1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.42
3 * enhancement: SB-EXT:WORD type is provided for use with SB-EXT:ATOMIC-INCF
5 * enhancement: CLOS effective method functions and defclass slot typechecking
6 function now have debug names for use in backtraces and profiles.
7 * enhancement: ASDF has been updated to version 2.004. (lp#605260, thanks to
9 * enhancement: symbols are printed using fully qualified names in several
10 error and warning messages which are often associated with package
11 conflicts or mixups (lp#622789, thanks to Attila Lendvai)
12 * optimization: use poll(2) instead of select(2) to check for blocking IO
14 * bug fix: SB-BSD-SOCKETS:SOCKET-CONNECT was not thread safe. (lp#505497,
15 thanks to Andrew Golding)
16 * bug fix: DOTIMES accepted literal non-integer reals. (lp#619393, thanks to
18 * bug fix: WRITE-TO-STRING compiler macro binding special variable names,
19 breaking code that tried to write the value of a printer control variable.
20 (lp#581564, thanks to Stas Boukarev)
21 * bug fix: WRITE compiler macro did not handle output stream designators
22 correctly. (lp#598374, thanks to Stas Boukarev)
23 * bug fix: better availability of names of foreign functions in backtraces
24 on Linux. (lp#626962, thanks to Stas Boukarev)
25 * bug fix: scripting and build for Solaris and FreeBSD. (lp#615497,
26 lp#627581, thanks to Josh Elsasser and Jim Wise)
27 * bug fix: build fixes for OpenBSD -current and 4.8 (lp#615489, lp#615492,
28 thanks to Josh Elsasser)
29 * bug fix: using aliases for builtin classes as defmethod specializers
30 without adding DEFTYPEs for them works. (lp#618387)
31 * bug fix: timetravel by getrusage() no longer causes type-errors during GC.
33 * bug fix: legally dynamic-extent lists and vectors used as
34 initialization arguments to MAKE-ARRAY can be stack allocated. (lp#586105)
35 * bug fix: inline-expansion creating references to dead lambda-variables
36 (lp#454681, thanks to Alexey Dejneka)
37 * bug fix: better error message for bogus numerical arguments to RANDOM.
38 (lp#598986, thanks to Stas Boukarev)
39 * bug fix: the compiler occasionally inlined references from incompatible
40 environments occurs. (lp#308951)
41 * bug fix: the compiler threw an error when trying to compile a local
42 function (labels or flet) known to take a specialized complex argument.
43 (not in launchpad, reported by sykopomp in #lispgames)
44 * bug fix: package-locks failed to protect against compile-time effects of
45 DEFUN when the symbol previously had a macro definition. (lp#576637)
46 * bug fix: spurious ignore warnings even given (DECLARE IGNORE) in methods
47 when parameter bindings mutated. (reported by Faré Rideau; lp #611361)
48 * bug fix: workaround for compiler hang in ORDER-UVL-SETS (lp#308914)
49 * bug fix: evaluation in debugger REPL works using the global context when
50 in frames that do not have sufficient debug information.
52 changes in sbcl-1.0.42 relative to sbcl-1.0.41
54 ** Cross-compilation host is now specified to make.sh using
55 command-line argument --xc-host=<command> instead of a positional
56 argument. (thanks to Daniel Herring)
57 ** Install location can be specified to make.sh using command-line
58 argument --prefix=<path>. (lp#550889s, thanks to Daniel Herring)
59 * optimization: The default implementation of
60 COMPUTE-DISCRIMINATING-FUNCTION does much less wasted work.
61 * enhancement: Explicit memory barrier operations are now available for use
62 by multithreaded code. See documentation for details.
63 * enhancement: Experimental support for threading on Linux/PPC.
64 * bug fix: RENAME-PACKAGE returns the package. (Thanks to Eric Marsden)
65 * bug fix: EXPT signals an error if first argument is a zero and second
66 argument is a floating point zero. (lp#571581, thanks to Roman Marynchak)
67 * bug fix: DEFTYPE signals an error for non-list lambda-lists.
68 (lp#576594, thanks to Roman Marynchak)
69 * bug fix: make ASDF-INSTALL compatible with the now-included ASDF2.
70 (lp#612998, reported by Phil Hargett; patch from Jim Wise)
72 changes in sbcl-1.0.41 relative to sbcl-1.0.40:
73 * optimization: validity of observed keyword initargs to MAKE-INSTANCE is
74 cached, leading to many fewer expensive calls to
75 COMPUTE-APPLICABLE-METHODS.
76 * optimization: in the (unoptimized) general method for MAKE-INSTANCE on a
77 CLASS argument, search for and call an appropriate optimized ctor function
79 * bug fix: WRITE always returns the correct value.
81 changes in sbcl-1.0.40 relative to sbcl-1.0.39:
82 * bug fix: readdir now works on :inode64 darwin builds (lp#592897)
83 * bug fix: Name conflicts between symbols passed as arguments to a single
84 call to IMPORT no longer add multiple symbols with the same name to the
85 package (detectable via DO-SYMBOLS).
86 * bug fix: support building without the dlshim on darwin x86 and x86-64
88 * bug fix: TRACE :ENCAPSULATE NIL now works on ppc/linux.
90 changes in sbcl-1.0.39 relative to sbcl-1.0.38:
91 * bug fix: Backtrace from undefined function on x86 and x86-64 now show
93 * bug fix: linkage-table entries on PPC now no longer overflow their
94 space allocation (potentially causing crashes if they are written out
96 * bug fix: Scrub control stack after scavenging in gencgc on non-x86oid
97 platforms, preventing the GC from seeing stale pointers on the control
98 stack in subsequent GCs (which would, and does, break invariants).
99 * bug fix: 32-bit unicode external formats now work on big-endian systems.
100 * bug fix: Literal characters with code points greater than about 32767
101 now work on PPC UNICODE builds.
102 * bug fix: Any noise left by SSE operations (sqrt and conversions) in the
103 high order bits are explicitly cleared out. In some contrived situations,
104 this could lead to wrong results in mixed real/complex float arithmetic.
105 * bug fix: Fix function/macro redefinition warnings when building with
106 clisp. (lp#576787, thanks to Josh Elsasser)
107 * new platform: experimental support for ppc/openbsd (thanks to Josh
109 * bug fix: Floating-point traps now work on ppc/linux.
111 changes in sbcl-1.0.38 relative to sbcl-1.0.37:
112 * incompatible change: Thread names are now restricted to SIMPLE-STRINGs
113 like for any other thread-related datastructure, MUTEX, etc. (lp#547095)
114 * deprecation: the SB-QUEUE contrib was merged into the SB-CONCURRENCY
115 contrib module. New code should depend on SB-CONCURRENCY, not SB-QUEUE.
116 * deprecation: SB-THEAD:GET-MUTEX was deprecated in favor of
117 SB-THREAD:GRAB-MUTEX.
118 * new contributed module: SB-CONCURRENCY is a new contrib; it's supposed to
119 contain additional data structures and tools for concurrent programming;
120 at the moment it contains a lock-free queue, and a lock-free mailbox
122 * new feature: added SB-THREAD:GRAB-MUTEX; it's like the now deprecated
123 GET-MUTEX but takes &key rather than &optional parameters. Also added
124 :TIMEOUT argument to GRAB-MUTEX on non-sb-lutex platforms like Linux and
126 * new feature: added SB-THREAD:TRY-SEMAPHORE, a non-blocking variant of
127 SB-THREAD:WAIT-ON-SEMAPHORE.
128 * new feature: SB-EXT:ATOMIC-DECF has been added as a companion to
130 * new feature: a CANCEL-DEADLINE is associated with DEADLINE-TIMEOUT
131 conditions to defer the deadline for forever.
132 * enhancement: *STANDARD-OUTPUT*, *STANDARD-INPUT*, and *ERROR-OUTPUT* are
134 * enhancement: errors from NO-APPLICABLE-METHOD and
135 NO-PRIMARY-METHOD now have a RETRY restart available to retry the
136 generic function call.
137 * enhancement: SB-BSD-SOCKET improvements
138 ** sockets and socket streams now have a more informative printed
139 representation based on the corresponding SOCKET-NAME and
141 ** SOCKET-MAKE-STREAM once more supports the :AUTO-CLOSE option.
143 ** SOCKET-CLOSE now accepts :ABORT argument, which is passed on to
144 CL:CLOSE when appropriate, and no longer disassociates the stream
145 from the socket if close failed. (lp#543951)
146 * improvements to the instrumenting profiler
147 ** new feature: report per-function GC overhead. (thanks to John Fremlin)
148 ** optimization: counters no longer use locks for the overflow mode.
149 ** bug fix: whenever a profiling counter wrapped into overflow mode, it
150 incurred an off-by-one miscount.
151 * enhancement: improved MAKE-HASH-TABLE documentation (lp#543473)
152 * enhancement: improved DEFMETHOD pretty-printing.
153 * enhancement: perform range reduction when arguments are too large for
154 x87's transcendentals (instead of returning 0). (lp#327192)
155 * enhancement: eliminate some spurious TYPE-WARNINGs. Should help with
156 some of CL-PPCRE's macros. (lp#570079)
157 * enhancement: our machine code is slightly less hostile to valgrind on
159 * enhancement: up-to-date versions of NetBSD-current are supported. (Thanks
160 to Robert Swindells and Aleksej Saushev)
161 * bug fix: correct restart text for the continuable error in MAKE-PACKAGE.
162 * bug fix: a rare case of startup-time page table corruption.
163 * bug fix: a semaphore with multiple waiters and some of them unwinding due
164 to timeouts could be left in an inconsistent state.
165 * bug fix: fix typo in "Reporting Bugs" section of the manual (lp#520366)
166 * bug fix: misoptimization of multiplication by one in
167 (SB-C::FLOAT-ACCURACY 0) policies.
168 * bug fix: miscounts in SB-PROFILE.
169 * bug fix: Fix lost wakeup bug between SB-THREAD:CONDITION-WAIT and
170 CONDITION-NOTIFY on Linux. See threads "lost wakeup in condition-wait /
171 condition-notify" (Feb 2010) and "Condition-Wait, Deadline handler, waking
172 up itself" (March 2010) for further details.
173 * bug fix: allow forward FIND and POSITION on lists to elide checking :END
174 against length of the list if the element is found before the specified
175 END is reached. (thanks to Alec Berryman, lp#554385)
176 * bug fix: errors signalled during package graph modification no longer
177 block FIND-SYMBOL and FIND-PACKAGE in other threads. (lp#511072)
178 * bug fix: SB-POSIX build was broken when SBCL was compiled without the
179 :SB-DOC feature. (lp#552564)
180 * bug fix: SB-INTROSPECT build issues on GENGC/PPC. (lp#490490)
181 * bug fix: more robust runtime executable path detection. (lp#375549)
182 * bug fix: GCD always returns positive values. (lp#413680)
183 * bug fix: Converting division to multiplication by reciprocal handles
185 * bug fix: We were too eager in eliding range reduction tests on x87.
186 The maximal magnitude is 2^63, not 2^64.
187 * bug fix: Transforms for TRUNCATE don't die when the result is completely
189 * bug fix: Maybe restore buildability on Alpha.
190 * bug fix: READ-BYTE isn't inline anymore, fixing weird streams failures.
192 * bug fix: RANDOM-STATE can be printed readably again.
193 * bug fix: Unreadable objects were sometimes printed like #<\nFoo>.
194 * bug fix: Using EQL with non-constant values of constant type (e.g. EQL
195 types) could result in type mismatches during compilation.
197 changes in sbcl-1.0.37 relative to sbcl-1.0.36:
198 * enhancement: Backtrace from THROW to uncaught tag on x86oids now shows
199 stack frame thrown from.
200 * enhancement: WITH-COMPILATION-UNIT :POLICY allows restricting changes to
201 compiler optimization qualities inside dynamic extent of its body.
202 * enhancement: LOAD-LOGICAL-PATHNAME-TRANSLATIONS can be used to load
203 translations from SYS:SITE;<HOST>.TRANSLATIONS.NEWEST (thanks to Michael
205 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) take advantage of
206 constraint propgation, allowing better compilation eg. when used to
207 access structures with WITH-SLOTS. (lp#520366)
208 * optimization: the compiler is now more aware of the type of the underlying
209 storage vector for multidimensional simple arrays resulting in better code
210 for accessing such arrays.
211 * optimization: passing NIL as the environment argument to TYPEP no longer
212 inhibits optimizing it. (lp#309788)
213 * optimization: more efficient register usage when handling single-float
214 arguments on x86-64. (thanks to Lutz Euler)
215 * optimization: ADJUST-ARRAY and STABLE-SORT on vectors no longer use
216 pre-allocated temporary vectors. (lp#496249)
217 * bug fix: Fix compiler error involving MAKE-ARRAY and IF forms
218 in :INITIAL-CONTENTS. (lp#523612)
219 * bug fix: FUNCTION-LAMBDA-EXPRESSION lost declarations from interpreted
220 functions. (lp#524707)
221 * bug fix: bogus style warnings from certain (SETF SLOT-VALUE) and
222 WITH-SLOTS usages during compilation.
223 * bug fix: SB-C::CLASS-INFO now prints correctly. (lp#514762)
224 * enhancement: Can now build with ud2 instead of int3 as trap instruction on
225 all x86oid platforms with :UD2-BREAKPOINTS target feature.
226 * bug fix: Breakpoints now work when using ud2 instead of int3 as trap
227 instruction (tested on x86oid linux with ud2-breakpoints).
228 * bug fix: slam.sh now works on win32.
229 * bug fix: better differences of numeric types. (lp#309124)
230 * bug fix: arrays declared intersection and union types can have their
231 upgraded element type derived. (lp#316078)
232 * bug fix: SB-SPROF allocation profiling for all threads failed to profile
233 threads started during profiling. (lp#472499)
234 * bug fix: SB-INTROSPECT test failure when building without SB-EVAL feature.
236 * bug fix: SB-CLTL2:DECLARATION-INFORMATION did not take
237 SB-EXT:RESTRICT-COMPILER-POLICY into account. (lp#313337)
238 * bug fix: Comma inside a backquoted array or structure resulted in nonsense
239 values instead of signaling an error. (lp#309093)
240 * bug fix: Spurious unused variable warning in a DEFSTRUCT edge case.
242 * bug fix: More consistent warnings and notes for ignored DYNAMIC-EXTENT
243 declarations (lp#497321)
244 * bug fix: FIND and POSITION on lists did not check sequence bounds properly
245 and failed to detect circular lists (lp#452008)
246 * bug fix: leakage from ~/.asdf-install into the ASDf-INSTALL contrib build
248 * bug fix: LOOP OF-TYPE VECTOR compile-time error. (lp#540186)
249 * bug fix: SIGNAL SB-SYS:INTERACTIVE-INTERRUPT before entering the debugger
250 due to it, so that handlers can run.
251 * bug fix: reparsing undefined types if they have become defined since
253 * bug fix: missing &REST type in a proclamation for a function with both
254 &REST and &KEY in lambda-list caused miscompilation (lp#458354)
255 * bug fix: WHO-CALLS information for source-transformed and compiler-macro
256 expanded calls (lp#542174)
257 * bug fix: more accurate WHO-MACROEXPANDS information; point into rather
258 than just at toplevel form.
260 changes in sbcl-1.0.36 relative to sbcl-1.0.35:
261 * new feature: SB-EXT:TYPEXPAND-1, SB-EXT:TYPEXPAND, and
262 SB-EXT:TYPEXPAND-ALL behave exactly like their MACROEXPAND counterparts
263 but work on type specifiers.
264 * new feature: SB-EXT:DEFINED-TYPE-NAME-P returns whether a symbol is known
265 to name a type specifier.
266 * new feature: SB-EXT:VALID-TYPE-SPECIFIER-P returns whether a given type
267 specifier is valid where "valid" basically means "would be accepted as
268 second argument of TYPEP".
269 * new feature: SB-INTROSPECT:FUNCTION-TYPE takes a function-designator and
270 returns the function's declared, or derived FTYPE.
271 * new feature: SB-POSIX now supports accessing the d_ino member of
272 dirent structures. (Thanks to Philipp Marek and Pierre THEIRRY)
273 * new feature: The function SB-EXT:SEED-RANDOM-STATE has been added to
274 provide for seeding a RANDOM-STATE object with user-provided data or
275 from the operating system's PRNG. Also, (MAKE-RANDOM-STATE T) will
276 attempt to initialize the returned state from the operating system's
277 PRNG where possible. (Thanks to Fare Rideau; launchpad bug lp#310116)
278 * bug fix: Fix SB-SIMPLE-STREAMS:READ-VECTOR to correctly set the
279 FILE-POSITION of the stream being read from. (launchpad bug lp#491087)
280 * bug fix: Fix grammar and style issues for the docstrings of
281 printer-related variables and functions. (Thanks to mon_key; launchpad
283 * bug fix: Fix compilation on chenygc platforms. Thanks to Larry Valkama and
285 * bug fix: SB-THREAD:CONDITION-WAIT sometimes signaled a deadline twice
286 in a row even though a handler defered the deadline long into the
288 * bug fix: A deadline handler was run without interrupts enabled for a
289 deadline signaled within SB-THREAD:CONDITION-WAIT. That could result
290 in infinitely spinning, non-killable threads.
291 * bug fix: Backtrace from internal-errors on x86-64 os x was truncated
292 before reaching the erring stack frame.
293 * bug fix: Fix type derivation for EXPT when raising a fixnum to a
294 real power. (launchpad bug lp#525949)
295 * bug fix: Fix SB-EXT:GENERATION-* accessors for generations > 0 on
296 GENCGC platforms. (launchpad bug lp#529014)
297 * bug fix: More robust checks for invalid DEFMETHOD argument specializers.
298 (launchpad bug lp#525916)
299 * bug fix: Fix building on Darwin when sysctl is not in the user's PATH.
300 (Thanks to Robert Goldman)
302 changes in sbcl-1.0.35 relative to sbcl-1.0.34:
303 * optimization: ROUND with a single single-float or double-float argument
304 is properly inlined when possible.
305 * optimization: Slightly better code is generated for integer<->float
306 conversions and for single-float<->double-float conversions on x86-64.
307 * optimization: SB-ROTATE-BYTE:ROTATE-BYTE now generates more efficient
308 code for 32-bit and 64-bit rotations on x86-64.
309 * bug fix: The install script changes the ownership of directories as well
310 as files for contrib modules using asdf. (thanks to Eugene Ossintsev;
311 launchpad bug lp#508485)
312 * bug fix: TRUNCATE with a single single-float or double-float argument is
313 properly inlined when possible. (launchpad bug lp#489388)
314 * bug fix: Passing a rotation count of zero to SB-ROTATE-BYTE:ROTATE-BYTE
315 no longer causes a compiler error on x86 and ppc.
316 * bug fix: GET-MACRO-CHARACTER bogusly computed its second return value
317 always relative to *READTABLE* rather than the passed argument.
319 changes in sbcl-1.0.34 relative to sbcl-1.0.33:
320 * minor incompatible change: threading support is now enabled by default
322 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
324 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
325 are now caught and reported just like errors during macroexpansion.
326 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
327 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
328 * enhancement: ASDF systems can now depends on SB-QUEUE.
329 * fixes and improvements related to Unicode and external formats:
330 ** bug fix: error handling and restart usage in the ucs-2 external format
332 ** there is now an implementation of the ucs-4 external format.
333 ** the utf-16 and utf-32 external formats are supported.
334 * bug fix: SB-POSIX wrapper for putenv no longer tries to put lisp strings
335 in the environment. setenv() and unsetenv() are also provided. (reported by
336 Fare Rideau; launchpad bug lp#460455)
337 * bug fix: LOAD of both .fasl and .FASL type files now forces fasl-style
338 loading. This ensures sensible errors for .FASL files from other
339 implementations on case-insensitive filesystems. (reported by Willem
340 Broekema; launchpad bug lp#489417)
341 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
342 Willem Broekema; launchpad bug lp#489698).
343 * bug fix: some minor code rearrangements to reenable warning-free building
344 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
345 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
347 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
350 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
351 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
352 * improvement: support O_LARGEFILE access to files larger than 2GB on
353 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
354 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
355 definitions for methods specializing on the passed class itself.
356 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
357 definitions for methods specializing on the passed class itself, or on
359 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
360 about itself during the build (e.g. for M-? in Slime), if this flag is
361 enabled in customize-target-features.lisp. This will increase the core
362 size by about 5-6mb, though, so it's mostly interesting to SBCL
364 * new feature: various GENCGC tuning parameters have been experimentally
365 documented and exported from SB-EXT. See documentation for details.
366 * fixes and improvements related to Unicode and external formats:
367 ** the Unicode character database has been upgraded to the
368 Unicode 5.2 standard, giving names and properties to a number of new
369 characters, and providing a few extra characters with case
371 ** improvement: restarts for providing replacement input/output on coding
372 errors for fd-stream external formats.
373 ** improvement: where :<encoding> is a keyword corresponding to an
374 external format the system supports, it is now possible to specify
375 (:<encoding> :replacement <character>) as an external format which will
376 automatically substitute <character> on encoding or decoding errors for
377 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
379 ** improvement: the file streams underlying the standard streams (such as
380 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
381 which uses the replacement mechanism to handle encoding errors,
382 preventing various infinite error chains and unrecoverable I/O
384 ** minor incompatible change: the utf-8 external format now correctly
385 refuses to encode Lisp characters in the surrogate range (char-codes
386 between #xd800 and #xdfff).
387 ** fix a typo preventing conversion of strings into octet vectors
388 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
390 ** fix a bug in the octet multibyte handling of decoding errors and the
391 USE-VALUE restart. (launchpad bug lp#314939)
392 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
393 restart on fd-stream decoding errors.
394 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
395 error is near the end of file.
396 ** fix a double-error case in unibyte octet conversions, when the first
397 use of USE-VALUE is ignored.
398 ** fix bugs in handling of undefined code points in unibyte encodings.
399 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
400 streams after an UNREAD-CHAR.
401 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
402 object is allocated in a boxed region of dynamic space.
403 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
404 fork with multiple Lisp threads running is made, instead of going
405 ahead with unpredictable consequences. (reported by Leslie Polzer)
406 * bug fix: uses of slot accessors on specialized method parameters within
407 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
408 error while finalizing the class. This fix may cause classes with slot
409 accessors to be finalized later than previously. (reported by Lars Rune
410 Nøstdal; launchpad bug lp#473699)
411 * bug fix: restore buildability on the MIPS platform. (regression from
412 1.0.30.38, reported by Samium Gromoff)
413 * bug fix: inspecting closures is less likely to fail with a type error.
414 * bug fix: no timer starvation when setting the system clock back.
415 (launchpad bug lp#460283)
416 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
417 standard pprint dispatch table as specified by CLHS.
418 * bug fix: give CLISP a hint about a type declaration to enable it to build
419 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
421 * bug fix: correctly dump literal objects in defaulting forms of arglists.
422 (reported by Attila Lendvai; launchpad bug lp#310132)
423 * bug fix: distinguish in type specifiers between arrays that might be
424 complex and arrays that are definitely complex. (launchpad bug lp#309129)
425 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
426 type. (reported by Levente Mészáros; launchpad bug lp#485972)
427 * bug fix: setting the value of a symbol-macro within a method in the
428 presence of type declarations works properly again. (reported by Iban
429 Hatchondo; launchpad bug lp#485019)
431 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
432 * optimization: faster FIND and POSITION on strings of unknown element type
433 in high SPEED policies. (thanks to Karol Swietlicki)
434 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
436 * improvement: better error signalling for bogus parameter specializer names
437 in DEFMETHOD forms (reported by Pluijzer)
438 * bug fix: DELETE-FILE once again works on logical pathnames (regression
440 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
442 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
443 removes previous default initargs (reported by Lars Rune Nøstdal and
445 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
446 lambda-lists. (reported by Peter Seibel)
447 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
448 correctly when starting from an executable core without saved runtime
449 options (reported by Faré Rideau, thanks to Zach Beane)
450 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
451 an optimization note instead. (reported by Martin Cracauer)
452 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
453 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
454 new process group if it doesn't need to share stdin with the sbcl
455 process. (thanks to Leslie Polzer)
456 * bug fix: SATISFIES could be misoptimized to refer to a local function.
457 (reported by Stanislaw Halik)
459 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
460 * improvement: stack allocation is should now be possible in all nested
461 inlining cases: failure to stack allocate when equivalent code is manually
462 open coded is now considered a bug.
463 * improvements related to Unicode:
464 ** the Unicode character database has been upgraded to the
465 Unicode 5.1 standard, giving names and properties to a number of new
466 characters, and providing a few extra characters with case
468 ** the system now recognizes and produces names for Unicode Hangul
470 ** the EBCDIC-US external-format is now supported for octet operations
471 (as well as for stream operations).
472 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
473 be used to output toplevel compile-time effects into a separate .CFASL
475 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
476 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
478 * optimization: TRUNCATE on known single- and double-floats is upto 25%
480 * optimization: division of floating point numbers by constants uses
481 multiplication by reciprocal when an exact reciprocal exists.
482 * optimization: multiplication of single- and double-floats floats by
483 constant two has been optimized.
484 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
485 sufficient type information is available. (thanks to Leslie Polzer)
486 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
487 known structure objects are as efficient as defstruct generated accessors.
488 * optimization: unused vector creation can now be optimized away.
489 * improvement: ASDF systems can now depends on SB-INTROSPECT.
490 * improvement: a STYLE-WARNING is signalled when a generic function
491 clobbers an earlier FTYPE proclamation.
492 * improvement: the compiler is able to track the effective type of
493 generic function across method addition and removal even in the
494 absence of an explicit DEFGENERIC.
495 * improvement: DESCRIBE now reports on symbols naming undefined
496 but assumed or declared function as well.
497 * improvement: recompilation of systems using SB-GROVEL now works
498 (thanks to Leslie Polzer)
499 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
500 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
502 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
503 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
504 well as user defined declaration names.
505 ** VARIABLE-INFORMATION is now aware of alien variables.
506 * improvement: improved address space layout on OpenBSD (thanks to Josh
508 * improvement: pretty-printing of various Lisp forms has been improved
509 (thanks to Tobias Rittweiler)
510 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
511 unused were deleted in safe code. (reported by John Fremlin)
512 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
514 * bug fix: some out-of-line array predicates were missing (reported by
516 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
518 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
519 (thanks to Larry D'Anna)
520 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
521 as well. (reported by Josh Elsasser)
522 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
523 leading to possible attempts to delete the same file twice. See docstring
524 on DELETE-FILE for details. (reported by John Fremlin)
525 * bug fix: DELETE-FILE once again deletes the file named by the pathname
526 designator argument, rather than its truename. (reported by Luis
528 * bug fix: the low-level debugger had 32-bit assumptions and was missing
529 information about some array types. (thanks to Luis Oliveira)
530 * bug fix: moderately complex combinations of inline expansions could
531 be miscompiled if the result was declared to be dynamic extent.
532 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
533 form (+ <variable> <integer>) were miscompiled under certain
535 * bug fix: in some cases no compiler note about failure to stack allocate
536 was emitted, even if the objects were in fact heap allocated.
537 * bug fix: minor violation of "otherwise inaccessible" rule for stack
538 allocation could cause objects users might reasonably expect to
539 be heap allocated to be stack allocated.
540 * bug fix: DESCRIBE signalled an error for generic functions under
541 certain circumstances. (thanks to Leslie Polzer)
542 * bug fix: Fixed spelling of an error message.
544 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
545 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
546 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
547 of SB-THREAD:THREAD-ERROR-THREAD.
548 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
549 * new feature: docstrings for local and anonymous functions are no longer
550 discarded. (thanks to Leslie Polzer)
551 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
552 values in other threads.
553 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
554 about object allocation.
555 * optimization: division of a real float by a complex float is implemented
556 with a specialised code sequence.
557 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
558 keywords is an order of magnitude faster.
559 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
560 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
561 methods, and similarly for non-standard metaclass classes as long as there
562 are no methods additional on MAKE-INSTANCE.
563 * optimization: more efficient type-checks for FIXNUMs when the value
564 is known to be a signed word on x86 and x86-64.
565 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
566 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
568 * optimization: compiler is smarter about delegating argument type checks to
570 * optimization: several character functions are now compiled somewhat more
571 efficiently. (reported by Lynn Quam)
572 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
573 * improvement: less unsafe constant folding in floating point arithmetic,
574 especially for mixed complex/real -float operations.
575 * optimization: constant double and single floats are stored in native
576 unboxed format on x86[-64].
577 * optimization: smarter code for arithmetic operations with constant floats,
578 complex floats, or integers on x86[-64].
579 * optimization: smarter code for conjugate/multiplication of float complexes
580 and abs/negate of floats on x86-64.
581 * optimization: more efficient complex float and real float operations on
583 * improvement: complex float division is slightly more stable.
584 * improvement: DESCRIBE output has been reworked to be easier to read and
585 contains more pertinent information.
586 * improvement: failure to provide requested stack allocation compiler notes
587 provided in all cases (requested stack allocation not happening without a
588 note being issued is now considered a bug.)
