior which is undefined under the ANSI spec. Note: This
5340 is specified by ANSI, but it's not very popular with programmers.
5341 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
5342 macro in the SBCL sources for an example of a workaround which you
5343 might use to make portable ANSI-standard code which does what you want.
5344 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
5345 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
5346 when it's used as a non-toplevel form. (This is required in order
5347 to implement the DEFCONSTANT-EQX macro.)
5348 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
5349 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
5350 funny things, due to bugs in the implementation of EVAL-WHEN
5351 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
5352 non-toplevel DEFCONSTANTs are uncommon.)
5353 * The core file version number and fasl file version number have been
5354 incremented, because the old noncompliant DEFCONSTANT behavior involved
5355 calling functions which no longer exist, and because I also took the
5356 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
5357 * fixed bug 1 (error handling before read-eval-print loop starts), and
5358 redid debugger restarts and related debugger commands somewhat while
5360 ** The QUIT debugger command is gone, since it did something
5361 rather different than the SB-EXT:QUIT command, and since it never
5362 worked properly outside the main toplevel read/eval/print loop.
5363 Invoking the new TOPLEVEL restart provides the same functionality.
5364 ** The GO debugger command is also gone, since you can just invoke
5365 the CONTINUE restart directly instead.
5366 ** The TOP debugger command is also gone, since it's redundant with the
5367 FRAME 0 command, and since it interfered with abbreviations for the
5369 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
5371 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
5372 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
5373 some porting work by Martin Atzmueller)
5374 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
5375 with a boatload of patches which he ported to SBCL. Now that those
5377 ** The system tries to make sure that its low-priority messages
5378 are prefixed by semicolons, to help people who like to use
5379 syntax highlighting in their ILISP buffer. (This patch
5380 was originally due to Raymond Toy.)
5381 ** The system now optimizes INTEGER-LENGTH better, thanks to more
5382 patches originally written by Raymond Toy.
5383 ** The compiler understands coercion between single-value and
5384 multiple-VALUES type expressions better, getting rid of some very
5385 weird behavior, thanks to patches originally by Robert MacLachlan
5386 and Douglas Crosher.
5387 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
5388 lambda lists, thanks to a patch originally by Pierre Mai.
5389 ** The system no longer bogusly warns about "abbreviated type
5391 ** The compiler gets less confused by inlining and RETURN-FROM,
5392 thanks to some patches originally by Tim Moore.
5393 ** The system no longer hangs when dumping circular lists to fasl
5394 files, thanks to a patch originally from Douglas Crosher.
5395 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
5396 undefined function error.
5397 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
5398 in structure slot initforms) for now, documented workaround instead:-|
5399 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
5400 * fixed bug 5: added stubs for various Gray stream functions called
5401 in the not-a-CL:STREAM case, so that even when Gray streams aren't
5402 installed, at least appropriate type errors are generated
5403 * fixed bug 8: better reporting of various PROGRAM-ERRORs
5404 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
5405 consistently in DEFMETHOD forms.
5406 * removed bug 21 from BUGS, since Martin Atzmueller points out that
5407 it doesn't seem to affect SBCL after all
5408 * The C runtime system now builds with better optimization and many
5409 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
5411 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
5413 * The system is now under CVS at SourceForge (instead of the
5414 CVS repository on my home machine).
5415 * The new signal handling code has been tweaked to treat register
5416 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
5417 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
5418 downstream, e.g. in debug-int.lisp, has implicit dependencies
5419 on the unsignedness of integer representation of machine words,
5420 and that caused the system to bomb out with infinite regress
5421 when trying to recover from type errors involving signed values,
5422 e.g. (BUTLAST '(1 2 3) -1).)
5423 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
5424 (This was one of the bugs Peter Van Eynde reported back in July.)
5425 * The system now uses code inspired by Colin Walters' O(N)
5426 implementation of MAP (from the cmucl-imp@cons.org mailing
5427 list, 2 September 2000) when it can't use a DEFTRANSFORM to
5428 inline the MAP operation, and there is more than one
5429 sequence argument to the MAP call (so that it can't just
5430 do ETYPECASE once and for all based on the type of the
5431 single sequence argument). (The old non-inline implementation
5432 of the general M-argument sequence-of-length-N case required
5433 O(M*N*N) time when any of the sequence arguments were LISTs.)
5434 * The QUIT :UNIX-CODE keyword argument has been renamed to
5435 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5436 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5438 * Raymond Wiker's patches to port dynamic loading from Linux to
5439 FreeBSD have been added.
5440 * The BUGS file is now more nearly up to date, thanks in large part
5441 to Martin Atzmueller's review of it.
