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 * bug fix: compiler failed to derive the result-type of MAKE-ARRAY as
10 (AND VECTOR (NOT SIMPLE-ARRAY)) when appropriate. (lp#309130)
11 * bug fix: (THE (VALUES ...)) in LOAD-TIME-VALUE caused a compiler-error.
13 * bug fix: interrupts arriving due to CL:OPEN caused an error.
14 * bug fix: overeager character buffering could cause input to block
15 spuriously when reading from a pipe (lp#643686)
16 * bug fix: more efficient timer expiry should avoid starvation on systems
17 where number of SIGALRMs per second is restricted. (lp#375515)
18 * improvements to the Windows port:
19 ** bug fix: stackoverwriting due to incorrect usage of PeekConsoleInput
20 on Windows. (thanks to Kalyanov Dmitry)
21 ** bug fix: build now works on cygwin with GCC 4.x installed. (thanks to
23 ** bug fix: run-sbcl.sh now works on Cygwin. (thanks to Kalyanov Dmitry)
24 * bug fix: non-unicode builds no longer fail (broken since 1.0.36.15).
25 * bug fix: compile-times no longer scale linearly with the size of
26 quoted lists in source-code. (lp#654289)
28 changes in sbcl-1.0.43 relative to sbcl-1.0.42:
29 * incompatible change: FD-STREAMS no longer participate in the serve-event
30 event-loop by default. (lp#316072)
31 ** In addition to streams created by explicit calls to MAKE-FD-STREAM this
32 affects streams from CL:OPEN.
33 ** Streams from SOCKET-MAKE-STREAM still participate in serve-event by
34 default, but this is liable to change: applications needing serve-event
35 for socket streams should explicitly request it using :SERVE-EVENTS T
37 * enhancement: SB-EXT:WORD type is provided for use with SB-EXT:ATOMIC-INCF
39 * enhancement: CLOS effective method functions and defclass slot typechecking
40 function now have debug names for use in backtraces and profiles.
41 * enhancement: ASDF has been updated to version 2.004. (lp#605260, thanks to
43 * enhancement: symbols are printed using fully qualified names in several
44 error and warning messages which are often associated with package
45 conflicts or mixups (lp#622789, thanks to Attila Lendvai)
46 * bug fix: SB-BSD-SOCKETS:SOCKET-CONNECT was not thread safe. (lp#505497,
47 thanks to Andrew Golding)
48 * bug fix: reading /proc files on Linux works. (lp#425199)
49 * bug fix: DOTIMES accepted literal non-integer reals. (lp#619393, thanks to
51 * bug fix: WRITE-TO-STRING compiler macro binding special variable names,
52 breaking code that tried to write the value of a printer control variable.
53 (lp#581564, thanks to Stas Boukarev)
54 * bug fix: WRITE compiler macro did not handle output stream designators
55 correctly. (lp#598374, thanks to Stas Boukarev)
56 * bug fix: better availability of names of foreign functions in backtraces
57 on Linux. (lp#626962, thanks to Stas Boukarev)
58 * bug fix: scripting and build for Solaris and FreeBSD. (lp#615497,
59 lp#627581, thanks to Josh Elsasser and Jim Wise)
60 * bug fix: build fixes for OpenBSD -current and 4.8 (lp#615489, lp#615492,
61 thanks to Josh Elsasser)
62 * bug fix: using aliases for builtin classes as defmethod specializers
63 without adding DEFTYPEs for them works. (lp#618387)
64 * bug fix: timetravel by getrusage() no longer causes type-errors during GC.
66 * bug fix: legally dynamic-extent lists and vectors used as
67 initialization arguments to MAKE-ARRAY can be stack allocated. (lp#586105)
68 * bug fix: inline-expansion creating references to dead lambda-variables
69 (lp#454681, thanks to Alexey Dejneka)
70 * bug fix: better error message for bogus numerical arguments to RANDOM.
71 (lp#598986, thanks to Stas Boukarev)
72 * bug fix: the compiler occasionally inlined references from incompatible
73 environments occurs. (lp#308951)
74 * bug fix: the compiler threw an error when trying to compile a local
75 function (labels or flet) known to take a specialized complex argument.
76 (not in launchpad, reported by sykopomp in #lispgames)
77 * bug fix: package-locks failed to protect against compile-time effects of
78 DEFUN when the symbol previously had a macro definition. (lp#576637)
79 * bug fix: spurious ignore warnings even given (DECLARE IGNORE) in methods
80 when parameter bindings mutated. (reported by Faré Rideau; lp #611361)
81 * bug fix: workaround for compiler hang in ORDER-UVL-SETS (lp#308914)
82 * bug fix: evaluation in debugger REPL works using the global context when
83 in frames that do not have sufficient debug information.
84 * bug fix: exceeding FD_SETSIZE limit now results in an sensible error
86 * bug fix: (SETF DOCUMENTATION) of a macro works properly. (lp#643958, thanks
88 * bug fix: interrupt taking longer than the requested period caused SLEEP
89 to hang on Darwin. (lp#640516, thanks to Joe Lebroco for the analysis)
91 changes in sbcl-1.0.42 relative to sbcl-1.0.41
93 ** Cross-compilation host is now specified to make.sh using
94 command-line argument --xc-host=<command> instead of a positional
95 argument. (thanks to Daniel Herring)
96 ** Install location can be specified to make.sh using command-line
97 argument --prefix=<path>. (lp#550889s, thanks to Daniel Herring)
98 * optimization: The default implementation of
99 COMPUTE-DISCRIMINATING-FUNCTION does much less wasted work.
100 * enhancement: Explicit memory barrier operations are now available for use
101 by multithreaded code. See documentation for details.
102 * enhancement: Experimental support for threading on Linux/PPC.
103 * bug fix: RENAME-PACKAGE returns the package. (Thanks to Eric Marsden)
104 * bug fix: EXPT signals an error if first argument is a zero and second
105 argument is a floating point zero. (lp#571581, thanks to Roman Marynchak)
106 * bug fix: DEFTYPE signals an error for non-list lambda-lists.
107 (lp#576594, thanks to Roman Marynchak)
108 * bug fix: make ASDF-INSTALL compatible with the now-included ASDF2.
109 (lp#612998, reported by Phil Hargett; patch from Jim Wise)
111 changes in sbcl-1.0.41 relative to sbcl-1.0.40:
112 * optimization: validity of observed keyword initargs to MAKE-INSTANCE is
113 cached, leading to many fewer expensive calls to
114 COMPUTE-APPLICABLE-METHODS.
115 * optimization: in the (unoptimized) general method for MAKE-INSTANCE on a
116 CLASS argument, search for and call an appropriate optimized ctor function
118 * bug fix: WRITE always returns the correct value.
120 changes in sbcl-1.0.40 relative to sbcl-1.0.39:
121 * bug fix: readdir now works on :inode64 darwin builds (lp#592897)
122 * bug fix: Name conflicts between symbols passed as arguments to a single
123 call to IMPORT no longer add multiple symbols with the same name to the
124 package (detectable via DO-SYMBOLS).
125 * bug fix: support building without the dlshim on darwin x86 and x86-64
127 * bug fix: TRACE :ENCAPSULATE NIL now works on ppc/linux.
129 changes in sbcl-1.0.39 relative to sbcl-1.0.38:
130 * bug fix: Backtrace from undefined function on x86 and x86-64 now show
132 * bug fix: linkage-table entries on PPC now no longer overflow their
133 space allocation (potentially causing crashes if they are written out
135 * bug fix: Scrub control stack after scavenging in gencgc on non-x86oid
136 platforms, preventing the GC from seeing stale pointers on the control
137 stack in subsequent GCs (which would, and does, break invariants).
138 * bug fix: 32-bit unicode external formats now work on big-endian systems.
139 * bug fix: Literal characters with code points greater than about 32767
140 now work on PPC UNICODE builds.
141 * bug fix: Any noise left by SSE operations (sqrt and conversions) in the
142 high order bits are explicitly cleared out. In some contrived situations,
143 this could lead to wrong results in mixed real/complex float arithmetic.
144 * bug fix: Fix function/macro redefinition warnings when building with
145 clisp. (lp#576787, thanks to Josh Elsasser)
146 * new platform: experimental support for ppc/openbsd (thanks to Josh
148 * bug fix: Floating-point traps now work on ppc/linux.
150 changes in sbcl-1.0.38 relative to sbcl-1.0.37:
151 * incompatible change: Thread names are now restricted to SIMPLE-STRINGs
152 like for any other thread-related datastructure, MUTEX, etc. (lp#547095)
153 * deprecation: the SB-QUEUE contrib was merged into the SB-CONCURRENCY
154 contrib module. New code should depend on SB-CONCURRENCY, not SB-QUEUE.
155 * deprecation: SB-THEAD:GET-MUTEX was deprecated in favor of
156 SB-THREAD:GRAB-MUTEX.
157 * new contributed module: SB-CONCURRENCY is a new contrib; it's supposed to
158 contain additional data structures and tools for concurrent programming;
159 at the moment it contains a lock-free queue, and a lock-free mailbox
161 * new feature: added SB-THREAD:GRAB-MUTEX; it's like the now deprecated
162 GET-MUTEX but takes &key rather than &optional parameters. Also added
163 :TIMEOUT argument to GRAB-MUTEX on non-sb-lutex platforms like Linux and
165 * new feature: added SB-THREAD:TRY-SEMAPHORE, a non-blocking variant of
166 SB-THREAD:WAIT-ON-SEMAPHORE.
167 * new feature: SB-EXT:ATOMIC-DECF has been added as a companion to
169 * new feature: a CANCEL-DEADLINE is associated with DEADLINE-TIMEOUT
170 conditions to defer the deadline for forever.
171 * enhancement: *STANDARD-OUTPUT*, *STANDARD-INPUT*, and *ERROR-OUTPUT* are
173 * enhancement: errors from NO-APPLICABLE-METHOD and
174 NO-PRIMARY-METHOD now have a RETRY restart available to retry the
175 generic function call.
176 * enhancement: SB-BSD-SOCKET improvements
177 ** sockets and socket streams now have a more informative printed
178 representation based on the corresponding SOCKET-NAME and
180 ** SOCKET-MAKE-STREAM once more supports the :AUTO-CLOSE option.
182 ** SOCKET-CLOSE now accepts :ABORT argument, which is passed on to
183 CL:CLOSE when appropriate, and no longer disassociates the stream
184 from the socket if close failed. (lp#543951)
185 * improvements to the instrumenting profiler
186 ** new feature: report per-function GC overhead. (thanks to John Fremlin)
187 ** optimization: counters no longer use locks for the overflow mode.
188 ** bug fix: whenever a profiling counter wrapped into overflow mode, it
189 incurred an off-by-one miscount.
190 * enhancement: improved MAKE-HASH-TABLE documentation (lp#543473)
191 * enhancement: improved DEFMETHOD pretty-printing.
192 * enhancement: perform range reduction when arguments are too large for
193 x87's transcendentals (instead of returning 0). (lp#327192)
194 * enhancement: eliminate some spurious TYPE-WARNINGs. Should help with
195 some of CL-PPCRE's macros. (lp#570079)
196 * enhancement: our machine code is slightly less hostile to valgrind on
198 * enhancement: up-to-date versions of NetBSD-current are supported. (Thanks
199 to Robert Swindells and Aleksej Saushev)
200 * bug fix: correct restart text for the continuable error in MAKE-PACKAGE.
201 * bug fix: a rare case of startup-time page table corruption.
202 * bug fix: a semaphore with multiple waiters and some of them unwinding due
203 to timeouts could be left in an inconsistent state.
204 * bug fix: fix typo in "Reporting Bugs" section of the manual (lp#520366)
205 * bug fix: misoptimization of multiplication by one in
206 (SB-C::FLOAT-ACCURACY 0) policies.
207 * bug fix: miscounts in SB-PROFILE.
208 * bug fix: Fix lost wakeup bug between SB-THREAD:CONDITION-WAIT and
209 CONDITION-NOTIFY on Linux. See threads "lost wakeup in condition-wait /
210 condition-notify" (Feb 2010) and "Condition-Wait, Deadline handler, waking
211 up itself" (March 2010) for further details.
212 * bug fix: allow forward FIND and POSITION on lists to elide checking :END
213 against length of the list if the element is found before the specified
214 END is reached. (thanks to Alec Berryman, lp#554385)
215 * bug fix: errors signalled during package graph modification no longer
216 block FIND-SYMBOL and FIND-PACKAGE in other threads. (lp#511072)
217 * bug fix: SB-POSIX build was broken when SBCL was compiled without the
218 :SB-DOC feature. (lp#552564)
219 * bug fix: SB-INTROSPECT build issues on GENGC/PPC. (lp#490490)
220 * bug fix: more robust runtime executable path detection. (lp#375549)
221 * bug fix: GCD always returns positive values. (lp#413680)
222 * bug fix: Converting division to multiplication by reciprocal handles
224 * bug fix: We were too eager in eliding range reduction tests on x87.
225 The maximal magnitude is 2^63, not 2^64.
226 * bug fix: Transforms for TRUNCATE don't die when the result is completely
228 * bug fix: Maybe restore buildability on Alpha.
229 * bug fix: READ-BYTE isn't inline anymore, fixing weird streams failures.
231 * bug fix: RANDOM-STATE can be printed readably again.
232 * bug fix: Unreadable objects were sometimes printed like #<\nFoo>.
233 * bug fix: Using EQL with non-constant values of constant type (e.g. EQL
234 types) could result in type mismatches during compilation.
236 changes in sbcl-1.0.37 relative to sbcl-1.0.36:
237 * enhancement: Backtrace from THROW to uncaught tag on x86oids now shows
238 stack frame thrown from.
239 * enhancement: WITH-COMPILATION-UNIT :POLICY allows restricting changes to
240 compiler optimization qualities inside dynamic extent of its body.
241 * enhancement: LOAD-LOGICAL-PATHNAME-TRANSLATIONS can be used to load
242 translations from SYS:SITE;<HOST>.TRANSLATIONS.NEWEST (thanks to Michael
244 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) take advantage of
245 constraint propgation, allowing better compilation eg. when used to
246 access structures with WITH-SLOTS. (lp#520366)
247 * optimization: the compiler is now more aware of the type of the underlying
248 storage vector for multidimensional simple arrays resulting in better code
249 for accessing such arrays.
250 * optimization: passing NIL as the environment argument to TYPEP no longer
251 inhibits optimizing it. (lp#309788)
252 * optimization: more efficient register usage when handling single-float
253 arguments on x86-64. (thanks to Lutz Euler)
254 * optimization: ADJUST-ARRAY and STABLE-SORT on vectors no longer use
255 pre-allocated temporary vectors. (lp#496249)
256 * bug fix: Fix compiler error involving MAKE-ARRAY and IF forms
257 in :INITIAL-CONTENTS. (lp#523612)
258 * bug fix: FUNCTION-LAMBDA-EXPRESSION lost declarations from interpreted
259 functions. (lp#524707)
260 * bug fix: bogus style warnings from certain (SETF SLOT-VALUE) and
261 WITH-SLOTS usages during compilation.
262 * bug fix: SB-C::CLASS-INFO now prints correctly. (lp#514762)
263 * enhancement: Can now build with ud2 instead of int3 as trap instruction on
264 all x86oid platforms with :UD2-BREAKPOINTS target feature.
265 * bug fix: Breakpoints now work when using ud2 instead of int3 as trap
266 instruction (tested on x86oid linux with ud2-breakpoints).
267 * bug fix: slam.sh now works on win32.
268 * bug fix: better differences of numeric types. (lp#309124)
269 * bug fix: arrays declared intersection and union types can have their
270 upgraded element type derived. (lp#316078)
271 * bug fix: SB-SPROF allocation profiling for all threads failed to profile
272 threads started during profiling. (lp#472499)
273 * bug fix: SB-INTROSPECT test failure when building without SB-EVAL feature.
275 * bug fix: SB-CLTL2:DECLARATION-INFORMATION did not take
276 SB-EXT:RESTRICT-COMPILER-POLICY into account. (lp#313337)
277 * bug fix: Comma inside a backquoted array or structure resulted in nonsense
278 values instead of signaling an error. (lp#309093)
279 * bug fix: Spurious unused variable warning in a DEFSTRUCT edge case.
281 * bug fix: More consistent warnings and notes for ignored DYNAMIC-EXTENT
282 declarations (lp#497321)
283 * bug fix: FIND and POSITION on lists did not check sequence bounds properly
284 and failed to detect circular lists (lp#452008)
285 * bug fix: leakage from ~/.asdf-install into the ASDf-INSTALL contrib build
287 * bug fix: LOOP OF-TYPE VECTOR compile-time error. (lp#540186)
288 * bug fix: SIGNAL SB-SYS:INTERACTIVE-INTERRUPT before entering the debugger
289 due to it, so that handlers can run.
290 * bug fix: reparsing undefined types if they have become defined since
292 * bug fix: missing &REST type in a proclamation for a function with both
293 &REST and &KEY in lambda-list caused miscompilation (lp#458354)
294 * bug fix: WHO-CALLS information for source-transformed and compiler-macro
295 expanded calls (lp#542174)
296 * bug fix: more accurate WHO-MACROEXPANDS information; point into rather
297 than just at toplevel form.
299 changes in sbcl-1.0.36 relative to sbcl-1.0.35:
300 * new feature: SB-EXT:TYPEXPAND-1, SB-EXT:TYPEXPAND, and
301 SB-EXT:TYPEXPAND-ALL behave exactly like their MACROEXPAND counterparts
302 but work on type specifiers.
303 * new feature: SB-EXT:DEFINED-TYPE-NAME-P returns whether a symbol is known
304 to name a type specifier.
305 * new feature: SB-EXT:VALID-TYPE-SPECIFIER-P returns whether a given type
306 specifier is valid where "valid" basically means "would be accepted as
307 second argument of TYPEP".
308 * new feature: SB-INTROSPECT:FUNCTION-TYPE takes a function-designator and
309 returns the function's declared, or derived FTYPE.
310 * new feature: SB-POSIX now supports accessing the d_ino member of
311 dirent structures. (Thanks to Philipp Marek and Pierre THEIRRY)
312 * new feature: The function SB-EXT:SEED-RANDOM-STATE has been added to
313 provide for seeding a RANDOM-STATE object with user-provided data or
314 from the operating system's PRNG. Also, (MAKE-RANDOM-STATE T) will
315 attempt to initialize the returned state from the operating system's
316 PRNG where possible. (Thanks to Fare Rideau; launchpad bug lp#310116)
317 * bug fix: Fix SB-SIMPLE-STREAMS:READ-VECTOR to correctly set the
318 FILE-POSITION of the stream being read from. (launchpad bug lp#491087)
319 * bug fix: Fix grammar and style issues for the docstrings of
320 printer-related variables and functions. (Thanks to mon_key; launchpad
322 * bug fix: Fix compilation on chenygc platforms. Thanks to Larry Valkama and
324 * bug fix: SB-THREAD:CONDITION-WAIT sometimes signaled a deadline twice
325 in a row even though a handler defered the deadline long into the
327 * bug fix: A deadline handler was run without interrupts enabled for a
328 deadline signaled within SB-THREAD:CONDITION-WAIT. That could result
329 in infinitely spinning, non-killable threads.
330 * bug fix: Backtrace from internal-errors on x86-64 os x was truncated
331 before reaching the erring stack frame.
332 * bug fix: Fix type derivation for EXPT when raising a fixnum to a
333 real power. (launchpad bug lp#525949)
334 * bug fix: Fix SB-EXT:GENERATION-* accessors for generations > 0 on
335 GENCGC platforms. (launchpad bug lp#529014)
336 * bug fix: More robust checks for invalid DEFMETHOD argument specializers.
337 (launchpad bug lp#525916)
338 * bug fix: Fix building on Darwin when sysctl is not in the user's PATH.
339 (Thanks to Robert Goldman)
341 changes in sbcl-1.0.35 relative to sbcl-1.0.34:
342 * optimization: ROUND with a single single-float or double-float argument
343 is properly inlined when possible.
344 * optimization: Slightly better code is generated for integer<->float
345 conversions and for single-float<->double-float conversions on x86-64.
346 * optimization: SB-ROTATE-BYTE:ROTATE-BYTE now generates more efficient
347 code for 32-bit and 64-bit rotations on x86-64.
348 * bug fix: The install script changes the ownership of directories as well
349 as files for contrib modules using asdf. (thanks to Eugene Ossintsev;
350 launchpad bug lp#508485)
351 * bug fix: TRUNCATE with a single single-float or double-float argument is
352 properly inlined when possible. (launchpad bug lp#489388)
353 * bug fix: Passing a rotation count of zero to SB-ROTATE-BYTE:ROTATE-BYTE
354 no longer causes a compiler error on x86 and ppc.
355 * bug fix: GET-MACRO-CHARACTER bogusly computed its second return value
356 always relative to *READTABLE* rather than the passed argument.
358 changes in sbcl-1.0.34 relative to sbcl-1.0.33:
359 * minor incompatible change: threading support is now enabled by default
361 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
363 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
364 are now caught and reported just like errors during macroexpansion.
365 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
366 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
367 * enhancement: ASDF systems can now depends on SB-QUEUE.
368 * fixes and improvements related to Unicode and external formats:
369 ** bug fix: error handling and restart usage in the ucs-2 external format
371 ** there is now an implementation of the ucs-4 external format.
372 ** the utf-16 and utf-32 external formats are supported.
373 * bug fix: SB-POSIX wrapper for putenv no longer tries to put lisp strings
374 in the environment. setenv() and unsetenv() are also provided. (reported by
375 Fare Rideau; launchpad bug lp#460455)
376 * bug fix: LOAD of both .fasl and .FASL type files now forces fasl-style
377 loading. This ensures sensible errors for .FASL files from other
378 implementations on case-insensitive filesystems. (reported by Willem
379 Broekema; launchpad bug lp#489417)
380 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
381 Willem Broekema; launchpad bug lp#489698).
382 * bug fix: some minor code rearrangements to reenable warning-free building
383 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
384 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
386 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
389 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
390 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
391 * improvement: support O_LARGEFILE access to files larger than 2GB on
392 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
393 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
394 definitions for methods specializing on the passed class itself.
395 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
396 definitions for methods specializing on the passed class itself, or on
398 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
399 about itself during the build (e.g. for M-? in Slime), if this flag is
400 enabled in customize-target-features.lisp. This will increase the core
401 size by about 5-6mb, though, so it's mostly interesting to SBCL
403 * new feature: various GENCGC tuning parameters have been experimentally
404 documented and exported from SB-EXT. See documentation for details.
405 * fixes and improvements related to Unicode and external formats:
406 ** the Unicode character database has been upgraded to the
407 Unicode 5.2 standard, giving names and properties to a number of new
408 characters, and providing a few extra characters with case
410 ** improvement: restarts for providing replacement input/output on coding
411 errors for fd-stream external formats.
412 ** improvement: where :<encoding> is a keyword corresponding to an
413 external format the system supports, it is now possible to specify
414 (:<encoding> :replacement <character>) as an external format which will
415 automatically substitute <character> on encoding or decoding errors for
416 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
418 ** improvement: the file streams underlying the standard streams (such as
419 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
420 which uses the replacement mechanism to handle encoding errors,
421 preventing various infinite error chains and unrecoverable I/O
423 ** minor incompatible change: the utf-8 external format now correctly
424 refuses to encode Lisp characters in the surrogate range (char-codes
425 between #xd800 and #xdfff).
426 ** fix a typo preventing conversion of strings into octet vectors
427 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
429 ** fix a bug in the octet multibyte handling of decoding errors and the
430 USE-VALUE restart. (launchpad bug lp#314939)
431 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
432 restart on fd-stream decoding errors.
433 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
434 error is near the end of file.
435 ** fix a double-error case in unibyte octet conversions, when the first
436 use of USE-VALUE is ignored.
437 ** fix bugs in handling of undefined code points in unibyte encodings.
438 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
439 streams after an UNREAD-CHAR.
440 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
441 object is allocated in a boxed region of dynamic space.
442 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
443 fork with multiple Lisp threads running is made, instead of going
444 ahead with unpredictable consequences. (reported by Leslie Polzer)
445 * bug fix: uses of slot accessors on specialized method parameters within
446 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
447 error while finalizing the class. This fix may cause classes with slot
448 accessors to be finalized later than previously. (reported by Lars Rune
449 Nøstdal; launchpad bug lp#473699)
450 * bug fix: restore buildability on the MIPS platform. (regression from
451 1.0.30.38, reported by Samium Gromoff)
452 * bug fix: inspecting closures is less likely to fail with a type error.
453 * bug fix: no timer starvation when setting the system clock back.
454 (launchpad bug lp#460283)
455 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
456 standard pprint dispatch table as specified by CLHS.
457 * bug fix: give CLISP a hint about a type declaration to enable it to build
458 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
460 * bug fix: correctly dump literal objects in defaulting forms of arglists.
461 (reported by Attila Lendvai; launchpad bug lp#310132)
462 * bug fix: distinguish in type specifiers between arrays that might be
463 complex and arrays that are definitely complex. (launchpad bug lp#309129)
464 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
465 type. (reported by Levente Mészáros; launchpad bug lp#485972)
466 * bug fix: setting the value of a symbol-macro within a method in the
467 presence of type declarations works properly again. (reported by Iban
468 Hatchondo; launchpad bug lp#485019)
470 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
471 * optimization: faster FIND and POSITION on strings of unknown element type
472 in high SPEED policies. (thanks to Karol Swietlicki)
473 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
475 * improvement: better error signalling for bogus parameter specializer names
476 in DEFMETHOD forms (reported by Pluijzer)
477 * bug fix: DELETE-FILE once again works on logical pathnames (regression
479 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
481 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
482 removes previous default initargs (reported by Lars Rune Nøstdal and
484 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
485 lambda-lists. (reported by Peter Seibel)
486 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
487 correctly when starting from an executable core without saved runtime
488 options (reported by Faré Rideau, thanks to Zach Beane)
489 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
490 an optimization note instead. (reported by Martin Cracauer)
491 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
492 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
493 new process group if it doesn't need to share stdin with the sbcl
494 process. (thanks to Leslie Polzer)
495 * bug fix: SATISFIES could be misoptimized to refer to a local function.
496 (reported by Stanislaw Halik)
498 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
499 * improvement: stack allocation is should now be possible in all nested
500 inlining cases: failure to stack allocate when equivalent code is manually
501 open coded is now considered a bug.
502 * improvements related to Unicode:
503 ** the Unicode character database has been upgraded to the
504 Unicode 5.1 standard, giving names and properties to a number of new
505 characters, and providing a few extra characters with case
507 ** the system now recognizes and produces names for Unicode Hangul
509 ** the EBCDIC-US external-format is now supported for octet operations
510 (as well as for stream operations).
511 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
512 be used to output toplevel compile-time effects into a separate .CFASL
514 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
515 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
517 * optimization: TRUNCATE on known single- and double-floats is upto 25%
519 * optimization: division of floating point numbers by constants uses
520 multiplication by reciprocal when an exact reciprocal exists.
521 * optimization: multiplication of single- and double-floats floats by
522 constant two has been optimized.
523 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
524 sufficient type information is available. (thanks to Leslie Polzer)
525 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
526 known structure objects are as efficient as defstruct generated accessors.
527 * optimization: unused vector creation can now be optimized away.
528 * improvement: ASDF systems can now depends on SB-INTROSPECT.
529 * improvement: a STYLE-WARNING is signalled when a generic function
530 clobbers an earlier FTYPE proclamation.
531 * improvement: the compiler is able to track the effective type of
532 generic function across method addition and removal even in the
533 absence of an explicit DEFGENERIC.
534 * improvement: DESCRIBE now reports on symbols naming undefined
535 but assumed or declared function as well.
536 * improvement: recompilation of systems using SB-GROVEL now works
537 (thanks to Leslie Polzer)
538 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
539 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
541 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
542 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
543 well as user defined declaration names.
544 ** VARIABLE-INFORMATION is now aware of alien variables.
545 * improvement: improved address space layout on OpenBSD (thanks to Josh
547 * improvement: pretty-printing of various Lisp forms has been improved
548 (thanks to Tobias Rittweiler)
549 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
550 unused were deleted in safe code. (reported by John Fremlin)
551 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
553 * bug fix: some out-of-line array predicates were missing (reported by
555 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
557 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
558 (thanks to Larry D'Anna)
559 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
560 as well. (reported by Josh Elsasser)
561 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
562 leading to possible attempts to delete the same file twice. See docstring
563 on DELETE-FILE for details. (reported by John Fremlin)
564 * bug fix: DELETE-FILE once again deletes the file named by the pathname
565 designator argument, rather than its truename. (reported by Luis
567 * bug fix: the low-level debugger had 32-bit assumptions and was missing
568 information about some array types. (thanks to Luis Oliveira)
569 * bug fix: moderately complex combinations of inline expansions could
570 be miscompiled if the result was declared to be dynamic extent.
571 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
572 form (+ <variable> <integer>) were miscompiled under certain
574 * bug fix: in some cases no compiler note about failure to stack allocate
575 was emitted, even if the objects were in fact heap allocated.
576 * bug fix: minor violation of "otherwise inaccessible" rule for stack
577 allocation could cause objects users might reasonably expect to
578 be heap allocated to be stack allocated.
579 * bug fix: DESCRIBE signalled an error for generic functions under
580 certain circumstances. (thanks to Leslie Polzer)
581 * bug fix: Fixed spelling of an error message.
583 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
584 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
585 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
586 of SB-THREAD:THREAD-ERROR-THREAD.
587 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
588 * new feature: docstrings for local and anonymous functions are no longer
589 discarded. (thanks to Leslie Polzer)
590 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
591 values in other threads.
592 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
593 about object allocation.
594 * optimization: division of a real float by a complex float is implemented
595 with a specialised code sequence.
596 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
597 keywords is an order of magnitude faster.