589 * bug fix: SB-POSIX exports the documented types and functions
590 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
591 types. (reported by "abhi")
592 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
593 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
594 (thanks to Paul Khuong)
595 * bug fix: looping around HANDLER-CASE could silently consume stack space
596 on each iteration. (reported by "foobar")
597 * bug fix: better error signalling when calls to functions seeking elements
598 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
599 (reported by Tobias Rittweiler)
600 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
601 directory components now works as it used to. (various prolems reported by
602 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
603 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
604 (reported by Elliot Slaughter, thanks to Stas Boukarev)
605 * bug fix: bogus undefined variable warnings from fopcompiled references to
606 global variables. (thanks to Lars Rune Nøstdal)
607 * bug fix: foreign function names should now appear in backtraces on
608 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
609 * bug fix: SETF compiler macro documentation strings are not discarded
611 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
612 (thanks to Alex Plotnick)
613 * bug fix: SLEEP accepts large integer arguments, truncating them to
614 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
615 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
616 * bug fix: compiler notes for expensive slot type checks could be emitted
617 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
619 changes in sbcl-1.0.29 relative to 1.0.28:
620 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
621 https://bugs.launchpad.net/sbcl
622 Bugs can be reported directly there, or by sending email to
623 sbcl-bugs@lists.sourceforge.net
624 (no subscription required.)
625 * minor incompatible change: under weak type checking policy integer
626 types are weakened less aggressively.
627 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
628 allowed to return, which causes SBCL to quit with exit status 0. Previously
629 if the function returned with a small integer return value, that value
630 was accidentally reused as the exit status.
631 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
632 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
633 :HASH-FUNCTION argument. Refer to user manual for details.
634 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
636 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
637 microseconds since the Unix epoch on all platforms.
638 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
639 allows the compiler to safely elide boundedness checks for special
641 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
642 the symbol, prohibits both lexical and dynamic binding. This is mainly an
643 efficiency measure for threaded platforms, but also valueable in
645 * new feature: UNC pathnames are now understood by the system on Windows.
646 * optimization: the compiler uses a specialized version of FILL when the
647 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
649 * optimization: accesses to potentially non-simple arrays where element type
650 is known are 50% faster.
651 * optimization: compiler now generates faster array typechecking code.
652 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
654 * optimization: multidimensional array accesses in the absence of type
655 information regarding array rank are approximately 10% faster due to
656 open coding of ARRAY-RANK.
657 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
658 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
660 * optimization: result of call to VECTOR can now be stack allocated.
661 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
662 as long as the resulting array is one-dimensional and has a known
663 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
664 length of the list matches the known length of the vector does not
665 allocate the list as an intermediate step. Ditto for VECTOR and simple
667 * optimization: MAKE-ARRAY can now stack allocate in the presence of
668 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
669 known element type, and is known to be simple and one dimensional.
670 * improvement: SBCL now emits a compiler note where stack allocation was
671 requested but could not be provided (not in all cases, unfortunately)
672 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
673 * improvement: pretty-printing loop has been implemented properly. (thanks
674 to Tobias Rittweiler)
675 * documentation: CLOS slot typechecing policy has been documented.
676 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
677 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
678 pointer, leading to memory faults. (thanks to Bart Botta)
679 * bug fix: better floating point exception handling on x86/OpenBSD.
680 (thanks to Josh Elsasser)
681 * bug fix: exit status from QUIT when called under --script was lost
682 (reported by Hubert Kauker)
683 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
684 same implementation of FILL to initialize the array, even if a more
685 efficient one was available (reported by Stas Boukarev, thanks to
687 * bug fix: potential miscompilation of array stack allocation on x86 and
688 x86-64. (reported by Time Tossavainen)
689 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
690 result in their subforms being treated as top level forms. (reported by
692 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
693 than a machine register could leave garbage in the high bits of the
694 result register (bug 316325).
695 * bug fix: disable address space randomization Linux/x86-64 as well,
696 not just x86-64. (reported by Ken Olum)
697 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
698 generate incorrect code.
699 * bug fix: #201; type inference for CONS and ARRAY types could derive
700 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
701 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
702 could not be assigned to or bound with PROGV. (reported by Lorenz
704 * bug fix: the value of CL:- in the inspector was the previous expression
705 evaluated rather than the expression being evaluated.
706 * bug fix: constants can no longer be locally declared special.
707 * bug fix: signals delivered to threads started from foreign land (read:
708 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
709 thread by blocking all signals and resignalling.
710 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
711 from :INITFORM, if any.
713 changes in sbcl-1.0.28 relative to 1.0.27:
714 * a number of bugs in cross-compilation have been fixed, with the ultimate
715 result that building under (at least) clisp should be much more reliable.
716 * minor incompatible changes: echo-streams now propagate unread-char to the
717 underlying input stream, and no longer permit unreading more than one
719 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
720 frames, allowing for instance more reliable backtraces.
721 * improvement: the debugger REPL can now reference lexical variables
722 by name directly for code compiled with (DEBUG 3).
723 * improvement: errors from malformed declarations now have better source
724 paths associated with them. (thanks to Tobias Rittweiler)
725 * optimization: faster local calls on x86/x86-64
726 * bug fix: some error messages for out-of-bound array indexes confused the
727 index and the bound. (thanks to Stas Boukarev)
728 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
730 * bug fix: running regressions tests in shells without OSTYPE set now works.
731 (reported by Harald Hanche-Olsen)
732 * bug fix: more robust static space exhaustion signalling from
733 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
734 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
735 docstring away instead of storing it under names such as (LAMBDA (X)).
736 (reported by Leslie Polzer)
737 * bug fix: timers could go off in the wrong order, be delayed indefinitely
738 (thanks to Ole Arndt for the patch)
739 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
740 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
741 potentially-invalid effective methods in its cache.
742 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
743 instances as well (thanks to Paul Khuong)
744 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
745 SB-POSIX:WAITPID and their C equivalents.
746 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
748 changes in sbcl-1.0.27 relative to 1.0.26:
749 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
750 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
751 * improvement: the system either recovers from stack exhaustion or dies
752 properly as opposed to leaving the user uncertain of whether the handler
753 trampled on some random memory next to the stack or having to rely on
754 --lose-on-corruption (which is still a good idea to use in production
755 because stack exhaustion can happen in signal handlers which will likely
757 * bug fix: fix gc related interrupt handling bug on ppc (regression from
758 1.0.25.37, reported by Harald Hanche-Olsen)
759 * bug fix: work around signal delivery bug in darwin (regression from
760 1.0.25.44, reported by Sidney Markowitz)
761 * bug fix: fix ERROR leaking memory (reported by David Thompson)
763 changes in sbcl-1.0.26 relative to 1.0.25:
764 * incompatible change: an interruption (be it a function passed to
765 INTERRUPT-THREAD or a timer function) runs in an environment where
766 interrupts can be enabled. The interruption can use
767 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
768 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
769 potentially running out of stack. Keep in mind that in the absence
770 of WITHOUT-INTERRUPTS some potentially blocking operation such as
771 acquiring a lock can enable interrupts.
772 * incompatible change: GC-OFF and GC-ON are removed, as they were
773 always unsafe. Use WITHOUT-GCING instead.
774 * new feature: runtime option --disable-ldb
775 * new feature: runtime option --lose-on-corruption to die at the
776 slightest hint of possibly non-recoverable errors: running out of
777 memory, stack, alien stack, binding stack, encountering a memory
778 fault, etc. In the absence of --lose-on-corruption a warning is
780 * enhancement: detect binding stack exhaustion
781 * enhancement: detect alien stack exhaustion on x86/x86-64
782 * improvement: generally more stable and reliable interrupt handling
783 * improvement: there is a per thread interruption queue,
784 interruptions are executed in order of arrival
785 * improvement: a repeating timer reschedules itself when the it has
786 finished, but expiration times are spaced equally. If an
787 expiration time is in the past it will trigger after a short grace
788 period that may give a chance to other things to run.
789 * optimization: slightly faster gc on multithreaded builds
790 * optimization: faster WITHOUT-GCING
791 * bug fix: when JOIN-THREAD signals an error, do it when not holding
792 important locks so that the debugger/handler doesn't produce
793 recursive errors or deadlock.
794 * bug fix: real-time signals are not used anymore, so no more
795 hanging when the system wide real-time signal queue gets full.
796 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
798 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
799 * bug fix: fix random memory faults related to interrupts on alpha
800 * bug fix: fix random memory faults related to interrupts on sparc
801 * bug fix: fix deadlocks related to starting threads
802 * bug fix: fix deadlines on locks on futex platforms
803 * bug fix: restore errno in signal handlers
804 * bug fix: fix deadlocks related to hash tables
805 * bug fix: fix deadlocks in pcl
807 changes in sbcl-1.0.25 relative to 1.0.24:
808 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
809 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
810 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
811 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
812 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
813 better name and does not return values so stale on multiprocessor systems.
814 Also, HOLDING-MUTEX-P was added for about the only sane usage of
816 * improvement: unithread builds keep track of MUTEX-VALUE.
817 * improvement: reading from a TWO-WAY-STREAM does not touch the output
818 stream anymore making it thread safe to have a concurrent reader and
819 a writer, for instance, in a pipe.
820 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
821 arguments, making compiler-macros for SETF-functions able to inspect
822 their constant arguments.
823 * improvement: COMPILE-FILE reports times with millisecond accuracy
824 (thanks to Luis Oliveira)
825 * optimization: CHAR-CODE type derivation has been improved, making
826 TYPEP elimination on subtypes of CHARACTER work better. (reported
827 by Tobias Rittweiler, patch by Paul Khuong)
828 * bug fix: setting alien structure fields of type struct by value now
829 computes the right offset for the memory copy.
830 * bug fix: compilation problem involving inlined calls to aliens with
831 result type VOID. (reported by Ken Olum)
832 * bug fix: #235a; sequential inline expasion in different policies no
833 longer reuses the functional from the previous expansion site.
834 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
835 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
836 * bug fix: named ENUMs in multiply-referenced alien record types no longer
837 cause an implied type redefinition. Regression from 1.0.21.29.
838 * bug fix: sign-extension of small signed return values now works with C
839 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
840 * improvements to the Windows port:
841 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
842 Westfold, thanks to Rudi Schlatte)
844 changes in sbcl-1.0.24 relative to 1.0.23:
845 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
846 vector of a multidimensional SIMPLE-ARRAY.
847 * new feature: the system now signals a continuable error if standard
848 readtable modification is attempted. (thanks to Tobias Rittweiler)
849 * new feature: DIRECTORY has been extended with a non-standard keyword
850 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
851 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
852 has been robustified and documented better. (thanks to Robert Goldman)
853 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
854 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
855 * tradeoff: constant FORMAT control strings are now compiled unless
856 SPACE > SPEED (previously only when SPEED > SPACE.)
857 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
859 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
861 * bug fix: lexical type declarations are now correctly reported by
862 SB-CLTL2. (reported by Larry D'Anna)
863 * bug fix: STRING-TO-OCTETS did not handle :START properly when
864 using UTF-8 as external format. (thanks to Luis Oliveira)
865 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
866 are signalled correctly. (thanks to Stas Boukarev)
867 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
868 designator. (thanks to Tobias Rittweiler)
869 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
870 designator, and returns T instead of the function. (thanks to
872 * bug fix: direct superclasses of STANDARD-CLASS and
873 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
874 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
875 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
876 when no class with the named slot yet exists no longer causes a
877 compile-time style-warning.
878 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
879 in safe code. (reported by Didier Verna)
880 * bug fix: #430; nested structure constructors can stack allocate.
881 * bug fix: on some 64-bit platforms dynamic space size was truncated
882 to #xffff0000 bytes. (reported by Benjamin Lambert)
883 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
884 REPL unusable. (reported by Daniel Herring)
885 * bug fix: return values of READ-SEQUENCE did not take :START into
886 account on file streams, regressions since 1.0.12.22. (reported by
887 Thomas Russ, patch by Paul Khuong)
888 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
889 method specializer used to confuse permuation vector optimization.
890 * bug fix: system inserted bogus implicit type declarations for local
891 special variables in DEFMETHOD bodies.
892 * bug fix: #354; duplicated frames in backtraces due to
893 non-tail-call-optimized XEPs to functions with return type NIL
895 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
896 initializes structure object slots according to DEFSTRUCT initforms,
897 and DEFSTRUCT forms :INCLUDEind structure classes defined using
898 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
899 (reported by Bruno Haible and Stephen Wilson)
900 * bug fix: #395; fill-pointer output streams used now support
901 element-type BASE-CHAR as well.
902 * bug fix: compiler error when attempting to derive return value of
903 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
905 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
907 * bug fix: Mac OS X binaries should now be portable between Leopard
910 changes in sbcl-1.0.23 relative to 1.0.22:
911 * enhancement: when disassembling method functions, disassembly
912 for the associated fast function is also produced.
913 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
915 * optimization: printing with *PRINT-PRETTY* true is now more
916 efficient as long as the object being printed doesn't require
917 special handling by the pretty printer.
918 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
919 now interact correctly with type declarations.
920 * partial bug fix: PCL detects infinite recursion during wrapper
921 validation. (thanks to Attila Lendvai)
922 * bug fix: #426; nested function calls are inlined properly.
923 Previously if FOO was an inline function, in calls of the form
924 (FOO (FOO ...)) the outer call was not inlined.
925 * bug fix: long long arguments passed in registers to alien
926 functions on PowerPC Linux and NetBSD are now handled correctly
927 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
929 changes in sbcl-1.0.22 relative to 1.0.21:
930 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
931 for the shared object in the current directory, but passes the native
932 namestring of the designated pathname to the operation system's shared
933 object loading function as-it.
934 * minor incompatible change: --disable-debugger toplevel option now takes
935 effect before processing of initialization files and --eval or --load
937 * new feature: new commandline argument: --script, which supports
938 shebang lines. See documentation for details. (based on work by
940 * new feature: SAVE-LISP-AND-DIE can save current values of
941 --dynamic-space-size and --control-stack-size in the executable core,
942 causing it to skip normal runtime option processing. See documentation
943 for details. (thanks to Zach Beane)
944 * enhancement: inoccous calls to EVAL or generic functions dispatching
945 on subclasses of eg. STREAM no longer cause compiler notes to appear.
946 * enhancement: the system no longer resignals errors from --load and
947 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
948 associated with the original error to be lost. (thanks to Ariel
950 * enhancement: :DONT-SAVE keyword argument has been added to
951 LOAD-SHARED-OBJECT for controlling interaction with
952 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
953 the effects of an earlier LOAD-SHARED-OBJECT call.
954 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
955 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
956 provided. (reported by Cedric St-Jean)
957 * bug fix: circularity handling in the reader did not treat raw
958 structure slots correctly. (reported by Cedric St-Jean)
959 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
960 file descriptors when there were none.
961 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
962 was broken. (reporteed by Willem Broekema)
963 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
964 pathnames without a directory.
965 * bug fix: COMPILE-FILE did not warn about undefined variable
966 references at toplevel, and LOAD of the resulting fasl did
968 * bug fix: functions with non-required arguments used to end up with
969 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
970 * bug fix: redefining a function with non-required arguments didn't
971 update the system's knowledge about its call signature properly.
972 * bug fix: fixed #431; incompatible alien record type redefinitions
973 are detected and handled. (reported by Neil Haven)
974 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
975 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
977 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
978 after alien stack frames.
979 * bug fix: strings passed to mk{s,d}temp lacked null termination
981 changes in sbcl-1.0.21 relative to 1.0.20:
982 * new feature: the compiler is able to track the effective type of a
983 generic function across method addition and removal.
984 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
985 appropriately typed structure slots without locking.
986 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
987 information like those gathered by TIME using a programming-friendly
989 * new feature: TIME reports time taken even if the form performs a
990 non-local transfer of control.
991 * enhancement: reduced conservativism on GENCGC platforms: on
992 average 45% less pages pinned (measured from SBCL self build).
993 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
994 mutate constant symbols or violate declaimed type of the symbol.
995 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
996 sleeping on the semaphore in heavily contested semaphores.
997 * bug fix: semaphores and condition variables were not interrupt
999 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
1000 owned by other threads anymore.
1001 * bug fix: FIND on lists called KEY outside the specified
1002 subsequence. (reported by budden)
1003 * bug fix: LOG doesn't use single-float intermediate results when
1004 given mixed integer and double-float arguments, leading to better
1005 precision. (reported by Bob Felts)
1006 * bug fix: LOG with base zero returned values of inconsistent type.
1008 changes in sbcl-1.0.20 relative to 1.0.19:
1009 * minor incompatible change: OPTIMIZE qualities
1010 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
1011 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
1012 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
1013 * documentation: some slot access efficiency guidelines have been
1014 added to the user manual.
1015 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
1016 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
1017 as ASSOC and MEMEBER.
1018 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
1019 to more efficient EQ-comparison versions more often.
1020 * optimization: enhanced derivation of DOLIST iteration variable type
1022 * optimization: constant folding of simple (LIST ...) forms as DOLIST
1024 * optimization: runtime lookup of function definitions can be
1025 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
1026 * optimization: compiler is able to derive the return type of
1027 (AREF (THE STRING X) Y) as being CHARACTER.
1028 * optimization: CLRHASH on empty hash-tables no longer does pointless
1029 work. (thanks to Alec Berryman)
1030 * optimization: REPLACE deftransforms don't punt when :START1 or
1032 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
1033 no longer emits a code deletion note. (reported by Andrew
1035 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
1036 functions with an odd number of &OPTIONAL arguments, a &REST
1037 argument, and one or more &KEY arguments at the call site.
1038 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
1039 arguments appeared at call sites as well.
1040 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
1041 READ-LINE &co can no longer read from them afterwards. (reported
1043 * bug fix: fixed #427: unused local aliens no longer cause compiler
1044 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
1046 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
1047 alien-stack leakage. (reported by Andy Hefner)
1048 * bug fix: PROGV signals an error when an attempt to violate declared
1049 type of a variable or bind a constant is made.
1050 * bug fix: SET signals an error when an attempt to violate declared
1051 type of a variable is made.
1052 * bug fix: restart computation during the execution of a restart
1053 test function no longer causes infinite recursion. (reported by
1055 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
1056 signals an error. (thanks to Pierre Mai)
1057 * bug fix: EXPORT left symbol unexported in conflict situations.
1058 (thanks to Michael Weber)
1059 * bug fix: correctly handle name conflicts involving CL:NIL.
1060 (thanks to Michael Weber)
1061 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
1062 conflicts arising from USEing package with conflicting symbols
1063 correctly. (thanks to Michael Weber)
1065 changes in sbcl-1.0.19 relative to 1.0.18:
1066 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
1067 warnings that go otherwise unhandled will be muffled if they are
1068 of the type that's the value of this variable.
1069 * optimization: stack allocation is slightly more efficient on x86
1071 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
1072 where a raw slot always is initialized using the initform whose
1073 type is not know sufficiently well a compile-time are now compiled
1074 correctly. (reported by John Morrison)
1075 * bug fix: compiler no longer makes erronous assumptions in the
1076 presence of non-foldable SATISFIES types.
1077 * bug fix: stack analysis missed cleanups of dynamic-extent
1078 arguments in non-let-converted calls to local functions.
1079 * improvements to the Windows port:
1080 ** adjusted address spaces for building on both Win32 and
1081 Win64. (thanks for John Connors)
1082 * fixed some bugs revealed by Paul Dietz' test suite:
1083 ** interval arithmetic during type derivation used inexact integer
1084 to single-float coercions.
1085 ** arithmetic operations involving large integers and single
1086 floats give the same results in compiled and interpreted code.
1087 ** deriving the result type of COERCE no longer signals an error
1088 if the derived type of the second argument is a MEMBER type
1089 containing invalid type specifiers.
1090 ** ADJOIN with constant NIL as second argument works correctly.
1091 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
1093 changes in sbcl-1.0.18 relative to 1.0.17:
1094 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
1095 profiles only the current thread.
1096 * minor incompatible change: changes to SYMBOL-VALUE of constants
1097 defined with DEFCONSTANT now signal an error.
1098 * enhancement: SB-SPROF now has support for wallclock profiling,
1099 and is also able to profile specific threads. REPORT output
1100 has also additional sorting options.
1101 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
1103 * optimization: structure allocation has been improved
1104 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
1105 ** out of line constructors are ~10% faster.
1106 ** inline constructors are ~15% faster.
1107 ** inline constructors are capable of dynamic extent allocation
1108 (generally on x86 and x86-64, in some cases on other platforms
1110 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
1112 * optimization: file compiler is now able to coalesce non-circular
1113 lists, non-base strings, and bit-vectors. Additionally, constants
1114 are never referenced through SYMBOL-VALUE at runtime.
1115 * optimization: code defining methods on PRINT-OBJECT (and other
1116 generic functions in the COMMON-LISP package) now loads faster.
1117 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
1119 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
1120 fasl is now deleted. (reported by Attila Lendvai)
1121 * bug fix: READ-LINE always returned NIL for the last line in files.
1122 (reported by Yoshinori Tahara)
1123 * bug fix: more accurate disassembly annotations of foreign function
1124 calls. (thanks to Andy Hefner)
1125 * bug fix: trimming non-simple strings and non-string string
1126 designators when the there is nothing to trim works properly.
1127 (thanks to James Knight)
1128 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
1129 (thanks to Travis Cross)
1130 * fixed some bugs revealed by Paul Dietz' test suite:
1131 ** NIL is a valid function name (regression at 1.0.13.38)
1132 ** FILL on lists was missing its return value (regression at 1.0.12.27)
1133 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1134 fill pointers properly (regression at 1.0.12.23)
1135 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
1136 displacement indices properly (regression at 1.0.12.23)
1138 changes in sbcl-1.0.17 relative to 1.0.16:
1139 * temporary regression: user code can no longer allocate closure
1140 variable storage on stack, due to bug 419 without explicitly
1141 requesting it. Please consult sbcl-devel for advice if you need to
1142 use this feature in the meanwhile.
1143 * new feature: runtime argument --control-stack-size can be used to
1144 adjust thread default control stack size.
1145 * enhancement: improved TIME output
1146 ** all times are reported using the measured accuracy (milliseconds
1147 for real and GC times, microseconds for everything else.)
1148 ** processor cycle counts on x86 and x86-64.
1149 ** interpreted forms are counted for both evaluator modes.
1150 ** number of lambdas converted by the compiler is reported.
1151 ** CPU percentage report (computed from real and total run time.)
1152 ** more comprehensive run time reporting, using a condenced format
1153 ** interperted form, lambda, and page fault counts are omitted
1155 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
1157 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
1158 * optimization: two argument forms of LAST are upto ~10% faster
1159 in normal SPEED policies.
1160 * optimization: NCONC no longer needs to heap cons its &REST list
1161 in normal SPEED policies.
1162 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
1163 to Sidney Markowitz)
1164 * bug fix: UNION and NUNION work with :TEST-NOT once more,
1165 regression since 1.0.9.1. (thanks to Eric Marsden)
1166 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
1167 since 1.0.15.36. (thanks to Paul Khuong)
1168 * bug fix: LAST when always returned the whole list when given a bignum
1169 as the second argument.
1170 * bug fix: dynamic extent allocation of nested lists and vectors
1171 could leak to otherwise accessible parts.
1172 * bug fix: invalid optimization of heap-allocated alien variable
1174 * bug fix: fasl header checking is less vulnerable to different
1175 platform word lengths.
1176 * bug fix: more correct assembler syntax for GNU binutils
1177 2.18.50.0.4 support. (thanks to Marijn Schouten)
1178 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
1180 * bug fix: the fopcompiler can handle LOCALLY forms (with no
1181 declarations) successfully. (reported by Attila Lendvai)
1183 changes in sbcl-1.0.16 relative to 1.0.15:
1184 * minor incompatible change: revert the changes to sb-posix's error
1185 signaling added in 1.0.14.
1186 * minor incompatible change: change PROBE-FILE back to returning
1187 NIL whenever we can't get a truename, as was the case before 1.0.14.
1188 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
1189 inherits from ERROR instead of just CONDITION.
1190 * new feature: SB-INTROSPECT can provide source locations for instances
1191 as well. (thanks to Tobias Rittweiler)
1192 * optimization: binding special variables now generates smaller code
1193 on threaded platforms.
1194 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
1195 and cases where no :TEST is given but the compiler can infer that
1196 the element to search is of type (OR FIXNUM (NOT NUMBER)).
1197 * optimization: better LOGNOT on fixnums.
1198 * optimization: modular arithmetic for a particular requested width
1199 is implemented using a tagged representation unless a better
1200 representation is available.
1201 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
1202 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
1203 for wellformedness and returns a specialized vector. (reported by
1204 Francois-Rene Rideau)
1205 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
1206 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
1207 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
1208 unscheduling timer at the same time as another timer fires.
1209 * bug fix: don't reschedule timers for dead threads.
1210 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
1211 * bug fix: copying output from RUN-PROGRAM to a stream signalled
1212 bogus errors if select() was interrupted.
1213 * enhancement: add support for fcntl's struct flock to SB-POSIX.
1215 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
1216 * enhancement: cleaner backtraces for interactive interrupts, as
1217 well as other cases where the interesting frames used to be
1218 obscured by interrupt handling frames.
1219 * enhancement: untracing a whole package using (UNTRACE "FOO") is
1220 now supported, and tracing a whole package using (TRACE "FOO") now
1221 traces SETF-functions as well.
1222 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
1223 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
1224 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
1225 * unreadably printed representation of hash-tables now includes
1227 * bug fix: partially fixed #188: type propagation from assignments
1228 is now more efficient.
1229 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
1230 'DOUBLE-FLOAT) are not flushable.
1231 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
1232 full address of the object, and none of the tag bits.
1233 * bug fix: readably printing hash-tables now respects other printer
1234 control variables. (reported by Cedric St-Jean)
1235 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
1236 keyword with MAKE-HASH-TABLE.
1237 * bug fix: export SB-POSIX:MKSTEMP.
1238 * bug fix: SORT was not interrupt safe.
1239 * bug fix: XREF accounts for the last node of each basic-block as
1241 * bug fix: MAKE-INSTANCE optimizations interacted badly with
1242 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
1243 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
1244 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
1245 create obsolete instances in the case of redefinition or
1246 obsoletion of a superclass. (thanks to Andy Hefner)
1247 * bug fix: Support for the Alpha architecture has been revived; it had
1248 suffered somewhat from lack of maintenance since sbcl-1.0.
1249 * improvements to the Windows port:
1250 ** The system detects the codepage to use at startup, instead of
1251 using the value from the saved core. (thanks to Kei Suzuki)
1253 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
1254 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
1255 (see documentation for details.)
1256 * revived support for OpenBSD (contributed by Josh Elsasser)
1257 * partially fixed bug #108: ROOM no longer suffers from occasional
1258 (AVER (SAP= CURRENT END)) failures.
1259 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
1260 bodies no longer cause a WARNING to be signalled. (reported by
1262 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
1263 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
1264 the bug in TRUNCATE.) (reported by Michael Weber)
1265 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
1266 no samples. (reported by Andy Hefner)
1267 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
1268 no longer appear as (NIL ...) frames in backtraces.
1269 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
1270 is now actually exported from SB-EXT as documented. (reported by
1272 * bug fix: sb-aclrepl now correctly understands how to inspect
1273 single-floats on 64-bit platforms where single-floats are not boxed.
1274 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
1275 yet been finalized. (reported by Levente Meszaros)
1276 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
1277 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
1278 SBCL-specific optimize qualities.
1280 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
1281 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
1282 an executable in the search path, and does so in the child
1283 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
1284 been removed; it can be found in the CVS history, for somebody who
1285 needs that search behavior (see the manual).
1286 * minor incompatible change: compiler policy re. weakening type
1287 checks has changed: now type checks are weakened only if SAFETY < 2
1289 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
1290 unparsing of directory pathnames as files. Analogously,
1291 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
1292 filename to parse into a directory pathname.
1293 * enhancement: implicit generic function creation now signals a
1294 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
1295 users can bind handlers for and muffle around calls to LOAD.
1296 * enhancement: RUN-PROGRAM allows unicode arguments and environments
1297 to be used (using the default stream external format), and allows
1298 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
1299 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
1300 strings and vectors whose element-type or simplicity is not fully
1301 known at compile-time.
1302 * optimization: STRING-TRIM and related functions no longer allocate
1303 a new string when no trimming needs to be performed. These functions
1304 are also faster than before when the input string has been declared
1306 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
1307 * optimization: READ-LINE is significantly faster for files containing
1309 * optimization: non-open coded uses of character comparison operators
1310 (e.g. char=) no longer cons when called with more than one parameter
1311 on platforms supporting dynamic-extent allocation.
1312 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
1313 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
1314 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
1316 * bug fix: some sequence functions elided bounds checking when
1318 * bug fix: too liberal weakening of union-type checks when SPEED >
1320 * bug fix: more bogus fixnum declarations in ROOM implementation
1323 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
1324 * new feature: MAKE-HASH-TABLE now experimentally accepts a
1325 :SYNCHRONIZED argument, which makes the hash-table safe for
1326 concurrent accesses (but not iteration.) See also:
1327 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1328 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1329 * optimization: CONCATENATE on strings is an order of magnitude faster
1330 in code compiled with (> SPEED SPACE).
1331 * optimization: SUBSEQ is ~50% faster on lists.
1332 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1333 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1334 END is smaller then START.
1335 * bug fix: SB-PROFILE will no longer report extra consing for nested
1336 calls to profiled functions.
1337 * bug fix: ROOM implementation had bogus fixnum declarations which
1338 could cause type-errors when calling ROOM on large images.
1339 * bug fix: if file compilation is aborted, the partial fasl is now
1340 deleted, and COMPILE-FILE returns NIL as the primary value.
1341 * bug fix: number of thread safety issues relating to SBCL's internal
1342 hash-table usage have been fixed.
1343 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1344 be returned from its body when the values were being returned
1345 using unknown-values return convection and the W-P-O was wrapped
1346 inside an UNWIND-PROTECT.
