1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.36:
3 * enhancement: Backtrace from THROW to uncaught tag on x86oids now shows
4 stack frame thrown from.
5 * enhancement: WITH-COMPILATION-UNIT :POLICY allows restricting changes to
6 compiler optimization qualities inside dynamic extent of its body.
7 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) take advantage of
8 constraint propgation, allowing better compilation eg. when used to
9 access structures with WITH-SLOTS. (lp#520366)
10 * optimization: the compiler is now more aware of the type of the underlying
11 storage vector for multidimensional simple arrays resulting in better code
12 for accessing such arrays.
13 * optimization: passing NIL as the environment argument to TYPEP no longer
14 inhibits optimizing it. (lp#309788)
15 * optimization: ADJUST-ARRAY and STABLE-SORT on vectors no longer use
16 pre-allocated temporary vectors. (lp#496249)
17 * bug fix: Fix compiler error involving MAKE-ARRAY and IF forms
18 in :INITIAL-CONTENTS. (lp#523612)
19 * bug fix: FUNCTION-LAMBDA-EXPRESSION lost declarations from interpreted
20 functions. (lp#524707)
21 * bug fix: bogus style warnings from certain (SETF SLOT-VALUE) and
22 WITH-SLOTS usages during compilation.
23 * bug fix: SB-C::CLASS-INFO now prints correctly. (lp#514762)
24 * enhancement: Can now build with ud2 instead of int3 as trap instruction on
25 all x86oid platforms with :UD2-BREAKPOINTS target feature.
26 * bug fix: Breakpoints now work when using ud2 instead of int3 as trap
27 instruction (tested on x86oid linux with ud2-breakpoints).
28 * bug fix: slam.sh now works on win32.
29 * bug fix: better differences of numeric types. (lp#309124)
30 * bug fix: arrays declared intersection and union types can have their
31 upgraded element type derived. (lp#316078)
32 * bug fix: SB-SPROF allocation profiling for all threads failed to profile
33 threads started during profiling. (lp#472499)
34 * bug fix: SB-INTROSPECT test failure when building without SB-EVAL feature.
36 * bug fix: SB-CLTL2:DECLARATION-INFORMATION did not take
37 SB-EXT:RESTRICT-COMPILER-POLICY into account. (lp#313337)
38 * bug fix: Comma inside a backquoted array or structure resulted in nonsense
39 values instead of signaling an error. (lp#309093)
40 * bug fix: Spurious unused variale warning in a DEFSTRUCT edge case.
43 changes in sbcl-1.0.36 relative to sbcl-1.0.35:
44 * new feature: SB-EXT:TYPEXPAND-1, SB-EXT:TYPEXPAND, and
45 SB-EXT:TYPEXPAND-ALL behave exactly like their MACROEXPAND counterparts
46 but work on type specifiers.
47 * new feature: SB-EXT:DEFINED-TYPE-NAME-P returns whether a symbol is known
48 to name a type specifier.
49 * new feature: SB-EXT:VALID-TYPE-SPECIFIER-P returns whether a given type
50 specifier is valid where "valid" basically means "would be accepted as
51 second argument of TYPEP".
52 * new feature: SB-INTROSPECT:FUNCTION-TYPE takes a function-designator and
53 returns the function's declared, or derived FTYPE.
54 * new feature: SB-POSIX now supports accessing the d_ino member of
55 dirent structures. (Thanks to Philipp Marek and Pierre THEIRRY)
56 * new feature: The function SB-EXT:SEED-RANDOM-STATE has been added to
57 provide for seeding a RANDOM-STATE object with user-provided data or
58 from the operating system's PRNG. Also, (MAKE-RANDOM-STATE T) will
59 attempt to initialize the returned state from the operating system's
60 PRNG where possible. (Thanks to Fare Rideau; launchpad bug lp#310116)
61 * bug fix: Fix SB-SIMPLE-STREAMS:READ-VECTOR to correctly set the
62 FILE-POSITION of the stream being read from. (launchpad bug lp#491087)
63 * bug fix: Fix grammar and style issues for the docstrings of
64 printer-related variables and functions. (Thanks to mon_key; launchpad
66 * bug fix: Fix compilation on chenygc platforms. Thanks to Larry Valkama and
68 * bug fix: SB-THREAD:CONDITION-WAIT sometimes signaled a deadline twice
69 in a row even though a handler defered the deadline long into the
71 * bug fix: A deadline handler was run without interrupts enabled for a
72 deadline signaled within SB-THREAD:CONDITION-WAIT. That could result
73 in infinitely spinning, non-killable threads.
74 * bug fix: Backtrace from internal-errors on x86-64 os x was truncated
75 before reaching the erring stack frame.
76 * bug fix: Fix type derivation for EXPT when raising a fixnum to a
77 real power. (launchpad bug lp#525949)
78 * bug fix: Fix SB-EXT:GENERATION-* accessors for generations > 0 on
79 GENCGC platforms. (launchpad bug lp#529014)
80 * bug fix: More robust checks for invalid DEFMETHOD argument specializers.
81 (launchpad bug lp#525916)
82 * bug fix: Fix building on Darwin when sysctl is not in the user's PATH.
83 (Thanks to Robert Goldman)
85 changes in sbcl-1.0.35 relative to sbcl-1.0.34:
86 * optimization: ROUND with a single single-float or double-float argument
87 is properly inlined when possible.
88 * optimization: Slightly better code is generated for integer<->float
89 conversions and for single-float<->double-float conversions on x86-64.
90 * optimization: SB-ROTATE-BYTE:ROTATE-BYTE now generates more efficient
91 code for 32-bit and 64-bit rotations on x86-64.
92 * bug fix: The install script changes the ownership of directories as well
93 as files for contrib modules using asdf. (thanks to Eugene Ossintsev;
94 launchpad bug lp#508485)
95 * bug fix: TRUNCATE with a single single-float or double-float argument is
96 properly inlined when possible. (launchpad bug lp#489388)
97 * bug fix: Passing a rotation count of zero to SB-ROTATE-BYTE:ROTATE-BYTE
98 no longer causes a compiler error on x86 and ppc.
99 * bug fix: GET-MACRO-CHARACTER bogusly computed its second return value
100 always relative to *READTABLE* rather than the passed argument.
102 changes in sbcl-1.0.34 relative to sbcl-1.0.33:
103 * minor incompatible change: threading support is now enabled by default
105 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
107 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
108 are now caught and reported just like errors during macroexpansion.
109 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
110 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
111 * enhancement: ASDF systems can now depends on SB-QUEUE.
112 * fixes and improvements related to Unicode and external formats:
113 ** bug fix: error handling and restart usage in the ucs-2 external format
115 ** there is now an implementation of the ucs-4 external format.
116 ** the utf-16 and utf-32 external formats are supported.
117 * bug fix: SB-POSIX wrapper for putenv no longer tries to put lisp strings
118 in the environment. setenv() and unsetenv() are also provided. (reported by
119 Fare Rideau; launchpad bug lp#460455)
120 * bug fix: LOAD of both .fasl and .FASL type files now forces fasl-style
121 loading. This ensures sensible errors for .FASL files from other
122 implementations on case-insensitive filesystems. (reported by Willem
123 Broekema; launchpad bug lp#489417)
124 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
125 Willem Broekema; launchpad bug lp#489698).
126 * bug fix: some minor code rearrangements to reenable warning-free building
127 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
128 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
130 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
133 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
134 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
135 * improvement: support O_LARGEFILE access to files larger than 2GB on
136 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
137 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
138 definitions for methods specializing on the passed class itself.
139 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
140 definitions for methods specializing on the passed class itself, or on
142 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
143 about itself during the build (e.g. for M-? in Slime), if this flag is
144 enabled in customize-target-features.lisp. This will increase the core
145 size by about 5-6mb, though, so it's mostly interesting to SBCL
147 * new feature: various GENCGC tuning parameters have been experimentally
148 documented and exported from SB-EXT. See documentation for details.
149 * fixes and improvements related to Unicode and external formats:
150 ** the Unicode character database has been upgraded to the
151 Unicode 5.2 standard, giving names and properties to a number of new
152 characters, and providing a few extra characters with case
154 ** improvement: restarts for providing replacement input/output on coding
155 errors for fd-stream external formats.
156 ** improvement: where :<encoding> is a keyword corresponding to an
157 external format the system supports, it is now possible to specify
158 (:<encoding> :replacement <character>) as an external format which will
159 automatically substitute <character> on encoding or decoding errors for
160 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
162 ** improvement: the file streams underlying the standard streams (such as
163 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
164 which uses the replacement mechanism to handle encoding errors,
165 preventing various infinite error chains and unrecoverable I/O
167 ** minor incompatible change: the utf-8 external format now correctly
168 refuses to encode Lisp characters in the surrogate range (char-codes
169 between #xd800 and #xdfff).
170 ** fix a typo preventing conversion of strings into octet vectors
171 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
173 ** fix a bug in the octet multibyte handling of decoding errors and the
174 USE-VALUE restart. (launchpad bug lp#314939)
175 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
176 restart on fd-stream decoding errors.
177 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
178 error is near the end of file.
179 ** fix a double-error case in unibyte octet conversions, when the first
180 use of USE-VALUE is ignored.
181 ** fix bugs in handling of undefined code points in unibyte encodings.
182 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
183 streams after an UNREAD-CHAR.
184 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
185 object is allocated in a boxed region of dynamic space.
186 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
187 fork with multiple Lisp threads running is made, instead of going
188 ahead with unpredictable consequences. (reported by Leslie Polzer)
189 * bug fix: uses of slot accessors on specialized method parameters within
190 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
191 error while finalizing the class. This fix may cause classes with slot
192 accessors to be finalized later than previously. (reported by Lars Rune
193 Nøstdal; launchpad bug lp#473699)
194 * bug fix: restore buildability on the MIPS platform. (regression from
195 1.0.30.38, reported by Samium Gromoff)
196 * bug fix: inspecting closures is less likely to fail with a type error.
197 * bug fix: no timer starvation when setting the system clock back.
198 (launchpad bug lp#460283)
199 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
200 standard pprint dispatch table as specified by CLHS.
201 * bug fix: give CLISP a hint about a type declaration to enable it to build
202 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
204 * bug fix: correctly dump literal objects in defaulting forms of arglists.
205 (reported by Attila Lendvai; launchpad bug lp#310132)
206 * bug fix: distinguish in type specifiers between arrays that might be
207 complex and arrays that are definitely complex. (launchpad bug lp#309129)
208 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
209 type. (reported by Levente Mészáros; launchpad bug lp#485972)
210 * bug fix: setting the value of a symbol-macro within a method in the
211 presence of type declarations works properly again. (reported by Iban
212 Hatchondo; launchpad bug lp#485019)
214 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
215 * optimization: faster FIND and POSITION on strings of unknown element type
216 in high SPEED policies. (thanks to Karol Swietlicki)
217 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
219 * improvement: better error signalling for bogus parameter specializer names
220 in DEFMETHOD forms (reported by Pluijzer)
221 * bug fix: DELETE-FILE once again works on logical pathnames (regression
223 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
225 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
226 removes previous default initargs (reported by Lars Rune Nøstdal and
228 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
229 lambda-lists. (reported by Peter Seibel)
230 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
231 correctly when starting from an executable core without saved runtime
232 options (reported by Faré Rideau, thanks to Zach Beane)
233 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
234 an optimization note instead. (reported by Martin Cracauer)
235 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
236 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
237 new process group if it doesn't need to share stdin with the sbcl
238 process. (thanks to Leslie Polzer)
239 * bug fix: SATISFIES could be misoptimized to refer to a local function.