598 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
599 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
600 methods, and similarly for non-standard metaclass classes as long as there
601 are no methods additional on MAKE-INSTANCE.
602 * optimization: more efficient type-checks for FIXNUMs when the value
603 is known to be a signed word on x86 and x86-64.
604 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
605 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
607 * optimization: compiler is smarter about delegating argument type checks to
609 * optimization: several character functions are now compiled somewhat more
610 efficiently. (reported by Lynn Quam)
611 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
612 * improvement: less unsafe constant folding in floating point arithmetic,
613 especially for mixed complex/real -float operations.
614 * optimization: constant double and single floats are stored in native
615 unboxed format on x86[-64].
616 * optimization: smarter code for arithmetic operations with constant floats,
617 complex floats, or integers on x86[-64].
618 * optimization: smarter code for conjugate/multiplication of float complexes
619 and abs/negate of floats on x86-64.
620 * optimization: more efficient complex float and real float operations on
622 * improvement: complex float division is slightly more stable.
623 * improvement: DESCRIBE output has been reworked to be easier to read and
624 contains more pertinent information.
625 * improvement: failure to provide requested stack allocation compiler notes
626 provided in all cases (requested stack allocation not happening without a
627 note being issued is now considered a bug.)
628 * bug fix: SB-POSIX exports the documented types and functions
629 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
630 types. (reported by "abhi")
631 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
632 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
633 (thanks to Paul Khuong)
634 * bug fix: looping around HANDLER-CASE could silently consume stack space
635 on each iteration. (reported by "foobar")
636 * bug fix: better error signalling when calls to functions seeking elements
637 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
638 (reported by Tobias Rittweiler)
639 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
640 directory components now works as it used to. (various prolems reported by
641 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
642 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
643 (reported by Elliot Slaughter, thanks to Stas Boukarev)
644 * bug fix: bogus undefined variable warnings from fopcompiled references to
645 global variables. (thanks to Lars Rune Nøstdal)
646 * bug fix: foreign function names should now appear in backtraces on
647 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
648 * bug fix: SETF compiler macro documentation strings are not discarded
650 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
651 (thanks to Alex Plotnick)
652 * bug fix: SLEEP accepts large integer arguments, truncating them to
653 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
654 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
655 * bug fix: compiler notes for expensive slot type checks could be emitted
656 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
658 changes in sbcl-1.0.29 relative to 1.0.28:
659 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
660 https://bugs.launchpad.net/sbcl
661 Bugs can be reported directly there, or by sending email to
662 sbcl-bugs@lists.sourceforge.net
663 (no subscription required.)
664 * minor incompatible change: under weak type checking policy integer
665 types are weakened less aggressively.
666 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
667 allowed to return, which causes SBCL to quit with exit status 0. Previously
668 if the function returned with a small integer return value, that value
669 was accidentally reused as the exit status.
670 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
671 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
672 :HASH-FUNCTION argument. Refer to user manual for details.
673 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
675 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
676 microseconds since the Unix epoch on all platforms.
677 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
678 allows the compiler to safely elide boundedness checks for special
680 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
681 the symbol, prohibits both lexical and dynamic binding. This is mainly an
682 efficiency measure for threaded platforms, but also valueable in
684 * new feature: UNC pathnames are now understood by the system on Windows.
685 * optimization: the compiler uses a specialized version of FILL when the
686 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
688 * optimization: accesses to potentially non-simple arrays where element type
689 is known are 50% faster.
690 * optimization: compiler now generates faster array typechecking code.
691 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
693 * optimization: multidimensional array accesses in the absence of type
694 information regarding array rank are approximately 10% faster due to
695 open coding of ARRAY-RANK.
696 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
697 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
699 * optimization: result of call to VECTOR can now be stack allocated.
700 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
701 as long as the resulting array is one-dimensional and has a known
702 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
703 length of the list matches the known length of the vector does not
704 allocate the list as an intermediate step. Ditto for VECTOR and simple
706 * optimization: MAKE-ARRAY can now stack allocate in the presence of
707 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
708 known element type, and is known to be simple and one dimensional.
709 * improvement: SBCL now emits a compiler note where stack allocation was
710 requested but could not be provided (not in all cases, unfortunately)
711 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
712 * improvement: pretty-printing loop has been implemented properly. (thanks
713 to Tobias Rittweiler)
714 * documentation: CLOS slot typechecing policy has been documented.
715 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
716 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
717 pointer, leading to memory faults. (thanks to Bart Botta)
718 * bug fix: better floating point exception handling on x86/OpenBSD.
719 (thanks to Josh Elsasser)
720 * bug fix: exit status from QUIT when called under --script was lost
721 (reported by Hubert Kauker)
722 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
723 same implementation of FILL to initialize the array, even if a more
724 efficient one was available (reported by Stas Boukarev, thanks to
726 * bug fix: potential miscompilation of array stack allocation on x86 and
727 x86-64. (reported by Time Tossavainen)
728 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
729 result in their subforms being treated as top level forms. (reported by
731 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
732 than a machine register could leave garbage in the high bits of the
733 result register (bug 316325).
734 * bug fix: disable address space randomization Linux/x86-64 as well,
735 not just x86-64. (reported by Ken Olum)
736 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
737 generate incorrect code.
738 * bug fix: #201; type inference for CONS and ARRAY types could derive
739 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
740 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
741 could not be assigned to or bound with PROGV. (reported by Lorenz
743 * bug fix: the value of CL:- in the inspector was the previous expression
744 evaluated rather than the expression being evaluated.
745 * bug fix: constants can no longer be locally declared special.
746 * bug fix: signals delivered to threads started from foreign land (read:
747 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
748 thread by blocking all signals and resignalling.
749 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
750 from :INITFORM, if any.
752 changes in sbcl-1.0.28 relative to 1.0.27:
753 * a number of bugs in cross-compilation have been fixed, with the ultimate
754 result that building under (at least) clisp should be much more reliable.
755 * minor incompatible changes: echo-streams now propagate unread-char to the
756 underlying input stream, and no longer permit unreading more than one
758 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
759 frames, allowing for instance more reliable backtraces.
760 * improvement: the debugger REPL can now reference lexical variables
761 by name directly for code compiled with (DEBUG 3).
762 * improvement: errors from malformed declarations now have better source
763 paths associated with them. (thanks to Tobias Rittweiler)
764 * optimization: faster local calls on x86/x86-64
765 * bug fix: some error messages for out-of-bound array indexes confused the
766 index and the bound. (thanks to Stas Boukarev)
767 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
769 * bug fix: running regressions tests in shells without OSTYPE set now works.
770 (reported by Harald Hanche-Olsen)
771 * bug fix: more robust static space exhaustion signalling from
772 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
773 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
774 docstring away instead of storing it under names such as (LAMBDA (X)).
775 (reported by Leslie Polzer)
776 * bug fix: timers could go off in the wrong order, be delayed indefinitely
777 (thanks to Ole Arndt for the patch)
778 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
779 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
780 potentially-invalid effective methods in its cache.
781 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
782 instances as well (thanks to Paul Khuong)
783 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
784 SB-POSIX:WAITPID and their C equivalents.
785 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
787 changes in sbcl-1.0.27 relative to 1.0.26:
788 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
789 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
790 * improvement: the system either recovers from stack exhaustion or dies
791 properly as opposed to leaving the user uncertain of whether the handler
792 trampled on some random memory next to the stack or having to rely on
793 --lose-on-corruption (which is still a good idea to use in production
794 because stack exhaustion can happen in signal handlers which will likely
796 * bug fix: fix gc related interrupt handling bug on ppc (regression from
797 1.0.25.37, reported by Harald Hanche-Olsen)
798 * bug fix: work around signal delivery bug in darwin (regression from
799 1.0.25.44, reported by Sidney Markowitz)
800 * bug fix: fix ERROR leaking memory (reported by David Thompson)
802 changes in sbcl-1.0.26 relative to 1.0.25:
803 * incompatible change: an interruption (be it a function passed to
804 INTERRUPT-THREAD or a timer function) runs in an environment where
805 interrupts can be enabled. The interruption can use
806 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
807 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
808 potentially running out of stack. Keep in mind that in the absence
809 of WITHOUT-INTERRUPTS some potentially blocking operation such as
810 acquiring a lock can enable interrupts.
811 * incompatible change: GC-OFF and GC-ON are removed, as they were
812 always unsafe. Use WITHOUT-GCING instead.
813 * new feature: runtime option --disable-ldb
814 * new feature: runtime option --lose-on-corruption to die at the
815 slightest hint of possibly non-recoverable errors: running out of
816 memory, stack, alien stack, binding stack, encountering a memory
817 fault, etc. In the absence of --lose-on-corruption a warning is
819 * enhancement: detect binding stack exhaustion
820 * enhancement: detect alien stack exhaustion on x86/x86-64
821 * improvement: generally more stable and reliable interrupt handling
822 * improvement: there is a per thread interruption queue,
823 interruptions are executed in order of arrival
824 * improvement: a repeating timer reschedules itself when the it has
825 finished, but expiration times are spaced equally. If an
826 expiration time is in the past it will trigger after a short grace
827 period that may give a chance to other things to run.
828 * optimization: slightly faster gc on multithreaded builds
829 * optimization: faster WITHOUT-GCING
830 * bug fix: when JOIN-THREAD signals an error, do it when not holding
831 important locks so that the debugger/handler doesn't produce
832 recursive errors or deadlock.
833 * bug fix: real-time signals are not used anymore, so no more
834 hanging when the system wide real-time signal queue gets full.
835 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
837 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
838 * bug fix: fix random memory faults related to interrupts on alpha
839 * bug fix: fix random memory faults related to interrupts on sparc
840 * bug fix: fix deadlocks related to starting threads
841 * bug fix: fix deadlines on locks on futex platforms
842 * bug fix: restore errno in signal handlers
843 * bug fix: fix deadlocks related to hash tables
844 * bug fix: fix deadlocks in pcl
846 changes in sbcl-1.0.25 relative to 1.0.24:
847 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
848 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
849 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
850 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
851 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
852 better name and does not return values so stale on multiprocessor systems.
853 Also, HOLDING-MUTEX-P was added for about the only sane usage of
855 * improvement: unithread builds keep track of MUTEX-VALUE.
856 * improvement: reading from a TWO-WAY-STREAM does not touch the output
857 stream anymore making it thread safe to have a concurrent reader and
858 a writer, for instance, in a pipe.
859 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
860 arguments, making compiler-macros for SETF-functions able to inspect
861 their constant arguments.
862 * improvement: COMPILE-FILE reports times with millisecond accuracy
863 (thanks to Luis Oliveira)
864 * optimization: CHAR-CODE type derivation has been improved, making
865 TYPEP elimination on subtypes of CHARACTER work better. (reported
866 by Tobias Rittweiler, patch by Paul Khuong)
867 * bug fix: setting alien structure fields of type struct by value now
868 computes the right offset for the memory copy.
869 * bug fix: compilation problem involving inlined calls to aliens with
870 result type VOID. (reported by Ken Olum)
871 * bug fix: #235a; sequential inline expasion in different policies no
872 longer reuses the functional from the previous expansion site.
873 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
874 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
875 * bug fix: named ENUMs in multiply-referenced alien record types no longer
876 cause an implied type redefinition. Regression from 1.0.21.29.
877 * bug fix: sign-extension of small signed return values now works with C
878 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
879 * improvements to the Windows port:
880 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
881 Westfold, thanks to Rudi Schlatte)
883 changes in sbcl-1.0.24 relative to 1.0.23:
884 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
885 vector of a multidimensional SIMPLE-ARRAY.
886 * new feature: the system now signals a continuable error if standard
887 readtable modification is attempted. (thanks to Tobias Rittweiler)
888 * new feature: DIRECTORY has been extended with a non-standard keyword
889 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
890 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
891 has been robustified and documented better. (thanks to Robert Goldman)
892 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
893 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
894 * tradeoff: constant FORMAT control strings are now compiled unless
895 SPACE > SPEED (previously only when SPEED > SPACE.)
896 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
898 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
900 * bug fix: lexical type declarations are now correctly reported by
901 SB-CLTL2. (reported by Larry D'Anna)
902 * bug fix: STRING-TO-OCTETS did not handle :START properly when
903 using UTF-8 as external format. (thanks to Luis Oliveira)
904 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
905 are signalled correctly. (thanks to Stas Boukarev)
906 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
907 designator. (thanks to Tobias Rittweiler)
908 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
909 designator, and returns T instead of the function. (thanks to
911 * bug fix: direct superclasses of STANDARD-CLASS and
912 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
913 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
914 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
915 when no class with the named slot yet exists no longer causes a
916 compile-time style-warning.
917 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
918 in safe code. (reported by Didier Verna)
919 * bug fix: #430; nested structure constructors can stack allocate.
920 * bug fix: on some 64-bit platforms dynamic space size was truncated
921 to #xffff0000 bytes. (reported by Benjamin Lambert)
922 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
923 REPL unusable. (reported by Daniel Herring)
924 * bug fix: return values of READ-SEQUENCE did not take :START into
925 account on file streams, regressions since 1.0.12.22. (reported by
926 Thomas Russ, patch by Paul Khuong)
927 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
928 method specializer used to confuse permuation vector optimization.
929 * bug fix: system inserted bogus implicit type declarations for local
930 special variables in DEFMETHOD bodies.
931 * bug fix: #354; duplicated frames in backtraces due to
932 non-tail-call-optimized XEPs to functions with return type NIL
934 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
935 initializes structure object slots according to DEFSTRUCT initforms,
936 and DEFSTRUCT forms :INCLUDEind structure classes defined using
937 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
938 (reported by Bruno Haible and Stephen Wilson)
939 * bug fix: #395; fill-pointer output streams used now support
940 element-type BASE-CHAR as well.
941 * bug fix: compiler error when attempting to derive return value of
942 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
944 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
946 * bug fix: Mac OS X binaries should now be portable between Leopard
949 changes in sbcl-1.0.23 relative to 1.0.22:
950 * enhancement: when disassembling method functions, disassembly
951 for the associated fast function is also produced.
952 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
954 * optimization: printing with *PRINT-PRETTY* true is now more
955 efficient as long as the object being printed doesn't require
956 special handling by the pretty printer.
957 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
958 now interact correctly with type declarations.
959 * partial bug fix: PCL detects infinite recursion during wrapper
960 validation. (thanks to Attila Lendvai)
961 * bug fix: #426; nested function calls are inlined properly.
962 Previously if FOO was an inline function, in calls of the form
963 (FOO (FOO ...)) the outer call was not inlined.
964 * bug fix: long long arguments passed in registers to alien
965 functions on PowerPC Linux and NetBSD are now handled correctly
966 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
968 changes in sbcl-1.0.22 relative to 1.0.21:
969 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
970 for the shared object in the current directory, but passes the native
971 namestring of the designated pathname to the operation system's shared
972 object loading function as-it.
973 * minor incompatible change: --disable-debugger toplevel option now takes
974 effect before processing of initialization files and --eval or --load
976 * new feature: new commandline argument: --script, which supports
977 shebang lines. See documentation for details. (based on work by
979 * new feature: SAVE-LISP-AND-DIE can save current values of
980 --dynamic-space-size and --control-stack-size in the executable core,
981 causing it to skip normal runtime option processing. See documentation
982 for details. (thanks to Zach Beane)
983 * enhancement: inoccous calls to EVAL or generic functions dispatching
984 on subclasses of eg. STREAM no longer cause compiler notes to appear.
985 * enhancement: the system no longer resignals errors from --load and
986 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
987 associated with the original error to be lost. (thanks to Ariel
989 * enhancement: :DONT-SAVE keyword argument has been added to
990 LOAD-SHARED-OBJECT for controlling interaction with
991 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
992 the effects of an earlier LOAD-SHARED-OBJECT call.
993 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
994 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
995 provided. (reported by Cedric St-Jean)
996 * bug fix: circularity handling in the reader did not treat raw
997 structure slots correctly. (reported by Cedric St-Jean)
998 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
999 file descriptors when there were none.
1000 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
1001 was broken. (reporteed by Willem Broekema)
1002 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
1003 pathnames without a directory.
1004 * bug fix: COMPILE-FILE did not warn about undefined variable
1005 references at toplevel, and LOAD of the resulting fasl did
1006 not signal an error.
1007 * bug fix: functions with non-required arguments used to end up with
1008 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
1009 * bug fix: redefining a function with non-required arguments didn't
1010 update the system's knowledge about its call signature properly.
1011 * bug fix: fixed #431; incompatible alien record type redefinitions
1012 are detected and handled. (reported by Neil Haven)
1013 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
1014 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
1016 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
1017 after alien stack frames.
1018 * bug fix: strings passed to mk{s,d}temp lacked null termination
1020 changes in sbcl-1.0.21 relative to 1.0.20:
1021 * new feature: the compiler is able to track the effective type of a
1022 generic function across method addition and removal.
1023 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
1024 appropriately typed structure slots without locking.
1025 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
1026 information like those gathered by TIME using a programming-friendly
1028 * new feature: TIME reports time taken even if the form performs a
1029 non-local transfer of control.
1030 * enhancement: reduced conservativism on GENCGC platforms: on
1031 average 45% less pages pinned (measured from SBCL self build).
1032 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
1033 mutate constant symbols or violate declaimed type of the symbol.
1034 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
1035 sleeping on the semaphore in heavily contested semaphores.
1036 * bug fix: semaphores and condition variables were not interrupt
1038 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
1039 owned by other threads anymore.
1040 * bug fix: FIND on lists called KEY outside the specified
1041 subsequence. (reported by budden)
1042 * bug fix: LOG doesn't use single-float intermediate results when
1043 given mixed integer and double-float arguments, leading to better
1044 precision. (reported by Bob Felts)
1045 * bug fix: LOG with base zero returned values of inconsistent type.
1047 changes in sbcl-1.0.20 relative to 1.0.19:
1048 * minor incompatible change: OPTIMIZE qualities
1049 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
1050 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
1051 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
1052 * documentation: some slot access efficiency guidelines have been
1053 added to the user manual.
1054 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
1055 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
1056 as ASSOC and MEMEBER.
1057 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
1058 to more efficient EQ-comparison versions more often.
1059 * optimization: enhanced derivation of DOLIST iteration variable type
1061 * optimization: constant folding of simple (LIST ...) forms as DOLIST
1063 * optimization: runtime lookup of function definitions can be
1064 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
1065 * optimization: compiler is able to derive the return type of
1066 (AREF (THE STRING X) Y) as being CHARACTER.
1067 * optimization: CLRHASH on empty hash-tables no longer does pointless
1068 work. (thanks to Alec Berryman)
1069 * optimization: REPLACE deftransforms don't punt when :START1 or
1071 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
1072 no longer emits a code deletion note. (reported by Andrew
1074 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
1075 functions with an odd number of &OPTIONAL arguments, a &REST
1076 argument, and one or more &KEY arguments at the call site.
1077 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
1078 arguments appeared at call sites as well.
1079 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
1080 READ-LINE &co can no longer read from them afterwards. (reported
1082 * bug fix: fixed #427: unused local aliens no longer cause compiler
1083 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
1085 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
1086 alien-stack leakage. (reported by Andy Hefner)
1087 * bug fix: PROGV signals an error when an attempt to violate declared
1088 type of a variable or bind a constant is made.
1089 * bug fix: SET signals an error when an attempt to violate declared
1090 type of a variable is made.
1091 * bug fix: restart computation during the execution of a restart
1092 test function no longer causes infinite recursion. (reported by
1094 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
1095 signals an error. (thanks to Pierre Mai)
1096 * bug fix: EXPORT left symbol unexported in conflict situations.
1097 (thanks to Michael Weber)
1098 * bug fix: correctly handle name conflicts involving CL:NIL.
1099 (thanks to Michael Weber)
1100 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
1101 conflicts arising from USEing package with conflicting symbols
1102 correctly. (thanks to Michael Weber)
1104 changes in sbcl-1.0.19 relative to 1.0.18:
1105 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
1106 warnings that go otherwise unhandled will be muffled if they are
1107 of the type that's the value of this variable.
1108 * optimization: stack allocation is slightly more efficient on x86
1110 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
1111 where a raw slot always is initialized using the initform whose
1112 type is not know sufficiently well a compile-time are now compiled
1113 correctly. (reported by John Morrison)
1114 * bug fix: compiler no longer makes erronous assumptions in the
1115 presence of non-foldable SATISFIES types.
1116 * bug fix: stack analysis missed cleanups of dynamic-extent
1117 arguments in non-let-converted calls to local functions.
1118 * improvements to the Windows port:
1119 ** adjusted address spaces for building on both Win32 and
1120 Win64. (thanks for John Connors)
1121 * fixed some bugs revealed by Paul Dietz' test suite:
1122 ** interval arithmetic during type derivation used inexact integer
1123 to single-float coercions.
1124 ** arithmetic operations involving large integers and single
1125 floats give the same results in compiled and interpreted code.
1126 ** deriving the result type of COERCE no longer signals an error
1127 if the derived type of the second argument is a MEMBER type
1128 containing invalid type specifiers.
1129 ** ADJOIN with constant NIL as second argument works correctly.
1130 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
1132 changes in sbcl-1.0.18 relative to 1.0.17:
1133 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
1134 profiles only the current thread.
1135 * minor incompatible change: changes to SYMBOL-VALUE of constants
1136 defined with DEFCONSTANT now signal an error.
1137 * enhancement: SB-SPROF now has support for wallclock profiling,
1138 and is also able to profile specific threads. REPORT output
1139 has also additional sorting options.
1140 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
1142 * optimization: structure allocation has been improved
1143 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
1144 ** out of line constructors are ~10% faster.
1145 ** inline constructors are ~15% faster.
1146 ** inline constructors are capable of dynamic extent allocation
1147 (generally on x86 and x86-64, in some cases on other platforms
1149 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
1151 * optimization: file compiler is now able to coalesce non-circular
1152 lists, non-base strings, and bit-vectors. Additionally, constants
1153 are never referenced through SYMBOL-VALUE at runtime.
1154 * optimization: code defining methods on PRINT-OBJECT (and other
1155 generic functions in the COMMON-LISP package) now loads faster.
1156 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
1158 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
1159 fasl is now deleted. (reported by Attila Lendvai)
1160 * bug fix: READ-LINE always returned NIL for the last line in files.
1161 (reported by Yoshinori Tahara)
1162 * bug fix: more accurate disassembly annotations of foreign function
1163 calls. (thanks to Andy Hefner)
1164 * bug fix: trimming non-simple strings and non-string string
1165 designators when the there is nothing to trim works properly.
1166 (thanks to James Knight)
1167 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
1168 (thanks to Travis Cross)
1169 * fixed some bugs revealed by Paul Dietz' test suite:
1170 ** NIL is a valid function name (regression at 1.0.13.38)
1171 ** FILL on lists was missing its return value (regression at 1.0.12.27)
1172 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1173 fill pointers properly (regression at 1.0.12.23)
1174 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1175 displacement indices properly (regression at 1.0.12.23)
1177 changes in sbcl-1.0.17 relative to 1.0.16:
1178 * temporary regression: user code can no longer allocate closure
1179 variable storage on stack, due to bug 419 without explicitly
1180 requesting it. Please consult sbcl-devel for advice if you need to
1181 use this feature in the meanwhile.
1182 * new feature: runtime argument --control-stack-size can be used to
1183 adjust thread default control stack size.
1184 * enhancement: improved TIME output
1185 ** all times are reported using the measured accuracy (milliseconds
1186 for real and GC times, microseconds for everything else.)
1187 ** processor cycle counts on x86 and x86-64.
1188 ** interpreted forms are counted for both evaluator modes.
1189 ** number of lambdas converted by the compiler is reported.
1190 ** CPU percentage report (computed from real and total run time.)
1191 ** more comprehensive run time reporting, using a condenced format
1192 ** interperted form, lambda, and page fault counts are omitted
1194 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
1196 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
1197 * optimization: two argument forms of LAST are upto ~10% faster
1198 in normal SPEED policies.
1199 * optimization: NCONC no longer needs to heap cons its &REST list
1200 in normal SPEED policies.
1201 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
1202 to Sidney Markowitz)
1203 * bug fix: UNION and NUNION work with :TEST-NOT once more,
1204 regression since 1.0.9.1. (thanks to Eric Marsden)
1205 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
1206 since 1.0.15.36. (thanks to Paul Khuong)
1207 * bug fix: LAST when always returned the whole list when given a bignum
1208 as the second argument.
1209 * bug fix: dynamic extent allocation of nested lists and vectors
1210 could leak to otherwise accessible parts.
1211 * bug fix: invalid optimization of heap-allocated alien variable
1213 * bug fix: fasl header checking is less vulnerable to different
1214 platform word lengths.
1215 * bug fix: more correct assembler syntax for GNU binutils
1216 2.18.50.0.4 support. (thanks to Marijn Schouten)
1217 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
1219 * bug fix: the fopcompiler can handle LOCALLY forms (with no
1220 declarations) successfully. (reported by Attila Lendvai)
1222 changes in sbcl-1.0.16 relative to 1.0.15:
1223 * minor incompatible change: revert the changes to sb-posix's error
1224 signaling added in 1.0.14.
1225 * minor incompatible change: change PROBE-FILE back to returning
1226 NIL whenever we can't get a truename, as was the case before 1.0.14.
1227 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
1228 inherits from ERROR instead of just CONDITION.
1229 * new feature: SB-INTROSPECT can provide source locations for instances
1230 as well. (thanks to Tobias Rittweiler)
1231 * optimization: binding special variables now generates smaller code
1232 on threaded platforms.
1233 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
1234 and cases where no :TEST is given but the compiler can infer that
1235 the element to search is of type (OR FIXNUM (NOT NUMBER)).
1236 * optimization: better LOGNOT on fixnums.
1237 * optimization: modular arithmetic for a particular requested width
1238 is implemented using a tagged representation unless a better
1239 representation is available.
1240 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
1241 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
1242 for wellformedness and returns a specialized vector. (reported by
1243 Francois-Rene Rideau)
1244 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
1245 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
1246 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
1247 unscheduling timer at the same time as another timer fires.
1248 * bug fix: don't reschedule timers for dead threads.
1249 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
1250 * bug fix: copying output from RUN-PROGRAM to a stream signalled
1251 bogus errors if select() was interrupted.
1252 * enhancement: add support for fcntl's struct flock to SB-POSIX.
1254 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
1255 * enhancement: cleaner backtraces for interactive interrupts, as
1256 well as other cases where the interesting frames used to be
1257 obscured by interrupt handling frames.
1258 * enhancement: untracing a whole package using (UNTRACE "FOO") is
1259 now supported, and tracing a whole package using (TRACE "FOO") now
1260 traces SETF-functions as well.
1261 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
1262 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
1263 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
1264 * unreadably printed representation of hash-tables now includes
1266 * bug fix: partially fixed #188: type propagation from assignments
1267 is now more efficient.
1268 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
1269 'DOUBLE-FLOAT) are not flushable.
1270 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
1271 full address of the object, and none of the tag bits.
1272 * bug fix: readably printing hash-tables now respects other printer
1273 control variables. (reported by Cedric St-Jean)
1274 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
1275 keyword with MAKE-HASH-TABLE.
1276 * bug fix: export SB-POSIX:MKSTEMP.
1277 * bug fix: SORT was not interrupt safe.
1278 * bug fix: XREF accounts for the last node of each basic-block as
1280 * bug fix: MAKE-INSTANCE optimizations interacted badly with
1281 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
1282 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
1283 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
1284 create obsolete instances in the case of redefinition or
1285 obsoletion of a superclass. (thanks to Andy Hefner)
1286 * bug fix: Support for the Alpha architecture has been revived; it had
1287 suffered somewhat from lack of maintenance since sbcl-1.0.
1288 * improvements to the Windows port:
1289 ** The system detects the codepage to use at startup, instead of
1290 using the value from the saved core. (thanks to Kei Suzuki)
1292 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
1293 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
1294 (see documentation for details.)
1295 * revived support for OpenBSD (contributed by Josh Elsasser)
1296 * partially fixed bug #108: ROOM no longer suffers from occasional
1297 (AVER (SAP= CURRENT END)) failures.
1298 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
1299 bodies no longer cause a WARNING to be signalled. (reported by
1301 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
1302 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
1303 the bug in TRUNCATE.) (reported by Michael Weber)
1304 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
1305 no samples. (reported by Andy Hefner)
1306 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
1307 no longer appear as (NIL ...) frames in backtraces.
1308 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
1309 is now actually exported from SB-EXT as documented. (reported by
1311 * bug fix: sb-aclrepl now correctly understands how to inspect
1312 single-floats on 64-bit platforms where single-floats are not boxed.
1313 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
1314 yet been finalized. (reported by Levente Meszaros)
1315 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
1316 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
1317 SBCL-specific optimize qualities.
1319 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
1320 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
1321 an executable in the search path, and does so in the child
1322 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
1323 been removed; it can be found in the CVS history, for somebody who
1324 needs that search behavior (see the manual).
1325 * minor incompatible change: compiler policy re. weakening type
1326 checks has changed: now type checks are weakened only if SAFETY < 2
1328 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
1329 unparsing of directory pathnames as files. Analogously,
1330 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
1331 filename to parse into a directory pathname.
1332 * enhancement: implicit generic function creation now signals a
1333 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
1334 users can bind handlers for and muffle around calls to LOAD.
1335 * enhancement: RUN-PROGRAM allows unicode arguments and environments
1336 to be used (using the default stream external format), and allows
1337 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
1338 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
1339 strings and vectors whose element-type or simplicity is not fully
1340 known at compile-time.
1341 * optimization: STRING-TRIM and related functions no longer allocate
1342 a new string when no trimming needs to be performed. These functions
1343 are also faster than before when the input string has been declared
1345 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
1346 * optimization: READ-LINE is significantly faster for files containing
1348 * optimization: non-open coded uses of character comparison operators
1349 (e.g. char=) no longer cons when called with more than one parameter
1350 on platforms supporting dynamic-extent allocation.