1347 * bug fix: sb-posix should now compile again under Windows, enabling
1348 slime to work again.
1350 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1351 * incompatible change: hash-table accessor functions are no longer
1352 automatically protected by locks. Concurrent accesses on the same hash-table
1353 from multiple threads can give inconsistent results or even corrupt the
1354 hash-table completely. Multi-threaded applications should do their own
1355 locking at the correct granularity. In the current implementation it is
1356 still safe to have multiple readers access the same table, but it's not
1357 guaranteed that this property will be maintained in future releases.
1358 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1359 and will signal an error at runtime.
1360 * enhancement: SB-THREAD package now exports a semaphore interface.
1361 * enhancement: CONS can now stack-allocate on x86 and
1362 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1364 * enhancement: nested lists can now be stack allocated on
1365 platforms providing stack allocation support.
1366 * enhancement: dynamic-extent support has been extended to support
1367 cases where there are multiple possible sources for the stack
1369 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1370 if the mutex is uncontested on Linux.
1371 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1372 as specified by the CLHS. (thanks to Tobias Rittweiler)
1373 * bug fix: NaN comparison now works on x86-64.
1374 * bug fix: CLOSE :ABORT T on a stream with pending output now
1376 * bug fix: instances of non-standard metaclasses using standard
1377 instance structure protocol sometimes missed the slot type checks
1379 * bug fix: known functions can be inlined in high-debug code.
1381 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1382 * minor incompatible change: the MSI installer on Windows no longer
1383 associates .lisp and .fasl files with the installed SBCL.
1384 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1385 on Windows. (thanks to Luis Oliviera)
1386 * new feature: SB-CLTL2 contrib module now implements
1387 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1388 * optimization: scavenging weak pointers is now more efficient,
1389 requiring O(1) instead of O(N) per weak pointer to identify
1390 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1391 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1392 method bodies using constant slot-names when the first argument is
1393 a specializer parameter for the method.
1394 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1395 names, when the class of the instance is a direct instance
1396 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1397 as slow as the constant slot-name case.)
1398 * optimization: member type construction is now O(N) instead
1400 * optimization: UNION and NUNION are now O(N+M) for large
1401 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1403 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1404 now more readable in environments like Slime which display it.
1405 (thanks to Tobias C. Rittweiler)
1406 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1407 was able to infer the correct argument type for the object on which
1408 the CAS operation was being performed.
1409 * bug fix: copy propagation interfered with parallel assignment
1410 semantics in local calls. (reported by Paul Khuong)
1411 * bug fix: the signed modular fixnum shift compiled to wrong code on
1412 x86 and x86-64. (spotted by a slight modification to some of
1415 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1416 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1417 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1418 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1420 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1421 and generic function now print the object identity for anonymous
1422 instances. (thanks to Attila Lendvai)
1423 * enhancement: as an extension to MOP, SBCL now supports SETF of
1424 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1425 (thanks to Attila Lendvai)
1426 * enhancement: sb-cover has an alternate annotation mode which puts
1427 the form annotations at the car of the form, instead of annotating
1428 it completely. The benefit of this mode is that it shows better
1429 which forms were instrumented by the compiler.
1430 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1431 by printer control variables. (reported by Dan Corkill)
1432 * bug fix: system leaked memory when delayed output was performed by
1433 the OS in smaller chunks then expected. (thanks to David Smith)
1434 * bug fix: system leaked memory when file streams were not closed
1436 * bug fix: large objects written to slow streams that were modified
1437 after the write could end up with the modified state written to
1438 the underlying file descriptor.
1439 * bug fix: multiple threads operating in parallel on the same stream
1440 could cause buffer-overflows.
1441 * bug fix: source location information is stored correctly
1442 (broken since 1.0.6). This bug would generally show up as the
1443 Slime debugger highlighting the wrong form when the sldb-show-source
1445 * bug fix: Branch forms are again annotated as branches in the sb-cover
1447 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1448 on x86-64. (reported by Gregory Vanuxem)
1449 * bug fix: EQUALP could return wrong results for structures with raw
1450 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1451 word). (reported by Vjacheslav Fyodorov)
1452 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1455 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1456 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1457 atomic compare-and-swap operations on threaded platforms.
1458 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1459 allows assigning a global minimum value to optimization qualities
1460 (overriding proclamations and declarations).
1461 * enhancement: closed over variables can be stack-allocated on x86
1463 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1465 * optimization: slot definition lookup is now O(1). This speeds up
1466 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1467 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1468 * optimization: ASSOC and MEMBER can now be open-coded for all
1469 combinations of keyword arguments when second argument is constant
1470 and SPEED >= SPACE. In other cases a specialized version is
1472 * bug fix: using obsoleted structure instances with TYPEP and
1473 generic functions now signals a sensible error.
1474 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1475 (reported by Kristoffer Kvello)
1476 * bug fix: backtrace construction is now more careful when making
1477 lisp-objects from pointers on the stack, to avoid creating bogus
1478 objects that can be seen by the GC.
1479 * bug fix: defaulting of values in contexts expecting more than 7
1480 variables now works on x86-64. (reported by Christopher Laux)
1481 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1483 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1484 as the property-list of a symbol.
1485 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1486 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1487 situations requested, are once again file-compileable. (reported
1490 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1491 * MOP improvement: support for user-defined subclasses of
1492 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1493 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1494 the expansion of DEFMETHOD, is responsible for generating a form
1495 which creates a list of specializers when evaluated. Additional
1496 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1497 debugging and introspective support.
1498 * minor incompatible change: the (unsupported) spinlock interface
1499 has changed: free spinlock now has the value NIL, and a held spinlock
1500 has the owning thread as its value.
1501 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1502 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1504 * enhancement: name of a socket-stream is now "a socket" instead of
1505 "a constant string".
1506 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1507 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1508 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1509 for generic functions. (Thanks to Tobias C. Rittweiler)
1510 * optimization: bignum printing speed has been improved by 20-40%
1511 (depending on the bignum size.)
1512 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1514 * bug fix: the cache used by the CLOS to store precomputed effective
1515 methods, slot offsets, and constant return values is now thread and
1517 * bug fix: generic function dispatch function updating is now thread
1518 and interrupt safe (in the sense that the known issues have been
1520 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1521 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1522 * bug fix: asdf systems with dependencies to the SB-POSIX or
1523 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1524 * bug fix: interrupt safety of applicable method computation has been
1527 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1528 * new contrib: sb-cover, an experimental code coverage tool, is included
1529 as a contrib module.
1530 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1531 significantly faster.
1532 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1533 produce more efficient code on x86-64 if the compiler can determine
1534 that (+ INDEX <constant>) does not require a bounds check and FOO
1535 has an element type at least 8 bits wide.
1536 * enhancement: a new, experimental synchronous timeout facility is
1537 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1538 * enhancement: when a symbol name conflict error arises, the
1539 conflicting symbols are always printed with a package prefix.
1540 (thanks to Kevin Reid)
1541 * enhancement: stepping is now once again supported on the SPARC and
1542 MIPS platforms. (It is also now more likely to work on CheneyGC
1544 * enhancement: sb-sprof can now also track and report accurate call
1546 * bug fixes: the treatment of non-standard subclasses of
1547 SB-MOP:SPECIALIZER is more correct.
1548 * incompatible change: PURIFY no longer copies the data from the
1549 dynamic space into the static and read-only spaces on platforms
1550 that use the generational garbage collector
1551 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1553 * bug fix: GC race condition occasionally resulting in crashes with
1554 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1556 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1558 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1559 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1560 * bug fix: an asynchronous interrupt could previously leave the
1561 system running with GC inhibited.
1562 * bug fix: a DECLARE form evaluated at top-level now causes an error
1563 rather than silently (or verbosely) returning NIL.
1564 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1565 result in a error being signalled due to source form lookup errors.
1566 (reported by Peter Graves)
1568 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1569 * incompatible change: removed writer methods for host-ent-name,
1570 host-ent-addresses -- changing the values did not update the DNS
1572 * minor incompatible change: changed experimental JOIN-THREAD interface
1573 * documentation: the manual now lists reader and writer methods
1574 in class slot documentation sections. (thanks to Richard M Kreuter)
1575 * documentation: unwinding from asyncronous events has been
1576 documented as unsafe.
1577 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1578 in multithreaded application code.
1579 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1580 platforms. (thanks to James Anderson for the optimization hint)
1581 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1583 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1584 variants no longer cons.
1585 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1586 their NOT- variants no longer cons.
1587 * optimization: EQUAL hash tables no longer use SXHASH for objects
1588 of all data types, but instead use an EQL hash for types for which
1589 EQUAL is the same as EQL.
1590 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1591 are significantly faster.
1592 * optimization: new STRING-OUTPUT-STREAM implementation is much
1593 faster for large outputs and conses 30% less on average.
1594 * enhancement: XREF information is now collected for references made
1595 to global variables using SYMBOL-VALUE and a constant argument.
1596 * enhancement: SIGINT now causes a specific condition
1597 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1598 * bug fix: FILE-POSITION used to signal an error in some cases where
1599 ANSI requires it to return NIL.
1600 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1601 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1602 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1604 * bug fix: bad type declaration in the CLOS implementation has
1605 been fixed. (reported by James Anderson)
1606 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1607 (thanks to Lutz Euler)
1608 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1609 required by the CLHS. (reported by Samium Gromoff)
1610 * bug fix: dead unbound variable references now signal an error.
1611 (reported by Marco Monteiro)
1612 * bug fix: / with an unused value was being deleted in safe code.
1613 (thanks to Marco Monteiro and Kevin Reid)
1614 * bug fix: number of characters that can be written onto a single
1615 line in a file is unlimited.
1616 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1617 been fixed by inhibiting interrupts for when GC is disabled.
1618 * bug fix: some interrupt safety issues with GETHASH, (SETF
1619 GETHASH), CLRHASH and REMHASH have been fixed.
1620 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1621 specifier no longer causes infinite recursion.
1622 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1623 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1624 * bug fix: an &environment argument with the correct variable information
1625 is passed to macros that are expanded during byte compilation.
1626 (reported by Samium Gromoff)
1627 * bug fix: a logic error was causing the hash function for bignums to
1628 have excessive amounts of collisions. (reported by Faré Rideau)
1629 * bug fix: modifying the contents of an array could change the return
1630 value of SXHASH on that array, which is only allowed for strings
1631 and bit vectors. (bug introduced in 0.9.16)
1632 * improvement: the x86-64/darwin port now passes all tests (except
1633 for the debugger tests) but should still be considered
1634 experimental until this is fixed.
1635 * improvement: a style-warning is signaled for CASE (etc) clauses with
1636 duplicate keys. (thanks to Kevin Reid)
1637 * improvement: macos/x86 and macos/x86-64 now use mach exception
1638 handlers for dealing with illegal instructions (for trapping and
1639 error handling) and memory protection violations (for GC).
1641 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1642 * new platform: experimental support for x86-64/darwin (MacOS).
1643 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1644 and getnameinfo sockets functions are used instead of gethostbyaddr
1645 and gethostbyname, on platforms where the newer functions are available.
1646 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1648 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1649 don't print the startup banner, but behave as if --noinform was passed
1650 as a command line argument. (thanks to Kevin Reid)
1651 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1652 * optimization: code using alien values with undeclared types is much faster.
1653 * optimization: the compiler is now able to open code SEARCH in more cases.
1654 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1655 * bug fix: using standardized COMMON-LISP special variables as loop
1656 variables no longer signals bogus package lock violations. (reported
1658 * bug fix: declaring local loop variables to be of a range-limited type
1659 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1660 (reported by Andras Simon)
1661 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1662 bugs remain on x86-64.)
1663 * bug fix: the #= and ## reader macros now interact reasonably with
1664 funcallable instances.
1665 * bug fix: type-checks for function arguments were compiled using the
1666 compiler policy settings of the wrong lexical environment (in 1.0.2
1668 * bug fix: SHADOW accepts characters as string designators, as required
1669 by the spec (thanks to Eric Marsden)
1670 * bug fix: fixed GC safety issues when foreign functions are called with
1671 non-base strings as arguments
1672 * bug fix: more consistent error output for fatal-compiler-warnings, like
1674 * bug fix: functions with proper names were showing up as NIL in
1675 backtraces (reported by Edi Weitz, regression in 1.0.2)
1677 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1678 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1679 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1680 produce more efficient code on the x86 if the compiler can determine
1681 that (+ INDEX <constant>) does not require a bounds check and FOO
1682 has an element type at least 8 bits wide.
1683 * bug fix: references to undefined variables are handled the same way
1684 in toplevel forms as in the normal compiler
1685 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1687 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1688 (thanks to Jon Buller)
1689 * improvement: faster compilation times for complex functions
1690 * improvement: added readlink support to SB-POSIX (thanks to Richard
1693 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1694 * improvement: experimental support for mach exception handling on
1695 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1697 * improvement: support for GBK external format.
1698 (thanks to Chun Tian (binghe))
1699 * improvement: the debugger now displays variables that have been closed
1700 over, in code compiled with (DEBUG 2) or higher
1701 * improvement: support for executable cores on NetBSD (thanks to
1703 * new feature: added a RESTART-FRAME debugger command
1704 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1705 be used to provide an implementation for FILE-POSITION on Gray streams
1706 (thanks to Eric Marsden)
1707 * new feature: add syslog, openlog and closelog support to SB-POSIX
1708 (thanks to Richard Kreuter)
1709 * optimization: the function call overhead in code compiled with
1710 a high DEBUG optimization setting is significantly
1711 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1712 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1713 (thanks to Tony Martinez)
1714 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1715 evaluated once (reported by Kevin Reid)
1716 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1717 works again (thanks to Kilian Sprotte)
1718 * bug fix: an error is signaled for tagbodies with duplicate tags
1719 (thanks to Stephen Wilson)
1720 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1721 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1722 stack frames from alien callbacks.
1723 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1724 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1725 65k pages (thanks to David Woodhouse)
1726 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1728 changes in sbcl-1.0.1 relative to sbcl-1.0:
1729 * new platform: FreeBSD/x86-64, including support for threading.
1730 * new feature: the compiler stores cross-referencing information
1731 about function calls (who-calls), macroexpansion (who-macroexpands)
1732 and special variables (who-binds, who-sets, who-references) for code
1733 compiled with (< SPACE 3). This information is available through the
1734 sb-introspect contrib.
1735 * new feature: users may subclass SEQUENCE, and have instances of
1736 these classes interoperate with standard Common Lisp functions if
1737 a number of methods are defined. (This feature is experimental
1738 and the interface subject to change based on feedback from SBCL
1739 users and the general community)
1740 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1741 x86/x86-64, rather than the previous fixed depth of 8
1742 * improvement: another pthread back-end of mutex "pthread-futex".
1743 * bug fix: non-ascii command-line arguments are processed correctly
1744 (thanks to Yaroslav Kavenchuk)
1745 * bug fix: non-required arguments were not passed correctly when a method
1746 defined using DEFMETHOD was called from a mop-generated method using
1747 CALL-NEXT-METHOD (reported by Pascal Costanza)
1748 * bug fix: recursion is now permitted in accessors through
1749 SLOT-UNBOUND. (reported by Pascal Costanza)
1750 * bug fix: an error was signaled at startup if the HOME environment
1751 variable was defined, but had an empty value (reported by Peter Van Eynde)
1752 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1753 signal a PROGRAM-ERROR, not a BUG.
1754 * bug fix: SB-POSIX:READDIR works when built with large file support.
1755 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1757 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1758 for better automatic indentation support. (thanks to Matt Pillsbury)
1759 * optimization: loading generic functions no longer takes O(n^2) time,
1760 proportional to the amount of methods in the generic function
1761 (reported by Todd Sabin and Jeremy Brown)
1762 * optimization: the FIND and POSITION family of sequence functions
1763 are significantly faster on arrays whose element types have been
1765 * improvements to the Windows port:
1766 ** Intermittent heap corruption problems have been fixed. (thanks
1767 to Alastair Bridgewater)
1768 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1770 ** Lisp is able to unwind foreign exception frames from alien
1771 callbacks. (thanks to Alastair Bridgewater)
1773 changes in sbcl-1.0 relative to sbcl-0.9.18:
1774 * improvement: experimental support for threading on FreeBSD/x86.
1775 (thanks to NIIMI Satoshi)
1776 * improvement: runtime option --dynamic-space-size can be used
1777 to set the size of the dynamic space reserved on startup.
1778 * improvement: floating point modes in effect are now saved in
1779 core, and restored on startup.
1780 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1781 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1782 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1783 threads are not running after *SAVE-HOOKS* have run.
1784 * improvement: writes to CLOS instance slots are type-checked in code
1785 compiled with (SAFETY 3)
1786 * improvement: floating-point exception handling on FreeBSD (thanks to
1788 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1789 (thanks to Zach Beane)
1790 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1792 * improvement: added support for the Shift-JIS external format.
1793 (contributed by NIIMI Satoshi)
1794 * improvement: callbacks are supported on Linux/PPC. (thanks to
1796 * bug fix: compiler bug triggered by a (non-standard) VALUES
1797 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1798 * bug fix: file compiler no longer confuses validated and already
1799 dumped structurres. (reported by Kaersten Poeck)
1800 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1801 fill-pointer signals a type-error as required. (thanks to
1803 * bug fix: disassemly of funcallable instances works.
1804 * bug fix: single stepping on PPC.
1805 * bug fix: fix thread-safety problems in the type system (generally
1806 manifesting as nonsensical errors like "STRING is a bad type specifier
1807 for sequences" or "The value 1 is not of type FIXNUM")
1808 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1809 (reported by Josip Gracin)
1810 * bug fix: an error is signaled for attempts to displace arrays with
1811 incompatible element types (thanks to Mario Mommer)
1812 * bug fix: more correct handling of wide characters in debug info
1813 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1814 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1815 and don't cause extra consing
1816 * optimization: MAP and MAP-INTO are significantly faster on vectors
1817 whose elements types have been declared.
1818 * Improvements to SB-SPROF:
1819 ** Support for allocation profiling
1820 ** Reduced profiling overhead, especially for long profiling runs
1821 * Improvements to the Windows port:
1822 ** floating point exceptions are now reported correctly.
1823 ** stack exhaustion detection works partially.
1824 ** more accurate GET-INTERNAL-REAL-TIME.
1825 ** better RUN-PROGRAM behaviour in case of errors in the child
1827 ** PROBE-FILE now simplifies pathnames correctly.
1828 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1829 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1830 1st 1970 (although time timezone data will be unreliable).
1831 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1833 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1834 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1835 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1837 * bug fix: two potential GC deadlocks affecting threaded builds.
1838 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1839 returns T (reported by Anton Kazennikov)
1840 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1841 with non-variable places
1842 * bug fix: remove a race condition in the setting of
1843 funcallable-instance functions, this should make threaded CLOS
1844 code more stable against memory faults.
1845 * bug fix: corruption of specials when unbinding is interrupted by an
1846 asynchronous unwind (reported by Hannu Koivisto)
1847 * improvement: the debugger will now also display local variables that
1848 are only used once, for code compiled with a DEBUG optimization quality
1851 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1852 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1853 * incompatible change: External-format support for FFI calls. The
1854 SB-ALIEN:C-STRING no longer implies an ASCII
1855 external-format. Instead the string is subject to external-format
1856 conversion. Additionally return values of type SB-ALIEN:C-STRING
1857 are by default of type (SIMPLE-ARRAY CHARACTER), not
1858 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1859 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1860 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1861 * incompatible change: SB-EXT package no longer contains the
1862 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1863 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1865 * incompatible change: the single-stepper is no longer available
1866 on Alpha, Mips and Sparc platforms.
1867 * minor incompatible change: the direct superclasses of
1868 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1869 not (STANDARD-OBJECT FUNCTION). This makes the
1870 class-precedence-lists of GENERIC-FUNCTION and
1871 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1873 * minor incompatible change: the default stream external format on
1874 non-unicode SBCL is no longer detected from the locale, but is always
1876 * new feature: Add a version of evaluator that uses an interpreter instead
1877 of the compiler. EVAL still uses the compiler by default, to switch it
1878 to use the interpreter, set the value of the variable
1879 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1880 * minor incompatible change: the single-stepper REPL has been merged
1881 with the normal debugger (see the "Stepping" heading of the debugger help
1882 for more details). The debugger command STEP will no longer switch
1883 to the single-stepper REPL.
1884 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1885 for a type now works.
1886 * bug fix: loading of default sysinit file works. (thanks to Leonid
1888 * bug fix: better detection of circularities in the file-compiler.
1889 (reported by Marco Monteiro)
1890 * bug fix: the CL pathname functions now work with files that have
1891 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1892 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1893 systems with Unix98 pty semantics.
1894 * bug fix: ASDF-INSTALL will now work with bsd tar.
1895 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1897 * bug fix: timers expiring in dead threads no longer cause a
1898 type-error (reported by Paul "Nonny Mouse").
1899 * bug fix: thanks to more lightweight single-stepper instrumentation,
1900 code compiled with (DEBUG 3) will compile and execute significantly faster,
1901 and will have more accurate type-inferencing than before
1902 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1903 whose bindings are modified
1904 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1905 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1906 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1907 CL:SOFTWARE-VERSION with proper version
1909 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1910 * feature: implemented the READER-METHOD-CLASS and
1911 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1912 as specified by AMOP.
1913 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1915 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1916 (thanks to Lutz Euler)
1917 * optimization: hashing of general arrays and vectors has been
1918 improved. (reported by Any Fingerhut)
1919 * enhancement: SB-INTROSPECT is now able to find definitions of
1920 profiled functions. (thanks to Troels Henriksen)
1921 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1922 as well, allowing compiler-macros for SETF-functions to expand.
1923 * enhancement: step-instrumentation no longer wraps known
1924 single-value functions with multiple-value context, allowing
1925 better type inference.
1926 * fixed bug #337: use of MAKE-METHOD in method combination now works
1927 even in the presence of user-defined method classes. (reported by
1928 Bruno Haible and Pascal Costanza)
1929 * fixed bug #339(c): if there are applicable methods not part of any
1930 long-form method-combination group, call INVALID-METHOD-ERROR.
1931 (reported by Bruno Haible)
1932 * fixed bug #361: the :FUNCTION initarg in the protocol for
1933 initialization of methods can now be used to override
1934 internally-produced optimized functions. (reported by Bruno
1936 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1937 system-provided lambda expression no longer cause warnings about
1938 unbound #:|pv-table| symbols.
1939 * bug fix: improved the handling of type declarations and the
1940 detection of violations for keyword arguments with non-constant
1942 * bug fix: potentially erronous calls to PATHNAME and
1943 MERGE-PATHNAMES were being flushed in some cases.
1944 (reported by Richard Kreuter)
1945 * bug fix: compiled calls to TYPEP were mishandling obsolete
1946 instances. (reported by James Bielman and Attila Lendvai)
1947 * bug fix: format strings with ~> without matching ~< no longer
1948 trigger an AVER, but signal an understandable error instead.
1949 (reported by Antonio Martinez)
1950 * bug fix: specifying an output-file without a pathname-type for
1951 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1952 of input-file instead of "fasl". (reported by Robert Dodier)
1953 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1954 (reported by James Y Knight).
1955 * bug fix: compiler-macros-function did not consider the environment
1956 argument for shadowing by local functions.
1957 * bug fix: compiler-macros expansion was inhibited by local INLINE
1959 * bug fix: inline expansions of known functions were subject to
1960 step-instrumentation in high debug policies, leading to problems
1961 with type-inference.
1962 * bug fix: compiler failed to differentiate between different CONS
1963 types in some cases.
1964 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1965 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1966 * bug fix: #368: incorrect use of expressed vs. upgraded array
1968 * bug fix: #306a: more precise unions of array types.
1969 * thread-safety improvements:
1970 ** CONDITION-WAIT could return early on Linux, if the thread was
1971 interrupted and subsequently continued with SIGCONT.
1972 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1974 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1975 * added support for the ucs-2 external format. (contributed by Ivan
1977 * minor incompatible change: pretty printing of objects of type
1978 (cons symbol) is, in the default pprint-dispatch-table, now
1979 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1981 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1982 called later in a class's lifetime, possibly as late as when the
1983 first instance of the class is created. Previously,
1984 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1985 class became finalizeable.
1986 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1987 for multibyte external-format streams. (thanks to "vbzoli")
1988 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1989 with :ALLOCATION :CLASS inherited from superclasses of the
1991 * fixed bug: anonymous classes can now be created using the :NAME
1992 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1993 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1994 * fixed bug: core-files saved with :EXECUTABLE T can again be
1995 executed when SBCL_HOME isn't set. (reported by James Knight)
1996 * fixed bug: toplevel LOCALLY forms with declarations could
1997 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1998 * fixed bug: printing from several different threads using different
1999 values of *print-case* could cause invalid output, due to
2000 some internal special variables of the printer not being bound
2001 thread-locally (reported by Max Mikhanosha)
2002 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
2003 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
2004 (reported by Pascal Costanza)
2005 * fixed bug: FINALIZE-INHERITANCE is called from
2006 REINITIALIZE-INSTANCE on classes when the class has previously
2007 been finalized, as required by AMOP.
2008 * minor code generation optimizations:
2009 ** better register allocation in CLOS dispatching functions
2010 ** overflow detection when coercing signed bytes to fixnums on x86-64
2011 ** is now implemented with one IMUL instruction instead of three shifts
2012 ** more efficient bit-vector access on x86 and x86-64
2013 ** more efficient access to raw structure slots on x86 and x86-64
2014 * fixed some bugs revealed by Paul Dietz' test suite:
2015 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
2016 return its argument.
2018 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
2019 * feature: thread support on Solaris/x86, and experimental thread support
2021 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
2023 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
2024 default method for SLOT-UNBOUND.
2025 * minor incompatible change: prevent the user from specializing the
2026 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
2027 somewhat counter to the intent of the protocol, I (CSR) think, and
2028 additionally it just doesn't work in SBCL as currently
2029 implemented, thanks to optimizations (that are always valid for
2030 the other three SLOT-VALUEish functions, but not for the setter).
2031 * minor incompatible change: the :SB-LDB feature is now enabled by
2032 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
2033 the low-level debugger.
2034 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
2035 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
2036 down to an SB-SYS:FD-STREAM, allowing direct communication in
2037 more cases. Temporary files and pipes are still used as a fallback
2039 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
2040 on the non-reentrant C functions localtime(3) and gmtime(3).
2041 * bug fix: global optimization policy was not visible in LOCALLY and
2043 * bug fix: class objects can be used as specializers in methods.
2044 (reported by Pascal Costanza)
2045 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
2046 with a type error. (reported by blitz_ on #lisp)
2047 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
2048 failed with a type error. (reported by Pascal Bourguignon)
2049 * bug fix: merging pathnames against defaults with :DIRECTORY
2050 starting with '(:RELATIVE :BACK) should preserve the :BACK.
2051 (reported by James Y Knight)
2052 * bug fix: saving large (>2GB) cores on x86-64 now works
2053 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
2054 constant index and a (simple-array (signed-byte 32)) array
2055 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
2056 error (patch by Robert J. Macomber)
2057 * feature: TIME now displays the amount of run-time spent in GC
2058 * bug fix: The debugger now does a better job of respecting
2059 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
2060 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
2061 workaround for bug 403.)
2062 * bug fix: floating point exception handling now works on FreeBSD
2063 (thanks to NIIMI Satoshi)
2064 * fixed some bugs revealed by Paul Dietz' test suite:
2065 ** MISC.641: LET-conversion were not supposed to work in late
2067 * improvements to the Win32/x86 port:
2068 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
2069 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
2071 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
2072 * new feature: source path information is generated for macro-expansion
2073 errors for use in IDE's like Slime (thanks to Helmut Eller)
2074 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
2075 * bug fix: compiler does not loop forever on an invalid type in
2077 * improvement: compilation of most CLOS applications is significantly
2079 * optimization: added a limited bytecode compiler for simple toplevel
2080 forms, speeding up compilation and FASL loading
2081 * bug fix: the statistical profiler now properly distinguishes anonymous
2084 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
2085 * minor incompatible change: in sbcl-0.9.11 (but not earlier
2086 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
2087 into subclasses of the finalized class. Now user calls to
2088 FINALIZE-INHERITANCE finalize just the one class, and calls by the
2089 system return before any subclasses are finalized.
2090 * minor incompatible change: The reader no longer ignores errors
2091 regarding non-existent packages in #+ and #- feature tests.
2092 * new feature: command line options --no-sysinit, --no-userinit to
2093 inhibit loading the corresponding init files
2094 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
2095 for sending data through UDP sockets (thanks to François-René Rideau)
2096 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
2097 error is signalled instead (thanks to François-René Rideau)
2098 * minor incompatible change: Error signalling behaviour of lexical
2099 operations violating package locks has changed slightly. Refer to
2100 documentation on package locks for details.
2101 * enhancement: EVAL can process IF-expressions without resorting to the
2103 * bug fix: Some operations involving SIN, COS, or TAN weren't being
2104 constant-folded properly due to missing out-of-line functions.