240 (reported by Stanislaw Halik)
242 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
243 * improvement: stack allocation is should now be possible in all nested
244 inlining cases: failure to stack allocate when equivalent code is manually
245 open coded is now considered a bug.
246 * improvements related to Unicode:
247 ** the Unicode character database has been upgraded to the
248 Unicode 5.1 standard, giving names and properties to a number of new
249 characters, and providing a few extra characters with case
251 ** the system now recognizes and produces names for Unicode Hangul
253 ** the EBCDIC-US external-format is now supported for octet operations
254 (as well as for stream operations).
255 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
256 be used to output toplevel compile-time effects into a separate .CFASL
258 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
259 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
261 * optimization: TRUNCATE on known single- and double-floats is upto 25%
263 * optimization: division of floating point numbers by constants uses
264 multiplication by reciprocal when an exact reciprocal exists.
265 * optimization: multiplication of single- and double-floats floats by
266 constant two has been optimized.
267 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
268 sufficient type information is available. (thanks to Leslie Polzer)
269 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
270 known structure objects are as efficient as defstruct generated accessors.
271 * optimization: unused vector creation can now be optimized away.
272 * improvement: ASDF systems can now depends on SB-INTROSPECT.
273 * improvement: a STYLE-WARNING is signalled when a generic function
274 clobbers an earlier FTYPE proclamation.
275 * improvement: the compiler is able to track the effective type of
276 generic function across method addition and removal even in the
277 absence of an explicit DEFGENERIC.
278 * improvement: DESCRIBE now reports on symbols naming undefined
279 but assumed or declared function as well.
280 * improvement: recompilation of systems using SB-GROVEL now works
281 (thanks to Leslie Polzer)
282 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
283 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
285 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
286 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
287 well as user defined declaration names.
288 ** VARIABLE-INFORMATION is now aware of alien variables.
289 * improvement: improved address space layout on OpenBSD (thanks to Josh
291 * improvement: pretty-printing of various Lisp forms has been improved
292 (thanks to Tobias Rittweiler)
293 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
294 unused were deleted in safe code. (reported by John Fremlin)
295 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
297 * bug fix: some out-of-line array predicates were missing (reported by
299 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
301 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
302 (thanks to Larry D'Anna)
303 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
304 as well. (reported by Josh Elsasser)
305 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
306 leading to possible attempts to delete the same file twice. See docstring
307 on DELETE-FILE for details. (reported by John Fremlin)
308 * bug fix: DELETE-FILE once again deletes the file named by the pathname
309 designator argument, rather than its truename. (reported by Luis
311 * bug fix: the low-level debugger had 32-bit assumptions and was missing
312 information about some array types. (thanks to Luis Oliveira)
313 * bug fix: moderately complex combinations of inline expansions could
314 be miscompiled if the result was declared to be dynamic extent.
315 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
316 form (+ <variable> <integer>) were miscompiled under certain
318 * bug fix: in some cases no compiler note about failure to stack allocate
319 was emitted, even if the objects were in fact heap allocated.
320 * bug fix: minor violation of "otherwise inaccessible" rule for stack
321 allocation could cause objects users might reasonably expect to
322 be heap allocated to be stack allocated.
323 * bug fix: DESCRIBE signalled an error for generic functions under
324 certain circumstances. (thanks to Leslie Polzer)
325 * bug fix: Fixed spelling of an error message.
327 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
328 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
329 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
330 of SB-THREAD:THREAD-ERROR-THREAD.
331 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
332 * new feature: docstrings for local and anonymous functions are no longer
333 discarded. (thanks to Leslie Polzer)
334 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
335 values in other threads.
336 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
337 about object allocation.
338 * optimization: division of a real float by a complex float is implemented
339 with a specialised code sequence.
340 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
341 keywords is an order of magnitude faster.
342 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
343 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
344 methods, and similarly for non-standard metaclass classes as long as there
345 are no methods additional on MAKE-INSTANCE.
346 * optimization: more efficient type-checks for FIXNUMs when the value
347 is known to be a signed word on x86 and x86-64.
348 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
349 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
351 * optimization: compiler is smarter about delegating argument type checks to
353 * optimization: several character functions are now compiled somewhat more
354 efficiently. (reported by Lynn Quam)
355 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
356 * improvement: less unsafe constant folding in floating point arithmetic,
357 especially for mixed complex/real -float operations.
358 * optimization: constant double and single floats are stored in native
359 unboxed format on x86[-64].
360 * optimization: smarter code for arithmetic operations with constant floats,
361 complex floats, or integers on x86[-64].
362 * optimization: smarter code for conjugate/multiplication of float complexes
363 and abs/negate of floats on x86-64.
364 * optimization: more efficient complex float and real float operations on
366 * improvement: complex float division is slightly more stable.
367 * improvement: DESCRIBE output has been reworked to be easier to read and
368 contains more pertinent information.
369 * improvement: failure to provide requested stack allocation compiler notes
370 provided in all cases (requested stack allocation not happening without a
371 note being issued is now considered a bug.)
372 * bug fix: SB-POSIX exports the documented types and functions
373 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
374 types. (reported by "abhi")
375 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
376 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
377 (thanks to Paul Khuong)
378 * bug fix: looping around HANDLER-CASE could silently consume stack space
379 on each iteration. (reported by "foobar")
380 * bug fix: better error signalling when calls to functions seeking elements
381 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
382 (reported by Tobias Rittweiler)
383 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
384 directory components now works as it used to. (various prolems reported by
385 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
386 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
387 (reported by Elliot Slaughter, thanks to Stas Boukarev)
388 * bug fix: bogus undefined variable warnings from fopcompiled references to
389 global variables. (thanks to Lars Rune Nøstdal)
390 * bug fix: foreign function names should now appear in backtraces on
391 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
392 * bug fix: SETF compiler macro documentation strings are not discarded
394 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
395 (thanks to Alex Plotnick)
396 * bug fix: SLEEP accepts large integer arguments, truncating them to
397 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
398 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
399 * bug fix: compiler notes for expensive slot type checks could be emitted
400 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
402 changes in sbcl-1.0.29 relative to 1.0.28:
403 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
404 https://bugs.launchpad.net/sbcl
405 Bugs can be reported directly there, or by sending email to
406 sbcl-bugs@lists.sourceforge.net
407 (no subscription required.)
408 * minor incompatible change: under weak type checking policy integer
409 types are weakened less aggressively.
410 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
411 allowed to return, which causes SBCL to quit with exit status 0. Previously
412 if the function returned with a small integer return value, that value
413 was accidentally reused as the exit status.
414 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
415 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
416 :HASH-FUNCTION argument. Refer to user manual for details.
417 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
419 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
420 microseconds since the Unix epoch on all platforms.
421 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
422 allows the compiler to safely elide boundedness checks for special
424 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
425 the symbol, prohibits both lexical and dynamic binding. This is mainly an
426 efficiency measure for threaded platforms, but also valueable in
428 * new feature: UNC pathnames are now understood by the system on Windows.
429 * optimization: the compiler uses a specialized version of FILL when the
430 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
432 * optimization: accesses to potentially non-simple arrays where element type
433 is known are 50% faster.
434 * optimization: compiler now generates faster array typechecking code.
435 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
437 * optimization: multidimensional array accesses in the absence of type
438 information regarding array rank are approximately 10% faster due to
439 open coding of ARRAY-RANK.
440 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
441 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
443 * optimization: result of call to VECTOR can now be stack allocated.
444 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
445 as long as the resulting array is one-dimensional and has a known
446 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
447 length of the list matches the known length of the vector does not
448 allocate the list as an intermediate step. Ditto for VECTOR and simple
450 * optimization: MAKE-ARRAY can now stack allocate in the presence of
451 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
452 known element type, and is known to be simple and one dimensional.
453 * improvement: SBCL now emits a compiler note where stack allocation was
454 requested but could not be provided (not in all cases, unfortunately)
455 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
456 * improvement: pretty-printing loop has been implemented properly. (thanks
457 to Tobias Rittweiler)
458 * documentation: CLOS slot typechecing policy has been documented.
459 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
460 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
461 pointer, leading to memory faults. (thanks to Bart Botta)
462 * bug fix: better floating point exception handling on x86/OpenBSD.
463 (thanks to Josh Elsasser)
464 * bug fix: exit status from QUIT when called under --script was lost
465 (reported by Hubert Kauker)
466 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
467 same implementation of FILL to initialize the array, even if a more
468 efficient one was available (reported by Stas Boukarev, thanks to
470 * bug fix: potential miscompilation of array stack allocation on x86 and
471 x86-64. (reported by Time Tossavainen)
472 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
473 result in their subforms being treated as top level forms. (reported by
475 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
476 than a machine register could leave garbage in the high bits of the
477 result register (bug 316325).
478 * bug fix: disable address space randomization Linux/x86-64 as well,
479 not just x86-64. (reported by Ken Olum)
480 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
481 generate incorrect code.
482 * bug fix: #201; type inference for CONS and ARRAY types could derive
483 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
484 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
485 could not be assigned to or bound with PROGV. (reported by Lorenz
487 * bug fix: the value of CL:- in the inspector was the previous expression
488 evaluated rather than the expression being evaluated.
489 * bug fix: constants can no longer be locally declared special.
490 * bug fix: signals delivered to threads started from foreign land (read:
491 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
492 thread by blocking all signals and resignalling.
493 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
494 from :INITFORM, if any.
496 changes in sbcl-1.0.28 relative to 1.0.27:
497 * a number of bugs in cross-compilation have been fixed, with the ultimate
498 result that building under (at least) clisp should be much more reliable.
499 * minor incompatible changes: echo-streams now propagate unread-char to the
500 underlying input stream, and no longer permit unreading more than one
502 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
503 frames, allowing for instance more reliable backtraces.
504 * improvement: the debugger REPL can now reference lexical variables
505 by name directly for code compiled with (DEBUG 3).
506 * improvement: errors from malformed declarations now have better source
507 paths associated with them. (thanks to Tobias Rittweiler)
508 * optimization: faster local calls on x86/x86-64
509 * bug fix: some error messages for out-of-bound array indexes confused the
510 index and the bound. (thanks to Stas Boukarev)
511 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
513 * bug fix: running regressions tests in shells without OSTYPE set now works.
514 (reported by Harald Hanche-Olsen)
515 * bug fix: more robust static space exhaustion signalling from
516 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
517 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
518 docstring away instead of storing it under names such as (LAMBDA (X)).
519 (reported by Leslie Polzer)
520 * bug fix: timers could go off in the wrong order, be delayed indefinitely
521 (thanks to Ole Arndt for the patch)
522 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
523 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
524 potentially-invalid effective methods in its cache.