1351 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
1352 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
1353 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
1355 * bug fix: some sequence functions elided bounds checking when
1357 * bug fix: too liberal weakening of union-type checks when SPEED >
1359 * bug fix: more bogus fixnum declarations in ROOM implementation
1362 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
1363 * new feature: MAKE-HASH-TABLE now experimentally accepts a
1364 :SYNCHRONIZED argument, which makes the hash-table safe for
1365 concurrent accesses (but not iteration.) See also:
1366 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1367 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1368 * optimization: CONCATENATE on strings is an order of magnitude faster
1369 in code compiled with (> SPEED SPACE).
1370 * optimization: SUBSEQ is ~50% faster on lists.
1371 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1372 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1373 END is smaller then START.
1374 * bug fix: SB-PROFILE will no longer report extra consing for nested
1375 calls to profiled functions.
1376 * bug fix: ROOM implementation had bogus fixnum declarations which
1377 could cause type-errors when calling ROOM on large images.
1378 * bug fix: if file compilation is aborted, the partial fasl is now
1379 deleted, and COMPILE-FILE returns NIL as the primary value.
1380 * bug fix: number of thread safety issues relating to SBCL's internal
1381 hash-table usage have been fixed.
1382 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1383 be returned from its body when the values were being returned
1384 using unknown-values return convection and the W-P-O was wrapped
1385 inside an UNWIND-PROTECT.
1386 * bug fix: sb-posix should now compile again under Windows, enabling
1387 slime to work again.
1389 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1390 * incompatible change: hash-table accessor functions are no longer
1391 automatically protected by locks. Concurrent accesses on the same hash-table
1392 from multiple threads can give inconsistent results or even corrupt the
1393 hash-table completely. Multi-threaded applications should do their own
1394 locking at the correct granularity. In the current implementation it is
1395 still safe to have multiple readers access the same table, but it's not
1396 guaranteed that this property will be maintained in future releases.
1397 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1398 and will signal an error at runtime.
1399 * enhancement: SB-THREAD package now exports a semaphore interface.
1400 * enhancement: CONS can now stack-allocate on x86 and
1401 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1403 * enhancement: nested lists can now be stack allocated on
1404 platforms providing stack allocation support.
1405 * enhancement: dynamic-extent support has been extended to support
1406 cases where there are multiple possible sources for the stack
1408 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1409 if the mutex is uncontested on Linux.
1410 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1411 as specified by the CLHS. (thanks to Tobias Rittweiler)
1412 * bug fix: NaN comparison now works on x86-64.
1413 * bug fix: CLOSE :ABORT T on a stream with pending output now
1415 * bug fix: instances of non-standard metaclasses using standard
1416 instance structure protocol sometimes missed the slot type checks
1418 * bug fix: known functions can be inlined in high-debug code.
1420 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1421 * minor incompatible change: the MSI installer on Windows no longer
1422 associates .lisp and .fasl files with the installed SBCL.
1423 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1424 on Windows. (thanks to Luis Oliviera)
1425 * new feature: SB-CLTL2 contrib module now implements
1426 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1427 * optimization: scavenging weak pointers is now more efficient,
1428 requiring O(1) instead of O(N) per weak pointer to identify
1429 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1430 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1431 method bodies using constant slot-names when the first argument is
1432 a specializer parameter for the method.
1433 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1434 names, when the class of the instance is a direct instance
1435 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1436 as slow as the constant slot-name case.)
1437 * optimization: member type construction is now O(N) instead
1439 * optimization: UNION and NUNION are now O(N+M) for large
1440 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1442 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1443 now more readable in environments like Slime which display it.
1444 (thanks to Tobias C. Rittweiler)
1445 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1446 was able to infer the correct argument type for the object on which
1447 the CAS operation was being performed.
1448 * bug fix: copy propagation interfered with parallel assignment
1449 semantics in local calls. (reported by Paul Khuong)
1450 * bug fix: the signed modular fixnum shift compiled to wrong code on
1451 x86 and x86-64. (spotted by a slight modification to some of
1454 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1455 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1456 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1457 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1459 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1460 and generic function now print the object identity for anonymous
1461 instances. (thanks to Attila Lendvai)
1462 * enhancement: as an extension to MOP, SBCL now supports SETF of
1463 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1464 (thanks to Attila Lendvai)
1465 * enhancement: sb-cover has an alternate annotation mode which puts
1466 the form annotations at the car of the form, instead of annotating
1467 it completely. The benefit of this mode is that it shows better
1468 which forms were instrumented by the compiler.
1469 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1470 by printer control variables. (reported by Dan Corkill)
1471 * bug fix: system leaked memory when delayed output was performed by
1472 the OS in smaller chunks then expected. (thanks to David Smith)
1473 * bug fix: system leaked memory when file streams were not closed
1475 * bug fix: large objects written to slow streams that were modified
1476 after the write could end up with the modified state written to
1477 the underlying file descriptor.
1478 * bug fix: multiple threads operating in parallel on the same stream
1479 could cause buffer-overflows.
1480 * bug fix: source location information is stored correctly
1481 (broken since 1.0.6). This bug would generally show up as the
1482 Slime debugger highlighting the wrong form when the sldb-show-source
1484 * bug fix: Branch forms are again annotated as branches in the sb-cover
1486 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1487 on x86-64. (reported by Gregory Vanuxem)
1488 * bug fix: EQUALP could return wrong results for structures with raw
1489 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1490 word). (reported by Vjacheslav Fyodorov)
1491 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1494 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1495 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1496 atomic compare-and-swap operations on threaded platforms.
1497 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1498 allows assigning a global minimum value to optimization qualities
1499 (overriding proclamations and declarations).
1500 * enhancement: closed over variables can be stack-allocated on x86
1502 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1504 * optimization: slot definition lookup is now O(1). This speeds up
1505 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1506 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1507 * optimization: ASSOC and MEMBER can now be open-coded for all
1508 combinations of keyword arguments when second argument is constant
1509 and SPEED >= SPACE. In other cases a specialized version is
1511 * bug fix: using obsoleted structure instances with TYPEP and
1512 generic functions now signals a sensible error.
1513 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1514 (reported by Kristoffer Kvello)
1515 * bug fix: backtrace construction is now more careful when making
1516 lisp-objects from pointers on the stack, to avoid creating bogus
1517 objects that can be seen by the GC.
1518 * bug fix: defaulting of values in contexts expecting more than 7
1519 variables now works on x86-64. (reported by Christopher Laux)
1520 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1522 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1523 as the property-list of a symbol.
1524 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1525 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1526 situations requested, are once again file-compileable. (reported
1529 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1530 * MOP improvement: support for user-defined subclasses of
1531 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1532 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1533 the expansion of DEFMETHOD, is responsible for generating a form
1534 which creates a list of specializers when evaluated. Additional
1535 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1536 debugging and introspective support.
1537 * minor incompatible change: the (unsupported) spinlock interface
1538 has changed: free spinlock now has the value NIL, and a held spinlock
1539 has the owning thread as its value.
1540 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1541 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1543 * enhancement: name of a socket-stream is now "a socket" instead of
1544 "a constant string".
1545 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1546 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1547 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1548 for generic functions. (Thanks to Tobias C. Rittweiler)
1549 * optimization: bignum printing speed has been improved by 20-40%
1550 (depending on the bignum size.)
1551 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1553 * bug fix: the cache used by the CLOS to store precomputed effective
1554 methods, slot offsets, and constant return values is now thread and
1556 * bug fix: generic function dispatch function updating is now thread
1557 and interrupt safe (in the sense that the known issues have been
1559 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1560 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1561 * bug fix: asdf systems with dependencies to the SB-POSIX or
1562 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1563 * bug fix: interrupt safety of applicable method computation has been
1566 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1567 * new contrib: sb-cover, an experimental code coverage tool, is included
1568 as a contrib module.
1569 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1570 significantly faster.
1571 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1572 produce more efficient code on x86-64 if the compiler can determine
1573 that (+ INDEX <constant>) does not require a bounds check and FOO
1574 has an element type at least 8 bits wide.
1575 * enhancement: a new, experimental synchronous timeout facility is
1576 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1577 * enhancement: when a symbol name conflict error arises, the
1578 conflicting symbols are always printed with a package prefix.
1579 (thanks to Kevin Reid)
1580 * enhancement: stepping is now once again supported on the SPARC and
1581 MIPS platforms. (It is also now more likely to work on CheneyGC
1583 * enhancement: sb-sprof can now also track and report accurate call
1585 * bug fixes: the treatment of non-standard subclasses of
1586 SB-MOP:SPECIALIZER is more correct.
1587 * incompatible change: PURIFY no longer copies the data from the
1588 dynamic space into the static and read-only spaces on platforms
1589 that use the generational garbage collector
1590 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1592 * bug fix: GC race condition occasionally resulting in crashes with
1593 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1595 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1597 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1598 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1599 * bug fix: an asynchronous interrupt could previously leave the
1600 system running with GC inhibited.
1601 * bug fix: a DECLARE form evaluated at top-level now causes an error
1602 rather than silently (or verbosely) returning NIL.
1603 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1604 result in a error being signalled due to source form lookup errors.
1605 (reported by Peter Graves)
1607 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1608 * incompatible change: removed writer methods for host-ent-name,
1609 host-ent-addresses -- changing the values did not update the DNS
1611 * minor incompatible change: changed experimental JOIN-THREAD interface
1612 * documentation: the manual now lists reader and writer methods
1613 in class slot documentation sections. (thanks to Richard M Kreuter)
1614 * documentation: unwinding from asyncronous events has been
1615 documented as unsafe.
1616 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1617 in multithreaded application code.
1618 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1619 platforms. (thanks to James Anderson for the optimization hint)
1620 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1622 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1623 variants no longer cons.
1624 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1625 their NOT- variants no longer cons.
1626 * optimization: EQUAL hash tables no longer use SXHASH for objects
1627 of all data types, but instead use an EQL hash for types for which
1628 EQUAL is the same as EQL.
1629 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1630 are significantly faster.
1631 * optimization: new STRING-OUTPUT-STREAM implementation is much
1632 faster for large outputs and conses 30% less on average.
1633 * enhancement: XREF information is now collected for references made
1634 to global variables using SYMBOL-VALUE and a constant argument.
1635 * enhancement: SIGINT now causes a specific condition
1636 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1637 * bug fix: FILE-POSITION used to signal an error in some cases where
1638 ANSI requires it to return NIL.
1639 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1640 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1641 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1643 * bug fix: bad type declaration in the CLOS implementation has
1644 been fixed. (reported by James Anderson)
1645 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1646 (thanks to Lutz Euler)
1647 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1648 required by the CLHS. (reported by Samium Gromoff)
1649 * bug fix: dead unbound variable references now signal an error.
1650 (reported by Marco Monteiro)
1651 * bug fix: / with an unused value was being deleted in safe code.
1652 (thanks to Marco Monteiro and Kevin Reid)
1653 * bug fix: number of characters that can be written onto a single
1654 line in a file is unlimited.
1655 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1656 been fixed by inhibiting interrupts for when GC is disabled.
1657 * bug fix: some interrupt safety issues with GETHASH, (SETF
1658 GETHASH), CLRHASH and REMHASH have been fixed.
1659 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1660 specifier no longer causes infinite recursion.
1661 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1662 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1663 * bug fix: an &environment argument with the correct variable information
1664 is passed to macros that are expanded during byte compilation.
1665 (reported by Samium Gromoff)
1666 * bug fix: a logic error was causing the hash function for bignums to
1667 have excessive amounts of collisions. (reported by Faré Rideau)
1668 * bug fix: modifying the contents of an array could change the return
1669 value of SXHASH on that array, which is only allowed for strings
1670 and bit vectors. (bug introduced in 0.9.16)
1671 * improvement: the x86-64/darwin port now passes all tests (except
1672 for the debugger tests) but should still be considered
1673 experimental until this is fixed.
1674 * improvement: a style-warning is signaled for CASE (etc) clauses with
1675 duplicate keys. (thanks to Kevin Reid)
1676 * improvement: macos/x86 and macos/x86-64 now use mach exception
1677 handlers for dealing with illegal instructions (for trapping and
1678 error handling) and memory protection violations (for GC).
1680 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1681 * new platform: experimental support for x86-64/darwin (MacOS).
1682 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1683 and getnameinfo sockets functions are used instead of gethostbyaddr
1684 and gethostbyname, on platforms where the newer functions are available.
1685 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1687 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1688 don't print the startup banner, but behave as if --noinform was passed
1689 as a command line argument. (thanks to Kevin Reid)
1690 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1691 * optimization: code using alien values with undeclared types is much faster.
1692 * optimization: the compiler is now able to open code SEARCH in more cases.
1693 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1694 * bug fix: using standardized COMMON-LISP special variables as loop
1695 variables no longer signals bogus package lock violations. (reported
1697 * bug fix: declaring local loop variables to be of a range-limited type
1698 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1699 (reported by Andras Simon)
1700 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1701 bugs remain on x86-64.)
1702 * bug fix: the #= and ## reader macros now interact reasonably with
1703 funcallable instances.
1704 * bug fix: type-checks for function arguments were compiled using the
1705 compiler policy settings of the wrong lexical environment (in 1.0.2
1707 * bug fix: SHADOW accepts characters as string designators, as required
1708 by the spec (thanks to Eric Marsden)
1709 * bug fix: fixed GC safety issues when foreign functions are called with
1710 non-base strings as arguments
1711 * bug fix: more consistent error output for fatal-compiler-warnings, like
1713 * bug fix: functions with proper names were showing up as NIL in
1714 backtraces (reported by Edi Weitz, regression in 1.0.2)
1716 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1717 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1718 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1719 produce more efficient code on the x86 if the compiler can determine
1720 that (+ INDEX <constant>) does not require a bounds check and FOO
1721 has an element type at least 8 bits wide.
1722 * bug fix: references to undefined variables are handled the same way
1723 in toplevel forms as in the normal compiler
1724 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1726 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1727 (thanks to Jon Buller)
1728 * improvement: faster compilation times for complex functions
1729 * improvement: added readlink support to SB-POSIX (thanks to Richard
1732 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1733 * improvement: experimental support for mach exception handling on
1734 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1736 * improvement: support for GBK external format.
1737 (thanks to Chun Tian (binghe))
1738 * improvement: the debugger now displays variables that have been closed
1739 over, in code compiled with (DEBUG 2) or higher
1740 * improvement: support for executable cores on NetBSD (thanks to
1742 * new feature: added a RESTART-FRAME debugger command
1743 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1744 be used to provide an implementation for FILE-POSITION on Gray streams
1745 (thanks to Eric Marsden)
1746 * new feature: add syslog, openlog and closelog support to SB-POSIX
1747 (thanks to Richard Kreuter)
1748 * optimization: the function call overhead in code compiled with
1749 a high DEBUG optimization setting is significantly
1750 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1751 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1752 (thanks to Tony Martinez)
1753 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1754 evaluated once (reported by Kevin Reid)
1755 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1756 works again (thanks to Kilian Sprotte)
1757 * bug fix: an error is signaled for tagbodies with duplicate tags
1758 (thanks to Stephen Wilson)
1759 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1760 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1761 stack frames from alien callbacks.
1762 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1763 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1764 65k pages (thanks to David Woodhouse)
1765 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1767 changes in sbcl-1.0.1 relative to sbcl-1.0:
1768 * new platform: FreeBSD/x86-64, including support for threading.
1769 * new feature: the compiler stores cross-referencing information
1770 about function calls (who-calls), macroexpansion (who-macroexpands)
1771 and special variables (who-binds, who-sets, who-references) for code
1772 compiled with (< SPACE 3). This information is available through the
1773 sb-introspect contrib.
1774 * new feature: users may subclass SEQUENCE, and have instances of
1775 these classes interoperate with standard Common Lisp functions if
1776 a number of methods are defined. (This feature is experimental
1777 and the interface subject to change based on feedback from SBCL
1778 users and the general community)
1779 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1780 x86/x86-64, rather than the previous fixed depth of 8
1781 * improvement: another pthread back-end of mutex "pthread-futex".
1782 * bug fix: non-ascii command-line arguments are processed correctly
1783 (thanks to Yaroslav Kavenchuk)
1784 * bug fix: non-required arguments were not passed correctly when a method
1785 defined using DEFMETHOD was called from a mop-generated method using
1786 CALL-NEXT-METHOD (reported by Pascal Costanza)
1787 * bug fix: recursion is now permitted in accessors through
1788 SLOT-UNBOUND. (reported by Pascal Costanza)
1789 * bug fix: an error was signaled at startup if the HOME environment
1790 variable was defined, but had an empty value (reported by Peter Van Eynde)
1791 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1792 signal a PROGRAM-ERROR, not a BUG.
1793 * bug fix: SB-POSIX:READDIR works when built with large file support.
1794 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1796 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1797 for better automatic indentation support. (thanks to Matt Pillsbury)
1798 * optimization: loading generic functions no longer takes O(n^2) time,
1799 proportional to the amount of methods in the generic function
1800 (reported by Todd Sabin and Jeremy Brown)
1801 * optimization: the FIND and POSITION family of sequence functions
1802 are significantly faster on arrays whose element types have been
1804 * improvements to the Windows port:
1805 ** Intermittent heap corruption problems have been fixed. (thanks
1806 to Alastair Bridgewater)
1807 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1809 ** Lisp is able to unwind foreign exception frames from alien
1810 callbacks. (thanks to Alastair Bridgewater)
1812 changes in sbcl-1.0 relative to sbcl-0.9.18:
1813 * improvement: experimental support for threading on FreeBSD/x86.
1814 (thanks to NIIMI Satoshi)
1815 * improvement: runtime option --dynamic-space-size can be used
1816 to set the size of the dynamic space reserved on startup.
1817 * improvement: floating point modes in effect are now saved in
1818 core, and restored on startup.
1819 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1820 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1821 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1822 threads are not running after *SAVE-HOOKS* have run.
1823 * improvement: writes to CLOS instance slots are type-checked in code
1824 compiled with (SAFETY 3)
1825 * improvement: floating-point exception handling on FreeBSD (thanks to
1827 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1828 (thanks to Zach Beane)
1829 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1831 * improvement: added support for the Shift-JIS external format.
1832 (contributed by NIIMI Satoshi)
1833 * improvement: callbacks are supported on Linux/PPC. (thanks to
1835 * bug fix: compiler bug triggered by a (non-standard) VALUES
1836 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1837 * bug fix: file compiler no longer confuses validated and already
1838 dumped structurres. (reported by Kaersten Poeck)
1839 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1840 fill-pointer signals a type-error as required. (thanks to
1842 * bug fix: disassemly of funcallable instances works.
1843 * bug fix: single stepping on PPC.
1844 * bug fix: fix thread-safety problems in the type system (generally
1845 manifesting as nonsensical errors like "STRING is a bad type specifier
1846 for sequences" or "The value 1 is not of type FIXNUM")
1847 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1848 (reported by Josip Gracin)
1849 * bug fix: an error is signaled for attempts to displace arrays with
1850 incompatible element types (thanks to Mario Mommer)
1851 * bug fix: more correct handling of wide characters in debug info
1852 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1853 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1854 and don't cause extra consing
1855 * optimization: MAP and MAP-INTO are significantly faster on vectors
1856 whose elements types have been declared.
1857 * Improvements to SB-SPROF:
1858 ** Support for allocation profiling
1859 ** Reduced profiling overhead, especially for long profiling runs
1860 * Improvements to the Windows port:
1861 ** floating point exceptions are now reported correctly.
1862 ** stack exhaustion detection works partially.
1863 ** more accurate GET-INTERNAL-REAL-TIME.
1864 ** better RUN-PROGRAM behaviour in case of errors in the child
1866 ** PROBE-FILE now simplifies pathnames correctly.
1867 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1868 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1869 1st 1970 (although time timezone data will be unreliable).
1870 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1872 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1873 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1874 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1876 * bug fix: two potential GC deadlocks affecting threaded builds.
1877 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1878 returns T (reported by Anton Kazennikov)
1879 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1880 with non-variable places
1881 * bug fix: remove a race condition in the setting of
1882 funcallable-instance functions, this should make threaded CLOS
1883 code more stable against memory faults.
1884 * bug fix: corruption of specials when unbinding is interrupted by an
1885 asynchronous unwind (reported by Hannu Koivisto)
1886 * improvement: the debugger will now also display local variables that
1887 are only used once, for code compiled with a DEBUG optimization quality
1890 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1891 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1892 * incompatible change: External-format support for FFI calls. The
1893 SB-ALIEN:C-STRING no longer implies an ASCII
1894 external-format. Instead the string is subject to external-format
1895 conversion. Additionally return values of type SB-ALIEN:C-STRING
1896 are by default of type (SIMPLE-ARRAY CHARACTER), not
1897 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1898 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1899 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1900 * incompatible change: SB-EXT package no longer contains the
1901 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1902 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1904 * incompatible change: the single-stepper is no longer available
1905 on Alpha, Mips and Sparc platforms.
1906 * minor incompatible change: the direct superclasses of
1907 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1908 not (STANDARD-OBJECT FUNCTION). This makes the
1909 class-precedence-lists of GENERIC-FUNCTION and
1910 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1912 * minor incompatible change: the default stream external format on
1913 non-unicode SBCL is no longer detected from the locale, but is always
1915 * new feature: Add a version of evaluator that uses an interpreter instead
1916 of the compiler. EVAL still uses the compiler by default, to switch it
1917 to use the interpreter, set the value of the variable
1918 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1919 * minor incompatible change: the single-stepper REPL has been merged
1920 with the normal debugger (see the "Stepping" heading of the debugger help
1921 for more details). The debugger command STEP will no longer switch
1922 to the single-stepper REPL.
1923 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1924 for a type now works.
1925 * bug fix: loading of default sysinit file works. (thanks to Leonid
1927 * bug fix: better detection of circularities in the file-compiler.
1928 (reported by Marco Monteiro)
1929 * bug fix: the CL pathname functions now work with files that have
1930 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1931 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1932 systems with Unix98 pty semantics.
1933 * bug fix: ASDF-INSTALL will now work with bsd tar.
1934 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1936 * bug fix: timers expiring in dead threads no longer cause a
1937 type-error (reported by Paul "Nonny Mouse").
1938 * bug fix: thanks to more lightweight single-stepper instrumentation,
1939 code compiled with (DEBUG 3) will compile and execute significantly faster,
1940 and will have more accurate type-inferencing than before
1941 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1942 whose bindings are modified
1943 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1944 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1945 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1946 CL:SOFTWARE-VERSION with proper version
1948 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1949 * feature: implemented the READER-METHOD-CLASS and
1950 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1951 as specified by AMOP.
1952 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1954 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1955 (thanks to Lutz Euler)
1956 * optimization: hashing of general arrays and vectors has been
1957 improved. (reported by Any Fingerhut)
1958 * enhancement: SB-INTROSPECT is now able to find definitions of
1959 profiled functions. (thanks to Troels Henriksen)
1960 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1961 as well, allowing compiler-macros for SETF-functions to expand.
1962 * enhancement: step-instrumentation no longer wraps known
1963 single-value functions with multiple-value context, allowing
1964 better type inference.
1965 * fixed bug #337: use of MAKE-METHOD in method combination now works
1966 even in the presence of user-defined method classes. (reported by
1967 Bruno Haible and Pascal Costanza)
1968 * fixed bug #339(c): if there are applicable methods not part of any
1969 long-form method-combination group, call INVALID-METHOD-ERROR.
1970 (reported by Bruno Haible)
1971 * fixed bug #361: the :FUNCTION initarg in the protocol for
1972 initialization of methods can now be used to override
1973 internally-produced optimized functions. (reported by Bruno
1975 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1976 system-provided lambda expression no longer cause warnings about
1977 unbound #:|pv-table| symbols.
1978 * bug fix: improved the handling of type declarations and the
1979 detection of violations for keyword arguments with non-constant
1981 * bug fix: potentially erronous calls to PATHNAME and
1982 MERGE-PATHNAMES were being flushed in some cases.
1983 (reported by Richard Kreuter)
1984 * bug fix: compiled calls to TYPEP were mishandling obsolete
1985 instances. (reported by James Bielman and Attila Lendvai)
1986 * bug fix: format strings with ~> without matching ~< no longer
1987 trigger an AVER, but signal an understandable error instead.
1988 (reported by Antonio Martinez)
1989 * bug fix: specifying an output-file without a pathname-type for
1990 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1991 of input-file instead of "fasl". (reported by Robert Dodier)
1992 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1993 (reported by James Y Knight).
1994 * bug fix: compiler-macros-function did not consider the environment
1995 argument for shadowing by local functions.
1996 * bug fix: compiler-macros expansion was inhibited by local INLINE
1998 * bug fix: inline expansions of known functions were subject to
1999 step-instrumentation in high debug policies, leading to problems
2000 with type-inference.
2001 * bug fix: compiler failed to differentiate between different CONS
2002 types in some cases.
2003 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
2004 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
2005 * bug fix: #368: incorrect use of expressed vs. upgraded array
2007 * bug fix: #306a: more precise unions of array types.
2008 * thread-safety improvements:
2009 ** CONDITION-WAIT could return early on Linux, if the thread was
2010 interrupted and subsequently continued with SIGCONT.
2011 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
2013 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
2014 * added support for the ucs-2 external format. (contributed by Ivan
2016 * minor incompatible change: pretty printing of objects of type
2017 (cons symbol) is, in the default pprint-dispatch-table, now
2018 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
2020 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
2021 called later in a class's lifetime, possibly as late as when the
2022 first instance of the class is created. Previously,
2023 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
2024 class became finalizeable.
2025 * fixed bug: FILE-POSITION sometimes returned inconsistent results
2026 for multibyte external-format streams. (thanks to "vbzoli")
2027 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
2028 with :ALLOCATION :CLASS inherited from superclasses of the
2030 * fixed bug: anonymous classes can now be created using the :NAME
2031 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
2032 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
2033 * fixed bug: core-files saved with :EXECUTABLE T can again be
2034 executed when SBCL_HOME isn't set. (reported by James Knight)
2035 * fixed bug: toplevel LOCALLY forms with declarations could
2036 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
2037 * fixed bug: printing from several different threads using different
2038 values of *print-case* could cause invalid output, due to
2039 some internal special variables of the printer not being bound
2040 thread-locally (reported by Max Mikhanosha)
2041 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
2042 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
2043 (reported by Pascal Costanza)
2044 * fixed bug: FINALIZE-INHERITANCE is called from
2045 REINITIALIZE-INSTANCE on classes when the class has previously
2046 been finalized, as required by AMOP.
2047 * minor code generation optimizations:
2048 ** better register allocation in CLOS dispatching functions
2049 ** overflow detection when coercing signed bytes to fixnums on x86-64
2050 ** is now implemented with one IMUL instruction instead of three shifts
2051 ** more efficient bit-vector access on x86 and x86-64
2052 ** more efficient access to raw structure slots on x86 and x86-64
2053 * fixed some bugs revealed by Paul Dietz' test suite:
2054 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
2055 return its argument.
2057 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
2058 * feature: thread support on Solaris/x86, and experimental thread support
2060 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
2062 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
2063 default method for SLOT-UNBOUND.
2064 * minor incompatible change: prevent the user from specializing the
2065 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
2066 somewhat counter to the intent of the protocol, I (CSR) think, and
2067 additionally it just doesn't work in SBCL as currently
2068 implemented, thanks to optimizations (that are always valid for
2069 the other three SLOT-VALUEish functions, but not for the setter).
2070 * minor incompatible change: the :SB-LDB feature is now enabled by
2071 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
2072 the low-level debugger.
2073 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
2074 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
2075 down to an SB-SYS:FD-STREAM, allowing direct communication in
2076 more cases. Temporary files and pipes are still used as a fallback
2078 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
2079 on the non-reentrant C functions localtime(3) and gmtime(3).
2080 * bug fix: global optimization policy was not visible in LOCALLY and
2082 * bug fix: class objects can be used as specializers in methods.
2083 (reported by Pascal Costanza)
2084 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
2085 with a type error. (reported by blitz_ on #lisp)
2086 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
2087 failed with a type error. (reported by Pascal Bourguignon)
2088 * bug fix: merging pathnames against defaults with :DIRECTORY
2089 starting with '(:RELATIVE :BACK) should preserve the :BACK.
2090 (reported by James Y Knight)
2091 * bug fix: saving large (>2GB) cores on x86-64 now works
2092 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
2093 constant index and a (simple-array (signed-byte 32)) array
2094 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
2095 error (patch by Robert J. Macomber)
2096 * feature: TIME now displays the amount of run-time spent in GC
2097 * bug fix: The debugger now does a better job of respecting
2098 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
2099 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
2100 workaround for bug 403.)
2101 * bug fix: floating point exception handling now works on FreeBSD
2102 (thanks to NIIMI Satoshi)
2103 * fixed some bugs revealed by Paul Dietz' test suite:
2104 ** MISC.641: LET-conversion were not supposed to work in late
2106 * improvements to the Win32/x86 port:
2107 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
2108 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
2110 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
2111 * new feature: source path information is generated for macro-expansion
2112 errors for use in IDE's like Slime (thanks to Helmut Eller)
2113 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
2114 * bug fix: compiler does not loop forever on an invalid type in
2116 * improvement: compilation of most CLOS applications is significantly
2118 * optimization: added a limited bytecode compiler for simple toplevel
2119 forms, speeding up compilation and FASL loading
2120 * bug fix: the statistical profiler now properly distinguishes anonymous
2123 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
2124 * minor incompatible change: in sbcl-0.9.11 (but not earlier
2125 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
2126 into subclasses of the finalized class. Now user calls to
2127 FINALIZE-INHERITANCE finalize just the one class, and calls by the
2128 system return before any subclasses are finalized.