2105 (reported by Mika Pihlajamäki)
2106 * bug fix: LISTEN sometimes returned T even in cases where no data was
2107 immediately available from the stream
2108 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
2109 were reversed. (reported by Levente Mészáros)
2110 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
2111 when given a vector displaced to an adjustable array. (reported
2113 * bug fix: derivation of float boundaries from numbers outside the
2114 appropriate float range (reported by John Wiseman)
2115 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
2117 * bug fix: precomputing cache entries for generic functions with
2118 some subclasses of specializers as yet invalid does not attempt to
2119 fill a cache line with a negative offset. (reported by Levente
2121 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
2122 allow condition class objects as arguments to DOCUMENTATION and
2123 (SETF DOCUMENTATION); only find and set documentation for
2124 structure names for the STRUCTURE doc-type. (suggested by Gary
2126 * improvements to the Win32/x86 port:
2127 ** (user-homedir-pathname) and default initialization file
2128 locations now know about the user's "Documents and Settings"
2129 directory (thanks to Yaroslav Kavenchuk)
2130 ** run-program is implemented (thanks to Mike Thomas)
2131 ** sockets support (thanks to Timothy Ritchey)
2132 ** better backtrace support (thanks to Alastair Bridgewater)
2133 ** sb-grovel supported
2134 ** asdf-install and sb-posix work somewhat
2135 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
2136 * improvements to the Solaris/x86 port:
2137 ** works on Solaris 11/Solaris Express
2138 ** floating-point exception handling support
2139 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
2140 * fixed some bugs revealed by Paul Dietz' test suite:
2141 ** REMOVE-METHOD returns its generic function argument even when
2142 the method is not one of the generic functions' methods.
2143 ** objects declared as MEMBER types can be admissible arguments to
2144 structure accessors.
2145 ** printing characters should simply be printed by the FORMAT ~:C
2147 ** compiler failure when compiling functions with hairy constant
2148 defaults for optional parameters.
2149 ** compiler produces wrong code when MAYBE-INLINE-expanding a
2150 function, which is already optimized.
2152 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
2153 * new platform: experimental support for SBCL x86/Darwin, including
2154 MacOS X 10.4.5 on Intel.
2155 * new feature: Unicode character names are now known to the system
2156 (through CHAR-NAME and NAME-CHAR).
2157 * new feature: the filesystem location of SBCL's core file is
2158 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
2159 * minor incompatible change: the contrib modules SB-POSIX and
2160 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
2161 this change is to make it easier to distribute
2162 location-independent binaries.
2163 * bug fix: as implied by AMOP, standardized classes no longer have
2164 slots named by external symbols of public packages. (reported by
2166 * optimization: calling functions via constant symbols -- as in
2167 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
2168 via the function object as in (FUNCALL #'FOO).
2169 * optimization: on x86 and x86-64, the calling convention for the
2170 case of dealing with an unknown number of values has been altered
2171 to be friendlier to the prediction heuristics implemented,
2172 particularly with reference to CALL and RET pairing. (thanks to
2173 Alastair Bridgewater)
2174 * optimization: on x86, the code for access to thread-local symbol
2175 values and binding/unbinding of thread-local symbols is smaller.
2176 (thanks to Alastair Bridgewater)
2177 * enhancement: CONSTANTP is now able to determine constantness of
2178 more complex forms, including calls to constant-foldable standardized
2179 functions and some special forms beyond QUOTE.
2180 * fixed bug: occasional GC crashes on Solaris/x86
2181 * optimization: x86-64 supports stack allocation of results of simple
2182 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
2183 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
2184 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
2185 (thanks to James Knight)
2186 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
2187 by the standard (thanks to Stephen Compall)
2189 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
2190 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
2191 be used for bundling the runtime and the core file into one
2192 executable binary. This feature is not currently supported on all SBCL
2193 platforms. (thanks to James Bielman and NIIMI Satoshi)
2194 * new feature: a generational or ephemeral garbage collector is now
2195 the default on the PowerPC platform (both Linux and Darwin). The
2196 old Cheney (stop and copy) collector is a build-time option.
2197 (thanks to Cyrus Harmon, after Raymond Toy)
2198 * minor incompatible change: the method by which SBCL finds its
2199 contributed modules has changed; it no longer relies on symbolic
2200 links from an $SBCL_HOME/systems directory, but searches directly
2201 in the subdirectories of $SBCL_HOME.
2202 * enhancement: the dynamic heap size on the Linux/PPC platform is
2203 markedly larger, even using the older Cheney garbage collector.
2204 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
2205 certain complicated string types. (reported by Gary King)
2206 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
2207 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
2208 * fixed bug: DOCUMENTATION on structures no longer signals an
2209 error if no structure type of the right name exists. (reported by
2211 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
2212 applicable non-standard primary or :AROUND method on
2213 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
2214 no longer cause unbound variable errors. (reported by Kevin Reid)
2215 * fixed bug: metacircle resolution in cases where methods have slots
2216 added before the slots from STANDARD-METHOD. (reported by Jean
2218 * fixed bug: the Power PC assembly code for calling into lisp
2219 sometimes computed its return address wrongly (depending on the
2220 15th bit of the address, fixed at link time). This used to
2221 manifest itself in a segmentation violation while building PCL.
2222 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
2223 many others over the years)
2224 * fixed bug: ROOM no longer reports silly numbers for stack usage if
2225 the stack pointer (treated as a signed integer) is negative.
2226 (thanks to Peter van Eynde)
2228 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
2229 * new platform: experimental support for the Windows operating
2230 system has been added. (thanks to Alastair Bridgewater)
2231 * fixed several bugs in and robustified the PPC FFI (including
2232 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
2233 * bug fix: allow non-simple string symbol names (reported by Paul
2235 * bug fix: interrupt handling on NetBSD (thanks to Richard M
2237 * bug fix: saving a core corrupted callbacks on x86/x86-64
2238 * bug fix: closed a loophole in metacircularity detection and
2239 grounding in the PCL implementation of CLOS.
2240 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
2241 (reported by tomppa on #lisp)
2242 * bug fix: clear the direction flag on Lisp -> C transitions, as
2243 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
2244 (reported by Andrej Grozin and Hendrik Maryns)
2245 * optimization: major improvements to GC efficiency on GENCGC platforms
2246 * optimization: faster implementation of EQUAL
2247 * optimization: emit more efficient opcodes for some common
2248 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
2249 * optimization: several other minor code-generation improvements on x86-64
2250 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
2252 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
2253 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2254 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
2255 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
2256 by ANSI as a generic function, and for consistency (SETF
2257 GENERIC-FUNCTION-NAME) is treated likewise.
2258 * fixed bug #233.b: make constraint propagation notice when a variable
2259 value is changed after it is referenced but before it is used
2260 * fixed bug #296: no more arbitrary behaviour from filenames with
2261 odd characters as --load arguments. (reported by Adam Warner)
2262 * fixed bug #390: :CHARACTER-SET pathname components now work as
2263 expected. (reported by Tim Daly Jr)
2264 * fixed bug #391: complicated :TYPE intersections in slot
2265 definitions no longer cause an error in PCL internals.
2266 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
2267 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
2268 returning the number of octets which would be written to the
2269 file-stream. (thanks to Robert J. Macomber)
2270 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
2271 arguments to the internal stream functions. (thanks to David
2273 * bug fix: stack exhaustion is now handled on the x86/SunOS
2274 platform. (thanks to Daisuke Homma)
2275 * bug fix: more accurate ROOM results on GENCGC platforms
2276 * optimization: improved type inference for arithmetic-for
2277 index variables in LOOP
2278 * optimization: faster floating-point SQRT on x86-64
2279 * fixed some bugs revealed by Paul Dietz' test suite:
2280 ** DOCUMENTATION returns NIL instead of "" for method combinations
2281 that don't have a docstring
2283 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
2284 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2285 GENERIC-FUNCTION-NAME) are no longer generic functions, and
2286 therefore are not customizeable by user code (as seems to be at
2287 least permitted and maybe required by AMOP). As a consolation,
2288 however, the SBCL implementation of these functions now calls
2289 REINITIALIZE-INSTANCE as specified by AMOP.
2290 * bug fix: slot-definition documentation is propagated as per ANSI
2291 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
2292 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
2293 argument, as do eslotd initialization methods. (from Pascal
2294 Costanza's "Closer" project)
2295 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
2296 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
2298 * bug fix: it is now possible to have more than one subclass of
2299 STANDARD-GENERIC-FUNCTION without causing stack overflow.
2300 (reported by Bruno Haible, Pascal Costanza and others)
2301 * bug fix: the dependent update protocol now works for generic
2302 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
2303 and Pascal Costanza)
2304 * bug fix: condition-class instances corresponding to
2305 DEFINE-CONDITION forms are now created eagerly. (reported by
2306 Kalle Olavi Niemitalo on comp.lang.lisp)
2307 * bug fix: floating point printing is more accurate in some
2308 circumstances. (thanks to Simon Alexander)
2309 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
2310 merged with *DEFAULT-PATHNAME-DEFAULTS*.
2311 * bug fix: callbacks on OS X now preserve stack-alignment.
2312 (thanks to Cyrus Harmon)
2313 * enhancement: the x86-64 disassembler is much better at
2314 disassembling SSE instructions. (thanks to Lutz Euler)
2315 * enhancement: CHECK-TYPE on a variable now allows type-inference
2316 (thanks to Vincent Arkesteijn)
2317 * optimization: improved performance of EUC-JP external format.
2318 (thanks to NIIMI Satoshi)
2319 * optimization: performance improvements to IO on file streams of
2320 :ELEMENT-TYPE CHARACTER
2321 * optimization: much faster memory allocation on x86-64
2322 * optimization: faster garbage collections (latency and throughput)
2324 * optimization: faster bignum subtraction and division on x86 and x86-64
2325 (thanks to Lutz Euler)
2326 * optimization: more accurate type-inference for loops that use a
2327 floating point index variable or a negative step.
2329 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2330 * bug fix: add a workaround to SBCL looping infinitely at startup on
2331 Linux kernels with apparently buggy implementations of personality().
2332 (thanks to Svein Ove Aas)
2333 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2334 (thanks to David Lichteblau)
2335 * bug fix: local bindings shadow global symbol macros for
2336 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2337 * optimization: non-open coded uses of numeric comparison operators
2338 (e.g. >) no longer cons when called with more than one parameter
2339 on platforms supporting dynamic-extent allocation.
2340 * enhancement: saving cores with foreign code loaded is now supported
2341 on MIPS/Linux in addition to the previously supported platforms.
2342 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2343 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2344 * bug fix: division by zero in sb-sprof when no samples were collected
2345 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2346 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2347 (thanks to Alexander Kjeldaas)
2348 * new feature: ignoring the return values of destructive functions
2349 like NREVERSE signals a compile-time style-warning.
2350 (thanks to Kevin Reid)
2351 * new feature: the alignment of alien structure fields can be
2352 explicitly specified. (thanks to Cyrus Harmon)
2353 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2354 and dump core on SIGQUIT
2356 ** incompatible change: threads do not inherit values of specials
2357 from their parents (see manual)
2358 ** bug fix: threads stacks belonging to dead threads are freed by the
2359 next exiting thread, no need to gc to collect thread stacks anymore
2360 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2361 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2362 ** bug fix: dynamic variable and thread start related gc lossage
2363 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2365 * fixed some bugs revealed by Paul Dietz' test suite:
2366 ** SUBTYPEP is slightly more accurate on heinously complicated
2367 CONS types where some of the members have uncertain (in the
2368 NIL, NIL sense) type relationships to each other.
2370 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2371 no longer purified unless :PURIFY T is explicitly specified.
2372 ** Non-purified cores are significantly smaller than before
2374 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2375 * new feature: timers based on Zach Beane's excellent timer package
2376 * added support for the following external formats: koi8-u,
2377 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2378 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2379 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2380 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2381 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2382 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2383 * incompatible change: a threaded SBCL will no longer revert to
2384 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2385 * bug fix: interrupts are disabled until startup is complete; no
2386 more sigsegvs when receiving a signal to soon
2387 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2389 * bug fix: add a workaround for the memory randomization features in
2390 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2391 workaround will only be in effect on systems with the proc filesystem
2393 * bug fix: printing objects of type HASH-TABLE signals a
2394 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2396 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2397 have been running for over 50 days. (reported by Gilbert Baumann)
2398 * bug fix: the logic for getting names of functions gets less
2399 confused when confronded with alternate-metaclass
2400 funcallable-instances. (reported by Cyrus Harmon)
2401 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2402 consistent, even on internal alternate-metaclass objects.
2403 * bug fix: flush closure information collected by physical
2404 environment analysis prepass before the main pass. (bug reported
2406 * bug fix: compiler pack phase does not modify a hashtable, which is
2407 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2408 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2409 may be expanded. (reported by Alexander <malishev> on c.l.l)
2410 * bug fix: no more style-warnings when compiling code using LAST
2411 under high-SPEED low-SPACE settings. (reported by David Wallin
2413 * fixed some incompatibilities between SBCL's MOP and the MOP
2415 ** the METAOBJECT class is now implemented;
2416 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2417 STANDARD-OBJECT, as required;
2418 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2419 now compatible (as required by VALIDATE-SUPERCLASS); there
2420 remains a consistency requirement over the presence or absence
2421 of the FUNCTION class in the superclass list of finalized
2422 classes; see the manual for more details;
2423 ** the :AROUND method for COMPUTE-SLOTS on
2424 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2425 requested slot ordering.
2427 ** bug fix: parent thread now can be gc'ed even with a live
2429 ** bug fix: binding a special with PROGV to no value is not
2430 the same as that symbol not having been bound (thanks to
2432 ** bug fix: binding specials is thread safe (thanks to
2434 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2435 and SB-SPROF do not die with 'no handler for signal XX in
2436 interrupt_handle_now(..)' anymore
2437 ** bug fix: WITH-TIMEOUT works with multiple running threads
2438 * fixed some bugs revealed by Paul Dietz' test suite:
2439 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2440 the :method-class keyword argument.
2442 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2443 * new port: the Solaris operating system on x86 processors is now
2444 mostly supported, though some rough edges in the environment
2445 remain. (thanks to Daisuke Homma)
2446 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2447 space for its heap. (thanks to Thiemo Seufer)
2448 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2449 closures. (thanks to Thiemo Seufer)
2450 * minor incompatible change: eof selects abort in the debugger.
2451 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2452 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2454 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2455 deprecated, and will go away in a future revision of SBCL.
2456 * minor incompatible change: GC-ON and GC-OFF are no longer
2457 implemented with a counter, it does not matter how many times gc
2458 is switched on or off
2459 * bug fix: discriminating functions for generic function classes
2460 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2461 make invalid assumptions about method precedence order. (reported
2463 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2464 platforms. (thanks to Thiemo Seufer)
2465 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2466 * bug fix: correct pathname printing with printer escaping is on.
2467 (thanks to Kevin Reid)
2468 * bug fix: complex VOP definitions in "user-space" no longer trigger
2469 package locks. (reported by Zach Beane)
2470 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2471 for funcallable-instance functions, and is no different from
2473 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2474 characters in the current readtable. (reported by Nicholas Neuss)
2475 * bug fix: SUBTYPEP on various CONS types returns more a more
2476 accurate acknowledgment of its certainty.
2477 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2478 lists in some cases. This partially fixes bug 384.
2479 * flush all standard streams before prompting in the REPL and the
2481 * bug fix: signal handling and triggering gc do not conflict
2482 directly anymore, in particular a high frequency sb-sprof does
2483 not prevent gc from running
2484 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2485 approximation for timezone and DST information between the
2486 universal time epoch and the smallest negative 32-bit time_t.
2487 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2488 year 1899 so long as the encoded time is non-negative.
2489 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2490 the correct answer for non-constant rotate amounts when performing
2491 an inline 32-bit rotation.
2493 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2494 there is only one thread in the session
2495 ** bug fix: memory leak for streams created in one thread and
2496 written to in another
2497 ** bug fix: lockup when compiled with gcc4
2498 ** bug fix: race that allows the gc to be triggered when gc is
2500 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2501 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2503 ** bug fix: finished threads are gc'ed properly
2504 * fixed some bugs revealed by Paul Dietz' test suite:
2505 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2506 has arguments with a different set of applicable methods from
2507 the orignal arguments.
2508 ** The type error thrown by MAP now has a correct expected-type
2510 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2511 name a compiled function.
2512 ** Three MISC tests where a large bignum was improperly coerced to
2513 a float that couldn't represent that bignum during type
2514 derivation were fixed.
2515 ** SUBTYPEP can now handle types involving the negation of a
2516 list-form FUNCTION type.
2517 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2518 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2519 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2521 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2522 * New feature: Experimental support for bivalent streams: streams
2523 opened with :element-type :default now allow character and binary
2524 (unsigned-byte 8) I/O
2525 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2526 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2528 * bug fix: no more highly sporadic "couldn't check whether ~S is
2529 readable" when reading a stream and an interrupt hits in the middle
2530 of a select system call
2531 * compiler better recognizes complex arrays (reported by Eduardo
2533 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2534 for variables on SBCL built with linkage-tables. (reported by Luis
2536 * various error reporting improvements.
2537 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2538 (Thanks to Thiemo Seufer)
2539 * bug fix: nested reader invokations work correctly
2540 * bug fix: it is possible to have simultaneous references to foreign
2541 code and foreign data with the same name.
2543 ** added x86-64 support
2544 ** incompatible change: the threading api now works with thread
2545 objects instead of thread ids
2546 ** bug fix: threads are protected from signals and interruption when
2547 starting up or going down
2548 ** bug fix: a race where an exiting thread could lose its stack to gc
2549 ** bug fix: don't halt on infinite error in threads if possible
2550 ** fixed numerous gc deadlocks introduced in the pthread merge
2551 ** bug fix: fixed thread safety issues in read and print
2552 ** bug fix: debugger doesn't hang on session lock if interrupted at
2553 an inappropriate moment
2554 ** bug fix: run-program is now thread safe(r)
2555 ** bug fix: inner with-recursive-lock no longer releases the mutex
2556 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2557 * fixed some bugs revealed by Paul Dietz' test suite:
2558 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2559 EXPECTED-TYPE slots filled.
2560 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2562 ** MAKE-CONDITION accepts classes as type-designators.
2563 ** COMPILE may never return NIL.
2564 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2565 range before calling Unix time functions
2567 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2568 * numerous signal handling fixes to increase stability
2569 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2570 * minor incompatible change: we now correctly canonize default
2571 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2572 as per the MOP, rather than the historical (INITARG INITFUNCTION
2573 INITFORM). (reported by Bruno Haible)
2574 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2575 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2576 for more information.
2577 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2578 pathname is a directory pathname.
2579 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2580 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2582 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2583 :METACLASS argument in addition to a class name. (reported by
2584 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2585 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2586 (reported by Bruno Haible; patch from Gerd Moellmann)
2587 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2589 * bug fix: more cleanups to the floating point exception handling on
2590 x86-64 (thanks to James Knight)
2591 * bug fix: the compiler does not try to constant fold calls of
2592 COERCE and BIT-* functions when they return freshly constructed
2593 objects. (reported by Nikodemus Siivola and Paul Dietz)
2594 * optimization: improved the MIPS versions of generic (in the
2595 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2596 * optimization: direct conversion of (unsigned-byte 32) to floats on
2597 the PowerPC platform.
2598 * optimization: structure instances with raw slots now use less
2599 memory, and probably show better memory locality. (thanks to
2601 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2602 are treated as requests for stack allocation on the x86-64,
2603 Alpha-32, PPC and SPARC platforms.
2604 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2605 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2606 (thanks to Tony Martinez)
2608 ** gcing a dead thread can no longer lead to lockups
2609 ** threads block signals until they are set up properly
2610 ** errno is no longer shared by threads
2611 ** interrupt-thread restores the eflags register on x86
2612 ** fixed some lockups due to gc/thread interaction
2613 * fixed some bugs revealed by Paul Dietz' test suite:
2614 ** invalid dotted lists no longer raise a read error when
2615 *READ-SUPPRESS* is T
2616 ** signal an error if a symbol that names a declaration is used
2617 as the name of a type, or vice versa
2618 ** allow using the (declare (typespec var*)) abbreviation for
2619 (declare (type typespec var*)) with all type specifiers
2620 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2621 ** always return NIL from PPRINT-POP when OBJECT is NIL
2622 ** don't signal errors when pretty-printing malformed LABELS,
2623 FLET or MACROLET forms
2624 ** declarations in a DOLIST body are also in scope for the
2626 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2628 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2631 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2632 * fixed cross-compiler leakages that prevented building a 32-bit
2633 target with a 64-bit host compiler.
2634 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2635 opened with :IF-EXISTS :OVERWRITE.
2636 * fixed bug 281: error for an invalid qualifier in a short-form method
2637 combination method is not signalled until the faulty method is called.
2638 * bug fix: iteration variable type inferrer failed to deal with open
2639 intervals. (reported by Alan Shields)
2640 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2641 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2642 or more runtime options were provided to the sbcl binary.
2643 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2644 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2645 * optimization: there's now a fast-path for fixnum arguments in the
2646 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2647 * optimization: the code generated on x86-64 is more compact thanks
2648 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2649 * bug fix: floating-point exception handling now partly works on
2650 x86-64. (Thanks to James Knight)
2651 * improvement to the MIPS backend from Thiemo Seufer: C-style
2652 64-bit long long arguments and return values to alien functions
2654 * fixed some bugs revealed by Paul Dietz' test suite:
2655 ** the type-error signalled from WARN has a filled-in DATUM slot.
2656 ** the type-error required when a stream is not associated with
2657 a file has the stream as its datum.
2658 ** type-errors on single-floats on x86-64 no longer have
2659 :INVALID-OBJECT as the datum
2660 ** the type-errors signalled for invalid function names now have
2661 a correct expected type
2662 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2663 for structures defined with a :TYPE. Documentation strings for
2664 typed structures are no longer immediately discarded
2665 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2666 broadcast streams and synonym streams. FILE-LENGTH now also works
2667 on broadcast streams.
2669 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2670 * incompatible change: the --noprogrammer option, deprecated since
2671 version 0.7.5, has been removed. Please use the equivalent
2672 --disable-debugger option instead.
2673 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2675 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2676 inoperational for a while now) has been completely removed.
2677 * null lexical environments are now printed as #<NULL-LEXENV>,
2678 significantly reducing the amount of clutter in typical backtraces.
2679 * documentation on weak pointers, finalization, and after GC hooks
2680 has been added to the manual.
2681 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2682 as other specialized array types, is much faster. SUBSEQ and
2683 COPY-SEQ on such arrays have also been sped up.
2684 * optimization: EQL is now more efficient when at least other argument
2685 is known to be of type (OR FIXNUM (NOT NUMBER)).
2686 * fixed bug: the runtime is now less vulnerable to changes in the
2687 size of the SBCL object on OS X, and virtual memory is reserved for
2688 all spaces that need to be at a fixed address.
2689 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2690 * fixed bug: finalizers and after GC hooks that cause consing are now
2692 * fixed bug: compiler error messages and summaries are now printed to
2693 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2694 * fixed inference of the upper bound of an iteration variable.
2695 (reported by Rajat Datta).
2696 * fixed bug 211e: calling local functions with duplicated constant
2697 keyword argument no longer causes a bogus style warning about an
2699 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2700 ftype declaration to be disregarded. (reported by Dave Roberts)
2701 * fixed bug 373: caused by erronous compilation of references to alien
2702 variables in the runtime on ppc/darwin.
2703 * fixed bug 376: CONJUGATE type deriver.
2704 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2705 (reported by Baughn on #lisp)
2706 * a cross-compiler bug on non-x86 platforms has been identified and
2707 fixed. (thanks to Bruno Haible)
2708 * improvements to the MIPS runtime code for increased stability.
2709 (thanks to Thiemo Seufer)
2710 * increased the maximimum compact environment size to allow
2711 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2712 * partial workaround for bug 135: don't name the function we're
2713 calling for hairy cases of EVAL, so as not to accumulate one environment
2714 entry per call to EVAL. (reported by Kevin Reid)
2715 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2716 * optimization: type testing for non-vector arrays should be faster.
2717 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2719 * bug fix: setting 31st element of a bit vector to zero did not work
2721 * bug fix: redefining a class definition which failed due to a
2722 previous accessor / function clash now works (but see BUGS entry
2723 #380 for more problems in this area). (thanks to Zach Beane)
2724 * the long form of DEFINE-METHOD-COMBINATION disables method group
2725 checking when given a single method group with pattern *. (thanks
2727 * on x86 compiler supports stack allocation of results of simple
2728 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2729 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2730 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2731 * fixed some bugs related to Unicode integration:
2732 ** the restarts for recovering from input and output encoding
2733 errors only appear when there is in fact such an error to
2735 * fixed some bugs revealed by Paul Dietz' test suite:
2736 ** MISC.549 and similar: late transformation of unsafe type
2737 assertions into derived types caused unexpected code
2739 ** SCALE-FLOAT type deriver is less wrong.
2740 ** type derivers for EXP, LOG and similar functions compute result
2741 types for complex arguments better.
2742 ** (MISC.563) CONJUGATE type deriver works for very restricted
2744 ** out-of-line type testers for character strings are available.
2745 ** EQUAL compiler transform understands specialness of objects
2747 ** accessing double-floats stored on the stack now works on x86-64.
2748 ** debugger internals could sometimes create invalid lispobjs,
2749 resulting in GC crashes.
2750 ** MISC.548: type check weakening can convert required type into
2752 ** initialization forms for bindings are not in scope of free special
2755 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2756 * incompatible change: thread support for non-NPTL systems has
2757 been removed - locking is buggy and unreliable. A threaded
2758 SBCL build will now warn at startup and refuse to create
2759 new threads, unless futex support is detected in the kernel
2760 * incompatible change: the top level REPL now has only an ABORT
2761 restart associated with it, not TOPLEVEL and ABORT as it used to.
2762 TOP and TOPLEVEL are now available as debugger commands for
2763 returning to the top level.
2764 * incompatible change: forms evaluated in the REPL now use the
2765 global optimization policy.
2766 * incompatible change: user- and system-initialization files are
2767 no longer processed with LOAD, but by READ and EVAL; hence the
2768 global optimization policy, startup package, readtable, etc,
2770 * The .fasl file format number has been incremented because of
2771 various incompatible changes.
2772 * internal entry point details and argument counts no longer appear
2773 in backtraces unless explicitly requested by setting
2774 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2775 * built-in and standard functions no longer have names like "top
2776 level local call to FOO".
2777 * fixed bug 32: functions defined in non-null lexical environments
2778 now have more legible printed representation
2779 * fixed bug 33: functions defined in non-null lexical environemnts
2780 are now more amenable to inspection by INSPECT.
2781 * workaround for bug 354: XEPs no longer appear in backtraces unless
2782 explicitly requested.
2783 * fixed bug: receiving the signal which results from attempting to
2784 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2785 SIGBUS on other *BSDs) on architectures where the C stack is also
2786 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2787 to debugger confusion or wild instability. (reported by Cheuksan
2789 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2790 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2791 (reported by Lutz Euler)
2792 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2793 compiled code. (reported by Paul Dietz)
2794 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2795 specialized parameter to a method to something that is not TYPEP
2796 the specializer is now possible.
2797 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2798 face of package deletion.
2799 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2800 pathname. (reported by David Tolpin on c.l.l)
2801 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2802 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2803 STRING-STREAM and FILE-STREAM, respectively.
2804 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2805 than previously; in addition, its test suite is now run on build.
2806 (thanks to Luke Gorrie)
2807 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2808 * optimization: added a immediate representation for single-floats
2810 * fixed some bugs revealed by Paul Dietz' test suite:
2811 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2812 ** attempting to create a package with a colliding nickname causes
2813 correctable errors to be signalled.
2814 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2815 ** several x86-64 backend bugs related to sign-extension of immediate
2818 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2819 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2820 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2822 * fixed loading of multiply forward-referenced layouts.
2823 (thanks to Cheuksan Wang)
2824 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2825 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2826 (see COMPILE-FILE documentation for details of :PRINT option)
2827 * fixed bugs 19 and 317: fixed-format floating point printing is
2828 more accurate. This also fixes a bug reported by Adam Warner
2829 related to the ~@F format directive.
2830 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2832 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2833 dispatch table if the from-char is a dispatch macro character.
2834 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2835 by the wordsize no longer ignore the last word. (reported by Lutz
2837 * fixed bug in type checking of dynamic-extent variables. (reported
2839 * optimization: sequence traversal functions use their freedom to
2840 coerce function designators to functions.
2841 * optimization: code with many calls to CLOS methods specialized on
2842 CLOS classes has had redundant type checks removed.
2843 * optimization: added declarations to speed up operations that access
2844 the internal character database (for example STRING-UPCASE)
2845 * optimization: comparison operations between floats and sufficiently small
2846 fixnums no longer create extra rationals
2847 * fixed some bugs related to Unicode integration:
2848 ** portions of multibyte characters at the end of buffers for
2849 character-based file input are correctly transferred to the
2850 start of the buffer at the next read.
2851 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2852 passing it through to OPEN.
2853 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2854 argument given to internal calls to OPEN.
2855 ** the built-in comment readers (introduced by character sequences
2856 ";" and "#|") are more forgiving to encoding errors; they will
2857 STYLE-WARN and then attempt to resync the stream at a character
2858 boundary. (thanks to Teemu Kalvas)
2859 * fixed some bugs in the x86-64 port:
2860 ** Negative short int return values from c-calls are sign-extended
2862 ** The stack is aligned to 16-bytes for c-calls, as required by
2863 the ABI. (thanks to Cheuksan Wang)
2864 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2865 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2866 for enums. (thanks to Vincent Arkesteijn)
2867 ** Multiple small optimizations and bugfixes for floating point
2869 * fixed some bugs revealed by Paul Dietz' test suite:
2870 ** Space, Tab, Linefeed, Return and Page have the invalid
2871 secondary constituent character trait.