525 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
526 instances as well (thanks to Paul Khuong)
527 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
528 SB-POSIX:WAITPID and their C equivalents.
529 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
531 changes in sbcl-1.0.27 relative to 1.0.26:
532 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
533 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
534 * improvement: the system either recovers from stack exhaustion or dies
535 properly as opposed to leaving the user uncertain of whether the handler
536 trampled on some random memory next to the stack or having to rely on
537 --lose-on-corruption (which is still a good idea to use in production
538 because stack exhaustion can happen in signal handlers which will likely
540 * bug fix: fix gc related interrupt handling bug on ppc (regression from
541 1.0.25.37, reported by Harald Hanche-Olsen)
542 * bug fix: work around signal delivery bug in darwin (regression from
543 1.0.25.44, reported by Sidney Markowitz)
544 * bug fix: fix ERROR leaking memory (reported by David Thompson)
546 changes in sbcl-1.0.26 relative to 1.0.25:
547 * incompatible change: an interruption (be it a function passed to
548 INTERRUPT-THREAD or a timer function) runs in an environment where
549 interrupts can be enabled. The interruption can use
550 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
551 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
552 potentially running out of stack. Keep in mind that in the absence
553 of WITHOUT-INTERRUPTS some potentially blocking operation such as
554 acquiring a lock can enable interrupts.
555 * incompatible change: GC-OFF and GC-ON are removed, as they were
556 always unsafe. Use WITHOUT-GCING instead.
557 * new feature: runtime option --disable-ldb
558 * new feature: runtime option --lose-on-corruption to die at the
559 slightest hint of possibly non-recoverable errors: running out of
560 memory, stack, alien stack, binding stack, encountering a memory
561 fault, etc. In the absence of --lose-on-corruption a warning is
563 * enhancement: detect binding stack exhaustion
564 * enhancement: detect alien stack exhaustion on x86/x86-64
565 * improvement: generally more stable and reliable interrupt handling
566 * improvement: there is a per thread interruption queue,
567 interruptions are executed in order of arrival
568 * improvement: a repeating timer reschedules itself when the it has
569 finished, but expiration times are spaced equally. If an
570 expiration time is in the past it will trigger after a short grace
571 period that may give a chance to other things to run.
572 * optimization: slightly faster gc on multithreaded builds
573 * optimization: faster WITHOUT-GCING
574 * bug fix: when JOIN-THREAD signals an error, do it when not holding
575 important locks so that the debugger/handler doesn't produce
576 recursive errors or deadlock.
577 * bug fix: real-time signals are not used anymore, so no more
578 hanging when the system wide real-time signal queue gets full.
579 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
581 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
582 * bug fix: fix random memory faults related to interrupts on alpha
583 * bug fix: fix random memory faults related to interrupts on sparc
584 * bug fix: fix deadlocks related to starting threads
585 * bug fix: fix deadlines on locks on futex platforms
586 * bug fix: restore errno in signal handlers
587 * bug fix: fix deadlocks related to hash tables
588 * bug fix: fix deadlocks in pcl
590 changes in sbcl-1.0.25 relative to 1.0.24:
591 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
592 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
593 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
594 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
595 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
596 better name and does not return values so stale on multiprocessor systems.
597 Also, HOLDING-MUTEX-P was added for about the only sane usage of
599 * improvement: unithread builds keep track of MUTEX-VALUE.
600 * improvement: reading from a TWO-WAY-STREAM does not touch the output
601 stream anymore making it thread safe to have a concurrent reader and
602 a writer, for instance, in a pipe.
603 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
604 arguments, making compiler-macros for SETF-functions able to inspect
605 their constant arguments.
606 * improvement: COMPILE-FILE reports times with millisecond accuracy
607 (thanks to Luis Oliveira)
608 * optimization: CHAR-CODE type derivation has been improved, making
609 TYPEP elimination on subtypes of CHARACTER work better. (reported
610 by Tobias Rittweiler, patch by Paul Khuong)
611 * bug fix: setting alien structure fields of type struct by value now
612 computes the right offset for the memory copy.
613 * bug fix: compilation problem involving inlined calls to aliens with
614 result type VOID. (reported by Ken Olum)
615 * bug fix: #235a; sequential inline expasion in different policies no
616 longer reuses the functional from the previous expansion site.
617 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
618 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
619 * bug fix: named ENUMs in multiply-referenced alien record types no longer
620 cause an implied type redefinition. Regression from 1.0.21.29.
621 * bug fix: sign-extension of small signed return values now works with C
622 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
623 * improvements to the Windows port:
624 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
625 Westfold, thanks to Rudi Schlatte)
627 changes in sbcl-1.0.24 relative to 1.0.23:
628 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
629 vector of a multidimensional SIMPLE-ARRAY.
630 * new feature: the system now signals a continuable error if standard
631 readtable modification is attempted. (thanks to Tobias Rittweiler)
632 * new feature: DIRECTORY has been extended with a non-standard keyword
633 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
634 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
635 has been robustified and documented better. (thanks to Robert Goldman)
636 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
637 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
638 * tradeoff: constant FORMAT control strings are now compiled unless
639 SPACE > SPEED (previously only when SPEED > SPACE.)
640 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
642 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
644 * bug fix: lexical type declarations are now correctly reported by
645 SB-CLTL2. (reported by Larry D'Anna)
646 * bug fix: STRING-TO-OCTETS did not handle :START properly when
647 using UTF-8 as external format. (thanks to Luis Oliveira)
648 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
649 are signalled correctly. (thanks to Stas Boukarev)
650 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
651 designator. (thanks to Tobias Rittweiler)
652 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
653 designator, and returns T instead of the function. (thanks to
655 * bug fix: direct superclasses of STANDARD-CLASS and
656 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
657 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
658 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
659 when no class with the named slot yet exists no longer causes a
660 compile-time style-warning.
661 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
662 in safe code. (reported by Didier Verna)
663 * bug fix: #430; nested structure constructors can stack allocate.
664 * bug fix: on some 64-bit platforms dynamic space size was truncated
665 to #xffff0000 bytes. (reported by Benjamin Lambert)
666 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
667 REPL unusable. (reported by Daniel Herring)
668 * bug fix: return values of READ-SEQUENCE did not take :START into
669 account on file streams, regressions since 1.0.12.22. (reported by
670 Thomas Russ, patch by Paul Khuong)
671 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
672 method specializer used to confuse permuation vector optimization.
673 * bug fix: system inserted bogus implicit type declarations for local
674 special variables in DEFMETHOD bodies.
675 * bug fix: #354; duplicated frames in backtraces due to
676 non-tail-call-optimized XEPs to functions with return type NIL
678 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
679 initializes structure object slots according to DEFSTRUCT initforms,
680 and DEFSTRUCT forms :INCLUDEind structure classes defined using
681 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
682 (reported by Bruno Haible and Stephen Wilson)
683 * bug fix: #395; fill-pointer output streams used now support
684 element-type BASE-CHAR as well.
685 * bug fix: compiler error when attempting to derive return value of
686 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
688 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
690 * bug fix: Mac OS X binaries should now be portable between Leopard
693 changes in sbcl-1.0.23 relative to 1.0.22:
694 * enhancement: when disassembling method functions, disassembly
695 for the associated fast function is also produced.
696 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
698 * optimization: printing with *PRINT-PRETTY* true is now more
699 efficient as long as the object being printed doesn't require
700 special handling by the pretty printer.
701 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
702 now interact correctly with type declarations.
703 * partial bug fix: PCL detects infinite recursion during wrapper
704 validation. (thanks to Attila Lendvai)
705 * bug fix: #426; nested function calls are inlined properly.
706 Previously if FOO was an inline function, in calls of the form
707 (FOO (FOO ...)) the outer call was not inlined.
708 * bug fix: long long arguments passed in registers to alien
709 functions on PowerPC Linux and NetBSD are now handled correctly
710 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
712 changes in sbcl-1.0.22 relative to 1.0.21:
713 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
714 for the shared object in the current directory, but passes the native
715 namestring of the designated pathname to the operation system's shared
716 object loading function as-it.
717 * minor incompatible change: --disable-debugger toplevel option now takes
718 effect before processing of initialization files and --eval or --load
720 * new feature: new commandline argument: --script, which supports
721 shebang lines. See documentation for details. (based on work by
723 * new feature: SAVE-LISP-AND-DIE can save current values of
724 --dynamic-space-size and --control-stack-size in the executable core,
725 causing it to skip normal runtime option processing. See documentation
726 for details. (thanks to Zach Beane)
727 * enhancement: inoccous calls to EVAL or generic functions dispatching
728 on subclasses of eg. STREAM no longer cause compiler notes to appear.
729 * enhancement: the system no longer resignals errors from --load and
730 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
731 associated with the original error to be lost. (thanks to Ariel
733 * enhancement: :DONT-SAVE keyword argument has been added to
734 LOAD-SHARED-OBJECT for controlling interaction with
735 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
736 the effects of an earlier LOAD-SHARED-OBJECT call.
737 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
738 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
739 provided. (reported by Cedric St-Jean)
740 * bug fix: circularity handling in the reader did not treat raw
741 structure slots correctly. (reported by Cedric St-Jean)
742 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
743 file descriptors when there were none.
744 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
745 was broken. (reporteed by Willem Broekema)
746 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
747 pathnames without a directory.
748 * bug fix: COMPILE-FILE did not warn about undefined variable
749 references at toplevel, and LOAD of the resulting fasl did
751 * bug fix: functions with non-required arguments used to end up with
752 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
753 * bug fix: redefining a function with non-required arguments didn't
754 update the system's knowledge about its call signature properly.
755 * bug fix: fixed #431; incompatible alien record type redefinitions
756 are detected and handled. (reported by Neil Haven)
757 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
758 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
760 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
761 after alien stack frames.
762 * bug fix: strings passed to mk{s,d}temp lacked null termination
764 changes in sbcl-1.0.21 relative to 1.0.20:
765 * new feature: the compiler is able to track the effective type of a
766 generic function across method addition and removal.
767 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
768 appropriately typed structure slots without locking.
769 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
770 information like those gathered by TIME using a programming-friendly
772 * new feature: TIME reports time taken even if the form performs a
773 non-local transfer of control.
774 * enhancement: reduced conservativism on GENCGC platforms: on
775 average 45% less pages pinned (measured from SBCL self build).
776 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
777 mutate constant symbols or violate declaimed type of the symbol.
778 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
779 sleeping on the semaphore in heavily contested semaphores.
780 * bug fix: semaphores and condition variables were not interrupt
782 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
783 owned by other threads anymore.
784 * bug fix: FIND on lists called KEY outside the specified
785 subsequence. (reported by budden)
786 * bug fix: LOG doesn't use single-float intermediate results when
787 given mixed integer and double-float arguments, leading to better
788 precision. (reported by Bob Felts)
789 * bug fix: LOG with base zero returned values of inconsistent type.
791 changes in sbcl-1.0.20 relative to 1.0.19:
792 * minor incompatible change: OPTIMIZE qualities
793 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
794 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
795 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
796 * documentation: some slot access efficiency guidelines have been
797 added to the user manual.
798 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
799 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
800 as ASSOC and MEMEBER.
801 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
802 to more efficient EQ-comparison versions more often.
803 * optimization: enhanced derivation of DOLIST iteration variable type
805 * optimization: constant folding of simple (LIST ...) forms as DOLIST
807 * optimization: runtime lookup of function definitions can be
808 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
809 * optimization: compiler is able to derive the return type of
810 (AREF (THE STRING X) Y) as being CHARACTER.
811 * optimization: CLRHASH on empty hash-tables no longer does pointless
812 work. (thanks to Alec Berryman)
813 * optimization: REPLACE deftransforms don't punt when :START1 or
815 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
816 no longer emits a code deletion note. (reported by Andrew
818 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
819 functions with an odd number of &OPTIONAL arguments, a &REST
820 argument, and one or more &KEY arguments at the call site.