2129 * minor incompatible change: The reader no longer ignores errors
2130 regarding non-existent packages in #+ and #- feature tests.
2131 * new feature: command line options --no-sysinit, --no-userinit to
2132 inhibit loading the corresponding init files
2133 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
2134 for sending data through UDP sockets (thanks to François-René Rideau)
2135 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
2136 error is signalled instead (thanks to François-René Rideau)
2137 * minor incompatible change: Error signalling behaviour of lexical
2138 operations violating package locks has changed slightly. Refer to
2139 documentation on package locks for details.
2140 * enhancement: EVAL can process IF-expressions without resorting to the
2142 * bug fix: Some operations involving SIN, COS, or TAN weren't being
2143 constant-folded properly due to missing out-of-line functions.
2144 (reported by Mika Pihlajamäki)
2145 * bug fix: LISTEN sometimes returned T even in cases where no data was
2146 immediately available from the stream
2147 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
2148 were reversed. (reported by Levente Mészáros)
2149 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
2150 when given a vector displaced to an adjustable array. (reported
2152 * bug fix: derivation of float boundaries from numbers outside the
2153 appropriate float range (reported by John Wiseman)
2154 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
2156 * bug fix: precomputing cache entries for generic functions with
2157 some subclasses of specializers as yet invalid does not attempt to
2158 fill a cache line with a negative offset. (reported by Levente
2160 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
2161 allow condition class objects as arguments to DOCUMENTATION and
2162 (SETF DOCUMENTATION); only find and set documentation for
2163 structure names for the STRUCTURE doc-type. (suggested by Gary
2165 * improvements to the Win32/x86 port:
2166 ** (user-homedir-pathname) and default initialization file
2167 locations now know about the user's "Documents and Settings"
2168 directory (thanks to Yaroslav Kavenchuk)
2169 ** run-program is implemented (thanks to Mike Thomas)
2170 ** sockets support (thanks to Timothy Ritchey)
2171 ** better backtrace support (thanks to Alastair Bridgewater)
2172 ** sb-grovel supported
2173 ** asdf-install and sb-posix work somewhat
2174 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
2175 * improvements to the Solaris/x86 port:
2176 ** works on Solaris 11/Solaris Express
2177 ** floating-point exception handling support
2178 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
2179 * fixed some bugs revealed by Paul Dietz' test suite:
2180 ** REMOVE-METHOD returns its generic function argument even when
2181 the method is not one of the generic functions' methods.
2182 ** objects declared as MEMBER types can be admissible arguments to
2183 structure accessors.
2184 ** printing characters should simply be printed by the FORMAT ~:C
2186 ** compiler failure when compiling functions with hairy constant
2187 defaults for optional parameters.
2188 ** compiler produces wrong code when MAYBE-INLINE-expanding a
2189 function, which is already optimized.
2191 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
2192 * new platform: experimental support for SBCL x86/Darwin, including
2193 MacOS X 10.4.5 on Intel.
2194 * new feature: Unicode character names are now known to the system
2195 (through CHAR-NAME and NAME-CHAR).
2196 * new feature: the filesystem location of SBCL's core file is
2197 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
2198 * minor incompatible change: the contrib modules SB-POSIX and
2199 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
2200 this change is to make it easier to distribute
2201 location-independent binaries.
2202 * bug fix: as implied by AMOP, standardized classes no longer have
2203 slots named by external symbols of public packages. (reported by
2205 * optimization: calling functions via constant symbols -- as in
2206 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
2207 via the function object as in (FUNCALL #'FOO).
2208 * optimization: on x86 and x86-64, the calling convention for the
2209 case of dealing with an unknown number of values has been altered
2210 to be friendlier to the prediction heuristics implemented,
2211 particularly with reference to CALL and RET pairing. (thanks to
2212 Alastair Bridgewater)
2213 * optimization: on x86, the code for access to thread-local symbol
2214 values and binding/unbinding of thread-local symbols is smaller.
2215 (thanks to Alastair Bridgewater)
2216 * enhancement: CONSTANTP is now able to determine constantness of
2217 more complex forms, including calls to constant-foldable standardized
2218 functions and some special forms beyond QUOTE.
2219 * fixed bug: occasional GC crashes on Solaris/x86
2220 * optimization: x86-64 supports stack allocation of results of simple
2221 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
2222 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
2223 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
2224 (thanks to James Knight)
2225 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
2226 by the standard (thanks to Stephen Compall)
2228 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
2229 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
2230 be used for bundling the runtime and the core file into one
2231 executable binary. This feature is not currently supported on all SBCL
2232 platforms. (thanks to James Bielman and NIIMI Satoshi)
2233 * new feature: a generational or ephemeral garbage collector is now
2234 the default on the PowerPC platform (both Linux and Darwin). The
2235 old Cheney (stop and copy) collector is a build-time option.
2236 (thanks to Cyrus Harmon, after Raymond Toy)
2237 * minor incompatible change: the method by which SBCL finds its
2238 contributed modules has changed; it no longer relies on symbolic
2239 links from an $SBCL_HOME/systems directory, but searches directly
2240 in the subdirectories of $SBCL_HOME.
2241 * enhancement: the dynamic heap size on the Linux/PPC platform is
2242 markedly larger, even using the older Cheney garbage collector.
2243 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
2244 certain complicated string types. (reported by Gary King)
2245 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
2246 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
2247 * fixed bug: DOCUMENTATION on structures no longer signals an
2248 error if no structure type of the right name exists. (reported by
2250 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
2251 applicable non-standard primary or :AROUND method on
2252 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
2253 no longer cause unbound variable errors. (reported by Kevin Reid)
2254 * fixed bug: metacircle resolution in cases where methods have slots
2255 added before the slots from STANDARD-METHOD. (reported by Jean
2257 * fixed bug: the Power PC assembly code for calling into lisp
2258 sometimes computed its return address wrongly (depending on the
2259 15th bit of the address, fixed at link time). This used to
2260 manifest itself in a segmentation violation while building PCL.
2261 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
2262 many others over the years)
2263 * fixed bug: ROOM no longer reports silly numbers for stack usage if
2264 the stack pointer (treated as a signed integer) is negative.
2265 (thanks to Peter van Eynde)
2267 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
2268 * new platform: experimental support for the Windows operating
2269 system has been added. (thanks to Alastair Bridgewater)
2270 * fixed several bugs in and robustified the PPC FFI (including
2271 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
2272 * bug fix: allow non-simple string symbol names (reported by Paul
2274 * bug fix: interrupt handling on NetBSD (thanks to Richard M
2276 * bug fix: saving a core corrupted callbacks on x86/x86-64
2277 * bug fix: closed a loophole in metacircularity detection and
2278 grounding in the PCL implementation of CLOS.
2279 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
2280 (reported by tomppa on #lisp)
2281 * bug fix: clear the direction flag on Lisp -> C transitions, as
2282 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
2283 (reported by Andrej Grozin and Hendrik Maryns)
2284 * optimization: major improvements to GC efficiency on GENCGC platforms
2285 * optimization: faster implementation of EQUAL
2286 * optimization: emit more efficient opcodes for some common
2287 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
2288 * optimization: several other minor code-generation improvements on x86-64
2289 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
2291 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
2292 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2293 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
2294 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
2295 by ANSI as a generic function, and for consistency (SETF
2296 GENERIC-FUNCTION-NAME) is treated likewise.
2297 * fixed bug #233.b: make constraint propagation notice when a variable
2298 value is changed after it is referenced but before it is used
2299 * fixed bug #296: no more arbitrary behaviour from filenames with
2300 odd characters as --load arguments. (reported by Adam Warner)
2301 * fixed bug #390: :CHARACTER-SET pathname components now work as
2302 expected. (reported by Tim Daly Jr)
2303 * fixed bug #391: complicated :TYPE intersections in slot
2304 definitions no longer cause an error in PCL internals.
2305 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
2306 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
2307 returning the number of octets which would be written to the
2308 file-stream. (thanks to Robert J. Macomber)
2309 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
2310 arguments to the internal stream functions. (thanks to David
2312 * bug fix: stack exhaustion is now handled on the x86/SunOS
2313 platform. (thanks to Daisuke Homma)
2314 * bug fix: more accurate ROOM results on GENCGC platforms
2315 * optimization: improved type inference for arithmetic-for
2316 index variables in LOOP
2317 * optimization: faster floating-point SQRT on x86-64
2318 * fixed some bugs revealed by Paul Dietz' test suite:
2319 ** DOCUMENTATION returns NIL instead of "" for method combinations
2320 that don't have a docstring
2322 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
2323 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2324 GENERIC-FUNCTION-NAME) are no longer generic functions, and
2325 therefore are not customizeable by user code (as seems to be at
2326 least permitted and maybe required by AMOP). As a consolation,
2327 however, the SBCL implementation of these functions now calls
2328 REINITIALIZE-INSTANCE as specified by AMOP.
2329 * bug fix: slot-definition documentation is propagated as per ANSI
2330 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
2331 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
2332 argument, as do eslotd initialization methods. (from Pascal
2333 Costanza's "Closer" project)
2334 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
2335 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
2337 * bug fix: it is now possible to have more than one subclass of
2338 STANDARD-GENERIC-FUNCTION without causing stack overflow.
2339 (reported by Bruno Haible, Pascal Costanza and others)
2340 * bug fix: the dependent update protocol now works for generic
2341 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
2342 and Pascal Costanza)
2343 * bug fix: condition-class instances corresponding to
2344 DEFINE-CONDITION forms are now created eagerly. (reported by
2345 Kalle Olavi Niemitalo on comp.lang.lisp)
2346 * bug fix: floating point printing is more accurate in some
2347 circumstances. (thanks to Simon Alexander)
2348 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
2349 merged with *DEFAULT-PATHNAME-DEFAULTS*.
2350 * bug fix: callbacks on OS X now preserve stack-alignment.
2351 (thanks to Cyrus Harmon)
2352 * enhancement: the x86-64 disassembler is much better at
2353 disassembling SSE instructions. (thanks to Lutz Euler)
2354 * enhancement: CHECK-TYPE on a variable now allows type-inference
2355 (thanks to Vincent Arkesteijn)
2356 * optimization: improved performance of EUC-JP external format.
2357 (thanks to NIIMI Satoshi)
2358 * optimization: performance improvements to IO on file streams of
2359 :ELEMENT-TYPE CHARACTER
2360 * optimization: much faster memory allocation on x86-64
2361 * optimization: faster garbage collections (latency and throughput)
2363 * optimization: faster bignum subtraction and division on x86 and x86-64
2364 (thanks to Lutz Euler)
2365 * optimization: more accurate type-inference for loops that use a
2366 floating point index variable or a negative step.
2368 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2369 * bug fix: add a workaround to SBCL looping infinitely at startup on
2370 Linux kernels with apparently buggy implementations of personality().
2371 (thanks to Svein Ove Aas)
2372 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2373 (thanks to David Lichteblau)
2374 * bug fix: local bindings shadow global symbol macros for
2375 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2376 * optimization: non-open coded uses of numeric comparison operators
2377 (e.g. >) no longer cons when called with more than one parameter
2378 on platforms supporting dynamic-extent allocation.
2379 * enhancement: saving cores with foreign code loaded is now supported
2380 on MIPS/Linux in addition to the previously supported platforms.
2381 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2382 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2383 * bug fix: division by zero in sb-sprof when no samples were collected
2384 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2385 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2386 (thanks to Alexander Kjeldaas)
2387 * new feature: ignoring the return values of destructive functions
2388 like NREVERSE signals a compile-time style-warning.
2389 (thanks to Kevin Reid)
2390 * new feature: the alignment of alien structure fields can be
2391 explicitly specified. (thanks to Cyrus Harmon)
2392 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2393 and dump core on SIGQUIT
2395 ** incompatible change: threads do not inherit values of specials
2396 from their parents (see manual)
2397 ** bug fix: threads stacks belonging to dead threads are freed by the
2398 next exiting thread, no need to gc to collect thread stacks anymore
2399 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2400 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2401 ** bug fix: dynamic variable and thread start related gc lossage
2402 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2404 * fixed some bugs revealed by Paul Dietz' test suite:
2405 ** SUBTYPEP is slightly more accurate on heinously complicated
2406 CONS types where some of the members have uncertain (in the
2407 NIL, NIL sense) type relationships to each other.
2409 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2410 no longer purified unless :PURIFY T is explicitly specified.
2411 ** Non-purified cores are significantly smaller than before
2413 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2414 * new feature: timers based on Zach Beane's excellent timer package
2415 * added support for the following external formats: koi8-u,
2416 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2417 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2418 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2419 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2420 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2421 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2422 * incompatible change: a threaded SBCL will no longer revert to
2423 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2424 * bug fix: interrupts are disabled until startup is complete; no
2425 more sigsegvs when receiving a signal to soon
2426 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2428 * bug fix: add a workaround for the memory randomization features in
2429 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2430 workaround will only be in effect on systems with the proc filesystem
2432 * bug fix: printing objects of type HASH-TABLE signals a
2433 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2435 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2436 have been running for over 50 days. (reported by Gilbert Baumann)
2437 * bug fix: the logic for getting names of functions gets less
2438 confused when confronded with alternate-metaclass
2439 funcallable-instances. (reported by Cyrus Harmon)
2440 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2441 consistent, even on internal alternate-metaclass objects.
2442 * bug fix: flush closure information collected by physical
2443 environment analysis prepass before the main pass. (bug reported
2445 * bug fix: compiler pack phase does not modify a hashtable, which is
2446 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2447 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2448 may be expanded. (reported by Alexander <malishev> on c.l.l)
2449 * bug fix: no more style-warnings when compiling code using LAST
2450 under high-SPEED low-SPACE settings. (reported by David Wallin
2452 * fixed some incompatibilities between SBCL's MOP and the MOP
2454 ** the METAOBJECT class is now implemented;
2455 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2456 STANDARD-OBJECT, as required;
2457 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2458 now compatible (as required by VALIDATE-SUPERCLASS); there
2459 remains a consistency requirement over the presence or absence
2460 of the FUNCTION class in the superclass list of finalized
2461 classes; see the manual for more details;
2462 ** the :AROUND method for COMPUTE-SLOTS on
2463 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2464 requested slot ordering.
2466 ** bug fix: parent thread now can be gc'ed even with a live
2468 ** bug fix: binding a special with PROGV to no value is not
2469 the same as that symbol not having been bound (thanks to
2471 ** bug fix: binding specials is thread safe (thanks to
2473 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2474 and SB-SPROF do not die with 'no handler for signal XX in
2475 interrupt_handle_now(..)' anymore
2476 ** bug fix: WITH-TIMEOUT works with multiple running threads
2477 * fixed some bugs revealed by Paul Dietz' test suite:
2478 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2479 the :method-class keyword argument.
2481 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2482 * new port: the Solaris operating system on x86 processors is now
2483 mostly supported, though some rough edges in the environment
2484 remain. (thanks to Daisuke Homma)
2485 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2486 space for its heap. (thanks to Thiemo Seufer)
2487 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2488 closures. (thanks to Thiemo Seufer)
2489 * minor incompatible change: eof selects abort in the debugger.
2490 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2491 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2493 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2494 deprecated, and will go away in a future revision of SBCL.
2495 * minor incompatible change: GC-ON and GC-OFF are no longer
2496 implemented with a counter, it does not matter how many times gc
2497 is switched on or off
2498 * bug fix: discriminating functions for generic function classes
2499 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2500 make invalid assumptions about method precedence order. (reported
2502 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2503 platforms. (thanks to Thiemo Seufer)
2504 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2505 * bug fix: correct pathname printing with printer escaping is on.
2506 (thanks to Kevin Reid)
2507 * bug fix: complex VOP definitions in "user-space" no longer trigger
2508 package locks. (reported by Zach Beane)
2509 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2510 for funcallable-instance functions, and is no different from
2512 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2513 characters in the current readtable. (reported by Nicholas Neuss)
2514 * bug fix: SUBTYPEP on various CONS types returns more a more
2515 accurate acknowledgment of its certainty.
2516 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2517 lists in some cases. This partially fixes bug 384.
2518 * flush all standard streams before prompting in the REPL and the
2520 * bug fix: signal handling and triggering gc do not conflict
2521 directly anymore, in particular a high frequency sb-sprof does
2522 not prevent gc from running
2523 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2524 approximation for timezone and DST information between the
2525 universal time epoch and the smallest negative 32-bit time_t.
2526 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2527 year 1899 so long as the encoded time is non-negative.
2528 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2529 the correct answer for non-constant rotate amounts when performing
2530 an inline 32-bit rotation.
2532 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2533 there is only one thread in the session
2534 ** bug fix: memory leak for streams created in one thread and
2535 written to in another
2536 ** bug fix: lockup when compiled with gcc4
2537 ** bug fix: race that allows the gc to be triggered when gc is
2539 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2540 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2542 ** bug fix: finished threads are gc'ed properly
2543 * fixed some bugs revealed by Paul Dietz' test suite:
2544 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2545 has arguments with a different set of applicable methods from
2546 the orignal arguments.
2547 ** The type error thrown by MAP now has a correct expected-type
2549 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2550 name a compiled function.
2551 ** Three MISC tests where a large bignum was improperly coerced to
2552 a float that couldn't represent that bignum during type
2553 derivation were fixed.
2554 ** SUBTYPEP can now handle types involving the negation of a
2555 list-form FUNCTION type.
2556 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2557 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2558 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2560 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2561 * New feature: Experimental support for bivalent streams: streams
2562 opened with :element-type :default now allow character and binary
2563 (unsigned-byte 8) I/O
2564 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2565 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2567 * bug fix: no more highly sporadic "couldn't check whether ~S is
2568 readable" when reading a stream and an interrupt hits in the middle
2569 of a select system call
2570 * compiler better recognizes complex arrays (reported by Eduardo
2572 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2573 for variables on SBCL built with linkage-tables. (reported by Luis
2575 * various error reporting improvements.
2576 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2577 (Thanks to Thiemo Seufer)
2578 * bug fix: nested reader invokations work correctly
2579 * bug fix: it is possible to have simultaneous references to foreign
2580 code and foreign data with the same name.
2582 ** added x86-64 support
2583 ** incompatible change: the threading api now works with thread
2584 objects instead of thread ids
2585 ** bug fix: threads are protected from signals and interruption when
2586 starting up or going down
2587 ** bug fix: a race where an exiting thread could lose its stack to gc
2588 ** bug fix: don't halt on infinite error in threads if possible
2589 ** fixed numerous gc deadlocks introduced in the pthread merge
2590 ** bug fix: fixed thread safety issues in read and print
2591 ** bug fix: debugger doesn't hang on session lock if interrupted at
2592 an inappropriate moment
2593 ** bug fix: run-program is now thread safe(r)
2594 ** bug fix: inner with-recursive-lock no longer releases the mutex
2595 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2596 * fixed some bugs revealed by Paul Dietz' test suite:
2597 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2598 EXPECTED-TYPE slots filled.
2599 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2601 ** MAKE-CONDITION accepts classes as type-designators.
2602 ** COMPILE may never return NIL.
2603 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2604 range before calling Unix time functions
2606 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2607 * numerous signal handling fixes to increase stability
2608 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2609 * minor incompatible change: we now correctly canonize default
2610 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2611 as per the MOP, rather than the historical (INITARG INITFUNCTION
2612 INITFORM). (reported by Bruno Haible)
2613 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2614 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2615 for more information.
2616 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2617 pathname is a directory pathname.
2618 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2619 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2621 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2622 :METACLASS argument in addition to a class name. (reported by
2623 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2624 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2625 (reported by Bruno Haible; patch from Gerd Moellmann)
2626 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2628 * bug fix: more cleanups to the floating point exception handling on
2629 x86-64 (thanks to James Knight)
2630 * bug fix: the compiler does not try to constant fold calls of
2631 COERCE and BIT-* functions when they return freshly constructed
2632 objects. (reported by Nikodemus Siivola and Paul Dietz)
2633 * optimization: improved the MIPS versions of generic (in the
2634 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2635 * optimization: direct conversion of (unsigned-byte 32) to floats on
2636 the PowerPC platform.
2637 * optimization: structure instances with raw slots now use less
2638 memory, and probably show better memory locality. (thanks to
2640 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2641 are treated as requests for stack allocation on the x86-64,
2642 Alpha-32, PPC and SPARC platforms.
2643 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2644 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2645 (thanks to Tony Martinez)
2647 ** gcing a dead thread can no longer lead to lockups
2648 ** threads block signals until they are set up properly
2649 ** errno is no longer shared by threads
2650 ** interrupt-thread restores the eflags register on x86
2651 ** fixed some lockups due to gc/thread interaction
2652 * fixed some bugs revealed by Paul Dietz' test suite:
2653 ** invalid dotted lists no longer raise a read error when
2654 *READ-SUPPRESS* is T
2655 ** signal an error if a symbol that names a declaration is used
2656 as the name of a type, or vice versa
2657 ** allow using the (declare (typespec var*)) abbreviation for
2658 (declare (type typespec var*)) with all type specifiers
2659 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2660 ** always return NIL from PPRINT-POP when OBJECT is NIL
2661 ** don't signal errors when pretty-printing malformed LABELS,
2662 FLET or MACROLET forms
2663 ** declarations in a DOLIST body are also in scope for the
2665 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2667 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2670 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2671 * fixed cross-compiler leakages that prevented building a 32-bit
2672 target with a 64-bit host compiler.
2673 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2674 opened with :IF-EXISTS :OVERWRITE.
2675 * fixed bug 281: error for an invalid qualifier in a short-form method
2676 combination method is not signalled until the faulty method is called.
2677 * bug fix: iteration variable type inferrer failed to deal with open
2678 intervals. (reported by Alan Shields)
2679 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2680 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2681 or more runtime options were provided to the sbcl binary.
2682 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2683 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2684 * optimization: there's now a fast-path for fixnum arguments in the
2685 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2686 * optimization: the code generated on x86-64 is more compact thanks
2687 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2688 * bug fix: floating-point exception handling now partly works on
2689 x86-64. (Thanks to James Knight)
2690 * improvement to the MIPS backend from Thiemo Seufer: C-style
2691 64-bit long long arguments and return values to alien functions
2693 * fixed some bugs revealed by Paul Dietz' test suite:
2694 ** the type-error signalled from WARN has a filled-in DATUM slot.
2695 ** the type-error required when a stream is not associated with
2696 a file has the stream as its datum.
2697 ** type-errors on single-floats on x86-64 no longer have
2698 :INVALID-OBJECT as the datum
2699 ** the type-errors signalled for invalid function names now have
2700 a correct expected type
2701 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2702 for structures defined with a :TYPE. Documentation strings for
2703 typed structures are no longer immediately discarded
2704 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2705 broadcast streams and synonym streams. FILE-LENGTH now also works
2706 on broadcast streams.
2708 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2709 * incompatible change: the --noprogrammer option, deprecated since
2710 version 0.7.5, has been removed. Please use the equivalent
2711 --disable-debugger option instead.
2712 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2714 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2715 inoperational for a while now) has been completely removed.
2716 * null lexical environments are now printed as #<NULL-LEXENV>,
2717 significantly reducing the amount of clutter in typical backtraces.
2718 * documentation on weak pointers, finalization, and after GC hooks
2719 has been added to the manual.
2720 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2721 as other specialized array types, is much faster. SUBSEQ and
2722 COPY-SEQ on such arrays have also been sped up.
2723 * optimization: EQL is now more efficient when at least other argument
2724 is known to be of type (OR FIXNUM (NOT NUMBER)).
2725 * fixed bug: the runtime is now less vulnerable to changes in the
2726 size of the SBCL object on OS X, and virtual memory is reserved for
2727 all spaces that need to be at a fixed address.
2728 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2729 * fixed bug: finalizers and after GC hooks that cause consing are now
2731 * fixed bug: compiler error messages and summaries are now printed to
2732 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2733 * fixed inference of the upper bound of an iteration variable.
2734 (reported by Rajat Datta).
2735 * fixed bug 211e: calling local functions with duplicated constant
2736 keyword argument no longer causes a bogus style warning about an
2738 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2739 ftype declaration to be disregarded. (reported by Dave Roberts)
2740 * fixed bug 373: caused by erronous compilation of references to alien
2741 variables in the runtime on ppc/darwin.
2742 * fixed bug 376: CONJUGATE type deriver.
2743 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2744 (reported by Baughn on #lisp)
2745 * a cross-compiler bug on non-x86 platforms has been identified and
2746 fixed. (thanks to Bruno Haible)
2747 * improvements to the MIPS runtime code for increased stability.
2748 (thanks to Thiemo Seufer)
2749 * increased the maximimum compact environment size to allow
2750 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2751 * partial workaround for bug 135: don't name the function we're
2752 calling for hairy cases of EVAL, so as not to accumulate one environment
2753 entry per call to EVAL. (reported by Kevin Reid)
2754 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2755 * optimization: type testing for non-vector arrays should be faster.
2756 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2758 * bug fix: setting 31st element of a bit vector to zero did not work
2760 * bug fix: redefining a class definition which failed due to a
2761 previous accessor / function clash now works (but see BUGS entry
2762 #380 for more problems in this area). (thanks to Zach Beane)
2763 * the long form of DEFINE-METHOD-COMBINATION disables method group
2764 checking when given a single method group with pattern *. (thanks
2766 * on x86 compiler supports stack allocation of results of simple
2767 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2768 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2769 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2770 * fixed some bugs related to Unicode integration:
2771 ** the restarts for recovering from input and output encoding
2772 errors only appear when there is in fact such an error to
2774 * fixed some bugs revealed by Paul Dietz' test suite:
2775 ** MISC.549 and similar: late transformation of unsafe type
2776 assertions into derived types caused unexpected code
2778 ** SCALE-FLOAT type deriver is less wrong.
2779 ** type derivers for EXP, LOG and similar functions compute result
2780 types for complex arguments better.
2781 ** (MISC.563) CONJUGATE type deriver works for very restricted
2783 ** out-of-line type testers for character strings are available.
2784 ** EQUAL compiler transform understands specialness of objects
2786 ** accessing double-floats stored on the stack now works on x86-64.
2787 ** debugger internals could sometimes create invalid lispobjs,
2788 resulting in GC crashes.
2789 ** MISC.548: type check weakening can convert required type into
2791 ** initialization forms for bindings are not in scope of free special
2794 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2795 * incompatible change: thread support for non-NPTL systems has
2796 been removed - locking is buggy and unreliable. A threaded
2797 SBCL build will now warn at startup and refuse to create
2798 new threads, unless futex support is detected in the kernel
2799 * incompatible change: the top level REPL now has only an ABORT
2800 restart associated with it, not TOPLEVEL and ABORT as it used to.
2801 TOP and TOPLEVEL are now available as debugger commands for
2802 returning to the top level.
2803 * incompatible change: forms evaluated in the REPL now use the
2804 global optimization policy.
2805 * incompatible change: user- and system-initialization files are
2806 no longer processed with LOAD, but by READ and EVAL; hence the
2807 global optimization policy, startup package, readtable, etc,
2809 * The .fasl file format number has been incremented because of
2810 various incompatible changes.
2811 * internal entry point details and argument counts no longer appear
2812 in backtraces unless explicitly requested by setting
2813 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2814 * built-in and standard functions no longer have names like "top
2815 level local call to FOO".
2816 * fixed bug 32: functions defined in non-null lexical environments
2817 now have more legible printed representation
2818 * fixed bug 33: functions defined in non-null lexical environemnts
2819 are now more amenable to inspection by INSPECT.
2820 * workaround for bug 354: XEPs no longer appear in backtraces unless
2821 explicitly requested.
2822 * fixed bug: receiving the signal which results from attempting to
2823 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2824 SIGBUS on other *BSDs) on architectures where the C stack is also
2825 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2826 to debugger confusion or wild instability. (reported by Cheuksan
2828 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2829 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2830 (reported by Lutz Euler)
2831 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2832 compiled code. (reported by Paul Dietz)
2833 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2834 specialized parameter to a method to something that is not TYPEP
2835 the specializer is now possible.
2836 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2837 face of package deletion.
2838 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2839 pathname. (reported by David Tolpin on c.l.l)
2840 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2841 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2842 STRING-STREAM and FILE-STREAM, respectively.
2843 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2844 than previously; in addition, its test suite is now run on build.
2845 (thanks to Luke Gorrie)
2846 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2847 * optimization: added a immediate representation for single-floats
2849 * fixed some bugs revealed by Paul Dietz' test suite:
2850 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2851 ** attempting to create a package with a colliding nickname causes
2852 correctable errors to be signalled.
2853 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2854 ** several x86-64 backend bugs related to sign-extension of immediate
2857 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2858 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2859 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2861 * fixed loading of multiply forward-referenced layouts.
2862 (thanks to Cheuksan Wang)
2863 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2864 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2865 (see COMPILE-FILE documentation for details of :PRINT option)
2866 * fixed bugs 19 and 317: fixed-format floating point printing is
2867 more accurate. This also fixes a bug reported by Adam Warner
2868 related to the ~@F format directive.
2869 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2871 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2872 dispatch table if the from-char is a dispatch macro character.
2873 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2874 by the wordsize no longer ignore the last word. (reported by Lutz
2876 * fixed bug in type checking of dynamic-extent variables. (reported
2878 * optimization: sequence traversal functions use their freedom to
2879 coerce function designators to functions.
2880 * optimization: code with many calls to CLOS methods specialized on
2881 CLOS classes has had redundant type checks removed.
2882 * optimization: added declarations to speed up operations that access
2883 the internal character database (for example STRING-UPCASE)
2884 * optimization: comparison operations between floats and sufficiently small
2885 fixnums no longer create extra rationals
2886 * fixed some bugs related to Unicode integration:
2887 ** portions of multibyte characters at the end of buffers for
2888 character-based file input are correctly transferred to the
2889 start of the buffer at the next read.
2890 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2891 passing it through to OPEN.