5442 * The debugger now flushes standard output streams before it begins
5443 its output ("debugger invoked" and so forth).
5444 * The core version number and fasl file version number have both
5445 been incremented, because of incompatible changes in the layout
5447 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5448 used to not be called for a saved Lisp image.)
5449 * Martin Atzmueller's version of a patch to fix a compiler crash,
5450 as posted on sbcl-devel 13 September 2000, has been installed.
5451 * Instead of installing Martin Atzmueller's patch for the
5452 compiler transform for SUBSEQ, I deleted the compiler transform,
5453 and transforms for some similar consing operations.
5454 * A bug in signal handling which kept TRACE from working on OpenBSD
5456 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5457 away by constant folding
5458 * The system now defines its address space constants in one place
5459 (in the Lisp sources), and propagates them automatically elsewhere
5460 (through GENESIS and the sbcl.h file). Therefore, patching the
5461 address map is less unnecessarily tedious and error-prone. The
5462 Lisp names of address space constants have also been systematized.
5463 * CVS tags like dollar-Header-dollar have been removed from
5464 the sources, because they have never saved me trouble and
5465 they've been source of trouble working with patches and other
5466 diff-related operations.
5467 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5468 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5470 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5472 * The system has been ported to OpenBSD.
5473 * The system now compiles with a simple "sh make.sh" on the systems
5474 that it's supported on. I.e., now you no longer need to tweak
5475 text in the target-features.lisp-expr and symlinks in src/runtime/
5476 by hand, the make.sh takes care of it for you.
5477 * The system is no longer so grossly inefficient when compiling code
5478 involving vectors implemented as general (not simple) vectors (VECTOR T),
5479 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5480 which dares to use the various sequence functions on non-simple
5481 vectors, takes less of a performance hit.
5482 * There is now a primitive type predicate VECTOR-T-P
5483 to test for the (VECTOR T) type, so that e.g.
5484 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5485 can now be compiled with some semblance of efficiency. (The old code
5486 turned the type declaration into a full call to %TYPEP at runtime!)
5487 * AREF on (VECTOR T) is still not fast, since it's still compiled
5488 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5489 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5490 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5491 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5492 without making full calls to SUBTYPEP at runtime.
5493 (Some analogous efficiency issues for non-simple vectors specialized to
5494 element types other than T, or for non-simple multidimensional arrays,
5495 have not been addressed. They could almost certainly be handled the
5496 same way if anyone is motivated to do so.)
5497 * The changes in array handling break binary compatibility, so
5498 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5499 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5500 instead of (VALUES T T).
5501 * By following the instructions that Dan Barlow posted to sbcl-devel
5502 on 2 July 2000, I was able to enable primitive dynamic object
5503 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5504 functionality is not implemented (since it calls ld to resolve
5505 library references automatically, requiring RUN-PROGRAM for its
5506 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5507 try to resolve library references) is now supported.
5508 * The system now flushes the standard output streams when it terminates,
5509 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5510 them at several other probably-convenient times, e.g. in each pass of
5511 the toplevel read-eval-print loop, and after evaluating a form given
5512 as an "--eval" command-line option. (These changes were motivated by a
5513 discussion of stream flushing issues on cmucl-imp in August 2000.)
5514 * The source transform for TYPEP of array types no longer assumes
5515 that an array whose element type is a not-yet-defined type
5516 is implemented as an array of T, but instead punts, so that the
5517 type will be interpreted at runtime.
5518 * There is now some support for cross-compiling in make.sh: each of
5519 the phases of make.sh has its own script. (This should be transparent
5520 to people doing ordinary, non-cross-compile builds.)
5521 * Since my laptop doesn't have hundreds of megabytes of memory like
5522 my desktop machine, I became more motivated to do some items on
5523 my to-do list in order to reduce the size of the system a little:
5524 ** Arrange for various needed-only-at-cold-init things to be
5525 uninterned after cold init. To support this, those things have
5526 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5527 symbols with such names are now uninterned after cold init).
5528 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5529 fasl files for cold load.
5530 ** Remove the old compiler structure pooling code (which used to
5531 be conditional on the target feature :SB-ALLOC) completely.
5532 ** Redo the representation of some data in cold init to be more compact.