821 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
822 arguments appeared at call sites as well.
823 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
824 READ-LINE &co can no longer read from them afterwards. (reported
826 * bug fix: fixed #427: unused local aliens no longer cause compiler
827 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
829 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
830 alien-stack leakage. (reported by Andy Hefner)
831 * bug fix: PROGV signals an error when an attempt to violate declared
832 type of a variable or bind a constant is made.
833 * bug fix: SET signals an error when an attempt to violate declared
834 type of a variable is made.
835 * bug fix: restart computation during the execution of a restart
836 test function no longer causes infinite recursion. (reported by
838 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
839 signals an error. (thanks to Pierre Mai)
840 * bug fix: EXPORT left symbol unexported in conflict situations.
841 (thanks to Michael Weber)
842 * bug fix: correctly handle name conflicts involving CL:NIL.
843 (thanks to Michael Weber)
844 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
845 conflicts arising from USEing package with conflicting symbols
846 correctly. (thanks to Michael Weber)
848 changes in sbcl-1.0.19 relative to 1.0.18:
849 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
850 warnings that go otherwise unhandled will be muffled if they are
851 of the type that's the value of this variable.
852 * optimization: stack allocation is slightly more efficient on x86
854 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
855 where a raw slot always is initialized using the initform whose
856 type is not know sufficiently well a compile-time are now compiled
857 correctly. (reported by John Morrison)
858 * bug fix: compiler no longer makes erronous assumptions in the
859 presence of non-foldable SATISFIES types.
860 * bug fix: stack analysis missed cleanups of dynamic-extent
861 arguments in non-let-converted calls to local functions.
862 * improvements to the Windows port:
863 ** adjusted address spaces for building on both Win32 and
864 Win64. (thanks for John Connors)
865 * fixed some bugs revealed by Paul Dietz' test suite:
866 ** interval arithmetic during type derivation used inexact integer
867 to single-float coercions.
868 ** arithmetic operations involving large integers and single
869 floats give the same results in compiled and interpreted code.
870 ** deriving the result type of COERCE no longer signals an error
871 if the derived type of the second argument is a MEMBER type
872 containing invalid type specifiers.
873 ** ADJOIN with constant NIL as second argument works correctly.
874 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
876 changes in sbcl-1.0.18 relative to 1.0.17:
877 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
878 profiles only the current thread.
879 * minor incompatible change: changes to SYMBOL-VALUE of constants
880 defined with DEFCONSTANT now signal an error.
881 * enhancement: SB-SPROF now has support for wallclock profiling,
882 and is also able to profile specific threads. REPORT output
883 has also additional sorting options.
884 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
886 * optimization: structure allocation has been improved
887 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
888 ** out of line constructors are ~10% faster.
889 ** inline constructors are ~15% faster.
890 ** inline constructors are capable of dynamic extent allocation
891 (generally on x86 and x86-64, in some cases on other platforms
893 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
895 * optimization: file compiler is now able to coalesce non-circular
896 lists, non-base strings, and bit-vectors. Additionally, constants
897 are never referenced through SYMBOL-VALUE at runtime.
898 * optimization: code defining methods on PRINT-OBJECT (and other
899 generic functions in the COMMON-LISP package) now loads faster.
900 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
902 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
903 fasl is now deleted. (reported by Attila Lendvai)
904 * bug fix: READ-LINE always returned NIL for the last line in files.
905 (reported by Yoshinori Tahara)
906 * bug fix: more accurate disassembly annotations of foreign function
907 calls. (thanks to Andy Hefner)
908 * bug fix: trimming non-simple strings and non-string string
909 designators when the there is nothing to trim works properly.
910 (thanks to James Knight)
911 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
912 (thanks to Travis Cross)
913 * fixed some bugs revealed by Paul Dietz' test suite:
914 ** NIL is a valid function name (regression at 1.0.13.38)
915 ** FILL on lists was missing its return value (regression at 1.0.12.27)
916 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
917 fill pointers properly (regression at 1.0.12.23)
918 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
919 displacement indices properly (regression at 1.0.12.23)
921 changes in sbcl-1.0.17 relative to 1.0.16:
922 * temporary regression: user code can no longer allocate closure
923 variable storage on stack, due to bug 419 without explicitly
924 requesting it. Please consult sbcl-devel for advice if you need to
925 use this feature in the meanwhile.
926 * new feature: runtime argument --control-stack-size can be used to
927 adjust thread default control stack size.
928 * enhancement: improved TIME output
929 ** all times are reported using the measured accuracy (milliseconds
930 for real and GC times, microseconds for everything else.)
931 ** processor cycle counts on x86 and x86-64.
932 ** interpreted forms are counted for both evaluator modes.
933 ** number of lambdas converted by the compiler is reported.
934 ** CPU percentage report (computed from real and total run time.)
935 ** more comprehensive run time reporting, using a condenced format
936 ** interperted form, lambda, and page fault counts are omitted
938 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
940 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
941 * optimization: two argument forms of LAST are upto ~10% faster
942 in normal SPEED policies.
943 * optimization: NCONC no longer needs to heap cons its &REST list
944 in normal SPEED policies.
945 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
947 * bug fix: UNION and NUNION work with :TEST-NOT once more,
948 regression since 1.0.9.1. (thanks to Eric Marsden)
949 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
950 since 1.0.15.36. (thanks to Paul Khuong)
951 * bug fix: LAST when always returned the whole list when given a bignum
952 as the second argument.
953 * bug fix: dynamic extent allocation of nested lists and vectors
954 could leak to otherwise accessible parts.
955 * bug fix: invalid optimization of heap-allocated alien variable
957 * bug fix: fasl header checking is less vulnerable to different
958 platform word lengths.
959 * bug fix: more correct assembler syntax for GNU binutils
960 2.18.50.0.4 support. (thanks to Marijn Schouten)
961 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
963 * bug fix: the fopcompiler can handle LOCALLY forms (with no
964 declarations) successfully. (reported by Attila Lendvai)
966 changes in sbcl-1.0.16 relative to 1.0.15:
967 * minor incompatible change: revert the changes to sb-posix's error
968 signaling added in 1.0.14.
969 * minor incompatible change: change PROBE-FILE back to returning
970 NIL whenever we can't get a truename, as was the case before 1.0.14.
971 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
972 inherits from ERROR instead of just CONDITION.
973 * new feature: SB-INTROSPECT can provide source locations for instances
974 as well. (thanks to Tobias Rittweiler)
975 * optimization: binding special variables now generates smaller code
976 on threaded platforms.
977 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
978 and cases where no :TEST is given but the compiler can infer that
979 the element to search is of type (OR FIXNUM (NOT NUMBER)).
980 * optimization: better LOGNOT on fixnums.
981 * optimization: modular arithmetic for a particular requested width
982 is implemented using a tagged representation unless a better
983 representation is available.
984 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
985 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
986 for wellformedness and returns a specialized vector. (reported by
987 Francois-Rene Rideau)
988 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
989 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
990 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
991 unscheduling timer at the same time as another timer fires.
992 * bug fix: don't reschedule timers for dead threads.
993 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
994 * bug fix: copying output from RUN-PROGRAM to a stream signalled
995 bogus errors if select() was interrupted.
996 * enhancement: add support for fcntl's struct flock to SB-POSIX.
998 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
999 * enhancement: cleaner backtraces for interactive interrupts, as
1000 well as other cases where the interesting frames used to be
1001 obscured by interrupt handling frames.
1002 * enhancement: untracing a whole package using (UNTRACE "FOO") is
1003 now supported, and tracing a whole package using (TRACE "FOO") now
1004 traces SETF-functions as well.
1005 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
1006 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
1007 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
1008 * unreadably printed representation of hash-tables now includes
1010 * bug fix: partially fixed #188: type propagation from assignments
1011 is now more efficient.
1012 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
1013 'DOUBLE-FLOAT) are not flushable.
1014 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
1015 full address of the object, and none of the tag bits.
1016 * bug fix: readably printing hash-tables now respects other printer
1017 control variables. (reported by Cedric St-Jean)
1018 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
1019 keyword with MAKE-HASH-TABLE.
1020 * bug fix: export SB-POSIX:MKSTEMP.
1021 * bug fix: SORT was not interrupt safe.
1022 * bug fix: XREF accounts for the last node of each basic-block as
1024 * bug fix: MAKE-INSTANCE optimizations interacted badly with
1025 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
1026 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
1027 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
1028 create obsolete instances in the case of redefinition or
1029 obsoletion of a superclass. (thanks to Andy Hefner)
1030 * bug fix: Support for the Alpha architecture has been revived; it had
1031 suffered somewhat from lack of maintenance since sbcl-1.0.
1032 * improvements to the Windows port:
1033 ** The system detects the codepage to use at startup, instead of
1034 using the value from the saved core. (thanks to Kei Suzuki)
1036 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
1037 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
1038 (see documentation for details.)
1039 * revived support for OpenBSD (contributed by Josh Elsasser)
1040 * partially fixed bug #108: ROOM no longer suffers from occasional
1041 (AVER (SAP= CURRENT END)) failures.
1042 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
1043 bodies no longer cause a WARNING to be signalled. (reported by
1045 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
1046 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
1047 the bug in TRUNCATE.) (reported by Michael Weber)
1048 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
1049 no samples. (reported by Andy Hefner)
1050 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
1051 no longer appear as (NIL ...) frames in backtraces.
1052 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
1053 is now actually exported from SB-EXT as documented. (reported by
1055 * bug fix: sb-aclrepl now correctly understands how to inspect
1056 single-floats on 64-bit platforms where single-floats are not boxed.
1057 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
1058 yet been finalized. (reported by Levente Meszaros)
1059 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
1060 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
1061 SBCL-specific optimize qualities.
1063 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
1064 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
1065 an executable in the search path, and does so in the child
1066 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
1067 been removed; it can be found in the CVS history, for somebody who
1068 needs that search behavior (see the manual).
1069 * minor incompatible change: compiler policy re. weakening type
1070 checks has changed: now type checks are weakened only if SAFETY < 2
1072 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
1073 unparsing of directory pathnames as files. Analogously,
1074 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
1075 filename to parse into a directory pathname.
1076 * enhancement: implicit generic function creation now signals a
1077 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
1078 users can bind handlers for and muffle around calls to LOAD.
1079 * enhancement: RUN-PROGRAM allows unicode arguments and environments
1080 to be used (using the default stream external format), and allows
1081 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
1082 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
1083 strings and vectors whose element-type or simplicity is not fully
1084 known at compile-time.
1085 * optimization: STRING-TRIM and related functions no longer allocate
1086 a new string when no trimming needs to be performed. These functions
1087 are also faster than before when the input string has been declared
1089 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
1090 * optimization: READ-LINE is significantly faster for files containing
1092 * optimization: non-open coded uses of character comparison operators
1093 (e.g. char=) no longer cons when called with more than one parameter
1094 on platforms supporting dynamic-extent allocation.
1095 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
1096 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
1097 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
1099 * bug fix: some sequence functions elided bounds checking when
1101 * bug fix: too liberal weakening of union-type checks when SPEED >
1103 * bug fix: more bogus fixnum declarations in ROOM implementation
1106 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
1107 * new feature: MAKE-HASH-TABLE now experimentally accepts a
1108 :SYNCHRONIZED argument, which makes the hash-table safe for
1109 concurrent accesses (but not iteration.) See also:
1110 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1111 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1112 * optimization: CONCATENATE on strings is an order of magnitude faster
1113 in code compiled with (> SPEED SPACE).