2892 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2893 argument given to internal calls to OPEN.
2894 ** the built-in comment readers (introduced by character sequences
2895 ";" and "#|") are more forgiving to encoding errors; they will
2896 STYLE-WARN and then attempt to resync the stream at a character
2897 boundary. (thanks to Teemu Kalvas)
2898 * fixed some bugs in the x86-64 port:
2899 ** Negative short int return values from c-calls are sign-extended
2901 ** The stack is aligned to 16-bytes for c-calls, as required by
2902 the ABI. (thanks to Cheuksan Wang)
2903 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2904 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2905 for enums. (thanks to Vincent Arkesteijn)
2906 ** Multiple small optimizations and bugfixes for floating point
2908 * fixed some bugs revealed by Paul Dietz' test suite:
2909 ** Space, Tab, Linefeed, Return and Page have the invalid
2910 secondary constituent character trait.
2911 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2913 ** WITH-INPUT-FROM-STRING should only update the index place on
2915 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2916 works more reliably.
2917 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2918 with one-dimensional arrays with fill pointers.
2919 ** TYPE-OF failed on a complex with an integer realpart and a
2921 ** compiler failure during type inference for the code of form
2922 (IF (EQL X (THE ...)) ...) (MISC.535).
2924 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2925 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2926 platform. The port supports 61-bit fixnums, large memory spaces
2927 and reloading shared object files.
2928 * enhancement: saving cores with foreign code loaded is now
2929 supported on ppc/Darwin in addition to the previously supported
2931 * enhancement: the statistical profiler now walks deeper into the
2932 call stack for more meaningful call-graphs and accrued time
2933 reports (x86/x86-64 only). It also now reports time spent in
2935 * enhancement: it is now possible to trace most individual methods
2936 of a generic function in addition to tracing the generic function
2938 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2940 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2941 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2942 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2943 produces an error. (thanks to Vincent Arkesteijn)
2944 * bug fix: NAMESTRING on pathnames with :WILD components in their
2945 directories works correctly. (thanks to Artem V. Andreev)
2946 * fixed bug 125: compiler preserves identity of closures. (reported
2948 * bug fixed: functions with &REST arguments sometimes failed with
2949 "Undefined function" when compiled with (DEBUG 3). (reported by
2951 * bug fix: overflow during compiling of setting element of a bit
2952 vector with constant index and value. (reported by Timmy Douglas)
2953 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2954 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2955 * on x86 compiler partially supports stack allocation of dynamic-extent
2957 * GO and RETURN-FROM do not check the extent of their exit points
2958 when compiled with SAFETY 0.
2959 * fixed some bugs related to Unicode integration:
2960 ** encoding and decoding errors are now much more robustly
2961 handled; it should now be possible to recover even from invalid
2962 input or output to the terminal. (thanks to Teemu Kalvas)
2963 ** provided a first cut at implementing STRING-TO-OCTETS and
2964 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2965 ** altered the SB-MD5 contributed module slightly, changing the
2966 interface just enough for it to be supportable for builds where
2967 lisp characters are not eight bits.
2968 * fixed some bugs revealed by Paul Dietz' test suite:
2969 ** the FORMATTER-generated functions for ~V[ conditionals require
2970 the correct number of arguments.
2971 ** READ-FROM-STRING returns the mandated second value when applied
2972 to displaced strings.
2973 ** the #\Rubout and #\Backspace characters are treated as invalid
2974 constituent characters by the tokenizer.
2976 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2977 * new feature: reloading changed shared object files with
2978 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2979 * new feature: references to foreign variables and functions
2980 can now be compiled and loaded before the corresponding shared
2981 object file is loaded, as long as the foreign definitions are
2982 available at runtime.
2983 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2984 supported. (thanks to Dan Debertin)
2985 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2986 just keywords. (thanks to Vincent Arkesteijn)
2987 * fixed bug #331: structure-class instances corresponding to
2988 DEFSTRUCT forms are now created eagerly.
2989 * fixed bug #345: backtraces from calls to undefined functions work
2990 on x86 as well. Related bug #61 is now also partially fixed on x86
2991 (backtraces from throws to unknown catch tags.)
2992 * bug fix: lambda-list parsing is now stricter vrt. order and number
2993 of lambda-list keywords.
2994 * bug fix: as specified by AMOP, an error is signalled if a
2995 class-option appears multiple times in a DEFCLASS form. (reported
2997 * bug fix: RANDOM can be compiled when the compiler derives the type
2998 of its numeric argument as a disjoint set of small integers.
2999 (reported by Paul Dietz)
3000 * bug fix: starting a core saved with shared objects loaded when
3001 those objects are not available no longer causes threaded SBCL to
3002 hang. (reported by Sean Ross)
3003 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
3005 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
3006 argument, which must always be NIL. (reported by Kalle Niemitalo)
3007 * bug fix: printing 1.0d+23 no longer results in an error.
3008 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
3009 * bug fix: structure slot setters preserve evaluation order. (thanks
3011 * fixed some bugs related to Unicode integration:
3012 ** RUN-PROGRAM can allow its child to take input from a Lisp
3013 stream. (reported by Stefan Scholl)
3014 ** ASDF-INSTALL successfully downloads .tar.gz files in all
3015 locales. (reported by Ken Causey)
3016 * fixed some bugs revealed by Paul Dietz' test suite:
3017 ** INCF, DECF and REMF evaluate their place form as specified in
3019 ** FORMATTER expands ~{ ~} iteration directives with V or #
3020 parameters correctly.
3021 ** FORMATTER deals with the ~@[ ~] conditional directive where the
3022 consequent uses no arguments correctly.
3023 ** the system has a partial understanding of the (COMPLEX RATIO)
3025 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
3028 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
3029 * new feature: a build-time option (controlled by the :SB-UNICODE
3030 keyword feature, enabled by default) for building the system with
3031 support for the entire 21-bit character space defined by the
3033 * new feature: the system now has rudimentary external-format
3034 support; the primary user-visible change at this time is that
3035 characters with the high bit set (such as lower-case-e-acute) will
3036 print correctly to a terminal in a UTF-8 environment.
3037 * minor incompatible change: BASE-CHAR no longer names a class;
3038 however, CHARACTER continues to do so, as required by ANSI.
3039 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
3040 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
3041 has been moved to the SB-EXT package (temporarily re-exported from
3043 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
3044 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
3046 * fixed bug #351: better error-handlind and reporting for malformed
3048 * fixed bug #350: bignum-printing is now more memory-efficient,
3049 allowing printing of very large bignums, eg. (expt 2 10000000).
3050 (reported by Bruno Haible)
3051 * fixed bug #302: better primitive-type selection for intersection
3053 * fixed bug #308: non-graphic characters now all have names, as
3054 required. (reported by Bruno Haible)
3055 * bug fix: redefining a class with different superclasses now correctly
3056 removes it from the direct-subclasses of its previous superclasses.
3057 (reported by David Morse)
3058 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
3059 new value now works. (reported by Bruno Haible)
3060 * bug fix: correct canonicalization of multiple non-standard slot
3061 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
3062 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
3063 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
3064 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
3065 now exists, an signals an error.
3066 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
3067 during macroexpansion for non-iteration variables as well. (reported
3068 by Bruno Haible for CMUCL)
3069 * bug fix: Cyclic structures and unprintable objects in compiler
3070 messages no longer cause errors. (reported by Bruno Haible)
3071 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
3072 and READ-FROM-STRING all now return a primary value of NIL if
3073 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
3074 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
3075 (reported by Bruno Haible for CMUCL)
3076 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
3077 do not designate a condition. (reported by Bruno Haible for
3079 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
3080 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
3081 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
3082 permitting resolution in favour of any of the conflicting symbols.
3083 (reported by Bruno Haible for CMUCL)
3084 * bug fix: EQUAL compiler optimizations is less aggressive on
3085 strings which can potentially compare true despite having distinct
3086 specialized array element types.
3087 * bug fix: unit enumerations can be defined without dividing by
3088 zero. (thanks to Vincent Arkesteijn)
3089 * FORMAT compile-time argument count checking has been enhanced.
3090 (report from Bruno Haible for CMUCL)
3091 * a partial workaround for the bug 262: the compiler does not try to
3092 inline-expand a local function doing RETURN-FROM from a deleted
3093 BLOCK. (thanks to Peter Denno for the bug report and to David
3094 Wragg for the simple test case)
3095 * fixed some bugs revealed by Paul Dietz' test suite:
3096 ** RENAME-PACKAGE allows all package designators as new package
3098 ** constraint propagation and lambda variable substitution are
3099 more cautious in dealing with partially deleted code.
3100 ** compiler sometimes lost reoptimization passes.
3101 ** CERROR, given a condition as condition designator, passes its
3102 remaining arguments to the continue format control without
3104 ** Case-altering FORMAT directives work correctly on non-ASCII
3106 ** The REAL type specifier handles bounds outside the floating
3107 point ranges without signalling FLOATING-POINT-OVERFLOW.
3108 ** Functions with IR1-transformations can create intercomponent
3109 references to global functions.
3110 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
3112 ** FORMAT ~R treats a nil value for its first parameter correctly.
3114 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
3115 * enhancement: saving cores with foreign code loaded is now
3116 supported on x86/NetBSD and sparc/Linux in addition to the previously
3117 supported platforms.
3118 * bug fix: on some platforms repeated installations caused multiple
3119 copies of HTML documentation to be installed -- should not happen
3120 any more. (reported by Stefan Scholl)
3121 * bug fix: parsing self-recursive alien record types multiple times
3122 no longer causes infinite recursion. (reported by Thomas F. Burdick,
3123 original patch by Helmut Eller for CMUCL)
3124 * bug fix: stack-exhaustion detection works now on NetBSD as well.
3125 (thanks to Richard Kreuter)
3126 * bug fix: defining classes whose accessors are methods on existing
3127 generic functions in other (locked) packages no longer signals
3128 bogus package lock violations. (reported by François-René Rideau)
3129 * bug fix: special variables as DEFMETHOD parameters no longer have
3130 associated bogus type declarations. (reported by David Wragg and
3132 * bug fix: read-write consistency on streams of element-type
3133 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
3134 * bug fix: redefiniton of the only method of a generic function with
3135 no DEFGENERIC no longer emits a full WARNING. In addition,
3136 redefinition of generic functions with no DEFGENERIC to an
3137 incompatible lambda list now signals an error. (thanks to Zach
3139 * bug fix: DEFGENERIC now works even when there's a function of the
3140 same name in an enclosing lexical environment. (thanks to Zach
3142 * fixed compiler failure, caused by instrumenting code during
3143 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
3144 * optimization: added loop analysis and improved register allocation
3145 * fixed some bugs revealed by Paul Dietz' test suite:
3146 ** POSITION on displaced vectors with non-zero displacement
3147 returns the right answer.
3148 ** (SIMPLE-STRING) is a valid type specifier for sequence
3150 ** *PRINT-LEVEL* handling for slotless structures is pedantically
3152 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
3153 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
3155 ** The justification version of the FORMAT ~< directive treats
3156 non-zero minpad parameter correctly.
3158 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
3159 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
3160 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
3161 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
3162 the supported interface.
3163 * new feature: Single-stepping of code compiled with DEBUG 2 or
3164 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
3165 * new feature: saving cores with foreign code loaded is now
3166 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
3167 Timothy Moore's work for CMUCL)
3168 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
3169 parameters to * instead of NIL if no initform is supplied.
3170 (reported by Johan Bockgaard)
3171 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
3172 correctly when FUNCALL appears as the car of the form. Note:
3173 despite this FUNCALL forms are not currently subject to
3174 compiler-macro expansion. (port of Raymond Toy's fix for the
3175 same from CMUCL, reported by Johan Bockgaard)
3176 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
3177 (thanks for Teemu Kalvas)
3178 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
3179 case instead of terminating the iteration (thanks for Julian
3180 Squires, Sean Champ and Raymond Toy)
3181 * bug fix: incorrect expansion of defgeneric that caused a style
3182 warning. (thanks for Zach Beane)
3183 * bug fix: slot accessor effective method computation works properly
3184 for classes with multiple non-standard applicable methods on
3185 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
3186 * on x86 compiler supports stack allocation of results of LIST and
3187 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
3188 CMUCL implementation by Gerd Moellmann)
3189 * fixed some bugs revealed by Paul Dietz' test suite:
3190 ** FORMAT strings with both the ~<~:;~> form of the justification
3191 directive and pretty-printing directives cause an error.
3193 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
3194 * incompatible change: the internal functions
3195 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
3196 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
3197 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
3198 instead of the old functions.
3199 * new feature: on platforms where "dladdr" is available foreign
3200 function names now appear in backtraces. (based on Helmut Eller's
3202 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
3203 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
3205 * documentation: ASDF manual is now distributed with SBCL.
3206 * bug fix: SBCL can now load its contributed modules with REQUIRE
3207 even if the system-provided entries have been removed from
3208 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
3210 * bug fix: backtraces involving undefined functions or assembly
3211 routines are more informative. (thanks to Brian Downing)
3212 * bug fix: mutually referent alien structures now work correctly.
3213 (reported by Rick Taube)
3214 * bug fix: structures defined by WITH-ALIEN can be referred to
3215 within other definitions in the same WITH-ALIEN.
3216 * bug fix: division operators (MOD, TRUNCATE and the like) with
3217 constant zero divisors and integer dividends no longer generate
3219 * bug fix: provide default methods for INPUT-STREAM-P and
3220 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
3221 * bug fix: improve the detection and resolution of MOP metacycles
3222 (where effective method computation depends on the generic function
3223 having its effective method computed). (reported by Bruno Haible)
3224 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
3225 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
3226 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
3227 * bug fix: pathnames with relative directory components can be
3228 represented relative to default pathnames.
3229 * optimization: in taking the GCD of bignums, reduce the two bignums
3230 to approximately the same size (using Euclid's algorithm) before
3231 applying the more sophisticated binary GCD. (thanks to Juho
3233 * optimization: COUNT on bitvectors now operates word-at-a-time.
3234 * optimization: ASH with a positive, but not necessarily constant,
3235 (leftwards) shift, when in a modular context, is compiled to a
3237 * fixed some bugs revealed by Paul Dietz' test suite:
3238 ** FORMAT variable parameters ("~V<char>") are defaulted properly
3239 if the corresponding argument is NIL.
3240 ** FORMAT directives accepting a minpad parameter treat negative
3242 ** literal commas as character parameters to FORMAT directives are
3244 ** literal spaces directly after ~<Newline> directives within a
3245 format-logical-block (~:< ~@:>) do not induce :FILL-style
3246 conditional newlines.
3247 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
3248 as specified: it no longer includes conditional newlines).
3249 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
3251 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
3252 :PER-LINE-PREFIX argument does not evaluate to a string.
3254 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
3255 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3256 the manual for details; for now, package locks can be disabled by
3257 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
3258 if no major problems are found then it is likely that they will be
3259 compiled in unconditionally.
3260 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
3261 unsupported operators on all platforms. To load a shared library
3262 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
3263 object file, link it into a shared library outside of SBCL and
3264 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
3266 * bug workaround: ROOM T and the GC were not working together
3267 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
3268 reporting facility aren't preserved. That reporting has been
3269 disabled, so now until and unless someone figures out how to make it
3270 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
3271 * minor incompatible change: as threatened around sbcl-0.8.0, the
3272 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
3273 The symbols are also exported from SB-PCL for backwards
3274 compatibility, but more so than before SB-PCL should be treated as
3275 an implementation-internal package.
3276 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
3278 * the SB-SPROF contrib now works on (most) non-x86 architectures.
3279 It is known as of this release not to work on the Alpha, however.
3280 * fixed bug #167: errors signalled due to illegal syntax in method
3281 bodies are now more legible.
3282 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
3283 designators and can hence be used with TYPEP.
3284 * fixed bug #333: CHECK-TYPE now ensures that the type error
3285 signalled, if any, has the right object to be accessed by
3286 TYPE-ERROR-DATUM. (reported by Tony Martinez)
3287 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
3288 5.1.2.3 for multiple-value place subforms. (reported by Kalle
3290 * fixed bug #334: programmatic addition of slots using specialized
3291 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
3292 and :ALLOCATION :CLASS slots.
3293 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
3294 float exponents. (rereported by Peter Seibel)
3295 * fixed a bug: #\Space (and other whitespace characters) are no
3296 longer considered to be macro characters in standard syntax by
3297 GET-MACRO-CHARACTER.
3298 * fixed bug: initialization of condition class metaobjects no longer
3299 causes an instance of the condition to be created. (reported by
3301 * fixed bug: it is now possible to have slots such that
3302 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
3303 description is neither :INSTANCE nor :CLASS.
3304 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
3305 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
3306 SLOT-VALUE-USING-CLASS), and no longer causes errors with
3307 non-standard SLOT-DEFINITION-ALLOCATION values.
3308 * fixed bugs: various race conditions handling exiting threads.
3309 CL-PPCRE's thread torture test now passes.
3310 * fixed arguably-a-bug: GC time proportional to number of threads in
3311 system even when most of them are idle
3312 * optimization: improved performance of BIT and SBIT on bit-vectors.
3313 * .fasl file incompatibility: The fasl file version number has
3314 been incremented because of changes associated with package locks.
3316 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
3317 * minor incompatible change: the system no longer provides
3318 optimization hints (conditions which are TYPEP
3319 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
3320 mode; these hints are still emitted when higher SPEED optimization
3322 * new contrib module: a sampling profiler (profiling by statistical
3323 sampling, rather than by instrumenting functions) is available as
3324 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
3325 * the behaviour of the standard function ED is now customizeable by
3326 third parties through a hook variable: see ED's documentation
3327 string for information on the protocol.
3328 * the compiler no longer emits efficiency notes for (FUNCALL X)
3329 when the type of X is uncertain under default optimization
3331 * fixed bug 276: mutating a binding of a specialized parameter to a
3332 method to something that is not TYPEP the specializer is now
3334 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
3335 exhibit the required behaviour on the x86 platform. (thanks to
3336 Peter van Eynde, Eric Marsden and Bruno Haible)
3337 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
3338 even for difficult arguments. (reported by Peter Graves)
3339 * fixed bug 141a: the backquote printer now descends quoted
3341 * fixed another bug in backquote printing: no more destructive
3342 modification of the form's list structure. (reported by Brian
3344 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
3345 move between its address being taken and the call to
3346 interrupt_thread, fixing a crashing race condition.
3347 * the SB-POSIX contrib implementation has been adjusted so that it
3348 no longer exhibits ridiculously poor performance when constructing
3349 instances corresponding to C structs.
3351 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
3352 * minor incompatible change: the sb-grovel contrib now treats C
3353 structures as alien (in the sense of SB-ALIEN) objects rather than
3354 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
3355 has implications for memory management of client code
3356 (sb-grovel-returned objects must now be manually managed) and for
3357 type safety (alien objects now have full types).
3358 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
3359 used to control emission of compiler diagnostics, rather than the
3360 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
3361 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
3362 quality should be considered deprecated.
3363 * install.sh now installs the user manual as well
3364 * (not quite a new documentable feature, but worth considering in
3365 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3366 beginnings of a semantically meaningful condition hierarchy is
3367 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3369 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3370 designator as the defaults argument.
3371 * fixed bug: Displaced arrays whose displaced-to array has become
3372 too small now cause ARRAY-DIMENSION to signal an error, providing
3373 for safer bounds-checking. (reported by Bruno Haible)
3374 * fixed bug: DEFCLASS slot definitions with identical :READER and
3375 :WRITER names now signal a reasonable error. (reported by Thomas
3377 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3379 * fixed bug: Invalid slot specification errors now print correctly.
3380 (thanks to Zach Beane)
3381 * fixed bug 320: Shared to local slot value transfers in class
3382 redefinitions now happen corrently from superclasses as
3383 well. (reported by Bruno Haible)
3384 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3386 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3387 improper lists correctly. (reported by Bruno Haible)
3388 * fixed bug 313: source-transform for <fun-name> was erroneously
3389 applied to a call of a value of a variable with name <fun-name>.
3390 (reported by Antonio Menezes Leitao)
3391 * fixed bug 307: The obsolete instance protocol ensures that
3392 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3393 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3394 remove dead unknown-values globs from the middle of the stack.
3395 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3396 easier to resume long computations after using *BREAK-ON-SIGNALS*
3397 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3398 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3399 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3400 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3401 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3402 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3404 * fixed a (fairly theoretical) bug in string printing: if
3405 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3406 does not have array-element-type equal to the most general string
3408 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3409 function name. (thanks to Nikodemus Siivola)
3410 * fixed bug: compile-time format string checker failed on
3411 non-closed ~{. (reported by Thomas F Burdick)
3412 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3413 don't warn on legal loop code involving more than one aggregate
3414 boolean. (thanks to Nikodemus Siivola)
3415 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3416 in keyword position to MAKE-PACKAGE was not being reported as
3418 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3419 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3421 * fixed bugs 280 and 312: the checking for multiple definitions in a
3422 file is less likely to become confused by uses of inline
3424 * fixed bug: the #S reader macro performs the keyword coercion
3425 specified for slot names. (reported by Kalle Niemitalo)
3426 * fixed bug: lambda lists may contain symbols whose names start with
3427 & but are not lambda-list-keywords; their occurrence triggers a
3429 * fixed bug 321: define-method-combination argument lambda lists do
3430 not cause invalid code to be generated when &optional or &aux
3431 variables have default values. (reported by Bruno Haible)
3432 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3433 have CLOS classes; this appears also to have fixed reported
3434 instability in stack exhaustion detection.
3435 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3436 MUFFLE-WARNING when no associated restart is present is now
3438 * optimization: rearranged the expansion of various defining macros
3439 so that each expands into only one top-level form in a
3440 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3442 * optimization: used a previously little-used slot in symbols to
3443 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3445 * fixed some bugs revealed by Paul Dietz' test suite:
3446 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3448 ** Fixed an optimization bug related to inheritance of initforms
3449 from local to shared slots.
3450 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3451 ** CAST optimizer forgot to flush argument derived type.
3452 ** print/read consistency on floats is now orders of magnitude
3453 more likely. (thanks also to Bruno Haible for a similar report
3455 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3456 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3457 for objects of type REAL. Make it so.
3458 ** SXHASH is specified (infelicitously) to respect similarity,
3459 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3460 it so. (thanks to Markus Ziegler)
3461 ** on the Alpha, the compiler succeeds in compiling functions
3462 returning a known number of arguments greater than 63.
3463 ** fixed handling of invalid NIL arguments in keyword position
3465 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3467 ** provided a readably-printable representation for RANDOM-STATE
3469 ** ensured that pathnames, where they have a namestring, always
3470 print using #P"..." syntax.
3472 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3473 * Support for the forthcoming 2.0 version of the NetBSD kernel
3474 running on x86 hardware has been added. (thanks to Perry
3475 E. Metzger most immediately, and others for their past work)
3476 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3477 current release version; SBCL's OpenBSD support had been broken
3478 since about the time of OpenBSD's switch to ELF binary format.)
3479 * [placeholder for DX summary]
3480 ** user code with &REST lists declared dynamic-extent, under high
3481 speed or space and low safety and debug optimization policy.
3482 * The manual has been converted to Texinfo format and the debugger
3483 chapter from the cmucl manual has been added.
3484 * A facility has been added to extract documentation strings from
3485 sbcl and store them as Texinfo-formatted snippets for inclusion in
3486 the manual (via Texinfo's @include directive)
3487 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3488 the test case to Dave Roberts)
3489 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3490 pointers. (reported by Sean Ross)
3491 * bug fix: PROFILE output is printed nicely even for large numerical
3492 values. (thanks to Zach Beane)
3493 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3494 greater than 32 handle EOF correctly.
3495 * bug fix: on X86 an immediate argument of the IMUL instruction is
3496 correctly printed in disassembly. (thanks to Lutz Euler)
3497 * bug fix: class slots in redefined classes preserve their old
3498 values. (thanks to Bruno Haible and Nikodemus Siivola)
3499 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3500 succeeds. (reported by Marco Baringer)
3501 * fixed some bugs revealed by Paul Dietz' test suite:
3502 ** READ-SEQUENCE now works on ECHO-STREAMs.
3503 ** RATIONALIZE works more according to its specification. (thanks
3506 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3507 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3508 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3509 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3510 should matter to you only if you rebind the printer control
3511 variables and then find you want different bindings in the
3512 debugger than in the ordinary execution of your program.)
3513 * The runtime build system has been tweaked to support building
3514 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3515 assembler and linker. (thanks to Nikodemus Siivola)
3516 * Unbound, undefined, undeclared variables now trigger full
3517 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3518 more often programmer error than deliberate exploitation of undefined
3520 * optimization: the hash algorithm for strings has changed to one
3521 that is less vulnerable to spurious collisions. (thanks to Juho
3523 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3524 needless bounds checking. (thanks to Juho Snellman)
3525 * optimization: implemented multiplication as a modular
3526 (UNSIGNED-BYTE 32) operation on the PPC backend.
3527 * fixed some bugs revealed by Paul Dietz' test suite:
3528 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3530 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3531 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3532 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3534 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3536 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3538 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3539 their output stream on EOF from read.
3540 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3541 have been read to end-of-file.
3542 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3544 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3545 description of determination of which consecutive characters
3547 ** Printing the "Space" character with escaping on now yields "#\\ ",
3548 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3549 ** Reading floating-point numbers with *READ-BASE* set to a number
3550 less than 10 works correctly.
3551 ** Reading floating-point numbers with *READ-BASE* set to a number
3552 more than 10 works correctly.
3553 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3554 the readtable currently in effect.
3556 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3557 * minor incompatible change: parsing of namestrings on a physical
3558 (Unix) host has changed; numbers after the final #\. in a
3559 namestring are no longer interpreted as a version field. This is
3560 intented to be largely invisible to the user, except that the
3561 meaning of the namestring "*.*.*" has changed: it now refers to a
3562 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3563 should usually be replaced by
3564 (make-pathname :name :wild :type :wild :version :wild)
3565 with the added benefit that this is more likely to be portable.
3566 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3567 option to OPEN now signals an error if the file being opened
3568 exists; this may have an impact on existing code.
3569 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3570 signals received, on the PowerPC platforms (both Linux and
3571 Darwin). (thanks to Pierre Mai for pointing out the location of
3573 * several fixes on OS X: The system now builds and runs cleanly on
3574 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3575 (thanks to Brian Mastenbrook)
3576 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3577 second-resolution: integer multiples of 1/3600 between -24 and 24.
3578 (thanks to Vincent Arkesteijn)
3579 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3580 type when called with 1 argument; PEEK-CHAR checked type of
3581 PEEK-TYPE only after having read first character from a
3582 stream. (reported by Peter Graves)
3583 * bug fix: the garbage collector now has much better locality
3584 behaviour, and in particular no longer treats all memory as being
3585 exhausted when many small objects point to each other in a deeply
3587 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3588 recognized as being TYPEP their class.
3589 * bug fix: the PUSHNEW documentation string has been corrected.
3590 (thanks to Vincent Arkesteijn)
3591 * bug fix: defaulting of the value for the last of an atypically
3592 large number of multiple values being bound was not being
3593 performed correctly on the Alpha or PPC platforms
3594 * optimization: implemented multiplication as a modular
3595 (UNSIGNED-BYTE 32) operation on the x86 backend.
3596 * optimization: SEARCH on simple-base-strings can now be open-coded.
3597 (see also contrib/compiler-extras.lisp for inspiration for
3598 teaching the compiler about the Boyer-Moore algorithm).
3599 * value, returned by MAX (and MIN) called with several EQUALP, but
3600 not EQL, arguments now does not depend on compiler settings.
3601 * fixed some bugs revealed by Paul Dietz' test suite:
3602 ** in stack analysis liveness information is propagated from
3603 non-local entry points.
3604 ** pathwise CAST removing failed when the CAST node did not start
3606 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3607 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3609 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3610 host is already defined.
3611 ** RENAME-FILE works on streams instead of signalling an internal
3613 ** PEEK-CHAR uses the current readtable when determining whether
3614 or not a character is whitespace.
3615 ** MERGE-PATHNAMES handles the case when the pathname does not
3616 specify a name while the default-pathname specifies a version.
3617 ** Pathnames now stand a better chance of respecting print/read
3619 ** Attempting to use standardized file system operators with a
3620 pathname with invalid :DIRECTORY components signals a
3622 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3624 ** DIRECTORY on logical pathnames is more correct.
3625 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3626 signal a TYPE-ERROR if their argument is not a stream.
3627 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3628 designator argument does not designate a stream.
3629 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3630 examining the synonym.
3631 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3633 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3634 element-type larger than ([UN]SIGNED-BYTE 32).
3636 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3637 * When built with the :SB-FUTEX feature, threaded builds now take
3638 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3639 for faster/more reliable mutex and condition variable support.
3640 * Incompatible change (but one you probably shouldn't have been using
3641 anyway): the interface and code for arbitrating between multiple
3642 threads in the same user session has been redesigned.
3643 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3644 error when called without an explicit environment argument.
3645 (thanks to Nikodemus Siivola)
3646 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3647 confused about their position. (thanks to Adam Warner and Gerd
3649 * bug fix: Pretty printing backquoted forms with unquotations in the
3650 argument list position of various code constructs such as LAMBDA
3651 now works correctly. (reported by Paul Dietz)
3652 * bug fix: Pretty printing unquotations no longer loses all
3653 stream position information.
3654 * optimization: performance of string output streams is now less
3655 poor for multiple small sequence writes.
3656 * optimization: performance of CSUBTYPEP in the presence of complex
3657 expressions involving CONS and NOT many times has been improved.
3658 (reported by Paul Dietz)
3659 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3661 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3662 streams for the REPL, for the debugger, and so on. (thanks to
3664 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3665 * fixed some bugs revealed by Paul Dietz' test suite:
3666 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3667 is ignored if it is too small, rather than propagating through
3668 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3669 ** extremely complex negations of CONS types were not being
3670 sufficiently canonicalized, leading to inconsistencies in
3672 ** VALUES tranformer lost derived type.