5533 (I also looked into supporting byte compiled code at bootstrap time,
5534 which would probably reduce the size of the system a lot, but that
5535 looked too complicated, so I punted for now.)
5536 * The maximum signal nesting depth in the src/runtime/ support code has
5537 been reduced from 4096 to 256. (I don't know any reason for the very
5538 large old value. If the new smaller value turns out to break something,
5539 I'll probably just bump it back up.)
5540 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5542 * Many, many bugs reported by Peter Van Eynde have been added to
5543 the BUGS list; some have even been fixed.
5544 * While enabling dynamic object file loading, I tried to make the
5545 code easier to understand, renaming various functions and variables
5546 with less ambiguous names, and changing some function calling
5547 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5548 * While trying to figure out how to do the OpenBSD port, I tried to
5549 clean up some of the code in src/runtime/. In particular, I dropped
5550 support for non-POSIX signal handling, added various comments,
5551 tweaked the code to reduce the number of compilation warnings, and
5552 renamed some files to increase consistency.
5553 * To support the new automatic configuration functionality in make.sh,
5554 the source file target-features.lisp-expr has been replaced with the
5555 source file base-target-features.lisp-expr and the machine-generated
5556 file local-target-features.lisp-expr.
5557 * fixed a stupid quoting error in make.sh so that using CMU CL
5558 "lisp -batch" as cross-compilation host works again
5560 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5562 * DESCRIBE no longer tries to call itself recursively to describe
5563 bound/fbound values, so that it no longer fails on symbols which are
5564 bound to themselves (like keywords, T, and NIL).
5565 * DESCRIBE now works on generic functions.
5566 * The printer now prints less-screwed-up representations of closures
5567 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5568 * A private symbol is used instead of the :EMPTY keyword previously
5569 used to mark empty slots in hash tables. Thus
5570 (DEFVAR *HT* (MAKE-HASH-TABLE))
5571 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5572 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5573 now does what ANSI says that it should. (You can still get
5574 similar noncompliant behavior if bang on the hash table
5575 implementation with all the symbols you get back from
5576 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5577 This breaks binary compatibility, since tests for equality to
5578 :EMPTY are wired into things like the macroexpansion of
5579 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5581 * There's now a minimal placeholder implementation for CL:STEP,
5582 as required by ANSI.
5583 * An obscure bug in the interaction of the normal compiler, the byte
5584 compiler, inlining, and structure predicates has been patched
5585 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5586 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5588 * Missing ordinary arguments in a macro call are now detected even
5589 when the macro lambda list contains &KEY or &REST.
5590 * The debugger no longer complains about encountering the top of the
5591 stack when you type "FRAME 0" to explicitly instruct it to go to
5592 the top of the stack. And it now prints the frame you request even
5593 if it's the current frame (instead of saying "You are here.").
5594 * As specified by ANSI, the system now always prints keywords
5595 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5597 * The default initial SIZE of HASH-TABLEs is now smaller.
5598 * Type information from CLOS class dispatch is now propagated
5599 into DEFMETHOD bodies, so that e.g.
5600 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5602 is now basically equivalent to
5603 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5604 (DECLARE (TYPE SINGLE-FLOAT X))
5606 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5607 operation, without having to do run-time type dispatch.
5608 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5609 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5610 * Since I don't seem to be making big file reorganizations very often
5611 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5612 is overflowing my ability to conveniently back them up), I've finally
5613 checked the system into CVS. (The CVS repository is on my home system,
5614 not at SourceForge -- putting it on SourceForge might come later.)
5615 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5616 high-level GC-NOTIFY-FOO functions send their output. (There's
5617 still very little control of where low-level verbose GC functions
5618 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5619 less than it used to -- the GC-NOTIFY-FOO functions are now under
5620 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5621 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5622 in only one place in the source code, and propagates it automatically
5623 everywhere that it's needed. Thus e.g. when I bump the version from
5624 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5625 * The C source files now include boilerplate legalese and documentation
5626 at the head of each file (just as the Lisp source files already did).
5627 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5628 to his page will be replaced with a link to his new CLiki service.
5630 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5632 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5633 * The build process now looks for GNU make under the default name "gmake",
5634 instead of "make" as it used to. If GNU make is not available as "gmake"
5635 on your system, you can change this default behavior by setting the
5636 GNUMAKE environment variable.
5637 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5638 can build without error under CMU CL.
5640 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5642 * There is now a partial SBCL user manual (with some new text and some
5643 text cribbed from the CMU CL manual).