1114 * optimization: SUBSEQ is ~50% faster on lists.
1115 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1116 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1117 END is smaller then START.
1118 * bug fix: SB-PROFILE will no longer report extra consing for nested
1119 calls to profiled functions.
1120 * bug fix: ROOM implementation had bogus fixnum declarations which
1121 could cause type-errors when calling ROOM on large images.
1122 * bug fix: if file compilation is aborted, the partial fasl is now
1123 deleted, and COMPILE-FILE returns NIL as the primary value.
1124 * bug fix: number of thread safety issues relating to SBCL's internal
1125 hash-table usage have been fixed.
1126 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1127 be returned from its body when the values were being returned
1128 using unknown-values return convection and the W-P-O was wrapped
1129 inside an UNWIND-PROTECT.
1130 * bug fix: sb-posix should now compile again under Windows, enabling
1131 slime to work again.
1133 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1134 * incompatible change: hash-table accessor functions are no longer
1135 automatically protected by locks. Concurrent accesses on the same hash-table
1136 from multiple threads can give inconsistent results or even corrupt the
1137 hash-table completely. Multi-threaded applications should do their own
1138 locking at the correct granularity. In the current implementation it is
1139 still safe to have multiple readers access the same table, but it's not
1140 guaranteed that this property will be maintained in future releases.
1141 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1142 and will signal an error at runtime.
1143 * enhancement: SB-THREAD package now exports a semaphore interface.
1144 * enhancement: CONS can now stack-allocate on x86 and
1145 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1147 * enhancement: nested lists can now be stack allocated on
1148 platforms providing stack allocation support.
1149 * enhancement: dynamic-extent support has been extended to support
1150 cases where there are multiple possible sources for the stack
1152 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1153 if the mutex is uncontested on Linux.
1154 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1155 as specified by the CLHS. (thanks to Tobias Rittweiler)
1156 * bug fix: NaN comparison now works on x86-64.
1157 * bug fix: CLOSE :ABORT T on a stream with pending output now
1159 * bug fix: instances of non-standard metaclasses using standard
1160 instance structure protocol sometimes missed the slot type checks
1162 * bug fix: known functions can be inlined in high-debug code.
1164 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1165 * minor incompatible change: the MSI installer on Windows no longer
1166 associates .lisp and .fasl files with the installed SBCL.
1167 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1168 on Windows. (thanks to Luis Oliviera)
1169 * new feature: SB-CLTL2 contrib module now implements
1170 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1171 * optimization: scavenging weak pointers is now more efficient,
1172 requiring O(1) instead of O(N) per weak pointer to identify
1173 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1174 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1175 method bodies using constant slot-names when the first argument is
1176 a specializer parameter for the method.
1177 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1178 names, when the class of the instance is a direct instance
1179 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1180 as slow as the constant slot-name case.)
1181 * optimization: member type construction is now O(N) instead
1183 * optimization: UNION and NUNION are now O(N+M) for large
1184 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1186 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1187 now more readable in environments like Slime which display it.
1188 (thanks to Tobias C. Rittweiler)
1189 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1190 was able to infer the correct argument type for the object on which
1191 the CAS operation was being performed.
1192 * bug fix: copy propagation interfered with parallel assignment
1193 semantics in local calls. (reported by Paul Khuong)
1194 * bug fix: the signed modular fixnum shift compiled to wrong code on
1195 x86 and x86-64. (spotted by a slight modification to some of
1198 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1199 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1200 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1201 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1203 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1204 and generic function now print the object identity for anonymous
1205 instances. (thanks to Attila Lendvai)
1206 * enhancement: as an extension to MOP, SBCL now supports SETF of
1207 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1208 (thanks to Attila Lendvai)
1209 * enhancement: sb-cover has an alternate annotation mode which puts
1210 the form annotations at the car of the form, instead of annotating
1211 it completely. The benefit of this mode is that it shows better
1212 which forms were instrumented by the compiler.
1213 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1214 by printer control variables. (reported by Dan Corkill)
1215 * bug fix: system leaked memory when delayed output was performed by
1216 the OS in smaller chunks then expected. (thanks to David Smith)
1217 * bug fix: system leaked memory when file streams were not closed
1219 * bug fix: large objects written to slow streams that were modified
1220 after the write could end up with the modified state written to
1221 the underlying file descriptor.
1222 * bug fix: multiple threads operating in parallel on the same stream
1223 could cause buffer-overflows.
1224 * bug fix: source location information is stored correctly
1225 (broken since 1.0.6). This bug would generally show up as the
1226 Slime debugger highlighting the wrong form when the sldb-show-source
1228 * bug fix: Branch forms are again annotated as branches in the sb-cover
1230 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1231 on x86-64. (reported by Gregory Vanuxem)
1232 * bug fix: EQUALP could return wrong results for structures with raw
1233 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1234 word). (reported by Vjacheslav Fyodorov)
1235 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1238 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1239 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1240 atomic compare-and-swap operations on threaded platforms.
1241 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1242 allows assigning a global minimum value to optimization qualities
1243 (overriding proclamations and declarations).
1244 * enhancement: closed over variables can be stack-allocated on x86
1246 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1248 * optimization: slot definition lookup is now O(1). This speeds up
1249 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1250 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1251 * optimization: ASSOC and MEMBER can now be open-coded for all
1252 combinations of keyword arguments when second argument is constant
1253 and SPEED >= SPACE. In other cases a specialized version is
1255 * bug fix: using obsoleted structure instances with TYPEP and
1256 generic functions now signals a sensible error.
1257 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1258 (reported by Kristoffer Kvello)
1259 * bug fix: backtrace construction is now more careful when making
1260 lisp-objects from pointers on the stack, to avoid creating bogus
1261 objects that can be seen by the GC.
1262 * bug fix: defaulting of values in contexts expecting more than 7
1263 variables now works on x86-64. (reported by Christopher Laux)
1264 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1266 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1267 as the property-list of a symbol.
1268 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1269 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1270 situations requested, are once again file-compileable. (reported
1273 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1274 * MOP improvement: support for user-defined subclasses of
1275 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1276 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1277 the expansion of DEFMETHOD, is responsible for generating a form
1278 which creates a list of specializers when evaluated. Additional
1279 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1280 debugging and introspective support.
1281 * minor incompatible change: the (unsupported) spinlock interface
1282 has changed: free spinlock now has the value NIL, and a held spinlock
1283 has the owning thread as its value.
1284 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1285 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1287 * enhancement: name of a socket-stream is now "a socket" instead of
1288 "a constant string".
1289 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1290 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1291 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1292 for generic functions. (Thanks to Tobias C. Rittweiler)
1293 * optimization: bignum printing speed has been improved by 20-40%
1294 (depending on the bignum size.)
1295 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1297 * bug fix: the cache used by the CLOS to store precomputed effective
1298 methods, slot offsets, and constant return values is now thread and
1300 * bug fix: generic function dispatch function updating is now thread
1301 and interrupt safe (in the sense that the known issues have been
1303 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1304 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1305 * bug fix: asdf systems with dependencies to the SB-POSIX or
1306 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1307 * bug fix: interrupt safety of applicable method computation has been
1310 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1311 * new contrib: sb-cover, an experimental code coverage tool, is included
1312 as a contrib module.
1313 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1314 significantly faster.
1315 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1316 produce more efficient code on x86-64 if the compiler can determine
1317 that (+ INDEX <constant>) does not require a bounds check and FOO
1318 has an element type at least 8 bits wide.
1319 * enhancement: a new, experimental synchronous timeout facility is
1320 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1321 * enhancement: when a symbol name conflict error arises, the
1322 conflicting symbols are always printed with a package prefix.
1323 (thanks to Kevin Reid)
1324 * enhancement: stepping is now once again supported on the SPARC and
1325 MIPS platforms. (It is also now more likely to work on CheneyGC
1327 * enhancement: sb-sprof can now also track and report accurate call
1329 * bug fixes: the treatment of non-standard subclasses of
1330 SB-MOP:SPECIALIZER is more correct.
1331 * incompatible change: PURIFY no longer copies the data from the
1332 dynamic space into the static and read-only spaces on platforms
1333 that use the generational garbage collector
1334 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1336 * bug fix: GC race condition occasionally resulting in crashes with
1337 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1339 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1341 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1342 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1343 * bug fix: an asynchronous interrupt could previously leave the
1344 system running with GC inhibited.
1345 * bug fix: a DECLARE form evaluated at top-level now causes an error
1346 rather than silently (or verbosely) returning NIL.
1347 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1348 result in a error being signalled due to source form lookup errors.
1349 (reported by Peter Graves)
1351 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1352 * incompatible change: removed writer methods for host-ent-name,
1353 host-ent-addresses -- changing the values did not update the DNS
1355 * minor incompatible change: changed experimental JOIN-THREAD interface
1356 * documentation: the manual now lists reader and writer methods
1357 in class slot documentation sections. (thanks to Richard M Kreuter)
1358 * documentation: unwinding from asyncronous events has been
1359 documented as unsafe.
1360 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1361 in multithreaded application code.
1362 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1363 platforms. (thanks to James Anderson for the optimization hint)
1364 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1366 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1367 variants no longer cons.
1368 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1369 their NOT- variants no longer cons.
1370 * optimization: EQUAL hash tables no longer use SXHASH for objects
1371 of all data types, but instead use an EQL hash for types for which
1372 EQUAL is the same as EQL.
1373 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1374 are significantly faster.
1375 * optimization: new STRING-OUTPUT-STREAM implementation is much
1376 faster for large outputs and conses 30% less on average.
1377 * enhancement: XREF information is now collected for references made
1378 to global variables using SYMBOL-VALUE and a constant argument.
1379 * enhancement: SIGINT now causes a specific condition
1380 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1381 * bug fix: FILE-POSITION used to signal an error in some cases where
1382 ANSI requires it to return NIL.
1383 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1384 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1385 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1387 * bug fix: bad type declaration in the CLOS implementation has
1388 been fixed. (reported by James Anderson)
1389 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1390 (thanks to Lutz Euler)
1391 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1392 required by the CLHS. (reported by Samium Gromoff)
1393 * bug fix: dead unbound variable references now signal an error.
1394 (reported by Marco Monteiro)
1395 * bug fix: / with an unused value was being deleted in safe code.
1396 (thanks to Marco Monteiro and Kevin Reid)
1397 * bug fix: number of characters that can be written onto a single
1398 line in a file is unlimited.
1399 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1400 been fixed by inhibiting interrupts for when GC is disabled.
1401 * bug fix: some interrupt safety issues with GETHASH, (SETF
1402 GETHASH), CLRHASH and REMHASH have been fixed.
1403 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1404 specifier no longer causes infinite recursion.
1405 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1406 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1407 * bug fix: an &environment argument with the correct variable information
1408 is passed to macros that are expanded during byte compilation.
1409 (reported by Samium Gromoff)
1410 * bug fix: a logic error was causing the hash function for bignums to
1411 have excessive amounts of collisions. (reported by Faré Rideau)
1412 * bug fix: modifying the contents of an array could change the return
1413 value of SXHASH on that array, which is only allowed for strings
1414 and bit vectors. (bug introduced in 0.9.16)
1415 * improvement: the x86-64/darwin port now passes all tests (except
1416 for the debugger tests) but should still be considered
1417 experimental until this is fixed.