3674 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3675 * fixed a bootstrapping bug: the build process no longer assumes
3676 that the various BOOLE-related constants have the same value in
3677 host and target lisps. (noted by Paul Dietz' test suite on an
3678 SBCL binary built from CLISP)
3679 * The system can now be dynamically linked on the MIPS platform,
3680 which enables dynamic loading of foreign code from Lisp. (thanks
3681 to Ralf Baechle for discussions on the MIPS ABI)
3682 * The system now records debugging information for its own source
3683 files in a filesystem-position-independent manner, relative to
3684 the "SYS" logical host.
3685 * fixed a compiler bug: MV-LET convertion did not check references
3686 to the "max args" entry point. (reported by Brian Downing)
3687 * tweaked disassembly notes to be less confident about proclaiming
3688 some instruction as an LRA. (thanks to Brian Downing)
3689 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3690 now each have their own history, command character, and other
3691 characteristics. (thanks to David Lichteblau)
3692 * fixed some bugs revealed by Paul Dietz' test suite:
3693 ** compiler failure in compiling LOGAND expressions including a
3695 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3696 shift greater than 32.
3697 ** FLUSH-DEST did not mark blocks for type check regeneration.
3698 ** HANDLER-CASE failed to accept declarations in handler clauses
3699 in some circumstances.
3701 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3702 * New code in contrib/sb-introspect (still probably not entirely
3703 stable yet) provides some support for smart Lisp development
3704 environments like SLIME.
3705 * The conditions signalled for errors occurring when loading .fasl
3706 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3707 in a way which should help ASDF recover gracefully.
3708 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3709 stuff has been cleaned up. If you code contrib/ stuff, this might
3710 affect you, and you can look at contrib/README, contrib/STANDARDS,
3711 and/or the 0.8.4.27 diff to check.
3712 * In full calls the compiler now does not generate checks for declared
3713 argument types for all arguments.
3714 * various threading fixes
3715 ** and some experimental patches which didn't make it into
3716 the main tree for this release, but which are shipped in
3717 contrib/experimental-thread.patch as a possible fix for some
3718 failures (deadlock, spinning...) in GC-intensive multithreaded
3720 * fixed PPC build problem (source code incompatibility of different
3721 library versions): added offsetof() hackery which attempts to divine
3722 where glibc maintainers put uc_mcontext today
3723 * fixed bug 282: compiler does not trust type assertions while passing
3724 arguments to a full call.
3725 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3726 &OPTIONAL VALUES type parameter.
3727 * fix bug 214: algorithm for noting rejected templates is now more
3728 similar to that of template seletion. (also reported by rydis on
3730 * fixed bug 141b: printing backquoted information readably and prettily
3731 inserts a space where necessary.
3732 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3733 (CHARACTER 10) are now reported as errors, rather than propagated
3734 as unknown types. (reported by piso on #lisp)
3735 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3736 with values NIL and :ERROR. (thanks to Milan Zamazal)
3737 * fixed bug 191c: CLOS now does proper keyword argument checking as
3738 described in CLHS 7.6.5 and 7.6.5.1.
3739 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3740 longer raise an error; further, using a list as a for-as-arithmetic
3741 counter now raises a meaningful error.
3742 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3743 now understood by sequence creation functions such as MAKE-SEQUENCE
3745 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3746 a STRING-INPUT-STREAM.
3747 * compiler enhancement: SIGNUM is now better able to derive the type
3749 * type declarations inside WITH-SLOTS are checked. (reported by
3751 * fixed some bugs revealed by Paul Dietz' test suite:
3752 ** incorrect optimization of TRUNCATE for a positive first
3753 argument and negative second.
3754 ** compiler failure in let-convertion during flushing dead code.
3755 ** compiler failure while deriving type of TRUNCATE on an
3756 interval, containing 0.
3757 ** ASH of a negative bignum by a negative bignum count now returns
3759 ** intersection of CONS types now canonicalizes properly, fixing
3760 inconsistencies in SUBTYPEP.
3762 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3763 * incompatible change: The --disable-debugger command line
3764 option now clobbers the debugger at a more fundamental
3765 level, by redefining #'INVOKE-DEBUGGER instead of by
3766 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3767 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3768 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3769 enter the debugger REPL and then suffer recursive errors
3770 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3771 while under the new system, BREAK in --disable-debugger mode
3772 terminates the system just as an unhandled error would.
3773 * fixed compiler performance when processing loops with a step >1;
3774 * bug fix: DOCUMENTATION now retrieves generic function
3775 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3776 support has been systematized, and now supports the methods
3777 specified by ANSI, along with a default method and a method for
3778 slot documentation. (reported by Nathan Froyd)
3779 * bug fix: effective methods associated with a generic function are
3780 no longer cached over a change of that generic function's method
3781 combination. (reported by Andreas Fuchs)
3782 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3783 names a non-existent directory. (thanks to Andreas Fuchs)
3784 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3785 the PPC platform, lead to stack corruption; this has been fixed.
3786 (reported by Rainer Joswig)
3787 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3788 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3789 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3790 arithmetic in general, is now much more reliable on the PPC
3792 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3793 platform now returns the right answer.
3794 * optimization: restored some effective method precomputation in
3795 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3796 precomputation is now tunable.
3797 * optimization: compiler-internal data structure use has been
3798 reviewed, and changes have been made that should improve the
3799 performance of the compiler by about 20%.
3800 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3801 simple-base-strings and simple-bit-vectors is improved.
3802 * optimization: the optimization of 32-bit logical and arithmetic
3803 functions introduced in version 0.8.3 on the x86 has been
3804 implemented on the mips, ppc and sparc platforms; an
3805 implementation of the same facility, but for 64-bit arithmetic,
3806 has been added for the alpha.
3807 * microoptimization: the compiler is better able to make use of the
3808 x86 LEA instruction for multiplication by constants.
3809 * bug fix: in some situations compiler did not report usage of
3810 generic arithmetic in (SPEED 3) policy.
3811 * bug 145b fix: compiler used wrong type specifier while converting
3812 MEMBER-types to numeric.
3813 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3815 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3816 index argument now works properly on the Alpha platform.
3817 * bug fix: floating point exception treatment on the Alpha platform
3819 * bug fix: FILE-POSITION works much better on string input and
3820 output streams. (thanks to Nikodemus Siivola)
3821 * bug fix: many threading/garbage collection symptoms sorted.
3822 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3823 might be pseudo-atomic.
3824 * internal change: Stopping for GC is now done with signals not
3825 ptrace. GC is now done in whichever thread wanted it, instead of
3827 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3829 * bug fix: result form in DO is not contained in the implicit
3831 * incompatible change: ICR structure is changed; the value part of
3832 CONTINUATION is now called LVAR; corresponding functions are
3833 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3834 * added type deriver for ISQRT (thanks to Robert E. Brown).
3835 * bug fix: better support for loading from the command line when an
3836 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3838 * fixed some bugs revealed by Paul Dietz' test suite:
3839 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3840 ** ROUND and FROUND now give the right answer when given very
3841 small float arguments.
3842 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3844 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3845 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3846 ** LCM should return a non-negative integer.
3847 ** PARSE-INTEGER returned the index of a terminator instead of the
3848 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3849 ** PARSE-INTEGER returned an incorrect index being applied to a
3851 ** LCM with two arguments of 0 returns 0 rather than signalling
3853 ** unsigned addition of a 32-bit constant with the high bit set no
3854 longer causes an internal compiler error on the x86.
3855 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3856 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3857 with negative last argument.
3858 ** byte specifiers with zero size and position no longer cause
3859 an error during type derivation.
3860 ** bignum multiplication on the Alpha platform now returns the
3862 * porting: The system now builds on SuSE AMD64, although it still
3863 generates a 32-bit binary.
3864 * .fasl file incompatibility: The fasl file version number has
3865 been incremented (because of the changes to internal compiler
3866 data structures referred to above).
3868 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3869 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3870 more accurately, on the Darwin kernel running on PowerPC hardware.
3871 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3872 * Compiler code deletion notes now signal a condition of type
3873 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3874 an associated MUFFLE-WARNING restart.
3875 * The compiler now performs limited argument count validation of
3876 constant format strings in FORMAT, and where appropriate in ERROR,
3877 CERROR and WARN. (thanks to Gerd Moellmann)
3878 * New ASDF-INSTALL contrib can be used for automatic download and
3879 installation of third-party Lisp code from CCLAN or other sites
3881 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3882 which forces another thread to execute a function supplied by the
3884 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3885 now accept and act upon their :ELEMENT-TYPE keyword argument.
3886 (reported by Martin Atzmueller, Edi Weitz)
3887 * bug fix: FILE-POSITION now accepts position designators up to
3888 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3889 is the greater. (thanks to Patrik Nordebo)
3890 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3891 Kalle Olavi Niemitalo)
3892 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3893 printable. (reported by Eric Marsden)
3894 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3895 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3896 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3897 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3898 (thanks to Stig Sandoe)
3899 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3900 UNREAD-CHAR now correctly includes the unread character in the
3901 target sequence. (thanks to Gerd Moellmann)
3902 * bug fix in threaded builds: the system can now be suspended and
3903 resumed by shell job control with minimal disruption.
3904 * bug fixes in times and timezones >2038AD
3905 * better handling of "where is GNU make?" problem in build scripts
3906 (thanks to Nikodemus Siivola)
3907 * new optimization: inside a named function any reference to a
3908 function with the same name is considered to be a self-reference;
3909 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3910 optimization quality.
3911 * new optimization on x86: logical functions and + now have
3912 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3913 used when the result is truncated to 32 bits.
3914 * VALUES declaration is partially enabled.
3915 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3916 * fixed some bugs revealed by Paul Dietz' test suite:
3917 ** The system now obeys the constraint imposed by
3918 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3919 types form a lattice under type intersection.
3920 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3921 ** ASSOC now ignores NIL elements in an alist.
3922 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3923 and (1+ MOST-POSITIVE-FIXNUM) answers.
3924 ** The addition of a method with invalid qualifiers to a generic
3925 function does not cause an error to be signalled immediately;
3926 a warning is signalled, and the error is generated only on
3927 calling the generic function.
3928 * changed .fasl file version number, in part to add type codes for
3929 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3930 obscure ANSI requirements
3932 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3933 * fixed bug 148: failure to inline-expand a local function left
3934 garbage, confusing the compiler.
3935 * fixed bugs 3cd: structure slot readers perform type check if the
3936 slot can have an invalid value (i.e. it is either not initialized
3937 or can be written with a less specific slot writer).
3938 * bug fix: the compiler now traps array references to elements off
3939 the end of an array; previously, the bounds checking in some
3940 circumstances could go off-by-one.
3941 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3943 * type declarations for array element types now obey the description
3944 on the CLHS page "Declaration TYPE", as per discussions on
3945 sbcl-help around 2003-05-08. This means that a declaration
3946 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3947 declaration, all references to BAR will be asserted or assumed
3948 (with THE, so dependent on compiler policy) to involve objects of
3949 type FOO. Note that no such declaration is implied in
3950 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3951 * declared types of functions from the "Conditions"
3952 chapter. (reported by Paul Dietz)
3953 * bug fix: CERROR accepts a function as its first argument.
3954 * bug fix: NTH an NTHCDR accept a bignum as index
3955 arguments. (reported by Adam Warner)
3956 * optimization: character compare routines now optimize comparing
3957 against a constant character. (reported by Gilbert Baumann)
3958 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3959 argument now works properly on the MIPS platform.
3960 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3961 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3962 declared to have a functional type, does not cause an error
3964 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3965 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3966 * fixed bug: if last continuation of a deleted block has a
3967 destination, this destination should be deleted too. (reported by
3969 * fixed a bug in the bootstrap process: the host compiler's values
3970 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3971 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3972 test case from Patrik Nordebo)
3973 * improved the ability of the disassembler on the PPC platform to
3974 provide helpful disassembly notes.
3975 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3976 the class in more cases than previously.
3977 * bug fix: FILE-POSITION now understands :START and :END for
3978 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3979 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3980 8) data. (thanks to David Lichteblau)
3981 * bug fix: it is possible to add a method to a generic function
3982 without lambda list.
3983 * bug fix: reader failed to signal END-OF-FILE inside an
3984 object representation. (reported by Nikodemus Siivola)
3985 * fixed some bugs revealed by Paul Dietz' test suite:
3986 ** LAST and [N]BUTLAST should accept a bignum.
3987 ** condition slot accessors are methods.
3988 ** (VECTOR NIL) is a subtype of STRING.
3990 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3991 * minor incompatible change: some nonsensical specialized lambda
3992 lists (used in DEFMETHOD) which were previously ignored now signal
3994 * minor incompatible change: the system is now aware of the types of
3995 variables in the COMMON-LISP package, and will signal errors for
3996 most violations of these type constraints (where previously they
3997 were silently accepted).
3998 * minor incompatible change: COMPILE-FILE now uses the freedom
3999 afforded (ANSI 3.2.2.3) to use derived function types for
4000 functions defined in the same file. This also permits the system
4001 to warn on static type mismatches and function
4002 redefinition. (Currently it does not work with high DEBUG level.)
4003 * minor incompatible change: VALUES declaration is disabled.
4004 * When issuing notes, the compiler now signals a condition of type
4005 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
4006 restart for use in user handlers. It is expected that the
4007 COMPILER-NOTE condition will eventually become a condition
4008 supertype to a hierarchy of note types, which will then be
4009 handleable in a similar fashion. However, at the moment, no such
4010 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
4011 but it's an implementation detail, not a classification for the
4013 * Changes in type checking closed the following bugs:
4014 ** type checking of unused values (192b, 194d, 203);
4015 ** template selection based on unsafe type assertions (192c, 236);
4016 ** type checking in branches (194bc).
4017 * A short form of VALUES type specifier has ANSI meaning (it has
4018 increased the number of situations when SBCL cannot perform type
4020 * fixed bug in DEFSTRUCT: once again, naming structure slots with
4021 keywords or constants is permissible.
4022 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
4023 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
4024 classes. (thanks to Antonio Martinez)
4025 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
4026 outputs FRESH-LINE or TERPRI, and no longer converts its stream
4027 argument to a pretty-print stream. Instead, it leaves any such
4028 operations to DESCRIBE-OBJECT methods.
4029 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
4031 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
4032 lambda lists. (thanks to David Lichteblau)
4033 * bug fix: Defining a generic function with a :METHOD-CLASS being a
4034 subclass of STANDARD-METHOD no longer causes stack exhaustion.
4035 (thanks to Gerd Moellmann)
4036 * fixed bug 246: increased compilation speed of long
4037 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
4039 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
4041 * DEFCONSTANT now throws a condition of type
4042 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
4043 constant to a non-EQL value; CONTINUE and ABORT restarts
4044 respectively change and preserve the value.
4045 * fixed bug 63: The code walker, part of the implementation of CLOS,
4046 is now better at handling symbol macros.
4047 * bug fix: There is no longer an internal implementation type named
4048 CL:LENGTH. (reported by Raymond Toy)
4049 * bug fix: In macro-like defining macros/special operators the
4050 implicit block does not enclose the lambda list.
4051 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
4052 enclosed in parentheses) are not suitable as type specifiers, and
4053 their use properly signals an error now.
4054 * bug fix: An argument count mismatch for a type specifier in code
4055 being compiled no longer causes an unhandled error at compile
4056 time, but signals a compile-time warning.
4057 * fixed simple vector readable printing
4058 * bug fix: DESCRIBE takes more care over whether the class
4059 precedence list slot of a class is bound before accessing it.
4060 (reported by Markus Krummenacker)
4061 * bug fix: FORMATTER can successfully compile pretty-printer format
4062 strings which use variants of the ~* directive inside.
4063 * bug fix: SEARCH now applies its TEST predicate to the elements of
4064 the arguments in the correct order. (thanks to Wolfhard Buss)
4065 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
4066 in type inference. (thanks to Robert E. Brown)
4067 * bug fix: Reading in symbols with an explicit package name of ""
4068 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
4069 * fixed some bugs revealed by Paul Dietz' test suite:
4070 ** NIL is now allowed as a structure slot name.
4071 ** Arbitrary numbers, not just REALs, are allowed in certain
4072 circumstances in LOOP for-as-arithmetic clauses.
4073 ** Multiple class redefinitions before slot access no longer
4074 causes a type error.
4075 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
4076 association between the name and a class.
4077 ** Generic functions with non-standard method-combination and over
4078 five methods all of which return constants no longer return NIL
4079 after the first few invocations. (thanks to Gerd Moellmann)
4080 ** CALL-NEXT-METHOD with no arguments now passes the original
4081 values of the arguments, even in the presence of assignment.
4082 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
4084 ** DISASSEMBLE works with closures and funcallable instances.
4085 ** ADD-METHOD now returns the generic function, not the new method.
4086 ** FIND-METHOD signals an error if the lengths of the specializers
4087 is incompatible with the generic function, even if the ERRORP
4089 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
4090 which its argument is a member.
4091 ** DEFCLASS only redefines the class named by its class-name
4092 argument if that name is the proper name of the class;
4093 otherwise, it creates a new class.
4094 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
4095 of the UNBOUND-SLOT condition to the name of the slot.
4096 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
4097 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
4098 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
4099 SLOT-MAKUNBOUND in the specified fashion.
4101 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
4102 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
4103 cross-compilation host. As a consequence, we can now bootstrap our
4104 way up to SBCL starting with a bare gcc toolchain and human-readable
4105 source code (first the source to CLISP, then the source to SBCL).
4106 * A contributed module containing a partial implementation of the
4107 simple-streams interface has been included. (thanks to Rudi
4109 * A contributed module implementing the RFC1321 Message Digest
4110 Algorithm, known as MD5, has been included.
4111 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
4112 no longer has any effect, as the code controlled by this feature
4113 has been deleted. (As far as we know, no-one has ever built using
4114 this feature, and its semantics were confused in any case).
4115 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
4116 work on conditions (as required by the ANSI specification),
4117 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
4118 expected behaviour on conditions. Users should note, however,
4119 that such behaviour is not required by the ANSI specification,
4120 and so use of this behaviour may render their code unportable.
4121 * fixed some bugs revealed by Paul Dietz' test suite:
4122 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
4124 ** &ENVIRONMENT parameter in macro lambda list is bound first.
4125 ** SXHASH on condition objects no longer returns NIL.
4126 ** :ALLOCATION :CLASS slots are better treated; their values are
4127 updated on class redefinition, and initforms inherited from
4128 superclasses are applied.
4129 ** REMOVE-METHOD returns its generic function argument even when
4130 no method was removed.
4131 ** SHARED-INITIALIZE now initializes the values of the requested
4132 slots, including those with :ALLOCATION :CLASS.
4133 ** ALLOCATE-INSTANCE now works on structure classes defined via
4134 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
4136 ** SLOT-EXISTS-P now works on conditions, as well as structures
4138 ** MAKE-LOAD-FORM now has the required methods on
4139 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
4140 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
4141 keyword, and now implements the SLOT-NAMES argument.
4142 ** methods with &OPTIONAL arguments no longer allow too many
4143 arguments to be passed in the call without error.
4144 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
4145 option is consistent with the required arguments of the generic
4146 function lambda list.
4147 * bug fix: REQUIRE accepts a string designator. (Thanks to
4149 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
4150 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
4151 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
4152 * bug fix: The compiler now checks for duplicated variables in macro
4154 * bug fix: SETQ on globals returns the correct value.
4155 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
4156 not a COMPILER-ERROR (followed by some other strange error on
4157 choosing the CONTINUE restart).
4158 * bug fix: make.sh and friends are now more consistent in the way that
4159 they look for GNU "make".
4161 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
4162 * experimental native threads support (on x86 Linux >=2.4 only).
4163 This is not compiled in by default: you need to add :SB-THREAD to
4164 the target features. See the "Beyond ANSI" chapter of the manual
4166 * fix for longstanding nonANSIism: The old distinction between
4167 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
4168 The return value from CL:FIND-CLASS is now a CLOS class, and
4169 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
4170 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
4171 * An interface to the MetaObject Protocol, as described in Kiczales,
4172 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
4173 MIT Press, 1991, is available from the SB-MOP package.
4174 * incompatible change: the SB-PCL package should now be considered
4175 a private implementation detail, and no longer a semi-private MOP
4177 * minor incompatible change: due to rearrangement for threads, the
4178 control stack and binding stack are now allocated at arbitrary
4179 addresses instead of being hardcoded per-port. Users affected by
4180 this probably have to be doing advanced things with shared
4181 libraries, and will know who they are.
4182 * minor incompatible change: Previously, all --eval forms used were
4183 processed with READ before any of them were processed with EVAL.
4184 Now each --eval form is processed with both READ and EVAL before
4185 the next --eval form is processed. (Thus package operations like
4186 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
4187 work as the user might reasonably expect.)
4188 * minor incompatible change: *STANDARD-INPUT* is now only an
4189 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
4191 * minor incompatible change: Y-OR-N-P is now character-oriented,
4192 not line oriented. Also, YES-OR-NO-P now works without errors.
4193 (thanks to Antonio Martinez)
4194 * sb-aclrepl module improvements: an integrated inspector, added
4195 repl features, and a bug fix to :trace command.
4196 * Known functions, which cannot be open coded by the backend, are
4197 considered to be able to check types of their arguments. (fixing
4198 a bug report by Nathan J. Froyd)
4199 * fixed a bug in computing method discriminating functions: It is
4200 now possible to define methods specialized on classes which have
4201 forward-referenced superclasses. (thanks to Gerd Moellmann)
4202 * fixed evaluation order in optional entries (reported by Gilbert
4204 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
4205 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
4206 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
4207 required-by-AMOP NAME argument, as well as CLASS and
4208 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
4209 * fixed bug 20: DEFMETHOD can define methods using names that are
4210 not the proper names of classes to designate class specializers.
4211 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
4212 Unix file descriptors, instead of blowing up. (thanks to Antonio
4214 * Garbage collection refactoring: user-visible change is that a
4215 call to the GC function during WITHOUT-GCING will not do garbage
4216 collection until the end of the WITHOUT-GCING. If you were doing
4217 this you were probably losing anyway.
4218 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
4219 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
4220 TYPEP the latter but not the former.
4221 * The compiler issues a full WARNING for calls to undefined functions
4222 with names from the CL package.
4223 * MAP-INTO for a vector destination is open coded. (reported by
4224 Brian Downing on c.l.l)
4225 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
4226 documentation string.
4227 * fixed some bugs revealed by Paul Dietz' test suite:
4228 ** COPY-ALIST now signals an error if its argument is a dotted
4230 ** Condition slots are now accessed more correctly in the presence
4231 of multiple initargs for a given slot.
4232 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
4233 exclude restarts of the same name associated with a different
4235 ** DEFCLASS of forward-referenced classes with another
4236 forward-referenced class in the superclasses list no longer
4238 ** Condition slots are now initialized once each, not multiple
4239 times. (thanks to Gerd Moellmann)
4240 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
4242 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
4243 ** NSET-EXCLUSIVE-OR does not return extra elements when its
4244 arguments contain duplicated elements.
4245 ** RESTART-CASE understands local macros.
4246 ** RESTART-CASE associates exactly its own restarts with a condition.
4247 ** ENDP in safe mode checks its argument to be of type LIST.
4248 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
4249 in question is unbound.
4250 ** Optimized MAKE-INSTANCE functions no longer cause internal
4251 assertion failures in the presence of duplicate initargs.
4252 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
4253 ** Side-effectful :DEFAULT-INITARGS have their side-effects
4254 propagated even in the ctor optimized implementation of
4256 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
4258 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
4259 * a better implementation of SXHASH on (simple) bit vectors,
4260 measured both in execution speed and in distribution of results
4261 over the positive fixnums, has been installed. Likewise, a better
4262 implementation of EQUAL for simple bit vectors is now available.
4263 * fixed CEILING optimization for a divisor of form 2^k.
4264 * fixed bug 240 (emitting extra style warnings "using the lexical
4265 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
4266 by Antonio Martinez)
4267 * fixed SXHASH, giving different results for NIL depending on type
4268 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
4269 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
4270 variable. (found by Rolf Wester)
4271 * SBCL does not ignore type declarations for special
4272 variables. (reported by rif on c.l.l 2003-03-05)
4273 * some bug fixes in contrib/sb-aclrepl/
4274 * fixed some bugs revealed by Paul Dietz' test suite:
4275 ** a bug in the CONS type specifier, whereby the CAR and CDR
4276 types got intertwined, has been fixed;
4277 ** the type system is now able to reason about the interaction
4278 between INTEGER and RATIO types more completely;
4279 ** APPEND, [N]REVERSE and NRECONC check that those their
4280 arguments, which must be proper lists, are really so;
4281 ** An array specialized to be unable to hold elements has been
4282 implemented, as required -- yes, really -- by ANSI;
4283 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
4284 on malformed property lists;
4286 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
4287 * incompatible packaging change: in line with Unix convention,
4288 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
4289 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
4290 * REQUIRE and PROVIDE are now optionally capable of doing something
4291 useful. See the documentation string for REQUIRE.
4292 * infrastructure for a managed SBCL contrib system: contributed
4293 modules in this release include:
4294 ** the ASDF system definition facility;
4295 ** an interface to the BSD Sockets API;
4296 ** an ACL-like convenience interface to the repl;
4297 (thanks to Kevin Rosenberg)
4298 ** an implementation of ROTATE-BYTE, with efficient implementation
4300 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
4301 gives a closer approximation to the right answer than previously.
4302 (thanks to Raymond Toy)
4303 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
4304 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
4305 optional environment arguments, as required by ANSI.
4306 * fixed bugs in other functions taking environment objects, allowing
4307 calls with an explicit NIL environment argument to be compiled
4309 * fixed bug 228: primary return values from
4310 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
4311 COMPILE or FUNCTION.
4312 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
4313 :TYPE will no longer signal errors on innocuous objects.
4314 * fixed bug 231b: SETQ is better at respecting type declarations in
4315 the lexical environment.
4316 * fixed a bug in DEFCLASS: classes named by symbols with no or
4317 unprintable packages can now be defined.
4318 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
4319 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
4320 * fixed compiler failure related to checking types of functions.
4321 (reported by Robert E. Brown)
4322 * the compiler is now much more consistent in its error-checking
4323 treatment of bounding index arguments to sequence functions: in
4324 (SAFETY 3) code, errors will be signalled in almost all cases if
4325 invalid sequence bounding indices are passed to functions defined
4326 by ANSI to operate on sequences.
4327 * fixed a bug in the build procedure: documentation of SBCL-specific
4328 packages is now preserved and available in the final Lisp image.
4329 * lifted FDEFINITION lookup out of loops in the implementation of
4330 many list operations. (thanks to Robert E. Brown)
4331 * fixed a bug in the reader: the #n# reader macro now works for
4332 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
4333 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
4334 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
4335 E. Brown and Tony Martinez respectively)
4336 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
4337 SLOT-BOUNDP: in particular, it is now able to optimize them much
4338 better, and is now not vulnerable to having packages renamed.
4339 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
4340 always called when a slot is not present in an instance. (thanks
4342 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
4343 longer signal an error on unreading a character following EOF on
4344 the previous constituent stream. (thanks to Tony Martinez)
4345 * fixed some bugs revealed by Paul Dietz' test suite:
4346 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
4347 not just nonnegative fixnums;
4348 ** the logical bit-array operators such as BIT-AND now accept an
4349 explicit NIL for their "opt-arg" argument (to indicate a
4350 freshly-consed result bit-array);
4351 ** ELT now signals an error on an invalid sequence index in safe
4353 ** the type system is now cleverer about negations of numeric
4354 types, and consequently understands the BIGNUM and RATIO types
4356 ** the type system is now cleverer about the interaction between
4357 INTEGER and RATIO types: while bugs still remain, many more
4358 cases are accurately computed;
4359 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
4360 if it is in the last clause;
4361 ** CONSTANTLY now correctly returns a side-effect-free function in
4363 ** DECLARE is no longer treated as a special-operator; in
4364 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
4365 * incremented fasl file version number due to the change in the
4366 DEFSTRUCT-SLOT-DESCRIPTION structure.
4368 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4369 * minor incompatible change: code processed by the "interpreter" or
4370 EVAL now has a compilation optimization policy of (DEBUG 2)
4371 (changed from (DEBUG 1)) to improve debuggability of interactive
4372 development, and to allow the use of the debug RETURN command in
4374 * an experimental implementation of the RETURN command for the
4375 debugger has been included. (thanks to Frederik Kuivinen)
4376 * fixed bug 62: constraints were not propagated into a loop.
4377 * fixed bug in embedded calls of SORT (reported and investigated by
4379 * fixed some bugs revealed by Paul Dietz' test suite:
4380 ** printing and reading of arrays with some dimensions having
4381 length 0 (thanks to Gerd Moellmann);
4382 ** BOA constructor with &AUX argument without a default value does
4383 not cause a type error;
4384 ** CONSTANTP now returns true for all self-evaluating objects.
4386 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4387 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4388 accessors that are related by inheritance, as specified in the
4389 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4390 Valtteri Vuorikoski)
4391 * The compiler is now able to inline functions that were defined in
4392 a complex lexical environment (e.g. inside a MACROLET).
4393 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4395 * Support for the upcoming FreeBSD-5.0 release has been included.
4396 (thanks to Dag-Erling Smorgrav)
4397 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4398 effect when it is not in a toplevel context.
4399 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4400 stand a better chance of being correct. (thanks to Gerd
4402 * fixed bug in COERCE, which now signals an error on coercing a
4403 rational to a bounded real type which excludes the expected
4405 * The compiler is now able to derive types more accurately from the
4406 COERCE and COMPILE functions.
4407 * fixed bug 223: functional binding is considered to be constant
4408 only for symbols in the CL package.