5644 * The beginnings of a profiler have been added (starting with the
5645 CMU CL profiler and simplifying and cleaning up). Eventually the
5646 main interface should be through the TRACE macro, but for now,
5647 it's still accessed through vaguely CMU-CL-style functions and macros
5648 exported from the package SB-PROFILE.
5649 * Some problems left over from porting CMU CL to the new
5650 cross-compilation bootstrap process have been cleaned up:
5651 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5652 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5653 part when getting rid of the compiler *BACKEND* stuff.)
5654 ** The value of *NULL-TYPE* was screwed up, because it was
5655 being initialized before the type system knew the final
5656 definition of the 'NULL type. This screwed up several key
5657 optimizations in the compiler, causing inefficiency in all sorts
5658 of places. (I found it because I wanted to understand why
5659 GET-INTERNAL-RUN-TIME was consing.)
5660 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5661 PROCLAIM FTYPE information. Unfortunately this broke binary
5662 compatibility again, since now the forms output by DEFGENERIC
5663 to refer to functions which didn't exist in 0.6.3.
5664 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5665 can use the new (as of 0.6.3) transform for SOME into MAP into
5667 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5668 values to (INTEGER 0 1000000), so that the time code will no longer
5669 occasionally get blown up by Linux returning 1000000 microseconds
5670 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5671 its output conform to the ANSI spec. (Alas, this makes its output
5672 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5673 * A full call to MAP NIL with a single sequence argument no longer conses.
5674 * fixes to problems pointed out by Martin Atzmueller:
5675 * The manual page no longer talks about multiprocessing as though
5676 it were currently supported.
5677 * The ILISP support patches have been removed from the distribution,
5678 because as of version 5.10.1, ILISP now supports SBCL without us
5679 having to maintain patches.
5680 * added a modified version of Raymond Toy's recent CMU CL patch for
5681 EQUALP comparison of HASH-TABLE
5683 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5685 * The system still can't cross-compile itself with
5686 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5687 entails), but at least it can compile more of itself that way
5688 than it used to be able to, and various buglets which were uncovered
5689 by trying to cross-compile itself that way have now been fixed.
5690 * This release breaks binary compatibility again. This time
5691 at least I've incremented the FASL file format version to 2, so that the
5692 problem can be detected reliably instead of just causing weird errors.
5693 * various new style warnings:
5694 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5695 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5696 ** using the lexical binding of a variable named in the *FOO* style
5697 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5698 as specified by ANSI.
5699 * *RANDOM-STATE* is no longer automatically initialized from
5700 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5701 default behavior of the system is to repeat its behavior every time
5702 it's run. If you'd like to change this behavior, you can always
5703 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5704 old convention there was no comparably easy way to get the system to
5705 repeat its behavior every time it was run.
5706 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5707 FTYPE declarations has been removed, in favor of their ANSI
5708 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5709 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5710 the types of their sequence arguments can be determined at compile time.
5711 This is done through a new open code expansion for MAP which eliminates
5712 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5713 argument types can be determined at compile time.
5714 * The optimizer now transforms COERCE into an identity operation when it
5715 can prove that the coerced object is already of the correct type. (This
5716 can be a win for machine generated code, including the output of other
5717 optimization transforms, such as the MAP transform above.)
5718 * Credit information has been moved from source file headers into CREDITS.
5719 * Source file headers have been made more standard.
5720 * The CASE macro now compiles without complaining even when it has
5723 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5725 * (Note that the way that the PCL macroexpansions were rewritten
5726 to accommodate the change in DEFGENERIC below breaks binary
5727 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5728 not run under sbcl-0.6.2. Once we get out of alpha releases,
5729 i.e. hit release 1.0.0, we'll probably try to maintain binary
5730 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5731 and sbcl-1.4.4. Until then, however, it might be fairly common
5732 for maintenance releases to break binary compatibility.)
5733 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5735 * The reporting of unhandled signals has been changed to print some
5736 explanatory text as well as the report form. (Previously only
5737 the report form was printed.)
5738 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5739 it defines, so that the compiler no longer issues undefined function
5740 warnings for compiled-but-not-yet-loaded generic functions.
5741 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5742 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5743 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5744 section of the ANSI spec are supported.
5745 * The "" nickname for the "KEYWORD" package has been removed.