1418 * improvement: a style-warning is signaled for CASE (etc) clauses with
1419 duplicate keys. (thanks to Kevin Reid)
1420 * improvement: macos/x86 and macos/x86-64 now use mach exception
1421 handlers for dealing with illegal instructions (for trapping and
1422 error handling) and memory protection violations (for GC).
1424 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1425 * new platform: experimental support for x86-64/darwin (MacOS).
1426 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1427 and getnameinfo sockets functions are used instead of gethostbyaddr
1428 and gethostbyname, on platforms where the newer functions are available.
1429 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1431 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1432 don't print the startup banner, but behave as if --noinform was passed
1433 as a command line argument. (thanks to Kevin Reid)
1434 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1435 * optimization: code using alien values with undeclared types is much faster.
1436 * optimization: the compiler is now able to open code SEARCH in more cases.
1437 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1438 * bug fix: using standardized COMMON-LISP special variables as loop
1439 variables no longer signals bogus package lock violations. (reported
1441 * bug fix: declaring local loop variables to be of a range-limited type
1442 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1443 (reported by Andras Simon)
1444 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1445 bugs remain on x86-64.)
1446 * bug fix: the #= and ## reader macros now interact reasonably with
1447 funcallable instances.
1448 * bug fix: type-checks for function arguments were compiled using the
1449 compiler policy settings of the wrong lexical environment (in 1.0.2
1451 * bug fix: SHADOW accepts characters as string designators, as required
1452 by the spec (thanks to Eric Marsden)
1453 * bug fix: fixed GC safety issues when foreign functions are called with
1454 non-base strings as arguments
1455 * bug fix: more consistent error output for fatal-compiler-warnings, like
1457 * bug fix: functions with proper names were showing up as NIL in
1458 backtraces (reported by Edi Weitz, regression in 1.0.2)
1460 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1461 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1462 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1463 produce more efficient code on the x86 if the compiler can determine
1464 that (+ INDEX <constant>) does not require a bounds check and FOO
1465 has an element type at least 8 bits wide.
1466 * bug fix: references to undefined variables are handled the same way
1467 in toplevel forms as in the normal compiler
1468 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1470 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1471 (thanks to Jon Buller)
1472 * improvement: faster compilation times for complex functions
1473 * improvement: added readlink support to SB-POSIX (thanks to Richard
1476 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1477 * improvement: experimental support for mach exception handling on
1478 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1480 * improvement: support for GBK external format.
1481 (thanks to Chun Tian (binghe))
1482 * improvement: the debugger now displays variables that have been closed
1483 over, in code compiled with (DEBUG 2) or higher
1484 * improvement: support for executable cores on NetBSD (thanks to
1486 * new feature: added a RESTART-FRAME debugger command
1487 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1488 be used to provide an implementation for FILE-POSITION on Gray streams
1489 (thanks to Eric Marsden)
1490 * new feature: add syslog, openlog and closelog support to SB-POSIX
1491 (thanks to Richard Kreuter)
1492 * optimization: the function call overhead in code compiled with
1493 a high DEBUG optimization setting is significantly
1494 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1495 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1496 (thanks to Tony Martinez)
1497 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1498 evaluated once (reported by Kevin Reid)
1499 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1500 works again (thanks to Kilian Sprotte)
1501 * bug fix: an error is signaled for tagbodies with duplicate tags
1502 (thanks to Stephen Wilson)
1503 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1504 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1505 stack frames from alien callbacks.
1506 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1507 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1508 65k pages (thanks to David Woodhouse)
1509 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1511 changes in sbcl-1.0.1 relative to sbcl-1.0:
1512 * new platform: FreeBSD/x86-64, including support for threading.
1513 * new feature: the compiler stores cross-referencing information
1514 about function calls (who-calls), macroexpansion (who-macroexpands)
1515 and special variables (who-binds, who-sets, who-references) for code
1516 compiled with (< SPACE 3). This information is available through the
1517 sb-introspect contrib.
1518 * new feature: users may subclass SEQUENCE, and have instances of
1519 these classes interoperate with standard Common Lisp functions if
1520 a number of methods are defined. (This feature is experimental
1521 and the interface subject to change based on feedback from SBCL
1522 users and the general community)
1523 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1524 x86/x86-64, rather than the previous fixed depth of 8
1525 * improvement: another pthread back-end of mutex "pthread-futex".
1526 * bug fix: non-ascii command-line arguments are processed correctly
1527 (thanks to Yaroslav Kavenchuk)
1528 * bug fix: non-required arguments were not passed correctly when a method
1529 defined using DEFMETHOD was called from a mop-generated method using
1530 CALL-NEXT-METHOD (reported by Pascal Costanza)
1531 * bug fix: recursion is now permitted in accessors through
1532 SLOT-UNBOUND. (reported by Pascal Costanza)
1533 * bug fix: an error was signaled at startup if the HOME environment
1534 variable was defined, but had an empty value (reported by Peter Van Eynde)
1535 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1536 signal a PROGRAM-ERROR, not a BUG.
1537 * bug fix: SB-POSIX:READDIR works when built with large file support.
1538 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1540 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1541 for better automatic indentation support. (thanks to Matt Pillsbury)
1542 * optimization: loading generic functions no longer takes O(n^2) time,
1543 proportional to the amount of methods in the generic function
1544 (reported by Todd Sabin and Jeremy Brown)
1545 * optimization: the FIND and POSITION family of sequence functions
1546 are significantly faster on arrays whose element types have been
1548 * improvements to the Windows port:
1549 ** Intermittent heap corruption problems have been fixed. (thanks
1550 to Alastair Bridgewater)
1551 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1553 ** Lisp is able to unwind foreign exception frames from alien
1554 callbacks. (thanks to Alastair Bridgewater)
1556 changes in sbcl-1.0 relative to sbcl-0.9.18:
1557 * improvement: experimental support for threading on FreeBSD/x86.
1558 (thanks to NIIMI Satoshi)
1559 * improvement: runtime option --dynamic-space-size can be used
1560 to set the size of the dynamic space reserved on startup.
1561 * improvement: floating point modes in effect are now saved in
1562 core, and restored on startup.
1563 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1564 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1565 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1566 threads are not running after *SAVE-HOOKS* have run.
1567 * improvement: writes to CLOS instance slots are type-checked in code
1568 compiled with (SAFETY 3)
1569 * improvement: floating-point exception handling on FreeBSD (thanks to
1571 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1572 (thanks to Zach Beane)
1573 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1575 * improvement: added support for the Shift-JIS external format.
1576 (contributed by NIIMI Satoshi)
1577 * improvement: callbacks are supported on Linux/PPC. (thanks to
1579 * bug fix: compiler bug triggered by a (non-standard) VALUES
1580 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1581 * bug fix: file compiler no longer confuses validated and already
1582 dumped structurres. (reported by Kaersten Poeck)
1583 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1584 fill-pointer signals a type-error as required. (thanks to
1586 * bug fix: disassemly of funcallable instances works.
1587 * bug fix: single stepping on PPC.
1588 * bug fix: fix thread-safety problems in the type system (generally
1589 manifesting as nonsensical errors like "STRING is a bad type specifier
1590 for sequences" or "The value 1 is not of type FIXNUM")
1591 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1592 (reported by Josip Gracin)
1593 * bug fix: an error is signaled for attempts to displace arrays with
1594 incompatible element types (thanks to Mario Mommer)
1595 * bug fix: more correct handling of wide characters in debug info
1596 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1597 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1598 and don't cause extra consing
1599 * optimization: MAP and MAP-INTO are significantly faster on vectors
1600 whose elements types have been declared.
1601 * Improvements to SB-SPROF:
1602 ** Support for allocation profiling
1603 ** Reduced profiling overhead, especially for long profiling runs
1604 * Improvements to the Windows port:
1605 ** floating point exceptions are now reported correctly.
1606 ** stack exhaustion detection works partially.
1607 ** more accurate GET-INTERNAL-REAL-TIME.
1608 ** better RUN-PROGRAM behaviour in case of errors in the child
1610 ** PROBE-FILE now simplifies pathnames correctly.
1611 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1612 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1613 1st 1970 (although time timezone data will be unreliable).
1614 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1616 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1617 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1618 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1620 * bug fix: two potential GC deadlocks affecting threaded builds.
1621 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1622 returns T (reported by Anton Kazennikov)
1623 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1624 with non-variable places
1625 * bug fix: remove a race condition in the setting of
1626 funcallable-instance functions, this should make threaded CLOS
1627 code more stable against memory faults.
1628 * bug fix: corruption of specials when unbinding is interrupted by an
1629 asynchronous unwind (reported by Hannu Koivisto)
1630 * improvement: the debugger will now also display local variables that
1631 are only used once, for code compiled with a DEBUG optimization quality
1634 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1635 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1636 * incompatible change: External-format support for FFI calls. The
1637 SB-ALIEN:C-STRING no longer implies an ASCII
1638 external-format. Instead the string is subject to external-format
1639 conversion. Additionally return values of type SB-ALIEN:C-STRING
1640 are by default of type (SIMPLE-ARRAY CHARACTER), not
1641 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1642 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1643 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1644 * incompatible change: SB-EXT package no longer contains the
1645 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1646 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1648 * incompatible change: the single-stepper is no longer available
1649 on Alpha, Mips and Sparc platforms.
1650 * minor incompatible change: the direct superclasses of
1651 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1652 not (STANDARD-OBJECT FUNCTION). This makes the
1653 class-precedence-lists of GENERIC-FUNCTION and
1654 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1656 * minor incompatible change: the default stream external format on
1657 non-unicode SBCL is no longer detected from the locale, but is always
1659 * new feature: Add a version of evaluator that uses an interpreter instead
1660 of the compiler. EVAL still uses the compiler by default, to switch it
1661 to use the interpreter, set the value of the variable
1662 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1663 * minor incompatible change: the single-stepper REPL has been merged
1664 with the normal debugger (see the "Stepping" heading of the debugger help
1665 for more details). The debugger command STEP will no longer switch
1666 to the single-stepper REPL.
1667 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1668 for a type now works.
1669 * bug fix: loading of default sysinit file works. (thanks to Leonid
1671 * bug fix: better detection of circularities in the file-compiler.
1672 (reported by Marco Monteiro)
1673 * bug fix: the CL pathname functions now work with files that have
1674 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1675 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1676 systems with Unix98 pty semantics.
1677 * bug fix: ASDF-INSTALL will now work with bsd tar.
1678 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1680 * bug fix: timers expiring in dead threads no longer cause a
1681 type-error (reported by Paul "Nonny Mouse").
1682 * bug fix: thanks to more lightweight single-stepper instrumentation,
1683 code compiled with (DEBUG 3) will compile and execute significantly faster,
1684 and will have more accurate type-inferencing than before
1685 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1686 whose bindings are modified
1687 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1688 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1689 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1690 CL:SOFTWARE-VERSION with proper version
1692 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1693 * feature: implemented the READER-METHOD-CLASS and
1694 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1695 as specified by AMOP.
1696 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1698 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1699 (thanks to Lutz Euler)
1700 * optimization: hashing of general arrays and vectors has been
1701 improved. (reported by Any Fingerhut)
1702 * enhancement: SB-INTROSPECT is now able to find definitions of
1703 profiled functions. (thanks to Troels Henriksen)
1704 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1705 as well, allowing compiler-macros for SETF-functions to expand.