4409 * fixed bug 231: SETQ did not check the type of a variable being set
4410 (reported by Robert E. Brown)
4411 * A new optimization for MAKE-INSTANCE has been included, fixing
4412 various bugs (including relating to :ALLOCATION :CLASS slots and
4413 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4415 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4416 ** As required by ANSI, LOOP now disallows anonymous collection
4417 clauses such as COLLECT I in conjunction with aggregate boolean
4418 clauses such as THEREIS (= I 1);
4419 ** LOOP now signals an error when any variable is reused in the
4420 same loop (including the potentially useful construct analogous
4421 to WITH A = 1 WITH A = (1+ A);
4422 ** IT is only a special loop symbol within the first clause of a
4423 conditional loop clause;
4424 ** LOOP with a typed iteration variable over a hashtable now
4425 signals a type error iff it should.
4426 * fixed some other bugs revealed by Paul Dietz' test suite:
4427 ** FILE-STREAM now names the class previously known as FD-STREAM;
4428 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4429 argument) no longer signals an error;
4430 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4431 of the slot symbol, rather than using the current package
4432 ((:CONC-NAME "") continues to intern the slot's name in the
4434 * incremented fasl file version number, because of the incompatible
4435 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4436 of the new implementation of DEFINE-COMPILER-MACRO.
4438 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4439 * Support for building SBCL for MIPS platforms running in
4440 little-endian mode has now been checked in, and basic
4441 functionality on said platforms verified.
4442 * minor incompatible change: PCL now records the pathname of a file
4443 in which methods and the like are defined, rather than its
4445 * minor incompatible change: TRUENAME now considers the truename of
4446 a file naming a directory to be the pathname with :DIRECTORY
4447 component indicating that directory.
4448 * minor incompatible change: a NAMED clause in the extended form of
4449 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4450 reason for the previous behaviour is unclear.
4451 * more systematization and improvement of CLOS and MOP conformance
4452 in PCL (thanks to Gerd Moellman and Pierre Mai):
4453 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4455 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4456 primary methods with no specializers;
4457 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4459 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4460 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4461 FORWARD-REFERENCED-CLASSes; error reporting on
4462 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4464 ** SXHASH on CLOS instances now uses a slot internal to the
4465 instance to return different numbers on distinct instances,
4466 while preserving the same return value through invocations of
4468 ** DEFMETHOD signals errors when methods with longer incongruent
4469 lambda lists are added to generic functions;
4470 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4471 CLASS, as specified in AMOP;
4472 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4473 on the order returned by the primary method for classes of
4474 class STANDARD-CLASS;
4475 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4476 * fixed some bugs shown by Paul Dietz' test suite:
4477 ** DOLIST puts its body in TAGBODY;
4478 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4480 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4481 value producing form;
4482 ** if more variables are given to PROGV than values, extra
4483 variables are bound and made to have no value;
4484 ** NSUBSTITUTE on list arguments gets the right answer with
4486 ** ELT signals an error of type TYPE-ERROR when the index argument
4487 is not a valid sequence index;
4488 ** LOOP signals (at macroexpansion time) an error of type
4489 PROGRAM-ERROR when duplicate variable names are found;
4490 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4491 ** FUNCALL of special-operators now cause an error of type
4493 ** PSETQ now works as required in the presence of side-effecting
4494 symbol-macro places;
4495 ** NCONC accepts any object as its last argument;
4496 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4498 ** loop-for-as-package does not require a package to be explicitely
4500 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4502 * fixed bug 166: compiler preserves "there is a way to go"
4503 invariant when deleting code.
4504 * fixed bug 172: macro lambda lists with required arguments after
4505 &REST arguments now cause an error to be signalled. (thanks to
4507 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4508 bug. (thanks to Matthew Danish)
4509 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4511 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4512 arguments are lexically rebound. (thanks to Gerd Moellmann and
4514 * fixed bug 194: error messages are now more informative when there
4515 is no primary method applicable in a call to a generic
4516 function. (thanks to Gerd Moellmann)
4517 * fixed bug in command line argument checking (thanks to Julian
4519 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4520 (thanks to Matthew Danish)
4521 * incremented fasl file version number, because of the
4522 SXHASH-related changes in the layout of CLOS data structures
4524 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4525 * minor incompatible change: The runtime (the Unix executable named
4526 "sbcl") is now much pickier about the .core files it will load.
4527 Essentially it now requires .core files to descend from the same
4528 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4529 as the runtime does. (The intent is to prevent the crashes which
4530 can occur, and which can even be reported as mysterious failures,
4531 when people patch the sources or change the build parameters
4532 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4533 sbcl and .core files.)
4534 * fixed bug: VALUES-LIST is no longer optimized away.
4535 * fixed bug 142: The FFI conversion of C string values to Lisp
4536 string values no longer conses excessively. (thanks to Nathan
4537 Froyd porting Raymond Toy's fix to CMU CL)
4538 * began to systematize and improve MOP conformance in PCL (thanks to
4539 Nathan Froyd, Gerd Moellman and Pierre Mai):
4540 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4542 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4543 ** FINALIZE-INHERITANCE is now called on class finalization;
4544 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4545 argument precedence order.
4546 * fixed bug 202: The compiler no longer fails on functions whose
4547 derived types contradict their declared type.
4548 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4549 so it can be non-toplevel.
4550 * The fasl file version number has changed (because of the new
4551 implementation of DEFMACRO).
4552 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4553 safe code, that any length requirement by their type-specifier
4554 argument is valid. The exceptions to this are described in bug
4556 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4557 functions now check the types of their inputs as required by ANSI.
4558 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4559 introduced symbol is DECLAREd to be SPECIAL.
4560 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4561 * fixed bug: :COUNT argument to sequence functions may be negative
4562 * fixed bug: body of DO-SYMBOLS may contain declarations
4563 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4564 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4565 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4566 symbol macro only once
4567 * fixed printing of call frame when argument list is unavailable
4568 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4569 * compiler no longer signals WARNING on unknown keyword
4572 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4573 * A beta-quality port to the mips architecture running Linux,
4574 based on the old CMUCL backend, has been made. It has been tested
4575 on a big-endian kernel, and works sufficiently well to be able to
4576 rebuild itself; it has not been tested in little-endian mode.
4577 * fixed an inconsistency between gencgc.c and purify.c which made
4578 dumping/loading .core files unreliable
4579 * fixed bug 120a: The compiler now deals correctly with IFs where
4580 the consequent is the same as the alternative, instead of
4581 misderiving the return type. (thanks to Alexey Dejneka)
4582 * fixed bug 113: Logical pathnames are now dumpable (the logical
4583 host is resolved at load-time, throwing an error if it is not
4585 * fixed bug 174: FORMAT's error message is slightly clearer when a
4586 non-printing character is used in a format directive.
4587 * fixed several bugs in compiler checking of type declarations, i.e.
4588 violations of the Python "declarations are assertions" principle
4589 (thanks to Alexey Dejneka)
4590 * fixed several bugs in PCL's error checking (thanks to Gerd
4592 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4594 * fixed bug in compilation of functions as first class values
4595 (thanks to Antonio Martinez-Shotton)
4596 * The compiler's handling TYPE-ERRORs which it can prove will
4597 inevitably happen at runtime has been cleaned up and corrected
4598 in several ways. (thanks to Alexey Dejneka)
4599 * improved argument type checking for various basic arithmetic
4600 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4601 lavished on them in the past that they can be compiled in many
4602 ways in different special cases
4603 * fixed bug 181: compiler checks validity of user supplied type
4605 * cleaned up code flushing in optimization: Function calls which
4606 should signal errors for safety purposes (e.g. which ANSI says
4607 should signal errors when their arguments are of incorrect type)
4608 are no longer optimized away.
4609 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4610 * incremented fasl file version number, because changes in the
4611 implementation of sequence functions like COERCE caused
4612 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4613 in old inline expansions) to become undefined. (Actually these
4614 changes were later undone, so we might very well be binary
4615 compatible with 0.7.7 after all, but leaving the version number
4616 incremented seemed like the simplest and most conservative
4619 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4620 * An alpha-quality port to the parisc architecture running Linux,
4621 based on the old CMUCL backend, has been made. This, even more so
4622 than the other backends, should be considered still a work in
4623 progress; known problems include that the Linux kernel in 64-bit
4624 mode does not propagate the correct sigcontext structure to
4625 userspace, and consequently SBCL on a parisc64 kernel will not
4627 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4628 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4629 suppressed.) Also now that the compiler is looking at declarations
4630 in the environment, it checks optimization declarations as well,
4631 and suppresses inlining when (> DEBUG SPEED).
4632 * More fixes have been made to treatment of floating point exception
4633 treatment and other Unix signals. In particular, floating point
4634 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4635 * The detection and handling of control stack exhaustion (infinite
4636 or very deeply nested recursion) has changed. Stack exhaustion
4637 detection is now done by write-protecting pages at the OS level
4638 and applies at all optimization settings; when found, a
4639 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4640 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4641 be caught using IGNORE-ERRORS.
4642 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4643 correctly when transforming recursive LABELS functions to LETs.
4644 (thanks to Alexey Dejneka)
4645 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4646 that are names of constants or global variables.
4647 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4648 alien routines with docstrings.
4649 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4650 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4652 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4653 (thanks to Eric Marsden)
4654 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4655 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4656 * Bug fix: The SPARC backend can now compile functions involving
4657 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4658 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4659 to a functions expecting a PATHNAME-DESIGNATOR.
4660 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4661 lambda lists. (thanks to Alexey Dejneka)
4662 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4663 functions so that EXPORTing the name of the class doesn't cause
4664 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4665 to Antonio Martinez for reporting this)
4666 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4667 constant-folded without failing an assertion. (thanks to Einar
4668 Floystad Dorum for reporting this)
4669 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4670 types are now dealt with more correctly by the compiler.
4671 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4672 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4673 behaviour analogous to RENAME-FILE. This puts its behaviour more
4674 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4676 * The fasl file version number has changed again. (because of the
4677 bug fix involving the names of PCL MAKE-INSTANCE functions)
4679 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4680 * bug fix: Floating point exceptions are treated much more
4681 consistently on the x86/Linux and PPC/Linux platforms.
4682 * Array initialization with :INITIAL-ELEMENT is now much faster for
4683 cases when the compiler cannot open code the array creation, but
4684 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4685 array accesses have also seen a speed increase.
4686 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4687 specified. (This was at the root of some bad interactions between
4688 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4690 * bug fix: Internal error arguments for undefined functions are now
4691 computed correctly on the PPC/Linux platform.
4692 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4693 Raymond Toy's patch for CMU CL)
4694 * Support for the Solaris 9 operating environment has been included
4695 (thanks to Daniel Merritt)
4696 * A very ugly but hopefully complete draft of the missing FFI chapter
4697 of the manual has been created by reformatting the corresponding
4698 CMU CL manual chapter into (currently very ugly and incoherent)
4699 DocBook and bringing it up to date for SBCL behavior. Thus, the
4700 manual is now essentially complete, at least by my extreme
4701 once-and-only-once standards, whereby it's acceptable to refer to
4702 the doc strings of SB-EXT functions as the primary documentation.
4703 * The fasl file version number has changed again, due to cleanup of
4704 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4705 is no longer a static symbol.)
4707 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4708 * SBCL now builds with OpenMCL (version 0.12) as the
4709 cross-compilation host; also, more progress has been made toward
4710 bootstrapping under CLISP.
4711 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4713 * bug 158 fixed: The compiler can now deal with integer loop
4714 increments different from 1; fixing this turned out also to fix
4716 * bug 169 fixed: no more bogus warnings about using lexical bindings
4717 despite the presence of perfectly good SPECIAL declarations (thanks
4718 to David Lichteblau)
4719 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4720 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4721 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4723 * bug 180 fixed: Method combination specifications no longer ignore
4724 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4725 * bug fix: Structure type predicate functions now check their argument
4726 count as they should.
4727 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4728 correctly. (thanks to Pierre Mai)
4729 * minor incompatible change: The --noprogrammer option is deprecated
4730 in favor of the new --disable-debugger option, which is very similar.
4731 (The major difference is that it takes effect at a slightly different
4732 time at startup, causing handling of errors in --sysinit and
4733 --userinit files will be affected differently.) The
4734 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4735 been added to allow this functionality to be controlled from ordinary
4736 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4737 maintainers, who might want to run non-interactive scripts to
4738 build SBCL cores which will later be used interactively.)
4739 * minor incompatible change: The LOAD function no longer, when given
4740 a wild pathname to load, loads all files matching that pathname.
4741 Instead, an error of type FILE-ERROR is signalled.
4743 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4744 * bug 147 fixed: The compiler preserves its block link/count
4745 invariants more correctly now so that it doesn't crash. (thanks
4747 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4749 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4750 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4751 no-value-for-this-character as NIL (as specified by ANSI).
4752 * HOST-NAMESTRING on physical pathnames now returns a string that is
4753 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4754 (thanks to Christophe Rhodes)
4755 * The Alpha port handles icache flushing more correctly. (thanks to
4757 * More progress has been made toward bootstrapping under CLISP. (thanks
4758 to Christophe Rhodes)
4759 * The fasl file format has changed again, because dynamic loading
4760 on OpenBSD (which has non-ELF object files) motivated some cleanups
4761 in the way that foreign symbols are transformed and passed around.
4762 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4763 is no longer treated as whitespace by the reader, but instead as
4764 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4765 |A
\7fB|, instead of A as it used to.
4767 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4768 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4769 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4770 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4771 this as of 0.7.1.45, but was left out of the previous news section
4772 (thanks to Dan Barlow)
4773 * SBCL now runs on the Solaris operating system on SPARC architectures
4774 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4775 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4776 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4777 * SPARC backend cleanups, allowing builds of cores optimized for V8
4778 and V9 SPARCS, and also emission of code targeted to a particular
4779 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4781 * SBCL is closer to bootstrapping under CLISP, thanks to various
4782 fixes by Christophe Rhodes.
4783 * The fasl file format has changed again, to allow the compiler's
4784 INFO database to support symbol macros.
4785 * The user manual (in doc/) is formatted into HTML more nicely.
4786 (thanks to coreythomas)
4787 * The system is smarter about SUBTYPEP relationships, especially
4788 those involving NOT types (including types such as ATOM which are
4789 represented internally using NOT types). Thus SUBTYPEP is less
4790 likely to return (VALUES NIL NIL) in general, and in particular
4791 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4793 * The fasl file format has changed again, because the internal
4794 representation of types now includes a new slot to support the new
4795 SUBTYPEP-of-NOT-types logic.
4796 * (not a change in the main branch of SBCL, but a related prototype
4797 which can hopefully be merged into the main branch of SBCL in the
4798 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4799 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4800 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4801 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4802 dispatch macro character combinations. (thanks to Alexey Dejneka)
4803 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4804 that superficially look like logical namestrings correctly.
4805 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4807 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4808 default supplied-p) for &optional and &key arguments. (thanks to
4811 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4812 * incompatible change: The compiler is now less aggressive about
4813 tail call optimization, doing it only when (> SPACE DEBUG) or
4814 (> SPEED DEBUG). (This is an incompatible change because there are
4815 programs which relied on the old CMU-CL-style behavior to optimize
4816 away their unbounded recursion which will now die of stack overflow.)
4817 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4818 for non-GENCGC systems has been increased to 20M (since that
4819 seems much closer to the likely performance optimum for modern
4820 systems than the old 4M value was)
4821 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4822 and *DEBUG-PRINT-LEVEL*
4823 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4824 of CMU CL's support for SPARC, and various endianness and other
4825 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4826 * new syntactic sugar for the Unix command line: --load foo.bar is now
4827 an alternate notation for --eval '(load "foo.bar")'.
4829 ** The system now detects stack overflow and handles it gracefully,
4830 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4831 optimization settings. (This is a good thing in general, and
4832 its introduction in this version should be particularly timely
4833 for anyone whose code fails because of suppression of tail
4835 ** The system now hunts for the C variable "environ" in a more
4836 devious way, to avoid segfaults when the C library version differs
4837 between compile time and run time. (thanks to Christophe Rhodes)
4838 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4839 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4840 ** The compiler no longer issues bogus style warnings for undefined
4841 classes in the same source file as the DEFCLASSes which defined
4842 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4844 ** fixes in CONDITION class precedence list for undefined function
4845 errors (thanks to Alexei Dejneka)
4846 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4847 correctly. (thanks to Dan Barlow)
4848 ** portability fixes aiming at bootstrapping under CLISP (thanks
4849 to Dave McDonald and Christophe Rhodes)
4850 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4851 ** fixes in type translation and and type inference (thanks to
4853 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4854 ** various fixes in the new ports (thanks to Dan Barlow)
4855 * several changes related to debugging:
4856 ** suppression of tail recursion, as noted above
4857 ** stack overflow detection, as noted above
4858 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4859 is now the default. (For some time encapsulation has been more
4860 reliable than the breakpoint-based :ENCAPSULATE NIL
4861 implementation, at least on X86 systems; and I just noticed that
4862 encapsulation also seems closer to the spirit of the ANSI
4865 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4867 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4868 up properly again. (There was a packaging bug in 0.7.0 which
4869 left their definitions in SB-SYS::LOAD-FOREIGN and
4870 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4871 vital for most things which interface to C-level interfaces,
4872 like extensions working with sockets or databases or
4873 Perl-compatible regexes or whatever, and the need to fix
4874 this bug is the main reason that 0.7.1 was released so
4876 ** DEFGENERIC is now choosier about the methods it redefines, so that
4877 reLOADing a previously-LOADed file containing DEFGENERICs does
4878 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4879 cycle now works as it should. (thanks to Alexey Dejneka)
4880 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4881 fixed by Christophe Rhodes. (He actually submitted this patch
4882 months ago, and I delayed until after 0.7.0.)
4883 ** Bug 111 (internal compiler confusion about runtime checks on
4884 FUNCTION types) was fixed by Alexey Dejneka.
4885 * Some internal cleanups (getting rid of variables which aren't
4886 needed now that the byte interpreter is gone) caused the fasl
4887 file format number to change again.
4889 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4890 * major incompatible change: The default fasl file extension, i.e. the
4891 default extension for files produced by COMPILE-FILE, has changed
4892 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4894 ** There are many changes in the implementation of the compiler.
4895 SBCL is now essentially a compiler-only implementation of ANSI
4896 Common Lisp. EVAL still "interprets" a few special cases, but
4897 almost all the interesting cases are handled by creating
4898 a LAMBDA expression, calling COMPILE on it, then calling
4899 FUNCALL on the result.
4900 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4901 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4902 Since the code is newer, there might still be some new bugs
4903 (though not as many as before Martin Atzmueller's fixes:-). But
4904 the new code is substantially simpler and clearer, and hopefully
4905 any remaining bugs will be simpler, less fundamental, and more
4906 fixable then the bugs in the old code.
4907 ** The revised compiler is still a little unsteady on its feet.
4909 *** The debugging information it produces (particularly the names
4910 of FUNCTION objects) is sometimes much less useful than what
4911 the old compiler produced.
4912 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4913 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4914 within a MACROLET) has been temporarily weakened.
4915 ** There are new compiler optimizations for various functions:
4916 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4917 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4918 *** the math functions TRUNCATE, FLOOR, and CEILING
4919 *** the function-of-all-trades COERCE
4920 Mostly these should be transparent, but there's one
4921 potentially-annoying problem (bug 117): when the compiler
4922 inline-expands a function and does type analysis on the result,
4923 it can create control paths which have type mismatches, and
4924 when it can't prove that those control paths aren't taken,
4925 it will issue WARNINGs about the type mismatches. This is
4926 a particular problem in practice for the new sequence functions.
4927 It's not clear how this should be fixed, and for now, a
4928 workaround is given in the entry for 117 in the BUGS file.
4929 ** (Because of the interaction between the two previous items --
4930 occasional inlining problems and new inline expansions -- some
4931 of the new sequence function optimizations won't really kick in
4932 completely until debugging information, and then inlining, are
4933 straightened out in some future version.)
4934 * minor incompatible changes:
4935 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4936 DIRECTORY no longer implicitly promotes NIL slots of its
4937 pathname argument to :WILD. In particular, when you ask for the
4938 contents of a directory (which you used to be able to do without
4939 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4940 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4941 ** changes in behavior that ANSI explicitly defines to be
4942 implementation dependent:
4943 *** The new compiler-only implementation still conforms with ANSI,
4944 but acts a little different than before. Besides the obvious
4945 changes in performance tradeoffs (that the cost per form passed
4946 to EVAL has gone up, and the cost per form executed by EVAL
4947 has gone down), the behavior of the system changes a little
4948 because there are no longer any interpreted function objects.
4949 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4950 e.g. doing COMPILE on the output of interactive DEFUN is
4952 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4954 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4955 has changed from (:CL) to NIL.
4956 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4957 e.g. #\Newline and #\Tab, don't have names specified in the
4958 ANSI Common Lisp standard, is now based on their ASCII symbolic
4959 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4960 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4961 are no longer used for output.
4962 ** changes in internal implementation constants:
4963 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4964 4 million. (If your application spends a lot of time GCing and
4965 you have a lot of RAM, you might want to experiment with
4966 increasing it even more.)
4967 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4968 However, almost all old code should still continue to work without
4969 immediate update, as SB-C-CALL is now a (deprecated) nickname
4971 ** Old operator names in the style DEF-FOO are now deprecated in
4972 favor of new corresponding names DEFINE-FOO, for consistency with
4973 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4974 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4975 internal symbols, but a few supported extensions like
4976 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4977 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4978 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4979 etc. as you get deeper into recursive calls to the debugger
4980 command loop, instead of the old "5]", "5]]", "5]]]"
4981 sequence. (I was motivated to do this when squabbles between
4982 ILISP and SBCL left me very deeply nested in the debugger. In the
4983 short term, this change will probably provoke more ILISP/SBCL
4984 squabbles, but hopefully it will be an improvement in the long run.)
4985 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4986 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4987 function, since declaiming return types involving aliens is
4988 (1) annoyingly messy to do by hand and (2) vital to efficient
4989 compilation of code which calls such functions.
4990 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4991 longer reexported by the SB-EXT package. They're solely useful
4992 for alien code, so it seems more logical that you should get
4993 them from the SB-ALIEN package, not in SB-EXT.
4994 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4995 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4996 features. Instead, the code that they used to control is always
4997 built into the system.
4998 * many other bug fixes
4999 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
5000 advantage of the new EVAL-WHEN stuff and to clean them up in
5001 general, and they are now more ANSI-compliant in a number of
5002 ways. Martin Atzmueller is responsible for a lot of this.
5003 ** Besides the cleanups discussed above, Martin Atzmueller fixed
5005 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
5006 *** correct ERROR type for various file operations
5007 *** some fixes for Lisp streams
5008 *** DEFMETHOD syntax checking
5009 *** changing old weird representation of debug information as
5010 strings (which, among their other deficiencies, don't transform
5011 correctly when you rename packages, and don't change their
5012 print representation when you change things like *PACKAGE*
5013 and *PRINT-LENGTH*) to symbols and lists of symbols
5014 He also made several improvements and fixed several bugs in DESCRIBE.
5015 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
5017 *** misbehavior of WRITE-STRING/WRITE-LINE
5018 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
5019 and several other LOOP problems as well
5020 *** DIRECTORY when similar filenames are present
5021 *** DEFGENERIC with :METHOD options
5022 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
5023 *** bug in the optimization of ARRAY-ELEMENT-TYPE
5024 *** argument ordering in FIND with :TEST option
5025 *** mishandled package designator argument in APROPOS-LIST
5026 *** various problems in the backquote readmacro
5027 *** a bug in APROPOS
5028 *** probably some others that I'm not describing very well here,
5029 since the CVS log documents them by reference to sbcl-devel
5030 messages, and the SourceForge archives aren't working well.:-(
5031 ** Dan Barlow improved the Alpha port (and is making progress on the
5032 PPC port, for those of you who think different).
5033 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
5034 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
5035 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
5036 continues to work on the SPARC port (for those of you in a position
5037 to look down upon our little PC-compatible boxes from a great height).
5038 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
5039 and uses that copy, rather than the current dynamic value, when
5040 it's trying to decide whether to truncate output. Thus e.g.
5041 (let ((*print-lines* 50))
5042 (pprint-logical-block (stream nil)
5044 (let ((*print-lines* 8))
5045 (print (aref possiblybigthings i) stream)))))
5046 should now truncate the logical block only at 50 lines, instead of
5047 often truncating it at 8 lines, as it did before.
5048 * The doc/cmucl/ directory, containing old CMU CL documentation
5049 from the time of the fork, is no longer part of the base system.
5050 SourceForge has shut down its anonymous FTP service, and with it
5051 my original plan for distributing the old CMU CL documentation
5052 there. For now, if you need these files you can download an old
5053 SBCL source release and extract them from it.
5054 * The fasl file version number changed again, for dozens of reasons,
5055 some of which are apparent above.
5057 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
5058 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
5059 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
5060 patch, so that DISASSEMBLE output is much nicer.
5061 * The code in the SB-PROFILE package now seems reasonably stable.
5062 I still haven't decided what the final interface should look like
5063 (I'd like PROFILE to interact cleanly with TRACE, since both
5064 facilities use function encapsulation) but if you have a need
5065 for profiling now, you can probably use it successfully with
5066 the current CMU-CL-style interface.
5067 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
5068 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
5069 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
5070 link now returns the dangling link itself, and for similar
5071 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
5072 link itself. (In these cases the old code signalled an error and
5073 looped endlessly, respectively.) Thus, DIRECTORY now works even
5074 in the presence of dangling and cyclic symbolic links.
5075 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
5076 is now a supported extension again, since the consensus on
5077 sbcl-devel was that it can be useful for ordinary development
5078 work, not just for debugging SBCL itself.
5079 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
5080 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
5081 that currently-defined functions might be redefined later with
5082 different return types.
5083 * Hash tables can be printed readably, as inspired by CMU CL code
5084 of Eric Marsden and SBCL code of Martin Atzmueller.
5085 * better error handling in CLOS method combination, thanks to
5086 Martin Atzmueller porting Pierre Mai's CMU CL patches
5087 * more overflow fixes for >16Mbyte I/O buffers
5088 * A bug in READ has been fixed, so that now a single Ctrl-D
5089 character suffices to cause end-of-file on character streams.
5090 In particular, now you only need one Ctrl-D at the command
5091 line (not two) to exit SBCL.
5092 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
5094 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
5095 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
5096 does the right thing.
5097 * The implementation of some type tests, especially for CONDITION
5098 types, is now tidier and maybe faster, due to CMU CL code
5099 originally by Douglas Crosher, ported by Martin Atzmueller.
5100 * Some math functions have been fixed, and there are new
5101 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
5102 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
5103 * (There are also some new optimizers in contrib/*-extras.lisp. Those
5104 aren't built into sbcl-0.6.13, but are a sneak preview of what's
5105 likely to be built into sbcl-0.7.0.)
5106 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
5107 CL, ported to SBCL by Martin Atzmueller)
5108 * DESCRIBE now gives more information in some cases. (Pierre Mai's
5109 patch to CMU CL, ported to SBCL by Martin Atzmueller)
5110 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
5111 * There's a new slam.sh hack to shorten the edit/compile/debug
5112 cycle for low-level changes to SBCL itself, and a new
5113 :SB-AFTER-XC-CORE target feature to control the generation of
5114 the after-xc.core file needed by slam.sh.
5115 * minor incompatible change: The ENTRY-POINTS &KEY argument to
5116 COMPILE-FILE is no longer supported, so that now every function
5117 gets an entry point, so that block compilation looks a little
5118 more like the plain vanilla ANSI section 3.2.2.3 scheme.
5119 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
5120 returns the number of bytes consed since the system started,
5121 rather than the number consed since the first time the function
5122 was called. (The new definition parallels ANSI functions like
5123 CL:GET-INTERNAL-RUN-TIME.)
5124 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
5125 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
5126 supported. Now DIRECTORY always does the abstract Common-Lisp-y
5127 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
5128 * Fasl file version numbers are now independent of the target CPU,
5129 since historically most system changes which required version
5130 number changes have affected all CPUs equally. Similarly,
5131 the byte fasl file version is now equal to the ordinary
5134 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
5135 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
5136 is no longer recognized. I apologize for this, because it was
5137 listed in SB-EXT as a supported extension, but I found that
5138 its existing behavior was poorly specified, as well as incorrectly
5139 specified, and it looked like too much of a mess to straighten it
5140 out. I have enough on my hands trying to get ANSI stuff to work..
5141 * many patches ported from CMU CL by Martin Atzmueller, with
5142 half a dozen bug fixes in pretty-printing and the debugger, and
5143 half a dozen others elsewhere
5144 * fixed bug 13: Floating point infinities are now supported again.
5145 They might still be a little bit flaky, but thanks to bug reports
5146 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
5147 as flaky as they were.
5148 * The --noprogrammer command line option is now supported. (Its
5149 behavior is slightly different in detail from what the old man
5150 page claimed it would do, but it's still appropriate under the
5151 same circumstances that the man page talks about.)
5152 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
5153 are now supported, and enabled by default. Thus, the compiler can
5154 handle many floating point and complex operations much less
5155 inefficiently. (Thus e.g. you can implement a complex FFT
5157 * The compiler now detects type mismatches between DECLAIM FTYPE
5158 and DEFUN better, and implements CHECK-TYPE more correctly, and
5159 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
5160 thanks to patches from Martin Atzmueller.
5161 * various fixes to make the cross-compiler more portable to
5162 ANSI-conforming-but-different cross-compilation hosts (notably
5163 Lispworks for Windows, following bug reports from Arthur Lemmens)
5164 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
5165 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
5166 fixed thanks to Pierre Mai's CMU CL patches.
5167 * fixes to keep the system from overflowing internal counters when
5168 it tries to use i/o buffers larger than 16M bytes
5169 * fixed bug 45a: Various internal functions required to support
5170 complex special functions have been merged from CMU CL sources.