2872 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2874 ** WITH-INPUT-FROM-STRING should only update the index place on
2876 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2877 works more reliably.
2878 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2879 with one-dimensional arrays with fill pointers.
2880 ** TYPE-OF failed on a complex with an integer realpart and a
2882 ** compiler failure during type inference for the code of form
2883 (IF (EQL X (THE ...)) ...) (MISC.535).
2885 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2886 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2887 platform. The port supports 61-bit fixnums, large memory spaces
2888 and reloading shared object files.
2889 * enhancement: saving cores with foreign code loaded is now
2890 supported on ppc/Darwin in addition to the previously supported
2892 * enhancement: the statistical profiler now walks deeper into the
2893 call stack for more meaningful call-graphs and accrued time
2894 reports (x86/x86-64 only). It also now reports time spent in
2896 * enhancement: it is now possible to trace most individual methods
2897 of a generic function in addition to tracing the generic function
2899 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2901 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2902 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2903 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2904 produces an error. (thanks to Vincent Arkesteijn)
2905 * bug fix: NAMESTRING on pathnames with :WILD components in their
2906 directories works correctly. (thanks to Artem V. Andreev)
2907 * fixed bug 125: compiler preserves identity of closures. (reported
2909 * bug fixed: functions with &REST arguments sometimes failed with
2910 "Undefined function" when compiled with (DEBUG 3). (reported by
2912 * bug fix: overflow during compiling of setting element of a bit
2913 vector with constant index and value. (reported by Timmy Douglas)
2914 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2915 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2916 * on x86 compiler partially supports stack allocation of dynamic-extent
2918 * GO and RETURN-FROM do not check the extent of their exit points
2919 when compiled with SAFETY 0.
2920 * fixed some bugs related to Unicode integration:
2921 ** encoding and decoding errors are now much more robustly
2922 handled; it should now be possible to recover even from invalid
2923 input or output to the terminal. (thanks to Teemu Kalvas)
2924 ** provided a first cut at implementing STRING-TO-OCTETS and
2925 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2926 ** altered the SB-MD5 contributed module slightly, changing the
2927 interface just enough for it to be supportable for builds where
2928 lisp characters are not eight bits.
2929 * fixed some bugs revealed by Paul Dietz' test suite:
2930 ** the FORMATTER-generated functions for ~V[ conditionals require
2931 the correct number of arguments.
2932 ** READ-FROM-STRING returns the mandated second value when applied
2933 to displaced strings.
2934 ** the #\Rubout and #\Backspace characters are treated as invalid
2935 constituent characters by the tokenizer.
2937 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2938 * new feature: reloading changed shared object files with
2939 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2940 * new feature: references to foreign variables and functions
2941 can now be compiled and loaded before the corresponding shared
2942 object file is loaded, as long as the foreign definitions are
2943 available at runtime.
2944 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2945 supported. (thanks to Dan Debertin)
2946 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2947 just keywords. (thanks to Vincent Arkesteijn)
2948 * fixed bug #331: structure-class instances corresponding to
2949 DEFSTRUCT forms are now created eagerly.
2950 * fixed bug #345: backtraces from calls to undefined functions work
2951 on x86 as well. Related bug #61 is now also partially fixed on x86
2952 (backtraces from throws to unknown catch tags.)
2953 * bug fix: lambda-list parsing is now stricter vrt. order and number
2954 of lambda-list keywords.
2955 * bug fix: as specified by AMOP, an error is signalled if a
2956 class-option appears multiple times in a DEFCLASS form. (reported
2958 * bug fix: RANDOM can be compiled when the compiler derives the type
2959 of its numeric argument as a disjoint set of small integers.
2960 (reported by Paul Dietz)
2961 * bug fix: starting a core saved with shared objects loaded when
2962 those objects are not available no longer causes threaded SBCL to
2963 hang. (reported by Sean Ross)
2964 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2966 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2967 argument, which must always be NIL. (reported by Kalle Niemitalo)
2968 * bug fix: printing 1.0d+23 no longer results in an error.
2969 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2970 * bug fix: structure slot setters preserve evaluation order. (thanks
2972 * fixed some bugs related to Unicode integration:
2973 ** RUN-PROGRAM can allow its child to take input from a Lisp
2974 stream. (reported by Stefan Scholl)
2975 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2976 locales. (reported by Ken Causey)
2977 * fixed some bugs revealed by Paul Dietz' test suite:
2978 ** INCF, DECF and REMF evaluate their place form as specified in
2980 ** FORMATTER expands ~{ ~} iteration directives with V or #
2981 parameters correctly.
2982 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2983 consequent uses no arguments correctly.
2984 ** the system has a partial understanding of the (COMPLEX RATIO)
2986 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2989 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2990 * new feature: a build-time option (controlled by the :SB-UNICODE
2991 keyword feature, enabled by default) for building the system with
2992 support for the entire 21-bit character space defined by the
2994 * new feature: the system now has rudimentary external-format
2995 support; the primary user-visible change at this time is that
2996 characters with the high bit set (such as lower-case-e-acute) will
2997 print correctly to a terminal in a UTF-8 environment.
2998 * minor incompatible change: BASE-CHAR no longer names a class;
2999 however, CHARACTER continues to do so, as required by ANSI.
3000 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
3001 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
3002 has been moved to the SB-EXT package (temporarily re-exported from
3004 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
3005 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
3007 * fixed bug #351: better error-handlind and reporting for malformed
3009 * fixed bug #350: bignum-printing is now more memory-efficient,
3010 allowing printing of very large bignums, eg. (expt 2 10000000).
3011 (reported by Bruno Haible)
3012 * fixed bug #302: better primitive-type selection for intersection
3014 * fixed bug #308: non-graphic characters now all have names, as
3015 required. (reported by Bruno Haible)
3016 * bug fix: redefining a class with different superclasses now correctly
3017 removes it from the direct-subclasses of its previous superclasses.
3018 (reported by David Morse)
3019 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
3020 new value now works. (reported by Bruno Haible)
3021 * bug fix: correct canonicalization of multiple non-standard slot
3022 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
3023 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
3024 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
3025 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
3026 now exists, an signals an error.
3027 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
3028 during macroexpansion for non-iteration variables as well. (reported
3029 by Bruno Haible for CMUCL)
3030 * bug fix: Cyclic structures and unprintable objects in compiler
3031 messages no longer cause errors. (reported by Bruno Haible)
3032 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
3033 and READ-FROM-STRING all now return a primary value of NIL if
3034 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
3035 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
3036 (reported by Bruno Haible for CMUCL)
3037 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
3038 do not designate a condition. (reported by Bruno Haible for
3040 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
3041 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
3042 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
3043 permitting resolution in favour of any of the conflicting symbols.
3044 (reported by Bruno Haible for CMUCL)
3045 * bug fix: EQUAL compiler optimizations is less aggressive on
3046 strings which can potentially compare true despite having distinct
3047 specialized array element types.
3048 * bug fix: unit enumerations can be defined without dividing by
3049 zero. (thanks to Vincent Arkesteijn)
3050 * FORMAT compile-time argument count checking has been enhanced.
3051 (report from Bruno Haible for CMUCL)
3052 * a partial workaround for the bug 262: the compiler does not try to
3053 inline-expand a local function doing RETURN-FROM from a deleted
3054 BLOCK. (thanks to Peter Denno for the bug report and to David
3055 Wragg for the simple test case)
3056 * fixed some bugs revealed by Paul Dietz' test suite:
3057 ** RENAME-PACKAGE allows all package designators as new package
3059 ** constraint propagation and lambda variable substitution are
3060 more cautious in dealing with partially deleted code.
3061 ** compiler sometimes lost reoptimization passes.
3062 ** CERROR, given a condition as condition designator, passes its
3063 remaining arguments to the continue format control without
3065 ** Case-altering FORMAT directives work correctly on non-ASCII
3067 ** The REAL type specifier handles bounds outside the floating
3068 point ranges without signalling FLOATING-POINT-OVERFLOW.
3069 ** Functions with IR1-transformations can create intercomponent
3070 references to global functions.
3071 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
3073 ** FORMAT ~R treats a nil value for its first parameter correctly.
3075 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
3076 * enhancement: saving cores with foreign code loaded is now
3077 supported on x86/NetBSD and sparc/Linux in addition to the previously
3078 supported platforms.
3079 * bug fix: on some platforms repeated installations caused multiple
3080 copies of HTML documentation to be installed -- should not happen
3081 any more. (reported by Stefan Scholl)
3082 * bug fix: parsing self-recursive alien record types multiple times
3083 no longer causes infinite recursion. (reported by Thomas F. Burdick,
3084 original patch by Helmut Eller for CMUCL)
3085 * bug fix: stack-exhaustion detection works now on NetBSD as well.
3086 (thanks to Richard Kreuter)
3087 * bug fix: defining classes whose accessors are methods on existing
3088 generic functions in other (locked) packages no longer signals
3089 bogus package lock violations. (reported by François-René Rideau)
3090 * bug fix: special variables as DEFMETHOD parameters no longer have
3091 associated bogus type declarations. (reported by David Wragg and
3093 * bug fix: read-write consistency on streams of element-type
3094 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
3095 * bug fix: redefiniton of the only method of a generic function with
3096 no DEFGENERIC no longer emits a full WARNING. In addition,
3097 redefinition of generic functions with no DEFGENERIC to an
3098 incompatible lambda list now signals an error. (thanks to Zach
3100 * bug fix: DEFGENERIC now works even when there's a function of the
3101 same name in an enclosing lexical environment. (thanks to Zach
3103 * fixed compiler failure, caused by instrumenting code during
3104 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
3105 * optimization: added loop analysis and improved register allocation
3106 * fixed some bugs revealed by Paul Dietz' test suite:
3107 ** POSITION on displaced vectors with non-zero displacement
3108 returns the right answer.
3109 ** (SIMPLE-STRING) is a valid type specifier for sequence
3111 ** *PRINT-LEVEL* handling for slotless structures is pedantically
3113 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
3114 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
3116 ** The justification version of the FORMAT ~< directive treats
3117 non-zero minpad parameter correctly.
3119 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
3120 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
3121 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
3122 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
3123 the supported interface.
3124 * new feature: Single-stepping of code compiled with DEBUG 2 or
3125 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
3126 * new feature: saving cores with foreign code loaded is now
3127 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
3128 Timothy Moore's work for CMUCL)
3129 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
3130 parameters to * instead of NIL if no initform is supplied.
3131 (reported by Johan Bockgaard)
3132 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
3133 correctly when FUNCALL appears as the car of the form. Note:
3134 despite this FUNCALL forms are not currently subject to
3135 compiler-macro expansion. (port of Raymond Toy's fix for the
3136 same from CMUCL, reported by Johan Bockgaard)
3137 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
3138 (thanks for Teemu Kalvas)
3139 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
3140 case instead of terminating the iteration (thanks for Julian
3141 Squires, Sean Champ and Raymond Toy)
3142 * bug fix: incorrect expansion of defgeneric that caused a style
3143 warning. (thanks for Zach Beane)
3144 * bug fix: slot accessor effective method computation works properly
3145 for classes with multiple non-standard applicable methods on
3146 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
3147 * on x86 compiler supports stack allocation of results of LIST and
3148 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
3149 CMUCL implementation by Gerd Moellmann)
3150 * fixed some bugs revealed by Paul Dietz' test suite:
3151 ** FORMAT strings with both the ~<~:;~> form of the justification
3152 directive and pretty-printing directives cause an error.
3154 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
3155 * incompatible change: the internal functions
3156 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
3157 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
3158 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
3159 instead of the old functions.
3160 * new feature: on platforms where "dladdr" is available foreign
3161 function names now appear in backtraces. (based on Helmut Eller's
3163 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
3164 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
3166 * documentation: ASDF manual is now distributed with SBCL.
3167 * bug fix: SBCL can now load its contributed modules with REQUIRE
3168 even if the system-provided entries have been removed from
3169 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
3171 * bug fix: backtraces involving undefined functions or assembly
3172 routines are more informative. (thanks to Brian Downing)
3173 * bug fix: mutually referent alien structures now work correctly.
3174 (reported by Rick Taube)
3175 * bug fix: structures defined by WITH-ALIEN can be referred to
3176 within other definitions in the same WITH-ALIEN.
3177 * bug fix: division operators (MOD, TRUNCATE and the like) with
3178 constant zero divisors and integer dividends no longer generate
3180 * bug fix: provide default methods for INPUT-STREAM-P and
3181 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
3182 * bug fix: improve the detection and resolution of MOP metacycles
3183 (where effective method computation depends on the generic function
3184 having its effective method computed). (reported by Bruno Haible)
3185 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
3186 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
3187 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
3188 * bug fix: pathnames with relative directory components can be
3189 represented relative to default pathnames.
3190 * optimization: in taking the GCD of bignums, reduce the two bignums
3191 to approximately the same size (using Euclid's algorithm) before
3192 applying the more sophisticated binary GCD. (thanks to Juho
3194 * optimization: COUNT on bitvectors now operates word-at-a-time.
3195 * optimization: ASH with a positive, but not necessarily constant,
3196 (leftwards) shift, when in a modular context, is compiled to a
3198 * fixed some bugs revealed by Paul Dietz' test suite:
3199 ** FORMAT variable parameters ("~V<char>") are defaulted properly
3200 if the corresponding argument is NIL.
3201 ** FORMAT directives accepting a minpad parameter treat negative
3203 ** literal commas as character parameters to FORMAT directives are
3205 ** literal spaces directly after ~<Newline> directives within a
3206 format-logical-block (~:< ~@:>) do not induce :FILL-style
3207 conditional newlines.
3208 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
3209 as specified: it no longer includes conditional newlines).
3210 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
3212 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
3213 :PER-LINE-PREFIX argument does not evaluate to a string.
3215 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
3216 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3217 the manual for details; for now, package locks can be disabled by
3218 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
3219 if no major problems are found then it is likely that they will be
3220 compiled in unconditionally.
3221 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
3222 unsupported operators on all platforms. To load a shared library
3223 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
3224 object file, link it into a shared library outside of SBCL and
3225 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
3227 * bug workaround: ROOM T and the GC were not working together
3228 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
3229 reporting facility aren't preserved. That reporting has been
3230 disabled, so now until and unless someone figures out how to make it
3231 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
3232 * minor incompatible change: as threatened around sbcl-0.8.0, the
3233 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
3234 The symbols are also exported from SB-PCL for backwards
3235 compatibility, but more so than before SB-PCL should be treated as
3236 an implementation-internal package.
3237 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
3239 * the SB-SPROF contrib now works on (most) non-x86 architectures.
3240 It is known as of this release not to work on the Alpha, however.
3241 * fixed bug #167: errors signalled due to illegal syntax in method
3242 bodies are now more legible.
3243 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
3244 designators and can hence be used with TYPEP.
3245 * fixed bug #333: CHECK-TYPE now ensures that the type error
3246 signalled, if any, has the right object to be accessed by
3247 TYPE-ERROR-DATUM. (reported by Tony Martinez)
3248 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
3249 5.1.2.3 for multiple-value place subforms. (reported by Kalle
3251 * fixed bug #334: programmatic addition of slots using specialized
3252 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
3253 and :ALLOCATION :CLASS slots.
3254 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
3255 float exponents. (rereported by Peter Seibel)
3256 * fixed a bug: #\Space (and other whitespace characters) are no
3257 longer considered to be macro characters in standard syntax by
3258 GET-MACRO-CHARACTER.
3259 * fixed bug: initialization of condition class metaobjects no longer
3260 causes an instance of the condition to be created. (reported by
3262 * fixed bug: it is now possible to have slots such that
3263 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
3264 description is neither :INSTANCE nor :CLASS.
3265 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
3266 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
3267 SLOT-VALUE-USING-CLASS), and no longer causes errors with
3268 non-standard SLOT-DEFINITION-ALLOCATION values.
3269 * fixed bugs: various race conditions handling exiting threads.
3270 CL-PPCRE's thread torture test now passes.
3271 * fixed arguably-a-bug: GC time proportional to number of threads in
3272 system even when most of them are idle
3273 * optimization: improved performance of BIT and SBIT on bit-vectors.
3274 * .fasl file incompatibility: The fasl file version number has
3275 been incremented because of changes associated with package locks.
3277 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
3278 * minor incompatible change: the system no longer provides
3279 optimization hints (conditions which are TYPEP
3280 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
3281 mode; these hints are still emitted when higher SPEED optimization
3283 * new contrib module: a sampling profiler (profiling by statistical
3284 sampling, rather than by instrumenting functions) is available as
3285 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
3286 * the behaviour of the standard function ED is now customizeable by
3287 third parties through a hook variable: see ED's documentation
3288 string for information on the protocol.
3289 * the compiler no longer emits efficiency notes for (FUNCALL X)
3290 when the type of X is uncertain under default optimization
3292 * fixed bug 276: mutating a binding of a specialized parameter to a
3293 method to something that is not TYPEP the specializer is now
3295 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
3296 exhibit the required behaviour on the x86 platform. (thanks to
3297 Peter van Eynde, Eric Marsden and Bruno Haible)
3298 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
3299 even for difficult arguments. (reported by Peter Graves)
3300 * fixed bug 141a: the backquote printer now descends quoted
3302 * fixed another bug in backquote printing: no more destructive
3303 modification of the form's list structure. (reported by Brian
3305 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
3306 move between its address being taken and the call to
3307 interrupt_thread, fixing a crashing race condition.
3308 * the SB-POSIX contrib implementation has been adjusted so that it
3309 no longer exhibits ridiculously poor performance when constructing
3310 instances corresponding to C structs.
3312 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
3313 * minor incompatible change: the sb-grovel contrib now treats C
3314 structures as alien (in the sense of SB-ALIEN) objects rather than
3315 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
3316 has implications for memory management of client code
3317 (sb-grovel-returned objects must now be manually managed) and for
3318 type safety (alien objects now have full types).
3319 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
3320 used to control emission of compiler diagnostics, rather than the
3321 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
3322 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
3323 quality should be considered deprecated.
3324 * install.sh now installs the user manual as well
3325 * (not quite a new documentable feature, but worth considering in
3326 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3327 beginnings of a semantically meaningful condition hierarchy is
3328 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3330 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3331 designator as the defaults argument.
3332 * fixed bug: Displaced arrays whose displaced-to array has become
3333 too small now cause ARRAY-DIMENSION to signal an error, providing
3334 for safer bounds-checking. (reported by Bruno Haible)
3335 * fixed bug: DEFCLASS slot definitions with identical :READER and
3336 :WRITER names now signal a reasonable error. (reported by Thomas
3338 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3340 * fixed bug: Invalid slot specification errors now print correctly.
3341 (thanks to Zach Beane)
3342 * fixed bug 320: Shared to local slot value transfers in class
3343 redefinitions now happen corrently from superclasses as
3344 well. (reported by Bruno Haible)
3345 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3347 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3348 improper lists correctly. (reported by Bruno Haible)
3349 * fixed bug 313: source-transform for <fun-name> was erroneously
3350 applied to a call of a value of a variable with name <fun-name>.
3351 (reported by Antonio Menezes Leitao)
3352 * fixed bug 307: The obsolete instance protocol ensures that
3353 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3354 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3355 remove dead unknown-values globs from the middle of the stack.
3356 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3357 easier to resume long computations after using *BREAK-ON-SIGNALS*
3358 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3359 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3360 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3361 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3362 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3363 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3365 * fixed a (fairly theoretical) bug in string printing: if
3366 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3367 does not have array-element-type equal to the most general string
3369 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3370 function name. (thanks to Nikodemus Siivola)
3371 * fixed bug: compile-time format string checker failed on
3372 non-closed ~{. (reported by Thomas F Burdick)
3373 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3374 don't warn on legal loop code involving more than one aggregate
3375 boolean. (thanks to Nikodemus Siivola)
3376 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3377 in keyword position to MAKE-PACKAGE was not being reported as
3379 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3380 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3382 * fixed bugs 280 and 312: the checking for multiple definitions in a
3383 file is less likely to become confused by uses of inline
3385 * fixed bug: the #S reader macro performs the keyword coercion
3386 specified for slot names. (reported by Kalle Niemitalo)
3387 * fixed bug: lambda lists may contain symbols whose names start with
3388 & but are not lambda-list-keywords; their occurrence triggers a
3390 * fixed bug 321: define-method-combination argument lambda lists do
3391 not cause invalid code to be generated when &optional or &aux
3392 variables have default values. (reported by Bruno Haible)
3393 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3394 have CLOS classes; this appears also to have fixed reported
3395 instability in stack exhaustion detection.
3396 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3397 MUFFLE-WARNING when no associated restart is present is now
3399 * optimization: rearranged the expansion of various defining macros
3400 so that each expands into only one top-level form in a
3401 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3403 * optimization: used a previously little-used slot in symbols to
3404 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3406 * fixed some bugs revealed by Paul Dietz' test suite:
3407 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3409 ** Fixed an optimization bug related to inheritance of initforms
3410 from local to shared slots.
3411 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3412 ** CAST optimizer forgot to flush argument derived type.
3413 ** print/read consistency on floats is now orders of magnitude
3414 more likely. (thanks also to Bruno Haible for a similar report
3416 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3417 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3418 for objects of type REAL. Make it so.
3419 ** SXHASH is specified (infelicitously) to respect similarity,
3420 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3421 it so. (thanks to Markus Ziegler)
3422 ** on the Alpha, the compiler succeeds in compiling functions
3423 returning a known number of arguments greater than 63.
3424 ** fixed handling of invalid NIL arguments in keyword position
3426 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3428 ** provided a readably-printable representation for RANDOM-STATE
3430 ** ensured that pathnames, where they have a namestring, always
3431 print using #P"..." syntax.
3433 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3434 * Support for the forthcoming 2.0 version of the NetBSD kernel
3435 running on x86 hardware has been added. (thanks to Perry
3436 E. Metzger most immediately, and others for their past work)
3437 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3438 current release version; SBCL's OpenBSD support had been broken
3439 since about the time of OpenBSD's switch to ELF binary format.)
3440 * [placeholder for DX summary]
3441 ** user code with &REST lists declared dynamic-extent, under high
3442 speed or space and low safety and debug optimization policy.
3443 * The manual has been converted to Texinfo format and the debugger
3444 chapter from the cmucl manual has been added.
3445 * A facility has been added to extract documentation strings from
3446 sbcl and store them as Texinfo-formatted snippets for inclusion in
3447 the manual (via Texinfo's @include directive)
3448 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3449 the test case to Dave Roberts)
3450 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3451 pointers. (reported by Sean Ross)
3452 * bug fix: PROFILE output is printed nicely even for large numerical
3453 values. (thanks to Zach Beane)
3454 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3455 greater than 32 handle EOF correctly.
3456 * bug fix: on X86 an immediate argument of the IMUL instruction is
3457 correctly printed in disassembly. (thanks to Lutz Euler)
3458 * bug fix: class slots in redefined classes preserve their old
3459 values. (thanks to Bruno Haible and Nikodemus Siivola)
3460 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3461 succeeds. (reported by Marco Baringer)
3462 * fixed some bugs revealed by Paul Dietz' test suite:
3463 ** READ-SEQUENCE now works on ECHO-STREAMs.
3464 ** RATIONALIZE works more according to its specification. (thanks
3467 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3468 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3469 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3470 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3471 should matter to you only if you rebind the printer control
3472 variables and then find you want different bindings in the
3473 debugger than in the ordinary execution of your program.)
3474 * The runtime build system has been tweaked to support building
3475 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3476 assembler and linker. (thanks to Nikodemus Siivola)
3477 * Unbound, undefined, undeclared variables now trigger full
3478 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3479 more often programmer error than deliberate exploitation of undefined
3481 * optimization: the hash algorithm for strings has changed to one
3482 that is less vulnerable to spurious collisions. (thanks to Juho
3484 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3485 needless bounds checking. (thanks to Juho Snellman)
3486 * optimization: implemented multiplication as a modular
3487 (UNSIGNED-BYTE 32) operation on the PPC backend.
3488 * fixed some bugs revealed by Paul Dietz' test suite:
3489 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3491 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3492 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3493 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3495 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3497 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3499 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3500 their output stream on EOF from read.
3501 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3502 have been read to end-of-file.
3503 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3505 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3506 description of determination of which consecutive characters
3508 ** Printing the "Space" character with escaping on now yields "#\\ ",
3509 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3510 ** Reading floating-point numbers with *READ-BASE* set to a number
3511 less than 10 works correctly.
3512 ** Reading floating-point numbers with *READ-BASE* set to a number
3513 more than 10 works correctly.
3514 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3515 the readtable currently in effect.
3517 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3518 * minor incompatible change: parsing of namestrings on a physical
3519 (Unix) host has changed; numbers after the final #\. in a
3520 namestring are no longer interpreted as a version field. This is
3521 intented to be largely invisible to the user, except that the
3522 meaning of the namestring "*.*.*" has changed: it now refers to a
3523 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3524 should usually be replaced by
3525 (make-pathname :name :wild :type :wild :version :wild)
3526 with the added benefit that this is more likely to be portable.
3527 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3528 option to OPEN now signals an error if the file being opened
3529 exists; this may have an impact on existing code.
3530 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3531 signals received, on the PowerPC platforms (both Linux and
3532 Darwin). (thanks to Pierre Mai for pointing out the location of
3534 * several fixes on OS X: The system now builds and runs cleanly on
3535 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3536 (thanks to Brian Mastenbrook)
3537 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3538 second-resolution: integer multiples of 1/3600 between -24 and 24.
3539 (thanks to Vincent Arkesteijn)
3540 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3541 type when called with 1 argument; PEEK-CHAR checked type of
3542 PEEK-TYPE only after having read first character from a
3543 stream. (reported by Peter Graves)
3544 * bug fix: the garbage collector now has much better locality
3545 behaviour, and in particular no longer treats all memory as being
3546 exhausted when many small objects point to each other in a deeply
3548 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3549 recognized as being TYPEP their class.
3550 * bug fix: the PUSHNEW documentation string has been corrected.
3551 (thanks to Vincent Arkesteijn)
3552 * bug fix: defaulting of the value for the last of an atypically
3553 large number of multiple values being bound was not being
3554 performed correctly on the Alpha or PPC platforms
3555 * optimization: implemented multiplication as a modular
3556 (UNSIGNED-BYTE 32) operation on the x86 backend.
3557 * optimization: SEARCH on simple-base-strings can now be open-coded.
3558 (see also contrib/compiler-extras.lisp for inspiration for
3559 teaching the compiler about the Boyer-Moore algorithm).
3560 * value, returned by MAX (and MIN) called with several EQUALP, but
3561 not EQL, arguments now does not depend on compiler settings.
3562 * fixed some bugs revealed by Paul Dietz' test suite:
3563 ** in stack analysis liveness information is propagated from
3564 non-local entry points.
3565 ** pathwise CAST removing failed when the CAST node did not start
3567 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3568 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3570 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3571 host is already defined.
3572 ** RENAME-FILE works on streams instead of signalling an internal
3574 ** PEEK-CHAR uses the current readtable when determining whether
3575 or not a character is whitespace.
3576 ** MERGE-PATHNAMES handles the case when the pathname does not
3577 specify a name while the default-pathname specifies a version.
3578 ** Pathnames now stand a better chance of respecting print/read
3580 ** Attempting to use standardized file system operators with a
3581 pathname with invalid :DIRECTORY components signals a
3583 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3585 ** DIRECTORY on logical pathnames is more correct.
3586 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3587 signal a TYPE-ERROR if their argument is not a stream.
3588 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3589 designator argument does not designate a stream.
3590 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3591 examining the synonym.
3592 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3594 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3595 element-type larger than ([UN]SIGNED-BYTE 32).
3597 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3598 * When built with the :SB-FUTEX feature, threaded builds now take
3599 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3600 for faster/more reliable mutex and condition variable support.
3601 * Incompatible change (but one you probably shouldn't have been using
3602 anyway): the interface and code for arbitrating between multiple
3603 threads in the same user session has been redesigned.
3604 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3605 error when called without an explicit environment argument.
3606 (thanks to Nikodemus Siivola)
3607 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3608 confused about their position. (thanks to Adam Warner and Gerd
3610 * bug fix: Pretty printing backquoted forms with unquotations in the
3611 argument list position of various code constructs such as LAMBDA
3612 now works correctly. (reported by Paul Dietz)
3613 * bug fix: Pretty printing unquotations no longer loses all
3614 stream position information.
3615 * optimization: performance of string output streams is now less
3616 poor for multiple small sequence writes.
3617 * optimization: performance of CSUBTYPEP in the presence of complex
3618 expressions involving CONS and NOT many times has been improved.
3619 (reported by Paul Dietz)
3620 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3622 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3623 streams for the REPL, for the debugger, and so on. (thanks to
3625 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3626 * fixed some bugs revealed by Paul Dietz' test suite:
3627 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3628 is ignored if it is too small, rather than propagating through
3629 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3630 ** extremely complex negations of CONS types were not being
3631 sufficiently canonicalized, leading to inconsistencies in
3633 ** VALUES tranformer lost derived type.
3635 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3636 * fixed a bootstrapping bug: the build process no longer assumes
3637 that the various BOOLE-related constants have the same value in
3638 host and target lisps. (noted by Paul Dietz' test suite on an
3639 SBCL binary built from CLISP)
3640 * The system can now be dynamically linked on the MIPS platform,
3641 which enables dynamic loading of foreign code from Lisp. (thanks
3642 to Ralf Baechle for discussions on the MIPS ABI)
3643 * The system now records debugging information for its own source
3644 files in a filesystem-position-independent manner, relative to
3645 the "SYS" logical host.