5746 The reader still handles symbol tokens which begin with a package marker
5747 as keywords, but it doesn't expose its mechanism for doing so in the
5748 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5749 * The system now issues STYLE-WARNINGs for contradictory TYPE
5750 proclamations. (Warnings for contradictory FTYPE proclamations would
5751 be nice too, but those can't be done usefully unless the type system
5752 is made smarter about FUNCTION types.)
5753 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5754 systematized, so that "*target-*.lisp is supposed to exist only on the
5755 target and imply that there's a related file which exists on the
5756 host, and *host-*.lisp is supposed to exist only on the host and imply
5757 that there's a related file which exists on the target. This involves a
5758 lot of renaming. Hopefully the acute confusion caused by the renaming
5759 will be justified by the reduction in chronic confusion..
5760 ** runtime-type.lisp -> early-target-type.lisp
5761 ** target-type.lisp -> late-target-type.lisp
5762 ** early-host-format.lisp -> early-format.lisp
5763 ** late-host-format.lisp -> late-format.lisp
5764 ** host-error.lisp -> misc-error.lisp
5765 ** early-error.lisp -> early-target-error.lisp
5766 ** late-error.lisp -> late-target-error.lisp
5767 ** host-defboot.lisp -> early-defboot.lisp
5768 ** code/misc.lisp -> code/target-misc.lisp
5769 ** code/host-misc.lisp -> code/misc.lisp
5770 ** code/numbers.lisp -> code/target-numbers.lisp
5771 ** code/early-numbers.lisp -> numbers.lisp
5772 ** early-host-type.lisp -> early-type.lisp
5773 ** late-host-type.lisp -> late-type.lisp
5774 ** host-typep.lisp -> typep.lisp
5775 ** load.lisp -> target-load.lisp
5776 ** host-load.lisp -> load.lisp
5777 ** host-disassem.lisp -> disassem.lisp
5778 ** host-insts.lisp -> insts.lisp
5779 ** byte-comp.lisp -> target-byte-comp.lisp
5780 ** host-byte-comp.lisp -> byte-comp.lisp
5781 ** host-signal.lisp -> signal.lisp
5782 ** host-defstruct.lisp -> defstruct.lisp
5783 ** late-target-type.lisp -> deftypes-for-target.lisp
5784 Furthermore, several other previously target-only files foo.lisp (e.g.
5785 hash-table.lisp and random.lisp) have been split into a target-and-host
5786 foo.lisp file and a target-only target-foo.lisp file, with their key type
5787 definitions in the target-and-host part, so that the cross-compiler will
5788 know more about target types.
5789 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5790 gone away. In their place are various *BACKEND-FOO* variables
5791 corresponding to the slots of the old structure.
5792 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5793 into the target SBCL has been fixed.
5794 * The chill.lisp system for loading cold code into a running SBCL
5796 * Support for the CMU CL "scavenger hook" extension has been removed.
5797 (It was undocumented and unused in the CMU CL sources that SBCL was
5798 derived from, and stale in sbcl-0.6.1.)
5799 * Various errors in the cross-compiler type system were detected
5800 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5801 (enabling various consistency checks). Many of them were fixed,
5802 but some hard problems remain, so the compiler is back to
5803 running without *TYPE-SYSTEM-INITIALIZED* for now.
5804 * As part of the cross-compiler type system cleanup, I implemented
5805 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5806 * I have started adding UNCROSS calls throughout the type system
5807 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5808 on cross-compiler input files will be able to go away, and various
5810 * CONSTANTP now returns true for quoted forms (as explicitly required
5813 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5815 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5816 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5817 (MAKE-LIST -1). In the longer term, it ought to become true
5818 that declarations are assertions even at SAFETY 1. For now, it's not
5819 quite true even at SAFETY 3, but it's at least more nearly true..
5820 (Note that this change seems to increases the size of the system by
5821 O(5%) and to decrease the speed of the compiler by 20% or more.)
5822 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5823 for the problem of printing dozens of lines of distracting information
5824 about low-level system machinery as part of the top stack frame
5825 on entry to the debugger when an undefined function was called.
5826 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5827 is not reset to COMMON-LISP-USER.
5828 * Compilation of stuff related to dyncount.lisp has been made conditional
5829 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5830 smaller. The various dyncount-related symbols have been moved into
5831 a new "SB-DYNCOUNT" package.
5832 * tty-inspect.lisp has been renamed to inspect.lisp.
5833 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5834 feature has gone away. (When we eventually port to other flavors of
5835 libc and/or Unix, we'll try to make the differences between flavors
5836 invisible at the user level.)