1706 * enhancement: step-instrumentation no longer wraps known
1707 single-value functions with multiple-value context, allowing
1708 better type inference.
1709 * fixed bug #337: use of MAKE-METHOD in method combination now works
1710 even in the presence of user-defined method classes. (reported by
1711 Bruno Haible and Pascal Costanza)
1712 * fixed bug #339(c): if there are applicable methods not part of any
1713 long-form method-combination group, call INVALID-METHOD-ERROR.
1714 (reported by Bruno Haible)
1715 * fixed bug #361: the :FUNCTION initarg in the protocol for
1716 initialization of methods can now be used to override
1717 internally-produced optimized functions. (reported by Bruno
1719 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1720 system-provided lambda expression no longer cause warnings about
1721 unbound #:|pv-table| symbols.
1722 * bug fix: improved the handling of type declarations and the
1723 detection of violations for keyword arguments with non-constant
1725 * bug fix: potentially erronous calls to PATHNAME and
1726 MERGE-PATHNAMES were being flushed in some cases.
1727 (reported by Richard Kreuter)
1728 * bug fix: compiled calls to TYPEP were mishandling obsolete
1729 instances. (reported by James Bielman and Attila Lendvai)
1730 * bug fix: format strings with ~> without matching ~< no longer
1731 trigger an AVER, but signal an understandable error instead.
1732 (reported by Antonio Martinez)
1733 * bug fix: specifying an output-file without a pathname-type for
1734 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1735 of input-file instead of "fasl". (reported by Robert Dodier)
1736 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1737 (reported by James Y Knight).
1738 * bug fix: compiler-macros-function did not consider the environment
1739 argument for shadowing by local functions.
1740 * bug fix: compiler-macros expansion was inhibited by local INLINE
1742 * bug fix: inline expansions of known functions were subject to
1743 step-instrumentation in high debug policies, leading to problems
1744 with type-inference.
1745 * bug fix: compiler failed to differentiate between different CONS
1746 types in some cases.
1747 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1748 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1749 * bug fix: #368: incorrect use of expressed vs. upgraded array
1751 * bug fix: #306a: more precise unions of array types.
1752 * thread-safety improvements:
1753 ** CONDITION-WAIT could return early on Linux, if the thread was
1754 interrupted and subsequently continued with SIGCONT.
1755 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1757 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1758 * added support for the ucs-2 external format. (contributed by Ivan
1760 * minor incompatible change: pretty printing of objects of type
1761 (cons symbol) is, in the default pprint-dispatch-table, now
1762 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1764 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1765 called later in a class's lifetime, possibly as late as when the
1766 first instance of the class is created. Previously,
1767 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1768 class became finalizeable.
1769 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1770 for multibyte external-format streams. (thanks to "vbzoli")
1771 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1772 with :ALLOCATION :CLASS inherited from superclasses of the
1774 * fixed bug: anonymous classes can now be created using the :NAME
1775 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1776 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1777 * fixed bug: core-files saved with :EXECUTABLE T can again be
1778 executed when SBCL_HOME isn't set. (reported by James Knight)
1779 * fixed bug: toplevel LOCALLY forms with declarations could
1780 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1781 * fixed bug: printing from several different threads using different
1782 values of *print-case* could cause invalid output, due to
1783 some internal special variables of the printer not being bound
1784 thread-locally (reported by Max Mikhanosha)
1785 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1786 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1787 (reported by Pascal Costanza)
1788 * fixed bug: FINALIZE-INHERITANCE is called from
1789 REINITIALIZE-INSTANCE on classes when the class has previously
1790 been finalized, as required by AMOP.
1791 * minor code generation optimizations:
1792 ** better register allocation in CLOS dispatching functions
1793 ** overflow detection when coercing signed bytes to fixnums on x86-64
1794 ** is now implemented with one IMUL instruction instead of three shifts
1795 ** more efficient bit-vector access on x86 and x86-64
1796 ** more efficient access to raw structure slots on x86 and x86-64
1797 * fixed some bugs revealed by Paul Dietz' test suite:
1798 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1799 return its argument.
1801 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1802 * feature: thread support on Solaris/x86, and experimental thread support
1804 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1806 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1807 default method for SLOT-UNBOUND.
1808 * minor incompatible change: prevent the user from specializing the
1809 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1810 somewhat counter to the intent of the protocol, I (CSR) think, and
1811 additionally it just doesn't work in SBCL as currently
1812 implemented, thanks to optimizations (that are always valid for
1813 the other three SLOT-VALUEish functions, but not for the setter).
1814 * minor incompatible change: the :SB-LDB feature is now enabled by
1815 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1816 the low-level debugger.
1817 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1818 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1819 down to an SB-SYS:FD-STREAM, allowing direct communication in
1820 more cases. Temporary files and pipes are still used as a fallback
1822 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1823 on the non-reentrant C functions localtime(3) and gmtime(3).
1824 * bug fix: global optimization policy was not visible in LOCALLY and
1826 * bug fix: class objects can be used as specializers in methods.
1827 (reported by Pascal Costanza)
1828 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1829 with a type error. (reported by blitz_ on #lisp)
1830 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1831 failed with a type error. (reported by Pascal Bourguignon)
1832 * bug fix: merging pathnames against defaults with :DIRECTORY
1833 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1834 (reported by James Y Knight)
1835 * bug fix: saving large (>2GB) cores on x86-64 now works
1836 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1837 constant index and a (simple-array (signed-byte 32)) array
1838 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1839 error (patch by Robert J. Macomber)
1840 * feature: TIME now displays the amount of run-time spent in GC
1841 * bug fix: The debugger now does a better job of respecting
1842 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1843 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1844 workaround for bug 403.)
1845 * bug fix: floating point exception handling now works on FreeBSD
1846 (thanks to NIIMI Satoshi)
1847 * fixed some bugs revealed by Paul Dietz' test suite:
1848 ** MISC.641: LET-conversion were not supposed to work in late
1850 * improvements to the Win32/x86 port:
1851 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1852 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1854 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1855 * new feature: source path information is generated for macro-expansion
1856 errors for use in IDE's like Slime (thanks to Helmut Eller)
1857 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1858 * bug fix: compiler does not loop forever on an invalid type in
1860 * improvement: compilation of most CLOS applications is significantly
1862 * optimization: added a limited bytecode compiler for simple toplevel
1863 forms, speeding up compilation and FASL loading
1864 * bug fix: the statistical profiler now properly distinguishes anonymous
1867 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1868 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1869 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1870 into subclasses of the finalized class. Now user calls to
1871 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1872 system return before any subclasses are finalized.
1873 * minor incompatible change: The reader no longer ignores errors
1874 regarding non-existent packages in #+ and #- feature tests.
1875 * new feature: command line options --no-sysinit, --no-userinit to
1876 inhibit loading the corresponding init files
1877 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1878 for sending data through UDP sockets (thanks to François-René Rideau)
1879 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1880 error is signalled instead (thanks to François-René Rideau)
1881 * minor incompatible change: Error signalling behaviour of lexical
1882 operations violating package locks has changed slightly. Refer to
1883 documentation on package locks for details.
1884 * enhancement: EVAL can process IF-expressions without resorting to the
1886 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1887 constant-folded properly due to missing out-of-line functions.
1888 (reported by Mika Pihlajamäki)
1889 * bug fix: LISTEN sometimes returned T even in cases where no data was
1890 immediately available from the stream
1891 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1892 were reversed. (reported by Levente Mészáros)
1893 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1894 when given a vector displaced to an adjustable array. (reported
1896 * bug fix: derivation of float boundaries from numbers outside the
1897 appropriate float range (reported by John Wiseman)
1898 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1900 * bug fix: precomputing cache entries for generic functions with
1901 some subclasses of specializers as yet invalid does not attempt to
1902 fill a cache line with a negative offset. (reported by Levente
1904 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1905 allow condition class objects as arguments to DOCUMENTATION and
1906 (SETF DOCUMENTATION); only find and set documentation for
1907 structure names for the STRUCTURE doc-type. (suggested by Gary
1909 * improvements to the Win32/x86 port:
1910 ** (user-homedir-pathname) and default initialization file
1911 locations now know about the user's "Documents and Settings"
1912 directory (thanks to Yaroslav Kavenchuk)
1913 ** run-program is implemented (thanks to Mike Thomas)
1914 ** sockets support (thanks to Timothy Ritchey)
1915 ** better backtrace support (thanks to Alastair Bridgewater)
1916 ** sb-grovel supported
1917 ** asdf-install and sb-posix work somewhat
1918 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1919 * improvements to the Solaris/x86 port:
1920 ** works on Solaris 11/Solaris Express
1921 ** floating-point exception handling support
1922 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1923 * fixed some bugs revealed by Paul Dietz' test suite:
1924 ** REMOVE-METHOD returns its generic function argument even when
1925 the method is not one of the generic functions' methods.
1926 ** objects declared as MEMBER types can be admissible arguments to
1927 structure accessors.
1928 ** printing characters should simply be printed by the FORMAT ~:C
1930 ** compiler failure when compiling functions with hairy constant
1931 defaults for optional parameters.
1932 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1933 function, which is already optimized.
1935 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1936 * new platform: experimental support for SBCL x86/Darwin, including
1937 MacOS X 10.4.5 on Intel.
1938 * new feature: Unicode character names are now known to the system
1939 (through CHAR-NAME and NAME-CHAR).
1940 * new feature: the filesystem location of SBCL's core file is
1941 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1942 * minor incompatible change: the contrib modules SB-POSIX and
1943 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1944 this change is to make it easier to distribute
1945 location-independent binaries.
1946 * bug fix: as implied by AMOP, standardized classes no longer have
1947 slots named by external symbols of public packages. (reported by
1949 * optimization: calling functions via constant symbols -- as in
1950 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1951 via the function object as in (FUNCALL #'FOO).
1952 * optimization: on x86 and x86-64, the calling convention for the
1953 case of dealing with an unknown number of values has been altered
1954 to be friendlier to the prediction heuristics implemented,
1955 particularly with reference to CALL and RET pairing. (thanks to
1956 Alastair Bridgewater)
1957 * optimization: on x86, the code for access to thread-local symbol
1958 values and binding/unbinding of thread-local symbols is smaller.
1959 (thanks to Alastair Bridgewater)
1960 * enhancement: CONSTANTP is now able to determine constantness of
1961 more complex forms, including calls to constant-foldable standardized
1962 functions and some special forms beyond QUOTE.
1963 * fixed bug: occasional GC crashes on Solaris/x86
1964 * optimization: x86-64 supports stack allocation of results of simple
1965 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1966 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1967 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1968 (thanks to James Knight)
1969 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1970 by the standard (thanks to Stephen Compall)
1972 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1973 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1974 be used for bundling the runtime and the core file into one
1975 executable binary. This feature is not currently supported on all SBCL
1976 platforms. (thanks to James Bielman and NIIMI Satoshi)
1977 * new feature: a generational or ephemeral garbage collector is now
1978 the default on the PowerPC platform (both Linux and Darwin). The
1979 old Cheney (stop and copy) collector is a build-time option.