3646 * fixed a compiler bug: MV-LET convertion did not check references
3647 to the "max args" entry point. (reported by Brian Downing)
3648 * tweaked disassembly notes to be less confident about proclaiming
3649 some instruction as an LRA. (thanks to Brian Downing)
3650 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3651 now each have their own history, command character, and other
3652 characteristics. (thanks to David Lichteblau)
3653 * fixed some bugs revealed by Paul Dietz' test suite:
3654 ** compiler failure in compiling LOGAND expressions including a
3656 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3657 shift greater than 32.
3658 ** FLUSH-DEST did not mark blocks for type check regeneration.
3659 ** HANDLER-CASE failed to accept declarations in handler clauses
3660 in some circumstances.
3662 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3663 * New code in contrib/sb-introspect (still probably not entirely
3664 stable yet) provides some support for smart Lisp development
3665 environments like SLIME.
3666 * The conditions signalled for errors occurring when loading .fasl
3667 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3668 in a way which should help ASDF recover gracefully.
3669 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3670 stuff has been cleaned up. If you code contrib/ stuff, this might
3671 affect you, and you can look at contrib/README, contrib/STANDARDS,
3672 and/or the 0.8.4.27 diff to check.
3673 * In full calls the compiler now does not generate checks for declared
3674 argument types for all arguments.
3675 * various threading fixes
3676 ** and some experimental patches which didn't make it into
3677 the main tree for this release, but which are shipped in
3678 contrib/experimental-thread.patch as a possible fix for some
3679 failures (deadlock, spinning...) in GC-intensive multithreaded
3681 * fixed PPC build problem (source code incompatibility of different
3682 library versions): added offsetof() hackery which attempts to divine
3683 where glibc maintainers put uc_mcontext today
3684 * fixed bug 282: compiler does not trust type assertions while passing
3685 arguments to a full call.
3686 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3687 &OPTIONAL VALUES type parameter.
3688 * fix bug 214: algorithm for noting rejected templates is now more
3689 similar to that of template seletion. (also reported by rydis on
3691 * fixed bug 141b: printing backquoted information readably and prettily
3692 inserts a space where necessary.
3693 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3694 (CHARACTER 10) are now reported as errors, rather than propagated
3695 as unknown types. (reported by piso on #lisp)
3696 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3697 with values NIL and :ERROR. (thanks to Milan Zamazal)
3698 * fixed bug 191c: CLOS now does proper keyword argument checking as
3699 described in CLHS 7.6.5 and 7.6.5.1.
3700 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3701 longer raise an error; further, using a list as a for-as-arithmetic
3702 counter now raises a meaningful error.
3703 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3704 now understood by sequence creation functions such as MAKE-SEQUENCE
3706 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3707 a STRING-INPUT-STREAM.
3708 * compiler enhancement: SIGNUM is now better able to derive the type
3710 * type declarations inside WITH-SLOTS are checked. (reported by
3712 * fixed some bugs revealed by Paul Dietz' test suite:
3713 ** incorrect optimization of TRUNCATE for a positive first
3714 argument and negative second.
3715 ** compiler failure in let-convertion during flushing dead code.
3716 ** compiler failure while deriving type of TRUNCATE on an
3717 interval, containing 0.
3718 ** ASH of a negative bignum by a negative bignum count now returns
3720 ** intersection of CONS types now canonicalizes properly, fixing
3721 inconsistencies in SUBTYPEP.
3723 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3724 * incompatible change: The --disable-debugger command line
3725 option now clobbers the debugger at a more fundamental
3726 level, by redefining #'INVOKE-DEBUGGER instead of by
3727 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3728 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3729 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3730 enter the debugger REPL and then suffer recursive errors
3731 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3732 while under the new system, BREAK in --disable-debugger mode
3733 terminates the system just as an unhandled error would.
3734 * fixed compiler performance when processing loops with a step >1;
3735 * bug fix: DOCUMENTATION now retrieves generic function
3736 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3737 support has been systematized, and now supports the methods
3738 specified by ANSI, along with a default method and a method for
3739 slot documentation. (reported by Nathan Froyd)
3740 * bug fix: effective methods associated with a generic function are
3741 no longer cached over a change of that generic function's method
3742 combination. (reported by Andreas Fuchs)
3743 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3744 names a non-existent directory. (thanks to Andreas Fuchs)
3745 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3746 the PPC platform, lead to stack corruption; this has been fixed.
3747 (reported by Rainer Joswig)
3748 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3749 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3750 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3751 arithmetic in general, is now much more reliable on the PPC
3753 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3754 platform now returns the right answer.
3755 * optimization: restored some effective method precomputation in
3756 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3757 precomputation is now tunable.
3758 * optimization: compiler-internal data structure use has been
3759 reviewed, and changes have been made that should improve the
3760 performance of the compiler by about 20%.
3761 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3762 simple-base-strings and simple-bit-vectors is improved.
3763 * optimization: the optimization of 32-bit logical and arithmetic
3764 functions introduced in version 0.8.3 on the x86 has been
3765 implemented on the mips, ppc and sparc platforms; an
3766 implementation of the same facility, but for 64-bit arithmetic,
3767 has been added for the alpha.
3768 * microoptimization: the compiler is better able to make use of the
3769 x86 LEA instruction for multiplication by constants.
3770 * bug fix: in some situations compiler did not report usage of
3771 generic arithmetic in (SPEED 3) policy.
3772 * bug 145b fix: compiler used wrong type specifier while converting
3773 MEMBER-types to numeric.
3774 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3776 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3777 index argument now works properly on the Alpha platform.
3778 * bug fix: floating point exception treatment on the Alpha platform
3780 * bug fix: FILE-POSITION works much better on string input and
3781 output streams. (thanks to Nikodemus Siivola)
3782 * bug fix: many threading/garbage collection symptoms sorted.
3783 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3784 might be pseudo-atomic.
3785 * internal change: Stopping for GC is now done with signals not
3786 ptrace. GC is now done in whichever thread wanted it, instead of
3788 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3790 * bug fix: result form in DO is not contained in the implicit
3792 * incompatible change: ICR structure is changed; the value part of
3793 CONTINUATION is now called LVAR; corresponding functions are
3794 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3795 * added type deriver for ISQRT (thanks to Robert E. Brown).
3796 * bug fix: better support for loading from the command line when an
3797 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3799 * fixed some bugs revealed by Paul Dietz' test suite:
3800 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3801 ** ROUND and FROUND now give the right answer when given very
3802 small float arguments.
3803 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3805 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3806 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3807 ** LCM should return a non-negative integer.
3808 ** PARSE-INTEGER returned the index of a terminator instead of the
3809 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3810 ** PARSE-INTEGER returned an incorrect index being applied to a
3812 ** LCM with two arguments of 0 returns 0 rather than signalling
3814 ** unsigned addition of a 32-bit constant with the high bit set no
3815 longer causes an internal compiler error on the x86.
3816 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3817 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3818 with negative last argument.
3819 ** byte specifiers with zero size and position no longer cause
3820 an error during type derivation.
3821 ** bignum multiplication on the Alpha platform now returns the
3823 * porting: The system now builds on SuSE AMD64, although it still
3824 generates a 32-bit binary.
3825 * .fasl file incompatibility: The fasl file version number has
3826 been incremented (because of the changes to internal compiler
3827 data structures referred to above).
3829 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3830 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3831 more accurately, on the Darwin kernel running on PowerPC hardware.
3832 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3833 * Compiler code deletion notes now signal a condition of type
3834 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3835 an associated MUFFLE-WARNING restart.
3836 * The compiler now performs limited argument count validation of
3837 constant format strings in FORMAT, and where appropriate in ERROR,
3838 CERROR and WARN. (thanks to Gerd Moellmann)
3839 * New ASDF-INSTALL contrib can be used for automatic download and
3840 installation of third-party Lisp code from CCLAN or other sites
3842 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3843 which forces another thread to execute a function supplied by the
3845 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3846 now accept and act upon their :ELEMENT-TYPE keyword argument.
3847 (reported by Martin Atzmueller, Edi Weitz)
3848 * bug fix: FILE-POSITION now accepts position designators up to
3849 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3850 is the greater. (thanks to Patrik Nordebo)
3851 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3852 Kalle Olavi Niemitalo)
3853 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3854 printable. (reported by Eric Marsden)
3855 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3856 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3857 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3858 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3859 (thanks to Stig Sandoe)
3860 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3861 UNREAD-CHAR now correctly includes the unread character in the
3862 target sequence. (thanks to Gerd Moellmann)
3863 * bug fix in threaded builds: the system can now be suspended and
3864 resumed by shell job control with minimal disruption.
3865 * bug fixes in times and timezones >2038AD
3866 * better handling of "where is GNU make?" problem in build scripts
3867 (thanks to Nikodemus Siivola)
3868 * new optimization: inside a named function any reference to a
3869 function with the same name is considered to be a self-reference;
3870 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3871 optimization quality.
3872 * new optimization on x86: logical functions and + now have
3873 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3874 used when the result is truncated to 32 bits.
3875 * VALUES declaration is partially enabled.
3876 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3877 * fixed some bugs revealed by Paul Dietz' test suite:
3878 ** The system now obeys the constraint imposed by
3879 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3880 types form a lattice under type intersection.
3881 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3882 ** ASSOC now ignores NIL elements in an alist.
3883 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3884 and (1+ MOST-POSITIVE-FIXNUM) answers.
3885 ** The addition of a method with invalid qualifiers to a generic
3886 function does not cause an error to be signalled immediately;
3887 a warning is signalled, and the error is generated only on
3888 calling the generic function.
3889 * changed .fasl file version number, in part to add type codes for
3890 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3891 obscure ANSI requirements
3893 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3894 * fixed bug 148: failure to inline-expand a local function left
3895 garbage, confusing the compiler.
3896 * fixed bugs 3cd: structure slot readers perform type check if the
3897 slot can have an invalid value (i.e. it is either not initialized
3898 or can be written with a less specific slot writer).
3899 * bug fix: the compiler now traps array references to elements off
3900 the end of an array; previously, the bounds checking in some
3901 circumstances could go off-by-one.
3902 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3904 * type declarations for array element types now obey the description
3905 on the CLHS page "Declaration TYPE", as per discussions on
3906 sbcl-help around 2003-05-08. This means that a declaration
3907 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3908 declaration, all references to BAR will be asserted or assumed
3909 (with THE, so dependent on compiler policy) to involve objects of
3910 type FOO. Note that no such declaration is implied in
3911 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3912 * declared types of functions from the "Conditions"
3913 chapter. (reported by Paul Dietz)
3914 * bug fix: CERROR accepts a function as its first argument.
3915 * bug fix: NTH an NTHCDR accept a bignum as index
3916 arguments. (reported by Adam Warner)
3917 * optimization: character compare routines now optimize comparing
3918 against a constant character. (reported by Gilbert Baumann)
3919 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3920 argument now works properly on the MIPS platform.
3921 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3922 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3923 declared to have a functional type, does not cause an error
3925 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3926 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3927 * fixed bug: if last continuation of a deleted block has a
3928 destination, this destination should be deleted too. (reported by
3930 * fixed a bug in the bootstrap process: the host compiler's values
3931 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3932 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3933 test case from Patrik Nordebo)
3934 * improved the ability of the disassembler on the PPC platform to
3935 provide helpful disassembly notes.
3936 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3937 the class in more cases than previously.
3938 * bug fix: FILE-POSITION now understands :START and :END for
3939 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3940 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3941 8) data. (thanks to David Lichteblau)
3942 * bug fix: it is possible to add a method to a generic function
3943 without lambda list.
3944 * bug fix: reader failed to signal END-OF-FILE inside an
3945 object representation. (reported by Nikodemus Siivola)
3946 * fixed some bugs revealed by Paul Dietz' test suite:
3947 ** LAST and [N]BUTLAST should accept a bignum.
3948 ** condition slot accessors are methods.
3949 ** (VECTOR NIL) is a subtype of STRING.
3951 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3952 * minor incompatible change: some nonsensical specialized lambda
3953 lists (used in DEFMETHOD) which were previously ignored now signal
3955 * minor incompatible change: the system is now aware of the types of
3956 variables in the COMMON-LISP package, and will signal errors for
3957 most violations of these type constraints (where previously they
3958 were silently accepted).
3959 * minor incompatible change: COMPILE-FILE now uses the freedom
3960 afforded (ANSI 3.2.2.3) to use derived function types for
3961 functions defined in the same file. This also permits the system
3962 to warn on static type mismatches and function
3963 redefinition. (Currently it does not work with high DEBUG level.)
3964 * minor incompatible change: VALUES declaration is disabled.
3965 * When issuing notes, the compiler now signals a condition of type
3966 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3967 restart for use in user handlers. It is expected that the
3968 COMPILER-NOTE condition will eventually become a condition
3969 supertype to a hierarchy of note types, which will then be
3970 handleable in a similar fashion. However, at the moment, no such
3971 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3972 but it's an implementation detail, not a classification for the
3974 * Changes in type checking closed the following bugs:
3975 ** type checking of unused values (192b, 194d, 203);
3976 ** template selection based on unsafe type assertions (192c, 236);
3977 ** type checking in branches (194bc).
3978 * A short form of VALUES type specifier has ANSI meaning (it has
3979 increased the number of situations when SBCL cannot perform type
3981 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3982 keywords or constants is permissible.
3983 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3984 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3985 classes. (thanks to Antonio Martinez)
3986 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3987 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3988 argument to a pretty-print stream. Instead, it leaves any such
3989 operations to DESCRIBE-OBJECT methods.
3990 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3992 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3993 lambda lists. (thanks to David Lichteblau)
3994 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3995 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3996 (thanks to Gerd Moellmann)
3997 * fixed bug 246: increased compilation speed of long
3998 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
4000 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
4002 * DEFCONSTANT now throws a condition of type
4003 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
4004 constant to a non-EQL value; CONTINUE and ABORT restarts
4005 respectively change and preserve the value.
4006 * fixed bug 63: The code walker, part of the implementation of CLOS,
4007 is now better at handling symbol macros.
4008 * bug fix: There is no longer an internal implementation type named
4009 CL:LENGTH. (reported by Raymond Toy)
4010 * bug fix: In macro-like defining macros/special operators the
4011 implicit block does not enclose the lambda list.
4012 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
4013 enclosed in parentheses) are not suitable as type specifiers, and
4014 their use properly signals an error now.
4015 * bug fix: An argument count mismatch for a type specifier in code
4016 being compiled no longer causes an unhandled error at compile
4017 time, but signals a compile-time warning.
4018 * fixed simple vector readable printing
4019 * bug fix: DESCRIBE takes more care over whether the class
4020 precedence list slot of a class is bound before accessing it.
4021 (reported by Markus Krummenacker)
4022 * bug fix: FORMATTER can successfully compile pretty-printer format
4023 strings which use variants of the ~* directive inside.
4024 * bug fix: SEARCH now applies its TEST predicate to the elements of
4025 the arguments in the correct order. (thanks to Wolfhard Buss)
4026 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
4027 in type inference. (thanks to Robert E. Brown)
4028 * bug fix: Reading in symbols with an explicit package name of ""
4029 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
4030 * fixed some bugs revealed by Paul Dietz' test suite:
4031 ** NIL is now allowed as a structure slot name.
4032 ** Arbitrary numbers, not just REALs, are allowed in certain
4033 circumstances in LOOP for-as-arithmetic clauses.
4034 ** Multiple class redefinitions before slot access no longer
4035 causes a type error.
4036 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
4037 association between the name and a class.
4038 ** Generic functions with non-standard method-combination and over
4039 five methods all of which return constants no longer return NIL
4040 after the first few invocations. (thanks to Gerd Moellmann)
4041 ** CALL-NEXT-METHOD with no arguments now passes the original
4042 values of the arguments, even in the presence of assignment.
4043 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
4045 ** DISASSEMBLE works with closures and funcallable instances.
4046 ** ADD-METHOD now returns the generic function, not the new method.
4047 ** FIND-METHOD signals an error if the lengths of the specializers
4048 is incompatible with the generic function, even if the ERRORP
4050 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
4051 which its argument is a member.
4052 ** DEFCLASS only redefines the class named by its class-name
4053 argument if that name is the proper name of the class;
4054 otherwise, it creates a new class.
4055 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
4056 of the UNBOUND-SLOT condition to the name of the slot.
4057 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
4058 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
4059 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
4060 SLOT-MAKUNBOUND in the specified fashion.
4062 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
4063 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
4064 cross-compilation host. As a consequence, we can now bootstrap our
4065 way up to SBCL starting with a bare gcc toolchain and human-readable
4066 source code (first the source to CLISP, then the source to SBCL).
4067 * A contributed module containing a partial implementation of the
4068 simple-streams interface has been included. (thanks to Rudi
4070 * A contributed module implementing the RFC1321 Message Digest
4071 Algorithm, known as MD5, has been included.
4072 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
4073 no longer has any effect, as the code controlled by this feature
4074 has been deleted. (As far as we know, no-one has ever built using
4075 this feature, and its semantics were confused in any case).
4076 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
4077 work on conditions (as required by the ANSI specification),
4078 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
4079 expected behaviour on conditions. Users should note, however,
4080 that such behaviour is not required by the ANSI specification,
4081 and so use of this behaviour may render their code unportable.
4082 * fixed some bugs revealed by Paul Dietz' test suite:
4083 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
4085 ** &ENVIRONMENT parameter in macro lambda list is bound first.
4086 ** SXHASH on condition objects no longer returns NIL.
4087 ** :ALLOCATION :CLASS slots are better treated; their values are
4088 updated on class redefinition, and initforms inherited from
4089 superclasses are applied.
4090 ** REMOVE-METHOD returns its generic function argument even when
4091 no method was removed.
4092 ** SHARED-INITIALIZE now initializes the values of the requested
4093 slots, including those with :ALLOCATION :CLASS.
4094 ** ALLOCATE-INSTANCE now works on structure classes defined via
4095 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
4097 ** SLOT-EXISTS-P now works on conditions, as well as structures
4099 ** MAKE-LOAD-FORM now has the required methods on
4100 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
4101 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
4102 keyword, and now implements the SLOT-NAMES argument.
4103 ** methods with &OPTIONAL arguments no longer allow too many
4104 arguments to be passed in the call without error.
4105 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
4106 option is consistent with the required arguments of the generic
4107 function lambda list.
4108 * bug fix: REQUIRE accepts a string designator. (Thanks to
4110 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
4111 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
4112 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
4113 * bug fix: The compiler now checks for duplicated variables in macro
4115 * bug fix: SETQ on globals returns the correct value.
4116 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
4117 not a COMPILER-ERROR (followed by some other strange error on
4118 choosing the CONTINUE restart).
4119 * bug fix: make.sh and friends are now more consistent in the way that
4120 they look for GNU "make".
4122 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
4123 * experimental native threads support (on x86 Linux >=2.4 only).
4124 This is not compiled in by default: you need to add :SB-THREAD to
4125 the target features. See the "Beyond ANSI" chapter of the manual
4127 * fix for longstanding nonANSIism: The old distinction between
4128 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
4129 The return value from CL:FIND-CLASS is now a CLOS class, and
4130 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
4131 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
4132 * An interface to the MetaObject Protocol, as described in Kiczales,
4133 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
4134 MIT Press, 1991, is available from the SB-MOP package.
4135 * incompatible change: the SB-PCL package should now be considered
4136 a private implementation detail, and no longer a semi-private MOP
4138 * minor incompatible change: due to rearrangement for threads, the
4139 control stack and binding stack are now allocated at arbitrary
4140 addresses instead of being hardcoded per-port. Users affected by
4141 this probably have to be doing advanced things with shared
4142 libraries, and will know who they are.
4143 * minor incompatible change: Previously, all --eval forms used were
4144 processed with READ before any of them were processed with EVAL.
4145 Now each --eval form is processed with both READ and EVAL before
4146 the next --eval form is processed. (Thus package operations like
4147 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
4148 work as the user might reasonably expect.)
4149 * minor incompatible change: *STANDARD-INPUT* is now only an
4150 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
4152 * minor incompatible change: Y-OR-N-P is now character-oriented,
4153 not line oriented. Also, YES-OR-NO-P now works without errors.
4154 (thanks to Antonio Martinez)
4155 * sb-aclrepl module improvements: an integrated inspector, added
4156 repl features, and a bug fix to :trace command.
4157 * Known functions, which cannot be open coded by the backend, are
4158 considered to be able to check types of their arguments. (fixing
4159 a bug report by Nathan J. Froyd)
4160 * fixed a bug in computing method discriminating functions: It is
4161 now possible to define methods specialized on classes which have
4162 forward-referenced superclasses. (thanks to Gerd Moellmann)
4163 * fixed evaluation order in optional entries (reported by Gilbert
4165 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
4166 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
4167 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
4168 required-by-AMOP NAME argument, as well as CLASS and
4169 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
4170 * fixed bug 20: DEFMETHOD can define methods using names that are
4171 not the proper names of classes to designate class specializers.
4172 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
4173 Unix file descriptors, instead of blowing up. (thanks to Antonio
4175 * Garbage collection refactoring: user-visible change is that a
4176 call to the GC function during WITHOUT-GCING will not do garbage
4177 collection until the end of the WITHOUT-GCING. If you were doing
4178 this you were probably losing anyway.
4179 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
4180 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
4181 TYPEP the latter but not the former.
4182 * The compiler issues a full WARNING for calls to undefined functions
4183 with names from the CL package.
4184 * MAP-INTO for a vector destination is open coded. (reported by
4185 Brian Downing on c.l.l)
4186 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
4187 documentation string.
4188 * fixed some bugs revealed by Paul Dietz' test suite:
4189 ** COPY-ALIST now signals an error if its argument is a dotted
4191 ** Condition slots are now accessed more correctly in the presence
4192 of multiple initargs for a given slot.
4193 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
4194 exclude restarts of the same name associated with a different
4196 ** DEFCLASS of forward-referenced classes with another
4197 forward-referenced class in the superclasses list no longer
4199 ** Condition slots are now initialized once each, not multiple
4200 times. (thanks to Gerd Moellmann)
4201 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
4203 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
4204 ** NSET-EXCLUSIVE-OR does not return extra elements when its
4205 arguments contain duplicated elements.
4206 ** RESTART-CASE understands local macros.
4207 ** RESTART-CASE associates exactly its own restarts with a condition.
4208 ** ENDP in safe mode checks its argument to be of type LIST.
4209 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
4210 in question is unbound.
4211 ** Optimized MAKE-INSTANCE functions no longer cause internal
4212 assertion failures in the presence of duplicate initargs.
4213 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
4214 ** Side-effectful :DEFAULT-INITARGS have their side-effects
4215 propagated even in the ctor optimized implementation of
4217 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
4219 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
4220 * a better implementation of SXHASH on (simple) bit vectors,
4221 measured both in execution speed and in distribution of results
4222 over the positive fixnums, has been installed. Likewise, a better
4223 implementation of EQUAL for simple bit vectors is now available.
4224 * fixed CEILING optimization for a divisor of form 2^k.
4225 * fixed bug 240 (emitting extra style warnings "using the lexical
4226 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
4227 by Antonio Martinez)
4228 * fixed SXHASH, giving different results for NIL depending on type
4229 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
4230 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
4231 variable. (found by Rolf Wester)
4232 * SBCL does not ignore type declarations for special
4233 variables. (reported by rif on c.l.l 2003-03-05)
4234 * some bug fixes in contrib/sb-aclrepl/
4235 * fixed some bugs revealed by Paul Dietz' test suite:
4236 ** a bug in the CONS type specifier, whereby the CAR and CDR
4237 types got intertwined, has been fixed;
4238 ** the type system is now able to reason about the interaction
4239 between INTEGER and RATIO types more completely;
4240 ** APPEND, [N]REVERSE and NRECONC check that those their
4241 arguments, which must be proper lists, are really so;
4242 ** An array specialized to be unable to hold elements has been
4243 implemented, as required -- yes, really -- by ANSI;
4244 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
4245 on malformed property lists;
4247 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
4248 * incompatible packaging change: in line with Unix convention,
4249 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
4250 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
4251 * REQUIRE and PROVIDE are now optionally capable of doing something
4252 useful. See the documentation string for REQUIRE.
4253 * infrastructure for a managed SBCL contrib system: contributed
4254 modules in this release include:
4255 ** the ASDF system definition facility;
4256 ** an interface to the BSD Sockets API;
4257 ** an ACL-like convenience interface to the repl;
4258 (thanks to Kevin Rosenberg)
4259 ** an implementation of ROTATE-BYTE, with efficient implementation
4261 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
4262 gives a closer approximation to the right answer than previously.
4263 (thanks to Raymond Toy)
4264 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
4265 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
4266 optional environment arguments, as required by ANSI.
4267 * fixed bugs in other functions taking environment objects, allowing
4268 calls with an explicit NIL environment argument to be compiled
4270 * fixed bug 228: primary return values from
4271 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
4272 COMPILE or FUNCTION.
4273 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
4274 :TYPE will no longer signal errors on innocuous objects.
4275 * fixed bug 231b: SETQ is better at respecting type declarations in
4276 the lexical environment.
4277 * fixed a bug in DEFCLASS: classes named by symbols with no or
4278 unprintable packages can now be defined.
4279 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
4280 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
4281 * fixed compiler failure related to checking types of functions.
4282 (reported by Robert E. Brown)
4283 * the compiler is now much more consistent in its error-checking
4284 treatment of bounding index arguments to sequence functions: in
4285 (SAFETY 3) code, errors will be signalled in almost all cases if
4286 invalid sequence bounding indices are passed to functions defined
4287 by ANSI to operate on sequences.
4288 * fixed a bug in the build procedure: documentation of SBCL-specific
4289 packages is now preserved and available in the final Lisp image.
4290 * lifted FDEFINITION lookup out of loops in the implementation of
4291 many list operations. (thanks to Robert E. Brown)
4292 * fixed a bug in the reader: the #n# reader macro now works for
4293 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
4294 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
4295 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
4296 E. Brown and Tony Martinez respectively)
4297 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
4298 SLOT-BOUNDP: in particular, it is now able to optimize them much
4299 better, and is now not vulnerable to having packages renamed.
4300 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
4301 always called when a slot is not present in an instance. (thanks
4303 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
4304 longer signal an error on unreading a character following EOF on
4305 the previous constituent stream. (thanks to Tony Martinez)
4306 * fixed some bugs revealed by Paul Dietz' test suite:
4307 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
4308 not just nonnegative fixnums;
4309 ** the logical bit-array operators such as BIT-AND now accept an
4310 explicit NIL for their "opt-arg" argument (to indicate a
4311 freshly-consed result bit-array);
4312 ** ELT now signals an error on an invalid sequence index in safe
4314 ** the type system is now cleverer about negations of numeric
4315 types, and consequently understands the BIGNUM and RATIO types
4317 ** the type system is now cleverer about the interaction between
4318 INTEGER and RATIO types: while bugs still remain, many more
4319 cases are accurately computed;
4320 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
4321 if it is in the last clause;
4322 ** CONSTANTLY now correctly returns a side-effect-free function in
4324 ** DECLARE is no longer treated as a special-operator; in
4325 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
4326 * incremented fasl file version number due to the change in the
4327 DEFSTRUCT-SLOT-DESCRIPTION structure.
4329 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4330 * minor incompatible change: code processed by the "interpreter" or
4331 EVAL now has a compilation optimization policy of (DEBUG 2)
4332 (changed from (DEBUG 1)) to improve debuggability of interactive
4333 development, and to allow the use of the debug RETURN command in
4335 * an experimental implementation of the RETURN command for the
4336 debugger has been included. (thanks to Frederik Kuivinen)
4337 * fixed bug 62: constraints were not propagated into a loop.
4338 * fixed bug in embedded calls of SORT (reported and investigated by
4340 * fixed some bugs revealed by Paul Dietz' test suite:
4341 ** printing and reading of arrays with some dimensions having
4342 length 0 (thanks to Gerd Moellmann);
4343 ** BOA constructor with &AUX argument without a default value does
4344 not cause a type error;
4345 ** CONSTANTP now returns true for all self-evaluating objects.
4347 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4348 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4349 accessors that are related by inheritance, as specified in the
4350 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4351 Valtteri Vuorikoski)
4352 * The compiler is now able to inline functions that were defined in
4353 a complex lexical environment (e.g. inside a MACROLET).
4354 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4356 * Support for the upcoming FreeBSD-5.0 release has been included.
4357 (thanks to Dag-Erling Smorgrav)
4358 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4359 effect when it is not in a toplevel context.
4360 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4361 stand a better chance of being correct. (thanks to Gerd
4363 * fixed bug in COERCE, which now signals an error on coercing a
4364 rational to a bounded real type which excludes the expected
4366 * The compiler is now able to derive types more accurately from the
4367 COERCE and COMPILE functions.
4368 * fixed bug 223: functional binding is considered to be constant
4369 only for symbols in the CL package.
4370 * fixed bug 231: SETQ did not check the type of a variable being set
4371 (reported by Robert E. Brown)
4372 * A new optimization for MAKE-INSTANCE has been included, fixing
4373 various bugs (including relating to :ALLOCATION :CLASS slots and
4374 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4376 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4377 ** As required by ANSI, LOOP now disallows anonymous collection
4378 clauses such as COLLECT I in conjunction with aggregate boolean
4379 clauses such as THEREIS (= I 1);
4380 ** LOOP now signals an error when any variable is reused in the
4381 same loop (including the potentially useful construct analogous
4382 to WITH A = 1 WITH A = (1+ A);
4383 ** IT is only a special loop symbol within the first clause of a
4384 conditional loop clause;
4385 ** LOOP with a typed iteration variable over a hashtable now
4386 signals a type error iff it should.