5837 * Various other *FEATURES* tags, and/or their associated conditionals,
5838 have been removed if obsolescent, or given better documentation, or
5839 sometimes given more-mnemonic names.
5841 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5843 * tidied up "make.sh" script
5844 * tidied up system directory structure
5845 * better "clean.sh" behavior
5846 * added doc/FOR-CMUCL-DEVELOPERS
5847 * many many small tweaks to output format, e.g. removing possibly-confusing
5848 trailing #\. character in DESCRIBE-INSTANCE
5849 * (EQUALP #\A 'A) no longer signals an error.
5850 * new hashing code, including EQUALP hashing
5851 * tidied up Lisp initialization and toplevel
5852 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5853 * command line argument processing
5854 * added POSIX-GETENV function to deal with Unix-ish environment variables
5855 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5856 terminating SBCL on EOF
5857 * non-verbose GC by default
5858 * There is no more "sbcl" shell script; the sbcl file is now the C
5859 runtime executable (just like CMU CL).
5860 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5862 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5863 symbol and package information as Lisp native symbol and package objects
5864 instead of strings naming symbols and strings naming packages. This way,
5865 whenever packages are renamed (as in warm init), debug information is
5866 transformed along with everything else.
5867 * tweaked the optimization policy declarations which control the building
5868 of SBCL itself. Now, among other things, the system no longer saves
5869 source location debugging information. (This helps two problems at once
5870 by reducing SBCL size and by keeping SBCL from trying to look for its
5871 sources -- which may not exist -- when reporting errors.)
5872 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5873 debugging and testing purposes
5874 * cleaned up printing, making the printer call PRINT-OBJECT for
5875 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5876 methods, giving nearly-ANSI behavior
5877 * converted almost all special variables to use *FOO* naming convention
5878 * deleted PARSE-TIME functionality, since it can be done portably
5879 * moved some files out of cold init into warm init
5880 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5882 * regularized formatting of source files
5883 * added an install.sh script
5884 * fixed ridiculous memory usage of cross-compiler by making
5885 compiler/alloc.lisp not try to do pooling unless it can hook
5886 itself into the GC of the cross-compilation host. Now the system
5887 builds nicely on my old laptop.
5888 * added :SB-ALLOC in target-features.lisp-expr
5889 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5890 was not implemented)
5891 * re-did condition handling and note reporting in the compiler. Notes
5892 are no longer handled by signalling conditions. Style warnings
5893 and warnings are handled more correctly and reported in such a way
5894 that it's easy to find one or the other in your output (so that you
5895 can e.g. figure out which of many problems caused COMPILE-FILE to
5897 * changed the severity of several compiler warnings from full WARNING
5898 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5899 compiler note reporting so that it doesn't use the condition system
5900 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5901 COMPILE-FILE command)
5902 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5903 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5904 no longer supported, since their implementation was deeply intertwingled
5905 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5906 * removed bogus "support" for compiler macros named (SETF FOO), and
5907 removed the compiler macro for SETF INFO (but only after making a fool
5908 of myself on the cmucl-imp mailing list by posting a bogus patch for
5909 DEFINE-COMPILER-MACRO..)
5910 * Compiled files containing forms which have side effects on the Lisp
5911 reader (such as DEFPACKAGE forms) are now handled more correctly.
5912 (Compiler queuing of top level lambdas has been suppressed by setting
5913 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5914 * deleted various currently-unused source files, e.g. gengc.lisp. They
5915 may be added back at some point e.g. when porting to other architectures,
5916 but until they are it's distracting to distribute them and to try to
5918 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5919 there were so many of them they're just distractions, and UNCROSS is
5920 known to be able to handle the current sources
5921 * moved PROFILE functionality into TRACE, so that it will be clear
5922 how the wrapping and unwrapping of functions when you profile them
5923 interacts with the wrapping and unwrapping of functions when you
5924 trace them. (Actually, the functionality isn't there yet, but at least
5925 the interface specification is there. Hopefully, the functionality will
5926 arrive with some future maintenance release.)
5927 * removed host-oops.lisp
5928 * changed signature of QUIT function to allow UNIX-CODE argument
5929 * fixed READ-SEQUENCE bug
5930 * tweaked verbose GC output so that it looks more like the progress
5931 output that ANSI specifies for functions like LOAD
5932 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5933 * added <http://sbcl.sourceforge.com> to the banner information printed by