5171 (When I was first setting up SBCL, I misunderstood a compile-time
5172 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
5173 * improved support for type intersection and union, fixing bug 12
5174 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
5175 more obscure bugs as well
5176 * some steps toward byte-compiling non-performance-critical
5177 parts of the system, courtesy of patches from Martin Atzmueller
5178 * Christophe Rhodes has made some debian packages of sbcl at
5179 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
5180 From his sbcl-devel e-mail of 2001-04-08 they're not completely
5181 stable, but are nonetheless usable. When he's ready, I'd be happy
5182 to add them to the SourceForge "File Releases" section. (And if
5183 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
5184 * new fasl file format version number (because of changes in
5185 internal representation of (OR ..) types to accommodate the new
5186 support for (AND ..) types, among other things)
5188 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
5189 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
5191 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
5193 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
5194 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
5195 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
5196 * DESCRIBE now works on structure objects again.
5197 * Most function call argument type mismatches are now handled as
5198 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
5199 know whether the function will be redefined before the call is
5200 executed. (The compiler could flag local calls with full WARNINGs,
5201 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
5202 it doesn't keep track of enough information to know whether calls
5203 are local in this sense.)
5204 * Compiler output is now more verbose, with messages truncated
5205 later than before. (There should be some supported way for users
5206 to override the default verbosity, but I haven't decided how to
5207 provide it yet, so this behavior is still controlled by the internal
5208 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
5209 src/compiler/ir1util.lisp.)
5210 * Fasl file format version numbers have increased again, because
5211 support for the Gray streams extension changes the layout of the
5212 system's STREAM objects.
5213 * The Gray subclassable streams extension now works, thanks to a
5214 patch from Martin Atzmueller.
5215 * The full LOAD-FOREIGN extension (not just the primitive
5216 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
5217 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
5218 but like most other programs, it defaults to copying the Unix
5219 environment from the original process instead of starting the
5220 new process in an empty environment.
5221 * Extensions which manipulate the Unix environment now support
5222 an :ENVIRONMENT keyword option which doesn't smash case or
5223 do other bad things. The CMU-CL-style :ENV option is retained
5224 for porting convenience.
5225 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
5226 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
5228 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
5230 * A patch from Martin Atzmueller seems to have solved the SIGINT
5231 problem, and as far as we know, signal-handling now works cleanly.
5232 (If you find any new bugs, please report them!)
5233 * The system no longer defaults Lisp source file names to types
5234 ".l", ".cl", or ".lsp", but only to ".lisp".
5235 * The compiler no longer uses special default file extensions for
5236 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
5237 seems to expect a single default extension for all compiled code,
5238 and there's no compelling reason to try to stretch the standard
5239 to allow two different extensions.) Instead, byte-compiled files
5240 default to the same extension as native-compiled files.
5241 * Fasl file format version numbers have increased again, because
5242 a rearrangement of internal implementation packages made some
5243 dumped symbols in old fasl files unreadable in new cores.
5244 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
5245 many fewer weird special cases.
5246 * Bug #17 (differing COMPILE-FILE behavior between logical and
5247 physical pathnames) has been fixed, and some related misbehavior too,
5248 thanks to a patch from Martin Atzmueller.
5249 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
5250 by Tim Moore, ported to SBCL by Martin Atzmueller.
5251 * Martin Atzmueller fixed several filesystem-related problems,
5252 including bug #36, in part by porting CMU CL patches, which were
5253 written in part by Paul Werkowski.
5254 * More compiler warnings in src/runtime/ are gone, thanks to
5255 more patches from Martin Atzmueller.
5256 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
5259 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
5261 * DESCRIBE now works on CONDITION objects.
5262 * The debugger now handles errors which arise when trying to print
5263 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
5265 * The build system now uses an additional file, customize-target-features.lisp,
5266 to allow local modifications to the target *FEATURES* list. (The point of
5267 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
5268 debugging features enabled, without having to worry about propagating it
5269 into everyone's system when I do a "cvs update".) When no
5270 customize-target-features.lisp file exists, the target *FEATURES* list
5271 should be constructed the same way as before.
5272 * fixed bugs in DEFCONSTANT ANSI-compatibility:
5273 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
5274 warn about behavior which is undefined under the ANSI spec. Note: This
5275 is specified by ANSI, but it's not very popular with programmers.
5276 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
5277 macro in the SBCL sources for an example of a workaround which you
5278 might use to make portable ANSI-standard code which does what you want.
5279 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
5280 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
5281 when it's used as a non-toplevel form. (This is required in order
5282 to implement the DEFCONSTANT-EQX macro.)
5283 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
5284 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
5285 funny things, due to bugs in the implementation of EVAL-WHEN
5286 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
5287 non-toplevel DEFCONSTANTs are uncommon.)
5288 * The core file version number and fasl file version number have been
5289 incremented, because the old noncompliant DEFCONSTANT behavior involved
5290 calling functions which no longer exist, and because I also took the
5291 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
5292 * fixed bug 1 (error handling before read-eval-print loop starts), and
5293 redid debugger restarts and related debugger commands somewhat while
5295 ** The QUIT debugger command is gone, since it did something
5296 rather different than the SB-EXT:QUIT command, and since it never
5297 worked properly outside the main toplevel read/eval/print loop.
5298 Invoking the new TOPLEVEL restart provides the same functionality.
5299 ** The GO debugger command is also gone, since you can just invoke
5300 the CONTINUE restart directly instead.
5301 ** The TOP debugger command is also gone, since it's redundant with the
5302 FRAME 0 command, and since it interfered with abbreviations for the
5304 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
5306 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
5307 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
5308 some porting work by Martin Atzmueller)
5309 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
5310 with a boatload of patches which he ported to SBCL. Now that those
5312 ** The system tries to make sure that its low-priority messages
5313 are prefixed by semicolons, to help people who like to use
5314 syntax highlighting in their ILISP buffer. (This patch
5315 was originally due to Raymond Toy.)
5316 ** The system now optimizes INTEGER-LENGTH better, thanks to more
5317 patches originally written by Raymond Toy.
5318 ** The compiler understands coercion between single-value and
5319 multiple-VALUES type expressions better, getting rid of some very
5320 weird behavior, thanks to patches originally by Robert MacLachlan
5321 and Douglas Crosher.
5322 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
5323 lambda lists, thanks to a patch originally by Pierre Mai.
5324 ** The system no longer bogusly warns about "abbreviated type
5326 ** The compiler gets less confused by inlining and RETURN-FROM,
5327 thanks to some patches originally by Tim Moore.
5328 ** The system no longer hangs when dumping circular lists to fasl
5329 files, thanks to a patch originally from Douglas Crosher.
5330 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
5331 undefined function error.
5332 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
5333 in structure slot initforms) for now, documented workaround instead:-|
5334 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
5335 * fixed bug 5: added stubs for various Gray stream functions called
5336 in the not-a-CL:STREAM case, so that even when Gray streams aren't
5337 installed, at least appropriate type errors are generated
5338 * fixed bug 8: better reporting of various PROGRAM-ERRORs
5339 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
5340 consistently in DEFMETHOD forms.
5341 * removed bug 21 from BUGS, since Martin Atzmueller points out that
5342 it doesn't seem to affect SBCL after all
5343 * The C runtime system now builds with better optimization and many
5344 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
5346 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
5348 * The system is now under CVS at SourceForge (instead of the
5349 CVS repository on my home machine).
5350 * The new signal handling code has been tweaked to treat register
5351 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
5352 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
5353 downstream, e.g. in debug-int.lisp, has implicit dependencies
5354 on the unsignedness of integer representation of machine words,
5355 and that caused the system to bomb out with infinite regress
5356 when trying to recover from type errors involving signed values,
5357 e.g. (BUTLAST '(1 2 3) -1).)
5358 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
5359 (This was one of the bugs Peter Van Eynde reported back in July.)
5360 * The system now uses code inspired by Colin Walters' O(N)
5361 implementation of MAP (from the cmucl-imp@cons.org mailing
5362 list, 2 September 2000) when it can't use a DEFTRANSFORM to
5363 inline the MAP operation, and there is more than one
5364 sequence argument to the MAP call (so that it can't just
5365 do ETYPECASE once and for all based on the type of the
5366 single sequence argument). (The old non-inline implementation
5367 of the general M-argument sequence-of-length-N case required
5368 O(M*N*N) time when any of the sequence arguments were LISTs.)
5369 * The QUIT :UNIX-CODE keyword argument has been renamed to
5370 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5371 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5373 * Raymond Wiker's patches to port dynamic loading from Linux to
5374 FreeBSD have been added.
5375 * The BUGS file is now more nearly up to date, thanks in large part
5376 to Martin Atzmueller's review of it.
5377 * The debugger now flushes standard output streams before it begins
5378 its output ("debugger invoked" and so forth).
5379 * The core version number and fasl file version number have both
5380 been incremented, because of incompatible changes in the layout
5382 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5383 used to not be called for a saved Lisp image.)
5384 * Martin Atzmueller's version of a patch to fix a compiler crash,
5385 as posted on sbcl-devel 13 September 2000, has been installed.
5386 * Instead of installing Martin Atzmueller's patch for the
5387 compiler transform for SUBSEQ, I deleted the compiler transform,
5388 and transforms for some similar consing operations.
5389 * A bug in signal handling which kept TRACE from working on OpenBSD
5391 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5392 away by constant folding
5393 * The system now defines its address space constants in one place
5394 (in the Lisp sources), and propagates them automatically elsewhere
5395 (through GENESIS and the sbcl.h file). Therefore, patching the
5396 address map is less unnecessarily tedious and error-prone. The
5397 Lisp names of address space constants have also been systematized.
5398 * CVS tags like dollar-Header-dollar have been removed from
5399 the sources, because they have never saved me trouble and
5400 they've been source of trouble working with patches and other
5401 diff-related operations.
5402 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5403 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5405 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5407 * The system has been ported to OpenBSD.
5408 * The system now compiles with a simple "sh make.sh" on the systems
5409 that it's supported on. I.e., now you no longer need to tweak
5410 text in the target-features.lisp-expr and symlinks in src/runtime/
5411 by hand, the make.sh takes care of it for you.
5412 * The system is no longer so grossly inefficient when compiling code
5413 involving vectors implemented as general (not simple) vectors (VECTOR T),
5414 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5415 which dares to use the various sequence functions on non-simple
5416 vectors, takes less of a performance hit.
5417 * There is now a primitive type predicate VECTOR-T-P
5418 to test for the (VECTOR T) type, so that e.g.
5419 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5420 can now be compiled with some semblance of efficiency. (The old code
5421 turned the type declaration into a full call to %TYPEP at runtime!)
5422 * AREF on (VECTOR T) is still not fast, since it's still compiled
5423 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5424 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5425 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5426 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5427 without making full calls to SUBTYPEP at runtime.
5428 (Some analogous efficiency issues for non-simple vectors specialized to
5429 element types other than T, or for non-simple multidimensional arrays,
5430 have not been addressed. They could almost certainly be handled the
5431 same way if anyone is motivated to do so.)
5432 * The changes in array handling break binary compatibility, so
5433 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5434 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5435 instead of (VALUES T T).
5436 * By following the instructions that Dan Barlow posted to sbcl-devel
5437 on 2 July 2000, I was able to enable primitive dynamic object
5438 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5439 functionality is not implemented (since it calls ld to resolve
5440 library references automatically, requiring RUN-PROGRAM for its
5441 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5442 try to resolve library references) is now supported.
5443 * The system now flushes the standard output streams when it terminates,
5444 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5445 them at several other probably-convenient times, e.g. in each pass of
5446 the toplevel read-eval-print loop, and after evaluating a form given
5447 as an "--eval" command-line option. (These changes were motivated by a
5448 discussion of stream flushing issues on cmucl-imp in August 2000.)
5449 * The source transform for TYPEP of array types no longer assumes
5450 that an array whose element type is a not-yet-defined type
5451 is implemented as an array of T, but instead punts, so that the
5452 type will be interpreted at runtime.
5453 * There is now some support for cross-compiling in make.sh: each of
5454 the phases of make.sh has its own script. (This should be transparent
5455 to people doing ordinary, non-cross-compile builds.)
5456 * Since my laptop doesn't have hundreds of megabytes of memory like
5457 my desktop machine, I became more motivated to do some items on
5458 my to-do list in order to reduce the size of the system a little:
5459 ** Arrange for various needed-only-at-cold-init things to be
5460 uninterned after cold init. To support this, those things have
5461 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5462 symbols with such names are now uninterned after cold init).
5463 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5464 fasl files for cold load.
5465 ** Remove the old compiler structure pooling code (which used to
5466 be conditional on the target feature :SB-ALLOC) completely.
5467 ** Redo the representation of some data in cold init to be more compact.
5468 (I also looked into supporting byte compiled code at bootstrap time,
5469 which would probably reduce the size of the system a lot, but that
5470 looked too complicated, so I punted for now.)
5471 * The maximum signal nesting depth in the src/runtime/ support code has
5472 been reduced from 4096 to 256. (I don't know any reason for the very
5473 large old value. If the new smaller value turns out to break something,
5474 I'll probably just bump it back up.)
5475 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5477 * Many, many bugs reported by Peter Van Eynde have been added to
5478 the BUGS list; some have even been fixed.
5479 * While enabling dynamic object file loading, I tried to make the
5480 code easier to understand, renaming various functions and variables
5481 with less ambiguous names, and changing some function calling
5482 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5483 * While trying to figure out how to do the OpenBSD port, I tried to
5484 clean up some of the code in src/runtime/. In particular, I dropped
5485 support for non-POSIX signal handling, added various comments,
5486 tweaked the code to reduce the number of compilation warnings, and
5487 renamed some files to increase consistency.
5488 * To support the new automatic configuration functionality in make.sh,
5489 the source file target-features.lisp-expr has been replaced with the
5490 source file base-target-features.lisp-expr and the machine-generated
5491 file local-target-features.lisp-expr.
5492 * fixed a stupid quoting error in make.sh so that using CMU CL
5493 "lisp -batch" as cross-compilation host works again
5495 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5497 * DESCRIBE no longer tries to call itself recursively to describe
5498 bound/fbound values, so that it no longer fails on symbols which are
5499 bound to themselves (like keywords, T, and NIL).
5500 * DESCRIBE now works on generic functions.
5501 * The printer now prints less-screwed-up representations of closures
5502 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5503 * A private symbol is used instead of the :EMPTY keyword previously
5504 used to mark empty slots in hash tables. Thus
5505 (DEFVAR *HT* (MAKE-HASH-TABLE))
5506 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5507 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5508 now does what ANSI says that it should. (You can still get
5509 similar noncompliant behavior if bang on the hash table
5510 implementation with all the symbols you get back from
5511 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5512 This breaks binary compatibility, since tests for equality to
5513 :EMPTY are wired into things like the macroexpansion of
5514 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5516 * There's now a minimal placeholder implementation for CL:STEP,
5517 as required by ANSI.
5518 * An obscure bug in the interaction of the normal compiler, the byte
5519 compiler, inlining, and structure predicates has been patched
5520 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5521 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5523 * Missing ordinary arguments in a macro call are now detected even
5524 when the macro lambda list contains &KEY or &REST.
5525 * The debugger no longer complains about encountering the top of the
5526 stack when you type "FRAME 0" to explicitly instruct it to go to
5527 the top of the stack. And it now prints the frame you request even
5528 if it's the current frame (instead of saying "You are here.").
5529 * As specified by ANSI, the system now always prints keywords
5530 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5532 * The default initial SIZE of HASH-TABLEs is now smaller.
5533 * Type information from CLOS class dispatch is now propagated
5534 into DEFMETHOD bodies, so that e.g.
5535 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5537 is now basically equivalent to
5538 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5539 (DECLARE (TYPE SINGLE-FLOAT X))
5541 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5542 operation, without having to do run-time type dispatch.
5543 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5544 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5545 * Since I don't seem to be making big file reorganizations very often
5546 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5547 is overflowing my ability to conveniently back them up), I've finally
5548 checked the system into CVS. (The CVS repository is on my home system,
5549 not at SourceForge -- putting it on SourceForge might come later.)
5550 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5551 high-level GC-NOTIFY-FOO functions send their output. (There's
5552 still very little control of where low-level verbose GC functions
5553 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5554 less than it used to -- the GC-NOTIFY-FOO functions are now under
5555 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5556 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5557 in only one place in the source code, and propagates it automatically
5558 everywhere that it's needed. Thus e.g. when I bump the version from
5559 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5560 * The C source files now include boilerplate legalese and documentation
5561 at the head of each file (just as the Lisp source files already did).
5562 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5563 to his page will be replaced with a link to his new CLiki service.
5565 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5567 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5568 * The build process now looks for GNU make under the default name "gmake",
5569 instead of "make" as it used to. If GNU make is not available as "gmake"
5570 on your system, you can change this default behavior by setting the
5571 GNUMAKE environment variable.
5572 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5573 can build without error under CMU CL.
5575 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5577 * There is now a partial SBCL user manual (with some new text and some
5578 text cribbed from the CMU CL manual).
5579 * The beginnings of a profiler have been added (starting with the
5580 CMU CL profiler and simplifying and cleaning up). Eventually the
5581 main interface should be through the TRACE macro, but for now,
5582 it's still accessed through vaguely CMU-CL-style functions and macros
5583 exported from the package SB-PROFILE.
5584 * Some problems left over from porting CMU CL to the new
5585 cross-compilation bootstrap process have been cleaned up:
5586 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5587 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5588 part when getting rid of the compiler *BACKEND* stuff.)
5589 ** The value of *NULL-TYPE* was screwed up, because it was
5590 being initialized before the type system knew the final
5591 definition of the 'NULL type. This screwed up several key
5592 optimizations in the compiler, causing inefficiency in all sorts
5593 of places. (I found it because I wanted to understand why
5594 GET-INTERNAL-RUN-TIME was consing.)
5595 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5596 PROCLAIM FTYPE information. Unfortunately this broke binary
5597 compatibility again, since now the forms output by DEFGENERIC
5598 to refer to functions which didn't exist in 0.6.3.
5599 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5600 can use the new (as of 0.6.3) transform for SOME into MAP into
5602 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5603 values to (INTEGER 0 1000000), so that the time code will no longer
5604 occasionally get blown up by Linux returning 1000000 microseconds
5605 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5606 its output conform to the ANSI spec. (Alas, this makes its output
5607 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5608 * A full call to MAP NIL with a single sequence argument no longer conses.
5609 * fixes to problems pointed out by Martin Atzmueller:
5610 * The manual page no longer talks about multiprocessing as though
5611 it were currently supported.
5612 * The ILISP support patches have been removed from the distribution,
5613 because as of version 5.10.1, ILISP now supports SBCL without us
5614 having to maintain patches.
5615 * added a modified version of Raymond Toy's recent CMU CL patch for
5616 EQUALP comparison of HASH-TABLE
5618 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5620 * The system still can't cross-compile itself with
5621 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5622 entails), but at least it can compile more of itself that way
5623 than it used to be able to, and various buglets which were uncovered
5624 by trying to cross-compile itself that way have now been fixed.
5625 * This release breaks binary compatibility again. This time
5626 at least I've incremented the FASL file format version to 2, so that the
5627 problem can be detected reliably instead of just causing weird errors.
5628 * various new style warnings:
5629 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5630 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5631 ** using the lexical binding of a variable named in the *FOO* style
5632 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5633 as specified by ANSI.
5634 * *RANDOM-STATE* is no longer automatically initialized from
5635 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5636 default behavior of the system is to repeat its behavior every time
5637 it's run. If you'd like to change this behavior, you can always
5638 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5639 old convention there was no comparably easy way to get the system to
5640 repeat its behavior every time it was run.
5641 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5642 FTYPE declarations has been removed, in favor of their ANSI
5643 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5644 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5645 the types of their sequence arguments can be determined at compile time.
5646 This is done through a new open code expansion for MAP which eliminates
5647 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5648 argument types can be determined at compile time.
5649 * The optimizer now transforms COERCE into an identity operation when it
5650 can prove that the coerced object is already of the correct type. (This
5651 can be a win for machine generated code, including the output of other
5652 optimization transforms, such as the MAP transform above.)
5653 * Credit information has been moved from source file headers into CREDITS.
5654 * Source file headers have been made more standard.
5655 * The CASE macro now compiles without complaining even when it has
5658 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5660 * (Note that the way that the PCL macroexpansions were rewritten
5661 to accommodate the change in DEFGENERIC below breaks binary
5662 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5663 not run under sbcl-0.6.2. Once we get out of alpha releases,
5664 i.e. hit release 1.0.0, we'll probably try to maintain binary
5665 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5666 and sbcl-1.4.4. Until then, however, it might be fairly common
5667 for maintenance releases to break binary compatibility.)
5668 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5670 * The reporting of unhandled signals has been changed to print some
5671 explanatory text as well as the report form. (Previously only
5672 the report form was printed.)
5673 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5674 it defines, so that the compiler no longer issues undefined function
5675 warnings for compiled-but-not-yet-loaded generic functions.
5676 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5677 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5678 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5679 section of the ANSI spec are supported.
5680 * The "" nickname for the "KEYWORD" package has been removed.
5681 The reader still handles symbol tokens which begin with a package marker
5682 as keywords, but it doesn't expose its mechanism for doing so in the
5683 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5684 * The system now issues STYLE-WARNINGs for contradictory TYPE
5685 proclamations. (Warnings for contradictory FTYPE proclamations would
5686 be nice too, but those can't be done usefully unless the type system
5687 is made smarter about FUNCTION types.)
5688 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5689 systematized, so that "*target-*.lisp is supposed to exist only on the
5690 target and imply that there's a related file which exists on the
5691 host, and *host-*.lisp is supposed to exist only on the host and imply
5692 that there's a related file which exists on the target. This involves a
5693 lot of renaming. Hopefully the acute confusion caused by the renaming
5694 will be justified by the reduction in chronic confusion..
5695 ** runtime-type.lisp -> early-target-type.lisp
5696 ** target-type.lisp -> late-target-type.lisp
5697 ** early-host-format.lisp -> early-format.lisp
5698 ** late-host-format.lisp -> late-format.lisp
5699 ** host-error.lisp -> misc-error.lisp
5700 ** early-error.lisp -> early-target-error.lisp
5701 ** late-error.lisp -> late-target-error.lisp
5702 ** host-defboot.lisp -> early-defboot.lisp
5703 ** code/misc.lisp -> code/target-misc.lisp
5704 ** code/host-misc.lisp -> code/misc.lisp
5705 ** code/numbers.lisp -> code/target-numbers.lisp
5706 ** code/early-numbers.lisp -> numbers.lisp
5707 ** early-host-type.lisp -> early-type.lisp
5708 ** late-host-type.lisp -> late-type.lisp
5709 ** host-typep.lisp -> typep.lisp
5710 ** load.lisp -> target-load.lisp
5711 ** host-load.lisp -> load.lisp
5712 ** host-disassem.lisp -> disassem.lisp
5713 ** host-insts.lisp -> insts.lisp
5714 ** byte-comp.lisp -> target-byte-comp.lisp
5715 ** host-byte-comp.lisp -> byte-comp.lisp
5716 ** host-signal.lisp -> signal.lisp
5717 ** host-defstruct.lisp -> defstruct.lisp
5718 ** late-target-type.lisp -> deftypes-for-target.lisp
5719 Furthermore, several other previously target-only files foo.lisp (e.g.
5720 hash-table.lisp and random.lisp) have been split into a target-and-host
5721 foo.lisp file and a target-only target-foo.lisp file, with their key type
5722 definitions in the target-and-host part, so that the cross-compiler will
5723 know more about target types.
5724 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5725 gone away. In their place are various *BACKEND-FOO* variables
5726 corresponding to the slots of the old structure.
5727 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5728 into the target SBCL has been fixed.
5729 * The chill.lisp system for loading cold code into a running SBCL
5731 * Support for the CMU CL "scavenger hook" extension has been removed.
5732 (It was undocumented and unused in the CMU CL sources that SBCL was
5733 derived from, and stale in sbcl-0.6.1.)
5734 * Various errors in the cross-compiler type system were detected
5735 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5736 (enabling various consistency checks). Many of them were fixed,
5737 but some hard problems remain, so the compiler is back to
5738 running without *TYPE-SYSTEM-INITIALIZED* for now.
5739 * As part of the cross-compiler type system cleanup, I implemented
5740 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5741 * I have started adding UNCROSS calls throughout the type system
5742 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5743 on cross-compiler input files will be able to go away, and various
5745 * CONSTANTP now returns true for quoted forms (as explicitly required
5748 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5750 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5751 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5752 (MAKE-LIST -1). In the longer term, it ought to become true
5753 that declarations are assertions even at SAFETY 1. For now, it's not
5754 quite true even at SAFETY 3, but it's at least more nearly true..
5755 (Note that this change seems to increases the size of the system by
5756 O(5%) and to decrease the speed of the compiler by 20% or more.)
5757 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5758 for the problem of printing dozens of lines of distracting information
5759 about low-level system machinery as part of the top stack frame
5760 on entry to the debugger when an undefined function was called.
5761 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5762 is not reset to COMMON-LISP-USER.
5763 * Compilation of stuff related to dyncount.lisp has been made conditional
5764 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5765 smaller. The various dyncount-related symbols have been moved into
5766 a new "SB-DYNCOUNT" package.
5767 * tty-inspect.lisp has been renamed to inspect.lisp.
5768 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5769 feature has gone away. (When we eventually port to other flavors of
5770 libc and/or Unix, we'll try to make the differences between flavors
5771 invisible at the user level.)
5772 * Various other *FEATURES* tags, and/or their associated conditionals,
5773 have been removed if obsolescent, or given better documentation, or
5774 sometimes given more-mnemonic names.
5776 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5778 * tidied up "make.sh" script
5779 * tidied up system directory structure
5780 * better "clean.sh" behavior
5781 * added doc/FOR-CMUCL-DEVELOPERS
5782 * many many small tweaks to output format, e.g. removing possibly-confusing
5783 trailing #\. character in DESCRIBE-INSTANCE
5784 * (EQUALP #\A 'A) no longer signals an error.
5785 * new hashing code, including EQUALP hashing
5786 * tidied up Lisp initialization and toplevel
5787 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5788 * command line argument processing
5789 * added POSIX-GETENV function to deal with Unix-ish environment variables
5790 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5791 terminating SBCL on EOF
5792 * non-verbose GC by default
5793 * There is no more "sbcl" shell script; the sbcl file is now the C
5794 runtime executable (just like CMU CL).
5795 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5797 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5798 symbol and package information as Lisp native symbol and package objects
5799 instead of strings naming symbols and strings naming packages. This way,
5800 whenever packages are renamed (as in warm init), debug information is
5801 transformed along with everything else.
5802 * tweaked the optimization policy declarations which control the building
5803 of SBCL itself. Now, among other things, the system no longer saves
5804 source location debugging information. (This helps two problems at once
5805 by reducing SBCL size and by keeping SBCL from trying to look for its
5806 sources -- which may not exist -- when reporting errors.)
5807 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5808 debugging and testing purposes
5809 * cleaned up printing, making the printer call PRINT-OBJECT for
5810 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5811 methods, giving nearly-ANSI behavior
5812 * converted almost all special variables to use *FOO* naming convention
5813 * deleted PARSE-TIME functionality, since it can be done portably
5814 * moved some files out of cold init into warm init
5815 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5817 * regularized formatting of source files
5818 * added an install.sh script
5819 * fixed ridiculous memory usage of cross-compiler by making
5820 compiler/alloc.lisp not try to do pooling unless it can hook
5821 itself into the GC of the cross-compilation host. Now the system
5822 builds nicely on my old laptop.
5823 * added :SB-ALLOC in target-features.lisp-expr
5824 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5825 was not implemented)
5826 * re-did condition handling and note reporting in the compiler. Notes
5827 are no longer handled by signalling conditions. Style warnings
5828 and warnings are handled more correctly and reported in such a way
5829 that it's easy to find one or the other in your output (so that you
5830 can e.g. figure out which of many problems caused COMPILE-FILE to
5832 * changed the severity of several compiler warnings from full WARNING
5833 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5834 compiler note reporting so that it doesn't use the condition system
5835 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5836 COMPILE-FILE command)
5837 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5838 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5839 no longer supported, since their implementation was deeply intertwingled
5840 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5841 * removed bogus "support" for compiler macros named (SETF FOO), and
5842 removed the compiler macro for SETF INFO (but only after making a fool
5843 of myself on the cmucl-imp mailing list by posting a bogus patch for
5844 DEFINE-COMPILER-MACRO..)
5845 * Compiled files containing forms which have side effects on the Lisp
5846 reader (such as DEFPACKAGE forms) are now handled more correctly.
5847 (Compiler queuing of top level lambdas has been suppressed by setting
5848 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5849 * deleted various currently-unused source files, e.g. gengc.lisp. They
5850 may be added back at some point e.g. when porting to other architectures,
5851 but until they are it's distracting to distribute them and to try to
5853 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5854 there were so many of them they're just distractions, and UNCROSS is
5855 known to be able to handle the current sources
5856 * moved PROFILE functionality into TRACE, so that it will be clear
5857 how the wrapping and unwrapping of functions when you profile them
5858 interacts with the wrapping and unwrapping of functions when you
5859 trace them. (Actually, the functionality isn't there yet, but at least
5860 the interface specification is there. Hopefully, the functionality will
5861 arrive with some future maintenance release.)
5862 * removed host-oops.lisp
5863 * changed signature of QUIT function to allow UNIX-CODE argument
5864 * fixed READ-SEQUENCE bug
5865 * tweaked verbose GC output so that it looks more like the progress
5866 output that ANSI specifies for functions like LOAD
5867 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5868 * added <http://sbcl.sourceforge.com> to the banner information printed by