4387 * fixed some other bugs revealed by Paul Dietz' test suite:
4388 ** FILE-STREAM now names the class previously known as FD-STREAM;
4389 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4390 argument) no longer signals an error;
4391 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4392 of the slot symbol, rather than using the current package
4393 ((:CONC-NAME "") continues to intern the slot's name in the
4395 * incremented fasl file version number, because of the incompatible
4396 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4397 of the new implementation of DEFINE-COMPILER-MACRO.
4399 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4400 * Support for building SBCL for MIPS platforms running in
4401 little-endian mode has now been checked in, and basic
4402 functionality on said platforms verified.
4403 * minor incompatible change: PCL now records the pathname of a file
4404 in which methods and the like are defined, rather than its
4406 * minor incompatible change: TRUENAME now considers the truename of
4407 a file naming a directory to be the pathname with :DIRECTORY
4408 component indicating that directory.
4409 * minor incompatible change: a NAMED clause in the extended form of
4410 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4411 reason for the previous behaviour is unclear.
4412 * more systematization and improvement of CLOS and MOP conformance
4413 in PCL (thanks to Gerd Moellman and Pierre Mai):
4414 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4416 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4417 primary methods with no specializers;
4418 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4420 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4421 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4422 FORWARD-REFERENCED-CLASSes; error reporting on
4423 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4425 ** SXHASH on CLOS instances now uses a slot internal to the
4426 instance to return different numbers on distinct instances,
4427 while preserving the same return value through invocations of
4429 ** DEFMETHOD signals errors when methods with longer incongruent
4430 lambda lists are added to generic functions;
4431 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4432 CLASS, as specified in AMOP;
4433 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4434 on the order returned by the primary method for classes of
4435 class STANDARD-CLASS;
4436 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4437 * fixed some bugs shown by Paul Dietz' test suite:
4438 ** DOLIST puts its body in TAGBODY;
4439 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4441 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4442 value producing form;
4443 ** if more variables are given to PROGV than values, extra
4444 variables are bound and made to have no value;
4445 ** NSUBSTITUTE on list arguments gets the right answer with
4447 ** ELT signals an error of type TYPE-ERROR when the index argument
4448 is not a valid sequence index;
4449 ** LOOP signals (at macroexpansion time) an error of type
4450 PROGRAM-ERROR when duplicate variable names are found;
4451 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4452 ** FUNCALL of special-operators now cause an error of type
4454 ** PSETQ now works as required in the presence of side-effecting
4455 symbol-macro places;
4456 ** NCONC accepts any object as its last argument;
4457 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4459 ** loop-for-as-package does not require a package to be explicitely
4461 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4463 * fixed bug 166: compiler preserves "there is a way to go"
4464 invariant when deleting code.
4465 * fixed bug 172: macro lambda lists with required arguments after
4466 &REST arguments now cause an error to be signalled. (thanks to
4468 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4469 bug. (thanks to Matthew Danish)
4470 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4472 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4473 arguments are lexically rebound. (thanks to Gerd Moellmann and
4475 * fixed bug 194: error messages are now more informative when there
4476 is no primary method applicable in a call to a generic
4477 function. (thanks to Gerd Moellmann)
4478 * fixed bug in command line argument checking (thanks to Julian
4480 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4481 (thanks to Matthew Danish)
4482 * incremented fasl file version number, because of the
4483 SXHASH-related changes in the layout of CLOS data structures
4485 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4486 * minor incompatible change: The runtime (the Unix executable named
4487 "sbcl") is now much pickier about the .core files it will load.
4488 Essentially it now requires .core files to descend from the same
4489 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4490 as the runtime does. (The intent is to prevent the crashes which
4491 can occur, and which can even be reported as mysterious failures,
4492 when people patch the sources or change the build parameters
4493 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4494 sbcl and .core files.)
4495 * fixed bug: VALUES-LIST is no longer optimized away.
4496 * fixed bug 142: The FFI conversion of C string values to Lisp
4497 string values no longer conses excessively. (thanks to Nathan
4498 Froyd porting Raymond Toy's fix to CMU CL)
4499 * began to systematize and improve MOP conformance in PCL (thanks to
4500 Nathan Froyd, Gerd Moellman and Pierre Mai):
4501 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4503 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4504 ** FINALIZE-INHERITANCE is now called on class finalization;
4505 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4506 argument precedence order.
4507 * fixed bug 202: The compiler no longer fails on functions whose
4508 derived types contradict their declared type.
4509 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4510 so it can be non-toplevel.
4511 * The fasl file version number has changed (because of the new
4512 implementation of DEFMACRO).
4513 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4514 safe code, that any length requirement by their type-specifier
4515 argument is valid. The exceptions to this are described in bug
4517 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4518 functions now check the types of their inputs as required by ANSI.
4519 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4520 introduced symbol is DECLAREd to be SPECIAL.
4521 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4522 * fixed bug: :COUNT argument to sequence functions may be negative
4523 * fixed bug: body of DO-SYMBOLS may contain declarations
4524 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4525 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4526 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4527 symbol macro only once
4528 * fixed printing of call frame when argument list is unavailable
4529 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4530 * compiler no longer signals WARNING on unknown keyword
4533 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4534 * A beta-quality port to the mips architecture running Linux,
4535 based on the old CMUCL backend, has been made. It has been tested
4536 on a big-endian kernel, and works sufficiently well to be able to
4537 rebuild itself; it has not been tested in little-endian mode.
4538 * fixed an inconsistency between gencgc.c and purify.c which made
4539 dumping/loading .core files unreliable
4540 * fixed bug 120a: The compiler now deals correctly with IFs where
4541 the consequent is the same as the alternative, instead of
4542 misderiving the return type. (thanks to Alexey Dejneka)
4543 * fixed bug 113: Logical pathnames are now dumpable (the logical
4544 host is resolved at load-time, throwing an error if it is not
4546 * fixed bug 174: FORMAT's error message is slightly clearer when a
4547 non-printing character is used in a format directive.
4548 * fixed several bugs in compiler checking of type declarations, i.e.
4549 violations of the Python "declarations are assertions" principle
4550 (thanks to Alexey Dejneka)
4551 * fixed several bugs in PCL's error checking (thanks to Gerd
4553 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4555 * fixed bug in compilation of functions as first class values
4556 (thanks to Antonio Martinez-Shotton)
4557 * The compiler's handling TYPE-ERRORs which it can prove will
4558 inevitably happen at runtime has been cleaned up and corrected
4559 in several ways. (thanks to Alexey Dejneka)
4560 * improved argument type checking for various basic arithmetic
4561 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4562 lavished on them in the past that they can be compiled in many
4563 ways in different special cases
4564 * fixed bug 181: compiler checks validity of user supplied type
4566 * cleaned up code flushing in optimization: Function calls which
4567 should signal errors for safety purposes (e.g. which ANSI says
4568 should signal errors when their arguments are of incorrect type)
4569 are no longer optimized away.
4570 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4571 * incremented fasl file version number, because changes in the
4572 implementation of sequence functions like COERCE caused
4573 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4574 in old inline expansions) to become undefined. (Actually these
4575 changes were later undone, so we might very well be binary
4576 compatible with 0.7.7 after all, but leaving the version number
4577 incremented seemed like the simplest and most conservative
4580 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4581 * An alpha-quality port to the parisc architecture running Linux,
4582 based on the old CMUCL backend, has been made. This, even more so
4583 than the other backends, should be considered still a work in
4584 progress; known problems include that the Linux kernel in 64-bit
4585 mode does not propagate the correct sigcontext structure to
4586 userspace, and consequently SBCL on a parisc64 kernel will not
4588 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4589 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4590 suppressed.) Also now that the compiler is looking at declarations
4591 in the environment, it checks optimization declarations as well,
4592 and suppresses inlining when (> DEBUG SPEED).
4593 * More fixes have been made to treatment of floating point exception
4594 treatment and other Unix signals. In particular, floating point
4595 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4596 * The detection and handling of control stack exhaustion (infinite
4597 or very deeply nested recursion) has changed. Stack exhaustion
4598 detection is now done by write-protecting pages at the OS level
4599 and applies at all optimization settings; when found, a
4600 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4601 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4602 be caught using IGNORE-ERRORS.
4603 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4604 correctly when transforming recursive LABELS functions to LETs.
4605 (thanks to Alexey Dejneka)
4606 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4607 that are names of constants or global variables.
4608 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4609 alien routines with docstrings.
4610 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4611 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4613 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4614 (thanks to Eric Marsden)
4615 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4616 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4617 * Bug fix: The SPARC backend can now compile functions involving
4618 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4619 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4620 to a functions expecting a PATHNAME-DESIGNATOR.
4621 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4622 lambda lists. (thanks to Alexey Dejneka)
4623 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4624 functions so that EXPORTing the name of the class doesn't cause
4625 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4626 to Antonio Martinez for reporting this)
4627 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4628 constant-folded without failing an assertion. (thanks to Einar
4629 Floystad Dorum for reporting this)
4630 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4631 types are now dealt with more correctly by the compiler.
4632 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4633 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4634 behaviour analogous to RENAME-FILE. This puts its behaviour more
4635 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4637 * The fasl file version number has changed again. (because of the
4638 bug fix involving the names of PCL MAKE-INSTANCE functions)
4640 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4641 * bug fix: Floating point exceptions are treated much more
4642 consistently on the x86/Linux and PPC/Linux platforms.
4643 * Array initialization with :INITIAL-ELEMENT is now much faster for
4644 cases when the compiler cannot open code the array creation, but
4645 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4646 array accesses have also seen a speed increase.
4647 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4648 specified. (This was at the root of some bad interactions between
4649 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4651 * bug fix: Internal error arguments for undefined functions are now
4652 computed correctly on the PPC/Linux platform.
4653 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4654 Raymond Toy's patch for CMU CL)
4655 * Support for the Solaris 9 operating environment has been included
4656 (thanks to Daniel Merritt)
4657 * A very ugly but hopefully complete draft of the missing FFI chapter
4658 of the manual has been created by reformatting the corresponding
4659 CMU CL manual chapter into (currently very ugly and incoherent)
4660 DocBook and bringing it up to date for SBCL behavior. Thus, the
4661 manual is now essentially complete, at least by my extreme
4662 once-and-only-once standards, whereby it's acceptable to refer to
4663 the doc strings of SB-EXT functions as the primary documentation.
4664 * The fasl file version number has changed again, due to cleanup of
4665 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4666 is no longer a static symbol.)
4668 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4669 * SBCL now builds with OpenMCL (version 0.12) as the
4670 cross-compilation host; also, more progress has been made toward
4671 bootstrapping under CLISP.
4672 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4674 * bug 158 fixed: The compiler can now deal with integer loop
4675 increments different from 1; fixing this turned out also to fix
4677 * bug 169 fixed: no more bogus warnings about using lexical bindings
4678 despite the presence of perfectly good SPECIAL declarations (thanks
4679 to David Lichteblau)
4680 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4681 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4682 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4684 * bug 180 fixed: Method combination specifications no longer ignore
4685 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4686 * bug fix: Structure type predicate functions now check their argument
4687 count as they should.
4688 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4689 correctly. (thanks to Pierre Mai)
4690 * minor incompatible change: The --noprogrammer option is deprecated
4691 in favor of the new --disable-debugger option, which is very similar.
4692 (The major difference is that it takes effect at a slightly different
4693 time at startup, causing handling of errors in --sysinit and
4694 --userinit files will be affected differently.) The
4695 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4696 been added to allow this functionality to be controlled from ordinary
4697 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4698 maintainers, who might want to run non-interactive scripts to
4699 build SBCL cores which will later be used interactively.)
4700 * minor incompatible change: The LOAD function no longer, when given
4701 a wild pathname to load, loads all files matching that pathname.
4702 Instead, an error of type FILE-ERROR is signalled.
4704 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4705 * bug 147 fixed: The compiler preserves its block link/count
4706 invariants more correctly now so that it doesn't crash. (thanks
4708 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4710 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4711 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4712 no-value-for-this-character as NIL (as specified by ANSI).
4713 * HOST-NAMESTRING on physical pathnames now returns a string that is
4714 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4715 (thanks to Christophe Rhodes)
4716 * The Alpha port handles icache flushing more correctly. (thanks to
4718 * More progress has been made toward bootstrapping under CLISP. (thanks
4719 to Christophe Rhodes)
4720 * The fasl file format has changed again, because dynamic loading
4721 on OpenBSD (which has non-ELF object files) motivated some cleanups
4722 in the way that foreign symbols are transformed and passed around.
4723 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4724 is no longer treated as whitespace by the reader, but instead as
4725 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4726 |A
\7fB|, instead of A as it used to.
4728 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4729 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4730 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4731 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4732 this as of 0.7.1.45, but was left out of the previous news section
4733 (thanks to Dan Barlow)
4734 * SBCL now runs on the Solaris operating system on SPARC architectures
4735 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4736 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4737 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4738 * SPARC backend cleanups, allowing builds of cores optimized for V8
4739 and V9 SPARCS, and also emission of code targeted to a particular
4740 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4742 * SBCL is closer to bootstrapping under CLISP, thanks to various
4743 fixes by Christophe Rhodes.
4744 * The fasl file format has changed again, to allow the compiler's
4745 INFO database to support symbol macros.
4746 * The user manual (in doc/) is formatted into HTML more nicely.
4747 (thanks to coreythomas)
4748 * The system is smarter about SUBTYPEP relationships, especially
4749 those involving NOT types (including types such as ATOM which are
4750 represented internally using NOT types). Thus SUBTYPEP is less
4751 likely to return (VALUES NIL NIL) in general, and in particular
4752 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4754 * The fasl file format has changed again, because the internal
4755 representation of types now includes a new slot to support the new
4756 SUBTYPEP-of-NOT-types logic.
4757 * (not a change in the main branch of SBCL, but a related prototype
4758 which can hopefully be merged into the main branch of SBCL in the
4759 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4760 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4761 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4762 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4763 dispatch macro character combinations. (thanks to Alexey Dejneka)
4764 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4765 that superficially look like logical namestrings correctly.
4766 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4768 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4769 default supplied-p) for &optional and &key arguments. (thanks to
4772 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4773 * incompatible change: The compiler is now less aggressive about
4774 tail call optimization, doing it only when (> SPACE DEBUG) or
4775 (> SPEED DEBUG). (This is an incompatible change because there are
4776 programs which relied on the old CMU-CL-style behavior to optimize
4777 away their unbounded recursion which will now die of stack overflow.)
4778 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4779 for non-GENCGC systems has been increased to 20M (since that
4780 seems much closer to the likely performance optimum for modern
4781 systems than the old 4M value was)
4782 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4783 and *DEBUG-PRINT-LEVEL*
4784 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4785 of CMU CL's support for SPARC, and various endianness and other
4786 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4787 * new syntactic sugar for the Unix command line: --load foo.bar is now
4788 an alternate notation for --eval '(load "foo.bar")'.
4790 ** The system now detects stack overflow and handles it gracefully,
4791 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4792 optimization settings. (This is a good thing in general, and
4793 its introduction in this version should be particularly timely
4794 for anyone whose code fails because of suppression of tail
4796 ** The system now hunts for the C variable "environ" in a more
4797 devious way, to avoid segfaults when the C library version differs
4798 between compile time and run time. (thanks to Christophe Rhodes)
4799 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4800 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4801 ** The compiler no longer issues bogus style warnings for undefined
4802 classes in the same source file as the DEFCLASSes which defined
4803 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4805 ** fixes in CONDITION class precedence list for undefined function
4806 errors (thanks to Alexei Dejneka)
4807 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4808 correctly. (thanks to Dan Barlow)
4809 ** portability fixes aiming at bootstrapping under CLISP (thanks
4810 to Dave McDonald and Christophe Rhodes)
4811 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4812 ** fixes in type translation and and type inference (thanks to
4814 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4815 ** various fixes in the new ports (thanks to Dan Barlow)
4816 * several changes related to debugging:
4817 ** suppression of tail recursion, as noted above
4818 ** stack overflow detection, as noted above
4819 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4820 is now the default. (For some time encapsulation has been more
4821 reliable than the breakpoint-based :ENCAPSULATE NIL
4822 implementation, at least on X86 systems; and I just noticed that
4823 encapsulation also seems closer to the spirit of the ANSI
4826 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4828 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4829 up properly again. (There was a packaging bug in 0.7.0 which
4830 left their definitions in SB-SYS::LOAD-FOREIGN and
4831 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4832 vital for most things which interface to C-level interfaces,
4833 like extensions working with sockets or databases or
4834 Perl-compatible regexes or whatever, and the need to fix
4835 this bug is the main reason that 0.7.1 was released so
4837 ** DEFGENERIC is now choosier about the methods it redefines, so that
4838 reLOADing a previously-LOADed file containing DEFGENERICs does
4839 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4840 cycle now works as it should. (thanks to Alexey Dejneka)
4841 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4842 fixed by Christophe Rhodes. (He actually submitted this patch
4843 months ago, and I delayed until after 0.7.0.)
4844 ** Bug 111 (internal compiler confusion about runtime checks on
4845 FUNCTION types) was fixed by Alexey Dejneka.
4846 * Some internal cleanups (getting rid of variables which aren't
4847 needed now that the byte interpreter is gone) caused the fasl
4848 file format number to change again.
4850 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4851 * major incompatible change: The default fasl file extension, i.e. the
4852 default extension for files produced by COMPILE-FILE, has changed
4853 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4855 ** There are many changes in the implementation of the compiler.
4856 SBCL is now essentially a compiler-only implementation of ANSI
4857 Common Lisp. EVAL still "interprets" a few special cases, but
4858 almost all the interesting cases are handled by creating
4859 a LAMBDA expression, calling COMPILE on it, then calling
4860 FUNCALL on the result.
4861 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4862 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4863 Since the code is newer, there might still be some new bugs
4864 (though not as many as before Martin Atzmueller's fixes:-). But
4865 the new code is substantially simpler and clearer, and hopefully
4866 any remaining bugs will be simpler, less fundamental, and more
4867 fixable then the bugs in the old code.
4868 ** The revised compiler is still a little unsteady on its feet.
4870 *** The debugging information it produces (particularly the names
4871 of FUNCTION objects) is sometimes much less useful than what
4872 the old compiler produced.
4873 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4874 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4875 within a MACROLET) has been temporarily weakened.
4876 ** There are new compiler optimizations for various functions:
4877 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4878 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4879 *** the math functions TRUNCATE, FLOOR, and CEILING
4880 *** the function-of-all-trades COERCE
4881 Mostly these should be transparent, but there's one
4882 potentially-annoying problem (bug 117): when the compiler
4883 inline-expands a function and does type analysis on the result,
4884 it can create control paths which have type mismatches, and
4885 when it can't prove that those control paths aren't taken,
4886 it will issue WARNINGs about the type mismatches. This is
4887 a particular problem in practice for the new sequence functions.
4888 It's not clear how this should be fixed, and for now, a
4889 workaround is given in the entry for 117 in the BUGS file.
4890 ** (Because of the interaction between the two previous items --
4891 occasional inlining problems and new inline expansions -- some
4892 of the new sequence function optimizations won't really kick in
4893 completely until debugging information, and then inlining, are
4894 straightened out in some future version.)
4895 * minor incompatible changes:
4896 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4897 DIRECTORY no longer implicitly promotes NIL slots of its
4898 pathname argument to :WILD. In particular, when you ask for the
4899 contents of a directory (which you used to be able to do without
4900 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4901 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4902 ** changes in behavior that ANSI explicitly defines to be
4903 implementation dependent:
4904 *** The new compiler-only implementation still conforms with ANSI,
4905 but acts a little different than before. Besides the obvious
4906 changes in performance tradeoffs (that the cost per form passed
4907 to EVAL has gone up, and the cost per form executed by EVAL
4908 has gone down), the behavior of the system changes a little
4909 because there are no longer any interpreted function objects.
4910 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4911 e.g. doing COMPILE on the output of interactive DEFUN is
4913 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4915 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4916 has changed from (:CL) to NIL.
4917 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4918 e.g. #\Newline and #\Tab, don't have names specified in the
4919 ANSI Common Lisp standard, is now based on their ASCII symbolic
4920 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4921 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4922 are no longer used for output.
4923 ** changes in internal implementation constants:
4924 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4925 4 million. (If your application spends a lot of time GCing and
4926 you have a lot of RAM, you might want to experiment with
4927 increasing it even more.)
4928 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4929 However, almost all old code should still continue to work without
4930 immediate update, as SB-C-CALL is now a (deprecated) nickname
4932 ** Old operator names in the style DEF-FOO are now deprecated in
4933 favor of new corresponding names DEFINE-FOO, for consistency with
4934 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4935 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4936 internal symbols, but a few supported extensions like
4937 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4938 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4939 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4940 etc. as you get deeper into recursive calls to the debugger
4941 command loop, instead of the old "5]", "5]]", "5]]]"
4942 sequence. (I was motivated to do this when squabbles between
4943 ILISP and SBCL left me very deeply nested in the debugger. In the
4944 short term, this change will probably provoke more ILISP/SBCL
4945 squabbles, but hopefully it will be an improvement in the long run.)
4946 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4947 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4948 function, since declaiming return types involving aliens is
4949 (1) annoyingly messy to do by hand and (2) vital to efficient
4950 compilation of code which calls such functions.
4951 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4952 longer reexported by the SB-EXT package. They're solely useful
4953 for alien code, so it seems more logical that you should get
4954 them from the SB-ALIEN package, not in SB-EXT.
4955 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4956 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4957 features. Instead, the code that they used to control is always
4958 built into the system.
4959 * many other bug fixes
4960 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4961 advantage of the new EVAL-WHEN stuff and to clean them up in
4962 general, and they are now more ANSI-compliant in a number of
4963 ways. Martin Atzmueller is responsible for a lot of this.
4964 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4966 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4967 *** correct ERROR type for various file operations
4968 *** some fixes for Lisp streams
4969 *** DEFMETHOD syntax checking
4970 *** changing old weird representation of debug information as
4971 strings (which, among their other deficiencies, don't transform
4972 correctly when you rename packages, and don't change their
4973 print representation when you change things like *PACKAGE*
4974 and *PRINT-LENGTH*) to symbols and lists of symbols
4975 He also made several improvements and fixed several bugs in DESCRIBE.
4976 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4978 *** misbehavior of WRITE-STRING/WRITE-LINE
4979 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4980 and several other LOOP problems as well
4981 *** DIRECTORY when similar filenames are present
4982 *** DEFGENERIC with :METHOD options
4983 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4984 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4985 *** argument ordering in FIND with :TEST option
4986 *** mishandled package designator argument in APROPOS-LIST
4987 *** various problems in the backquote readmacro
4988 *** a bug in APROPOS
4989 *** probably some others that I'm not describing very well here,
4990 since the CVS log documents them by reference to sbcl-devel
4991 messages, and the SourceForge archives aren't working well.:-(
4992 ** Dan Barlow improved the Alpha port (and is making progress on the
4993 PPC port, for those of you who think different).
4994 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4995 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4996 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4997 continues to work on the SPARC port (for those of you in a position
4998 to look down upon our little PC-compatible boxes from a great height).
4999 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
5000 and uses that copy, rather than the current dynamic value, when
5001 it's trying to decide whether to truncate output. Thus e.g.
5002 (let ((*print-lines* 50))
5003 (pprint-logical-block (stream nil)
5005 (let ((*print-lines* 8))
5006 (print (aref possiblybigthings i) stream)))))
5007 should now truncate the logical block only at 50 lines, instead of
5008 often truncating it at 8 lines, as it did before.
5009 * The doc/cmucl/ directory, containing old CMU CL documentation
5010 from the time of the fork, is no longer part of the base system.
5011 SourceForge has shut down its anonymous FTP service, and with it
5012 my original plan for distributing the old CMU CL documentation
5013 there. For now, if you need these files you can download an old
5014 SBCL source release and extract them from it.
5015 * The fasl file version number changed again, for dozens of reasons,
5016 some of which are apparent above.
5018 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
5019 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
5020 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
5021 patch, so that DISASSEMBLE output is much nicer.
5022 * The code in the SB-PROFILE package now seems reasonably stable.
5023 I still haven't decided what the final interface should look like
5024 (I'd like PROFILE to interact cleanly with TRACE, since both
5025 facilities use function encapsulation) but if you have a need
5026 for profiling now, you can probably use it successfully with
5027 the current CMU-CL-style interface.
5028 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
5029 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
5030 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
5031 link now returns the dangling link itself, and for similar
5032 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
5033 link itself. (In these cases the old code signalled an error and
5034 looped endlessly, respectively.) Thus, DIRECTORY now works even
5035 in the presence of dangling and cyclic symbolic links.
5036 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
5037 is now a supported extension again, since the consensus on
5038 sbcl-devel was that it can be useful for ordinary development
5039 work, not just for debugging SBCL itself.
5040 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
5041 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
5042 that currently-defined functions might be redefined later with
5043 different return types.
5044 * Hash tables can be printed readably, as inspired by CMU CL code
5045 of Eric Marsden and SBCL code of Martin Atzmueller.
5046 * better error handling in CLOS method combination, thanks to
5047 Martin Atzmueller porting Pierre Mai's CMU CL patches
5048 * more overflow fixes for >16Mbyte I/O buffers
5049 * A bug in READ has been fixed, so that now a single Ctrl-D
5050 character suffices to cause end-of-file on character streams.
5051 In particular, now you only need one Ctrl-D at the command
5052 line (not two) to exit SBCL.
5053 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
5055 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
5056 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
5057 does the right thing.
5058 * The implementation of some type tests, especially for CONDITION
5059 types, is now tidier and maybe faster, due to CMU CL code
5060 originally by Douglas Crosher, ported by Martin Atzmueller.
5061 * Some math functions have been fixed, and there are new
5062 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
5063 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
5064 * (There are also some new optimizers in contrib/*-extras.lisp. Those
5065 aren't built into sbcl-0.6.13, but are a sneak preview of what's
5066 likely to be built into sbcl-0.7.0.)
5067 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
5068 CL, ported to SBCL by Martin Atzmueller)
5069 * DESCRIBE now gives more information in some cases. (Pierre Mai's
5070 patch to CMU CL, ported to SBCL by Martin Atzmueller)
5071 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
5072 * There's a new slam.sh hack to shorten the edit/compile/debug
5073 cycle for low-level changes to SBCL itself, and a new
5074 :SB-AFTER-XC-CORE target feature to control the generation of
5075 the after-xc.core file needed by slam.sh.
5076 * minor incompatible change: The ENTRY-POINTS &KEY argument to
5077 COMPILE-FILE is no longer supported, so that now every function
5078 gets an entry point, so that block compilation looks a little
5079 more like the plain vanilla ANSI section 3.2.2.3 scheme.
5080 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
5081 returns the number of bytes consed since the system started,
5082 rather than the number consed since the first time the function
5083 was called. (The new definition parallels ANSI functions like
5084 CL:GET-INTERNAL-RUN-TIME.)
5085 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
5086 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
5087 supported. Now DIRECTORY always does the abstract Common-Lisp-y
5088 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
5089 * Fasl file version numbers are now independent of the target CPU,
5090 since historically most system changes which required version
5091 number changes have affected all CPUs equally. Similarly,
5092 the byte fasl file version is now equal to the ordinary
5095 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
5096 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
5097 is no longer recognized. I apologize for this, because it was
5098 listed in SB-EXT as a supported extension, but I found that
5099 its existing behavior was poorly specified, as well as incorrectly
5100 specified, and it looked like too much of a mess to straighten it
5101 out. I have enough on my hands trying to get ANSI stuff to work..
5102 * many patches ported from CMU CL by Martin Atzmueller, with
5103 half a dozen bug fixes in pretty-printing and the debugger, and
5104 half a dozen others elsewhere
5105 * fixed bug 13: Floating point infinities are now supported again.
5106 They might still be a little bit flaky, but thanks to bug reports
5107 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
5108 as flaky as they were.
5109 * The --noprogrammer command line option is now supported. (Its
5110 behavior is slightly different in detail from what the old man
5111 page claimed it would do, but it's still appropriate under the
5112 same circumstances that the man page talks about.)
5113 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
5114 are now supported, and enabled by default. Thus, the compiler can
5115 handle many floating point and complex operations much less
5116 inefficiently. (Thus e.g. you can implement a complex FFT
5118 * The compiler now detects type mismatches between DECLAIM FTYPE
5119 and DEFUN better, and implements CHECK-TYPE more correctly, and
5120 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
5121 thanks to patches from Martin Atzmueller.
5122 * various fixes to make the cross-compiler more portable to
5123 ANSI-conforming-but-different cross-compilation hosts (notably
5124 Lispworks for Windows, following bug reports from Arthur Lemmens)
5125 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
5126 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
5127 fixed thanks to Pierre Mai's CMU CL patches.
5128 * fixes to keep the system from overflowing internal counters when
5129 it tries to use i/o buffers larger than 16M bytes
5130 * fixed bug 45a: Various internal functions required to support
5131 complex special functions have been merged from CMU CL sources.
5132 (When I was first setting up SBCL, I misunderstood a compile-time
5133 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
5134 * improved support for type intersection and union, fixing bug 12
5135 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
5136 more obscure bugs as well
5137 * some steps toward byte-compiling non-performance-critical
5138 parts of the system, courtesy of patches from Martin Atzmueller
5139 * Christophe Rhodes has made some debian packages of sbcl at
5140 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
5141 From his sbcl-devel e-mail of 2001-04-08 they're not completely
5142 stable, but are nonetheless usable. When he's ready, I'd be happy
5143 to add them to the SourceForge "File Releases" section. (And if
5144 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
5145 * new fasl file format version number (because of changes in
5146 internal representation of (OR ..) types to accommodate the new
5147 support for (AND ..) types, among other things)
5149 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
5150 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
5152 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
5154 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
5155 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
5156 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
5157 * DESCRIBE now works on structure objects again.
5158 * Most function call argument type mismatches are now handled as
5159 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
5160 know whether the function will be redefined before the call is
5161 executed. (The compiler could flag local calls with full WARNINGs,
5162 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
5163 it doesn't keep track of enough information to know whether calls
5164 are local in this sense.)
5165 * Compiler output is now more verbose, with messages truncated
5166 later than before. (There should be some supported way for users
5167 to override the default verbosity, but I haven't decided how to
5168 provide it yet, so this behavior is still controlled by the internal
5169 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
5170 src/compiler/ir1util.lisp.)
5171 * Fasl file format version numbers have increased again, because
5172 support for the Gray streams extension changes the layout of the
5173 system's STREAM objects.
5174 * The Gray subclassable streams extension now works, thanks to a
5175 patch from Martin Atzmueller.
5176 * The full LOAD-FOREIGN extension (not just the primitive
5177 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
5178 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
5179 but like most other programs, it defaults to copying the Unix
5180 environment from the original process instead of starting the
5181 new process in an empty environment.
5182 * Extensions which manipulate the Unix environment now support
5183 an :ENVIRONMENT keyword option which doesn't smash case or
5184 do other bad things. The CMU-CL-style :ENV option is retained
5185 for porting convenience.
5186 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
5187 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
5189 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
5191 * A patch from Martin Atzmueller seems to have solved the SIGINT
5192 problem, and as far as we know, signal-handling now works cleanly.
5193 (If you find any new bugs, please report them!)
5194 * The system no longer defaults Lisp source file names to types
5195 ".l", ".cl", or ".lsp", but only to ".lisp".
5196 * The compiler no longer uses special default file extensions for
5197 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
5198 seems to expect a single default extension for all compiled code,
5199 and there's no compelling reason to try to stretch the standard
5200 to allow two different extensions.) Instead, byte-compiled files
5201 default to the same extension as native-compiled files.
5202 * Fasl file format version numbers have increased again, because
5203 a rearrangement of internal implementation packages made some
5204 dumped symbols in old fasl files unreadable in new cores.
5205 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
5206 many fewer weird special cases.
5207 * Bug #17 (differing COMPILE-FILE behavior between logical and
5208 physical pathnames) has been fixed, and some related misbehavior too,
5209 thanks to a patch from Martin Atzmueller.
5210 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
5211 by Tim Moore, ported to SBCL by Martin Atzmueller.
5212 * Martin Atzmueller fixed several filesystem-related problems,
5213 including bug #36, in part by porting CMU CL patches, which were
5214 written in part by Paul Werkowski.
5215 * More compiler warnings in src/runtime/ are gone, thanks to
5216 more patches from Martin Atzmueller.
5217 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
5220 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
5222 * DESCRIBE now works on CONDITION objects.
5223 * The debugger now handles errors which arise when trying to print
5224 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
5226 * The build system now uses an additional file, customize-target-features.lisp,
5227 to allow local modifications to the target *FEATURES* list. (The point of
5228 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
5229 debugging features enabled, without having to worry about propagating it
5230 into everyone's system when I do a "cvs update".) When no
5231 customize-target-features.lisp file exists, the target *FEATURES* list
5232 should be constructed the same way as before.
5233 * fixed bugs in DEFCONSTANT ANSI-compatibility:
5234 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
5235 warn about behavior which is undefined under the ANSI spec. Note: This
5236 is specified by ANSI, but it's not very popular with programmers.
5237 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
5238 macro in the SBCL sources for an example of a workaround which you
5239 might use to make portable ANSI-standard code which does what you want.
5240 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
5241 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
5242 when it's used as a non-toplevel form. (This is required in order
5243 to implement the DEFCONSTANT-EQX macro.)
5244 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
5245 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
5246 funny things, due to bugs in the implementation of EVAL-WHEN
5247 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
5248 non-toplevel DEFCONSTANTs are uncommon.)
5249 * The core file version number and fasl file version number have been
5250 incremented, because the old noncompliant DEFCONSTANT behavior involved
5251 calling functions which no longer exist, and because I also took the
5252 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
5253 * fixed bug 1 (error handling before read-eval-print loop starts), and
5254 redid debugger restarts and related debugger commands somewhat while
5256 ** The QUIT debugger command is gone, since it did something
5257 rather different than the SB-EXT:QUIT command, and since it never
5258 worked properly outside the main toplevel read/eval/print loop.
5259 Invoking the new TOPLEVEL restart provides the same functionality.
5260 ** The GO debugger command is also gone, since you can just invoke
5261 the CONTINUE restart directly instead.
5262 ** The TOP debugger command is also gone, since it's redundant with the
5263 FRAME 0 command, and since it interfered with abbreviations for the
5265 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
5267 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
5268 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
5269 some porting work by Martin Atzmueller)
5270 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
5271 with a boatload of patches which he ported to SBCL. Now that those
5273 ** The system tries to make sure that its low-priority messages
5274 are prefixed by semicolons, to help people who like to use
5275 syntax highlighting in their ILISP buffer. (This patch
5276 was originally due to Raymond Toy.)
5277 ** The system now optimizes INTEGER-LENGTH better, thanks to more
5278 patches originally written by Raymond Toy.
5279 ** The compiler understands coercion between single-value and
5280 multiple-VALUES type expressions better, getting rid of some very
5281 weird behavior, thanks to patches originally by Robert MacLachlan
5282 and Douglas Crosher.
5283 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
5284 lambda lists, thanks to a patch originally by Pierre Mai.
5285 ** The system no longer bogusly warns about "abbreviated type
5287 ** The compiler gets less confused by inlining and RETURN-FROM,
5288 thanks to some patches originally by Tim Moore.
5289 ** The system no longer hangs when dumping circular lists to fasl
5290 files, thanks to a patch originally from Douglas Crosher.
5291 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
5292 undefined function error.
5293 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
5294 in structure slot initforms) for now, documented workaround instead:-|
5295 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
5296 * fixed bug 5: added stubs for various Gray stream functions called
5297 in the not-a-CL:STREAM case, so that even when Gray streams aren't
5298 installed, at least appropriate type errors are generated
5299 * fixed bug 8: better reporting of various PROGRAM-ERRORs
5300 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
5301 consistently in DEFMETHOD forms.
5302 * removed bug 21 from BUGS, since Martin Atzmueller points out that
5303 it doesn't seem to affect SBCL after all
5304 * The C runtime system now builds with better optimization and many
5305 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
5307 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
5309 * The system is now under CVS at SourceForge (instead of the
5310 CVS repository on my home machine).
5311 * The new signal handling code has been tweaked to treat register
5312 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
5313 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
5314 downstream, e.g. in debug-int.lisp, has implicit dependencies
5315 on the unsignedness of integer representation of machine words,
5316 and that caused the system to bomb out with infinite regress
5317 when trying to recover from type errors involving signed values,
5318 e.g. (BUTLAST '(1 2 3) -1).)
5319 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
5320 (This was one of the bugs Peter Van Eynde reported back in July.)
5321 * The system now uses code inspired by Colin Walters' O(N)
5322 implementation of MAP (from the cmucl-imp@cons.org mailing
5323 list, 2 September 2000) when it can't use a DEFTRANSFORM to
5324 inline the MAP operation, and there is more than one
5325 sequence argument to the MAP call (so that it can't just
5326 do ETYPECASE once and for all based on the type of the
5327 single sequence argument). (The old non-inline implementation
5328 of the general M-argument sequence-of-length-N case required
5329 O(M*N*N) time when any of the sequence arguments were LISTs.)
5330 * The QUIT :UNIX-CODE keyword argument has been renamed to
5331 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5332 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5334 * Raymond Wiker's patches to port dynamic loading from Linux to
5335 FreeBSD have been added.
5336 * The BUGS file is now more nearly up to date, thanks in large part
5337 to Martin Atzmueller's review of it.
5338 * The debugger now flushes standard output streams before it begins
5339 its output ("debugger invoked" and so forth).
5340 * The core version number and fasl file version number have both
5341 been incremented, because of incompatible changes in the layout
5343 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5344 used to not be called for a saved Lisp image.)
5345 * Martin Atzmueller's version of a patch to fix a compiler crash,
5346 as posted on sbcl-devel 13 September 2000, has been installed.
5347 * Instead of installing Martin Atzmueller's patch for the
5348 compiler transform for SUBSEQ, I deleted the compiler transform,
5349 and transforms for some similar consing operations.
5350 * A bug in signal handling which kept TRACE from working on OpenBSD
5352 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5353 away by constant folding
5354 * The system now defines its address space constants in one place
5355 (in the Lisp sources), and propagates them automatically elsewhere
5356 (through GENESIS and the sbcl.h file). Therefore, patching the
5357 address map is less unnecessarily tedious and error-prone. The
5358 Lisp names of address space constants have also been systematized.
5359 * CVS tags like dollar-Header-dollar have been removed from
5360 the sources, because they have never saved me trouble and
5361 they've been source of trouble working with patches and other
5362 diff-related operations.
5363 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5364 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5366 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5368 * The system has been ported to OpenBSD.
5369 * The system now compiles with a simple "sh make.sh" on the systems
5370 that it's supported on. I.e., now you no longer need to tweak
5371 text in the target-features.lisp-expr and symlinks in src/runtime/
5372 by hand, the make.sh takes care of it for you.
5373 * The system is no longer so grossly inefficient when compiling code
5374 involving vectors implemented as general (not simple) vectors (VECTOR T),
5375 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5376 which dares to use the various sequence functions on non-simple
5377 vectors, takes less of a performance hit.
5378 * There is now a primitive type predicate VECTOR-T-P
5379 to test for the (VECTOR T) type, so that e.g.
5380 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5381 can now be compiled with some semblance of efficiency. (The old code
5382 turned the type declaration into a full call to %TYPEP at runtime!)
5383 * AREF on (VECTOR T) is still not fast, since it's still compiled
5384 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5385 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5386 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5387 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5388 without making full calls to SUBTYPEP at runtime.
5389 (Some analogous efficiency issues for non-simple vectors specialized to
5390 element types other than T, or for non-simple multidimensional arrays,
5391 have not been addressed. They could almost certainly be handled the
5392 same way if anyone is motivated to do so.)
5393 * The changes in array handling break binary compatibility, so
5394 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5395 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5396 instead of (VALUES T T).
5397 * By following the instructions that Dan Barlow posted to sbcl-devel
5398 on 2 July 2000, I was able to enable primitive dynamic object
5399 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5400 functionality is not implemented (since it calls ld to resolve
5401 library references automatically, requiring RUN-PROGRAM for its
5402 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5403 try to resolve library references) is now supported.
5404 * The system now flushes the standard output streams when it terminates,
5405 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5406 them at several other probably-convenient times, e.g. in each pass of
5407 the toplevel read-eval-print loop, and after evaluating a form given
5408 as an "--eval" command-line option. (These changes were motivated by a
5409 discussion of stream flushing issues on cmucl-imp in August 2000.)
5410 * The source transform for TYPEP of array types no longer assumes
5411 that an array whose element type is a not-yet-defined type
5412 is implemented as an array of T, but instead punts, so that the
5413 type will be interpreted at runtime.
5414 * There is now some support for cross-compiling in make.sh: each of
5415 the phases of make.sh has its own script. (This should be transparent
5416 to people doing ordinary, non-cross-compile builds.)
5417 * Since my laptop doesn't have hundreds of megabytes of memory like
5418 my desktop machine, I became more motivated to do some items on
5419 my to-do list in order to reduce the size of the system a little:
5420 ** Arrange for various needed-only-at-cold-init things to be
5421 uninterned after cold init. To support this, those things have
5422 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5423 symbols with such names are now uninterned after cold init).
5424 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5425 fasl files for cold load.
5426 ** Remove the old compiler structure pooling code (which used to
5427 be conditional on the target feature :SB-ALLOC) completely.
5428 ** Redo the representation of some data in cold init to be more compact.
5429 (I also looked into supporting byte compiled code at bootstrap time,
5430 which would probably reduce the size of the system a lot, but that
5431 looked too complicated, so I punted for now.)
5432 * The maximum signal nesting depth in the src/runtime/ support code has
5433 been reduced from 4096 to 256. (I don't know any reason for the very
5434 large old value. If the new smaller value turns out to break something,
5435 I'll probably just bump it back up.)
5436 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5438 * Many, many bugs reported by Peter Van Eynde have been added to
5439 the BUGS list; some have even been fixed.
5440 * While enabling dynamic object file loading, I tried to make the
5441 code easier to understand, renaming various functions and variables
5442 with less ambiguous names, and changing some function calling
5443 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5444 * While trying to figure out how to do the OpenBSD port, I tried to
5445 clean up some of the code in src/runtime/. In particular, I dropped
5446 support for non-POSIX signal handling, added various comments,
5447 tweaked the code to reduce the number of compilation warnings, and
5448 renamed some files to increase consistency.
5449 * To support the new automatic configuration functionality in make.sh,
5450 the source file target-features.lisp-expr has been replaced with the
5451 source file base-target-features.lisp-expr and the machine-generated
5452 file local-target-features.lisp-expr.
5453 * fixed a stupid quoting error in make.sh so that using CMU CL
5454 "lisp -batch" as cross-compilation host works again
5456 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5458 * DESCRIBE no longer tries to call itself recursively to describe
5459 bound/fbound values, so that it no longer fails on symbols which are
5460 bound to themselves (like keywords, T, and NIL).
5461 * DESCRIBE now works on generic functions.
5462 * The printer now prints less-screwed-up representations of closures
5463 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5464 * A private symbol is used instead of the :EMPTY keyword previously
5465 used to mark empty slots in hash tables. Thus
5466 (DEFVAR *HT* (MAKE-HASH-TABLE))
5467 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5468 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5469 now does what ANSI says that it should. (You can still get
5470 similar noncompliant behavior if bang on the hash table
5471 implementation with all the symbols you get back from
5472 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5473 This breaks binary compatibility, since tests for equality to
5474 :EMPTY are wired into things like the macroexpansion of
5475 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5477 * There's now a minimal placeholder implementation for CL:STEP,
5478 as required by ANSI.
5479 * An obscure bug in the interaction of the normal compiler, the byte
5480 compiler, inlining, and structure predicates has been patched
5481 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5482 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5484 * Missing ordinary arguments in a macro call are now detected even
5485 when the macro lambda list contains &KEY or &REST.
5486 * The debugger no longer complains about encountering the top of the
5487 stack when you type "FRAME 0" to explicitly instruct it to go to
5488 the top of the stack. And it now prints the frame you request even
5489 if it's the current frame (instead of saying "You are here.").
5490 * As specified by ANSI, the system now always prints keywords
5491 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5493 * The default initial SIZE of HASH-TABLEs is now smaller.
5494 * Type information from CLOS class dispatch is now propagated
5495 into DEFMETHOD bodies, so that e.g.
5496 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5498 is now basically equivalent to
5499 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5500 (DECLARE (TYPE SINGLE-FLOAT X))
5502 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5503 operation, without having to do run-time type dispatch.
5504 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5505 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5506 * Since I don't seem to be making big file reorganizations very often
5507 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5508 is overflowing my ability to conveniently back them up), I've finally
5509 checked the system into CVS. (The CVS repository is on my home system,
5510 not at SourceForge -- putting it on SourceForge might come later.)
5511 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5512 high-level GC-NOTIFY-FOO functions send their output. (There's
5513 still very little control of where low-level verbose GC functions
5514 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5515 less than it used to -- the GC-NOTIFY-FOO functions are now under
5516 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5517 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5518 in only one place in the source code, and propagates it automatically
5519 everywhere that it's needed. Thus e.g. when I bump the version from
5520 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5521 * The C source files now include boilerplate legalese and documentation
5522 at the head of each file (just as the Lisp source files already did).
5523 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5524 to his page will be replaced with a link to his new CLiki service.
5526 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5528 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5529 * The build process now looks for GNU make under the default name "gmake",
5530 instead of "make" as it used to. If GNU make is not available as "gmake"
5531 on your system, you can change this default behavior by setting the
5532 GNUMAKE environment variable.
5533 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5534 can build without error under CMU CL.
5536 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5538 * There is now a partial SBCL user manual (with some new text and some
5539 text cribbed from the CMU CL manual).
5540 * The beginnings of a profiler have been added (starting with the
5541 CMU CL profiler and simplifying and cleaning up). Eventually the
5542 main interface should be through the TRACE macro, but for now,
5543 it's still accessed through vaguely CMU-CL-style functions and macros
5544 exported from the package SB-PROFILE.
5545 * Some problems left over from porting CMU CL to the new
5546 cross-compilation bootstrap process have been cleaned up:
5547 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5548 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5549 part when getting rid of the compiler *BACKEND* stuff.)
5550 ** The value of *NULL-TYPE* was screwed up, because it was
5551 being initialized before the type system knew the final
5552 definition of the 'NULL type. This screwed up several key
5553 optimizations in the compiler, causing inefficiency in all sorts
5554 of places. (I found it because I wanted to understand why
5555 GET-INTERNAL-RUN-TIME was consing.)
5556 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5557 PROCLAIM FTYPE information. Unfortunately this broke binary
5558 compatibility again, since now the forms output by DEFGENERIC
5559 to refer to functions which didn't exist in 0.6.3.
5560 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5561 can use the new (as of 0.6.3) transform for SOME into MAP into
5563 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5564 values to (INTEGER 0 1000000), so that the time code will no longer
5565 occasionally get blown up by Linux returning 1000000 microseconds
5566 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5567 its output conform to the ANSI spec. (Alas, this makes its output
5568 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5569 * A full call to MAP NIL with a single sequence argument no longer conses.
5570 * fixes to problems pointed out by Martin Atzmueller:
5571 * The manual page no longer talks about multiprocessing as though
5572 it were currently supported.
5573 * The ILISP support patches have been removed from the distribution,
5574 because as of version 5.10.1, ILISP now supports SBCL without us
5575 having to maintain patches.
5576 * added a modified version of Raymond Toy's recent CMU CL patch for
5577 EQUALP comparison of HASH-TABLE
5579 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5581 * The system still can't cross-compile itself with
5582 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5583 entails), but at least it can compile more of itself that way
5584 than it used to be able to, and various buglets which were uncovered
5585 by trying to cross-compile itself that way have now been fixed.
5586 * This release breaks binary compatibility again. This time
5587 at least I've incremented the FASL file format version to 2, so that the
5588 problem can be detected reliably instead of just causing weird errors.
5589 * various new style warnings:
5590 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5591 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5592 ** using the lexical binding of a variable named in the *FOO* style
5593 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5594 as specified by ANSI.
5595 * *RANDOM-STATE* is no longer automatically initialized from
5596 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5597 default behavior of the system is to repeat its behavior every time
5598 it's run. If you'd like to change this behavior, you can always
5599 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5600 old convention there was no comparably easy way to get the system to
5601 repeat its behavior every time it was run.
5602 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5603 FTYPE declarations has been removed, in favor of their ANSI
5604 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5605 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5606 the types of their sequence arguments can be determined at compile time.
5607 This is done through a new open code expansion for MAP which eliminates
5608 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5609 argument types can be determined at compile time.
5610 * The optimizer now transforms COERCE into an identity operation when it
5611 can prove that the coerced object is already of the correct type. (This
5612 can be a win for machine generated code, including the output of other
5613 optimization transforms, such as the MAP transform above.)
5614 * Credit information has been moved from source file headers into CREDITS.
5615 * Source file headers have been made more standard.
5616 * The CASE macro now compiles without complaining even when it has
5619 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5621 * (Note that the way that the PCL macroexpansions were rewritten
5622 to accommodate the change in DEFGENERIC below breaks binary
5623 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5624 not run under sbcl-0.6.2. Once we get out of alpha releases,
5625 i.e. hit release 1.0.0, we'll probably try to maintain binary
5626 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5627 and sbcl-1.4.4. Until then, however, it might be fairly common
5628 for maintenance releases to break binary compatibility.)
5629 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5631 * The reporting of unhandled signals has been changed to print some
5632 explanatory text as well as the report form. (Previously only
5633 the report form was printed.)
5634 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5635 it defines, so that the compiler no longer issues undefined function
5636 warnings for compiled-but-not-yet-loaded generic functions.
5637 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5638 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5639 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5640 section of the ANSI spec are supported.
5641 * The "" nickname for the "KEYWORD" package has been removed.
5642 The reader still handles symbol tokens which begin with a package marker
5643 as keywords, but it doesn't expose its mechanism for doing so in the
5644 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5645 * The system now issues STYLE-WARNINGs for contradictory TYPE
5646 proclamations. (Warnings for contradictory FTYPE proclamations would
5647 be nice too, but those can't be done usefully unless the type system
5648 is made smarter about FUNCTION types.)
5649 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5650 systematized, so that "*target-*.lisp is supposed to exist only on the
5651 target and imply that there's a related file which exists on the
5652 host, and *host-*.lisp is supposed to exist only on the host and imply
5653 that there's a related file which exists on the target. This involves a
5654 lot of renaming. Hopefully the acute confusion caused by the renaming
5655 will be justified by the reduction in chronic confusion..
5656 ** runtime-type.lisp -> early-target-type.lisp
5657 ** target-type.lisp -> late-target-type.lisp
5658 ** early-host-format.lisp -> early-format.lisp
5659 ** late-host-format.lisp -> late-format.lisp
5660 ** host-error.lisp -> misc-error.lisp
5661 ** early-error.lisp -> early-target-error.lisp
5662 ** late-error.lisp -> late-target-error.lisp
5663 ** host-defboot.lisp -> early-defboot.lisp
5664 ** code/misc.lisp -> code/target-misc.lisp
5665 ** code/host-misc.lisp -> code/misc.lisp
5666 ** code/numbers.lisp -> code/target-numbers.lisp
5667 ** code/early-numbers.lisp -> numbers.lisp
5668 ** early-host-type.lisp -> early-type.lisp
5669 ** late-host-type.lisp -> late-type.lisp
5670 ** host-typep.lisp -> typep.lisp
5671 ** load.lisp -> target-load.lisp
5672 ** host-load.lisp -> load.lisp
5673 ** host-disassem.lisp -> disassem.lisp
5674 ** host-insts.lisp -> insts.lisp
5675 ** byte-comp.lisp -> target-byte-comp.lisp
5676 ** host-byte-comp.lisp -> byte-comp.lisp
5677 ** host-signal.lisp -> signal.lisp
5678 ** host-defstruct.lisp -> defstruct.lisp
5679 ** late-target-type.lisp -> deftypes-for-target.lisp
5680 Furthermore, several other previously target-only files foo.lisp (e.g.
5681 hash-table.lisp and random.lisp) have been split into a target-and-host
5682 foo.lisp file and a target-only target-foo.lisp file, with their key type
5683 definitions in the target-and-host part, so that the cross-compiler will
5684 know more about target types.
5685 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5686 gone away. In their place are various *BACKEND-FOO* variables
5687 corresponding to the slots of the old structure.
5688 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5689 into the target SBCL has been fixed.
5690 * The chill.lisp system for loading cold code into a running SBCL
5692 * Support for the CMU CL "scavenger hook" extension has been removed.
5693 (It was undocumented and unused in the CMU CL sources that SBCL was
5694 derived from, and stale in sbcl-0.6.1.)
5695 * Various errors in the cross-compiler type system were detected
5696 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5697 (enabling various consistency checks). Many of them were fixed,
5698 but some hard problems remain, so the compiler is back to
5699 running without *TYPE-SYSTEM-INITIALIZED* for now.
5700 * As part of the cross-compiler type system cleanup, I implemented
5701 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5702 * I have started adding UNCROSS calls throughout the type system
5703 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5704 on cross-compiler input files will be able to go away, and various
5706 * CONSTANTP now returns true for quoted forms (as explicitly required
5709 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5711 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5712 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5713 (MAKE-LIST -1). In the longer term, it ought to become true
5714 that declarations are assertions even at SAFETY 1. For now, it's not
5715 quite true even at SAFETY 3, but it's at least more nearly true..
5716 (Note that this change seems to increases the size of the system by
5717 O(5%) and to decrease the speed of the compiler by 20% or more.)
5718 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5719 for the problem of printing dozens of lines of distracting information
5720 about low-level system machinery as part of the top stack frame
5721 on entry to the debugger when an undefined function was called.
5722 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5723 is not reset to COMMON-LISP-USER.
5724 * Compilation of stuff related to dyncount.lisp has been made conditional
5725 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5726 smaller. The various dyncount-related symbols have been moved into
5727 a new "SB-DYNCOUNT" package.
5728 * tty-inspect.lisp has been renamed to inspect.lisp.
5729 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5730 feature has gone away. (When we eventually port to other flavors of
5731 libc and/or Unix, we'll try to make the differences between flavors
5732 invisible at the user level.)
5733 * Various other *FEATURES* tags, and/or their associated conditionals,
5734 have been removed if obsolescent, or given better documentation, or
5735 sometimes given more-mnemonic names.
5737 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5739 * tidied up "make.sh" script
5740 * tidied up system directory structure
5741 * better "clean.sh" behavior
5742 * added doc/FOR-CMUCL-DEVELOPERS
5743 * many many small tweaks to output format, e.g. removing possibly-confusing
5744 trailing #\. character in DESCRIBE-INSTANCE
5745 * (EQUALP #\A 'A) no longer signals an error.
5746 * new hashing code, including EQUALP hashing
5747 * tidied up Lisp initialization and toplevel
5748 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5749 * command line argument processing
5750 * added POSIX-GETENV function to deal with Unix-ish environment variables
5751 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5752 terminating SBCL on EOF
5753 * non-verbose GC by default
5754 * There is no more "sbcl" shell script; the sbcl file is now the C
5755 runtime executable (just like CMU CL).
5756 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5758 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5759 symbol and package information as Lisp native symbol and package objects
5760 instead of strings naming symbols and strings naming packages. This way,
5761 whenever packages are renamed (as in warm init), debug information is
5762 transformed along with everything else.
5763 * tweaked the optimization policy declarations which control the building
5764 of SBCL itself. Now, among other things, the system no longer saves
5765 source location debugging information. (This helps two problems at once
5766 by reducing SBCL size and by keeping SBCL from trying to look for its
5767 sources -- which may not exist -- when reporting errors.)
5768 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5769 debugging and testing purposes
5770 * cleaned up printing, making the printer call PRINT-OBJECT for
5771 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5772 methods, giving nearly-ANSI behavior
5773 * converted almost all special variables to use *FOO* naming convention
5774 * deleted PARSE-TIME functionality, since it can be done portably
5775 * moved some files out of cold init into warm init
5776 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5778 * regularized formatting of source files
5779 * added an install.sh script
5780 * fixed ridiculous memory usage of cross-compiler by making
5781 compiler/alloc.lisp not try to do pooling unless it can hook
5782 itself into the GC of the cross-compilation host. Now the system
5783 builds nicely on my old laptop.
5784 * added :SB-ALLOC in target-features.lisp-expr
5785 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5786 was not implemented)
5787 * re-did condition handling and note reporting in the compiler. Notes
5788 are no longer handled by signalling conditions. Style warnings
5789 and warnings are handled more correctly and reported in such a way
5790 that it's easy to find one or the other in your output (so that you
5791 can e.g. figure out which of many problems caused COMPILE-FILE to
5793 * changed the severity of several compiler warnings from full WARNING
5794 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5795 compiler note reporting so that it doesn't use the condition system
5796 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5797 COMPILE-FILE command)
5798 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5799 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5800 no longer supported, since their implementation was deeply intertwingled
5801 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5802 * removed bogus "support" for compiler macros named (SETF FOO), and
5803 removed the compiler macro for SETF INFO (but only after making a fool
5804 of myself on the cmucl-imp mailing list by posting a bogus patch for
5805 DEFINE-COMPILER-MACRO..)
5806 * Compiled files containing forms which have side effects on the Lisp
5807 reader (such as DEFPACKAGE forms) are now handled more correctly.
5808 (Compiler queuing of top level lambdas has been suppressed by setting
5809 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5810 * deleted various currently-unused source files, e.g. gengc.lisp. They
5811 may be added back at some point e.g. when porting to other architectures,
5812 but until they are it's distracting to distribute them and to try to
5814 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5815 there were so many of them they're just distractions, and UNCROSS is
5816 known to be able to handle the current sources
5817 * moved PROFILE functionality into TRACE, so that it will be clear
5818 how the wrapping and unwrapping of functions when you profile them
5819 interacts with the wrapping and unwrapping of functions when you
5820 trace them. (Actually, the functionality isn't there yet, but at least
5821 the interface specification is there. Hopefully, the functionality will
5822 arrive with some future maintenance release.)
5823 * removed host-oops.lisp
5824 * changed signature of QUIT function to allow UNIX-CODE argument
5825 * fixed READ-SEQUENCE bug
5826 * tweaked verbose GC output so that it looks more like the progress
5827 output that ANSI specifies for functions like LOAD
5828 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5829 * added <http://sbcl.sourceforge.com> to the banner information printed by