1980 (thanks to Cyrus Harmon, after Raymond Toy)
1981 * minor incompatible change: the method by which SBCL finds its
1982 contributed modules has changed; it no longer relies on symbolic
1983 links from an $SBCL_HOME/systems directory, but searches directly
1984 in the subdirectories of $SBCL_HOME.
1985 * enhancement: the dynamic heap size on the Linux/PPC platform is
1986 markedly larger, even using the older Cheney garbage collector.
1987 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1988 certain complicated string types. (reported by Gary King)
1989 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1990 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1991 * fixed bug: DOCUMENTATION on structures no longer signals an
1992 error if no structure type of the right name exists. (reported by
1994 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1995 applicable non-standard primary or :AROUND method on
1996 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1997 no longer cause unbound variable errors. (reported by Kevin Reid)
1998 * fixed bug: metacircle resolution in cases where methods have slots
1999 added before the slots from STANDARD-METHOD. (reported by Jean
2001 * fixed bug: the Power PC assembly code for calling into lisp
2002 sometimes computed its return address wrongly (depending on the
2003 15th bit of the address, fixed at link time). This used to
2004 manifest itself in a segmentation violation while building PCL.
2005 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
2006 many others over the years)
2007 * fixed bug: ROOM no longer reports silly numbers for stack usage if
2008 the stack pointer (treated as a signed integer) is negative.
2009 (thanks to Peter van Eynde)
2011 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
2012 * new platform: experimental support for the Windows operating
2013 system has been added. (thanks to Alastair Bridgewater)
2014 * fixed several bugs in and robustified the PPC FFI (including
2015 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
2016 * bug fix: allow non-simple string symbol names (reported by Paul
2018 * bug fix: interrupt handling on NetBSD (thanks to Richard M
2020 * bug fix: saving a core corrupted callbacks on x86/x86-64
2021 * bug fix: closed a loophole in metacircularity detection and
2022 grounding in the PCL implementation of CLOS.
2023 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
2024 (reported by tomppa on #lisp)
2025 * bug fix: clear the direction flag on Lisp -> C transitions, as
2026 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
2027 (reported by Andrej Grozin and Hendrik Maryns)
2028 * optimization: major improvements to GC efficiency on GENCGC platforms
2029 * optimization: faster implementation of EQUAL
2030 * optimization: emit more efficient opcodes for some common
2031 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
2032 * optimization: several other minor code-generation improvements on x86-64
2033 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
2035 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
2036 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2037 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
2038 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
2039 by ANSI as a generic function, and for consistency (SETF
2040 GENERIC-FUNCTION-NAME) is treated likewise.
2041 * fixed bug #233.b: make constraint propagation notice when a variable
2042 value is changed after it is referenced but before it is used
2043 * fixed bug #296: no more arbitrary behaviour from filenames with
2044 odd characters as --load arguments. (reported by Adam Warner)
2045 * fixed bug #390: :CHARACTER-SET pathname components now work as
2046 expected. (reported by Tim Daly Jr)
2047 * fixed bug #391: complicated :TYPE intersections in slot
2048 definitions no longer cause an error in PCL internals.
2049 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
2050 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
2051 returning the number of octets which would be written to the
2052 file-stream. (thanks to Robert J. Macomber)
2053 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
2054 arguments to the internal stream functions. (thanks to David
2056 * bug fix: stack exhaustion is now handled on the x86/SunOS
2057 platform. (thanks to Daisuke Homma)
2058 * bug fix: more accurate ROOM results on GENCGC platforms
2059 * optimization: improved type inference for arithmetic-for
2060 index variables in LOOP
2061 * optimization: faster floating-point SQRT on x86-64
2062 * fixed some bugs revealed by Paul Dietz' test suite:
2063 ** DOCUMENTATION returns NIL instead of "" for method combinations
2064 that don't have a docstring
2066 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
2067 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2068 GENERIC-FUNCTION-NAME) are no longer generic functions, and
2069 therefore are not customizeable by user code (as seems to be at
2070 least permitted and maybe required by AMOP). As a consolation,
2071 however, the SBCL implementation of these functions now calls
2072 REINITIALIZE-INSTANCE as specified by AMOP.
2073 * bug fix: slot-definition documentation is propagated as per ANSI
2074 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
2075 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
2076 argument, as do eslotd initialization methods. (from Pascal
2077 Costanza's "Closer" project)
2078 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
2079 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
2081 * bug fix: it is now possible to have more than one subclass of
2082 STANDARD-GENERIC-FUNCTION without causing stack overflow.
2083 (reported by Bruno Haible, Pascal Costanza and others)
2084 * bug fix: the dependent update protocol now works for generic
2085 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
2086 and Pascal Costanza)
2087 * bug fix: condition-class instances corresponding to
2088 DEFINE-CONDITION forms are now created eagerly. (reported by
2089 Kalle Olavi Niemitalo on comp.lang.lisp)
2090 * bug fix: floating point printing is more accurate in some
2091 circumstances. (thanks to Simon Alexander)
2092 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
2093 merged with *DEFAULT-PATHNAME-DEFAULTS*.
2094 * bug fix: callbacks on OS X now preserve stack-alignment.
2095 (thanks to Cyrus Harmon)
2096 * enhancement: the x86-64 disassembler is much better at
2097 disassembling SSE instructions. (thanks to Lutz Euler)
2098 * enhancement: CHECK-TYPE on a variable now allows type-inference
2099 (thanks to Vincent Arkesteijn)
2100 * optimization: improved performance of EUC-JP external format.
2101 (thanks to NIIMI Satoshi)
2102 * optimization: performance improvements to IO on file streams of
2103 :ELEMENT-TYPE CHARACTER
2104 * optimization: much faster memory allocation on x86-64
2105 * optimization: faster garbage collections (latency and throughput)
2107 * optimization: faster bignum subtraction and division on x86 and x86-64
2108 (thanks to Lutz Euler)
2109 * optimization: more accurate type-inference for loops that use a
2110 floating point index variable or a negative step.
2112 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2113 * bug fix: add a workaround to SBCL looping infinitely at startup on
2114 Linux kernels with apparently buggy implementations of personality().
2115 (thanks to Svein Ove Aas)
2116 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2117 (thanks to David Lichteblau)
2118 * bug fix: local bindings shadow global symbol macros for
2119 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2120 * optimization: non-open coded uses of numeric comparison operators
2121 (e.g. >) no longer cons when called with more than one parameter
2122 on platforms supporting dynamic-extent allocation.
2123 * enhancement: saving cores with foreign code loaded is now supported
2124 on MIPS/Linux in addition to the previously supported platforms.
2125 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2126 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2127 * bug fix: division by zero in sb-sprof when no samples were collected
2128 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2129 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2130 (thanks to Alexander Kjeldaas)
2131 * new feature: ignoring the return values of destructive functions
2132 like NREVERSE signals a compile-time style-warning.
2133 (thanks to Kevin Reid)
2134 * new feature: the alignment of alien structure fields can be
2135 explicitly specified. (thanks to Cyrus Harmon)
2136 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2137 and dump core on SIGQUIT
2139 ** incompatible change: threads do not inherit values of specials
2140 from their parents (see manual)
2141 ** bug fix: threads stacks belonging to dead threads are freed by the
2142 next exiting thread, no need to gc to collect thread stacks anymore
2143 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2144 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2145 ** bug fix: dynamic variable and thread start related gc lossage
2146 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2148 * fixed some bugs revealed by Paul Dietz' test suite:
2149 ** SUBTYPEP is slightly more accurate on heinously complicated
2150 CONS types where some of the members have uncertain (in the
2151 NIL, NIL sense) type relationships to each other.
2153 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2154 no longer purified unless :PURIFY T is explicitly specified.
2155 ** Non-purified cores are significantly smaller than before
2157 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2158 * new feature: timers based on Zach Beane's excellent timer package
2159 * added support for the following external formats: koi8-u,
2160 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2161 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2162 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2163 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2164 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2165 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2166 * incompatible change: a threaded SBCL will no longer revert to
2167 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2168 * bug fix: interrupts are disabled until startup is complete; no
2169 more sigsegvs when receiving a signal to soon
2170 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2172 * bug fix: add a workaround for the memory randomization features in
2173 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2174 workaround will only be in effect on systems with the proc filesystem
2176 * bug fix: printing objects of type HASH-TABLE signals a
2177 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2179 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2180 have been running for over 50 days. (reported by Gilbert Baumann)
2181 * bug fix: the logic for getting names of functions gets less
2182 confused when confronded with alternate-metaclass
2183 funcallable-instances. (reported by Cyrus Harmon)
2184 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2185 consistent, even on internal alternate-metaclass objects.
2186 * bug fix: flush closure information collected by physical
2187 environment analysis prepass before the main pass. (bug reported
2189 * bug fix: compiler pack phase does not modify a hashtable, which is
2190 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2191 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2192 may be expanded. (reported by Alexander <malishev> on c.l.l)
2193 * bug fix: no more style-warnings when compiling code using LAST
2194 under high-SPEED low-SPACE settings. (reported by David Wallin
2196 * fixed some incompatibilities between SBCL's MOP and the MOP
2198 ** the METAOBJECT class is now implemented;
2199 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2200 STANDARD-OBJECT, as required;
2201 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2202 now compatible (as required by VALIDATE-SUPERCLASS); there
2203 remains a consistency requirement over the presence or absence
2204 of the FUNCTION class in the superclass list of finalized
2205 classes; see the manual for more details;
2206 ** the :AROUND method for COMPUTE-SLOTS on
2207 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2208 requested slot ordering.
2210 ** bug fix: parent thread now can be gc'ed even with a live
2212 ** bug fix: binding a special with PROGV to no value is not
2213 the same as that symbol not having been bound (thanks to
2215 ** bug fix: binding specials is thread safe (thanks to
2217 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2218 and SB-SPROF do not die with 'no handler for signal XX in
2219 interrupt_handle_now(..)' anymore
2220 ** bug fix: WITH-TIMEOUT works with multiple running threads
2221 * fixed some bugs revealed by Paul Dietz' test suite:
2222 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2223 the :method-class keyword argument.
2225 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2226 * new port: the Solaris operating system on x86 processors is now
2227 mostly supported, though some rough edges in the environment
2228 remain. (thanks to Daisuke Homma)
2229 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2230 space for its heap. (thanks to Thiemo Seufer)
2231 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2232 closures. (thanks to Thiemo Seufer)
2233 * minor incompatible change: eof selects abort in the debugger.
2234 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2235 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2237 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2238 deprecated, and will go away in a future revision of SBCL.
2239 * minor incompatible change: GC-ON and GC-OFF are no longer
2240 implemented with a counter, it does not matter how many times gc
2241 is switched on or off
2242 * bug fix: discriminating functions for generic function classes
2243 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2244 make invalid assumptions about method precedence order. (reported
2246 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2247 platforms. (thanks to Thiemo Seufer)
2248 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2249 * bug fix: correct pathname printing with printer escaping is on.
2250 (thanks to Kevin Reid)
2251 * bug fix: complex VOP definitions in "user-space" no longer trigger
2252 package locks. (reported by Zach Beane)
2253 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2254 for funcallable-instance functions, and is no different from
2256 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2257 characters in the current readtable. (reported by Nicholas Neuss)
2258 * bug fix: SUBTYPEP on various CONS types returns more a more
2259 accurate acknowledgment of its certainty.
2260 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2261 lists in some cases. This partially fixes bug 384.
2262 * flush all standard streams before prompting in the REPL and the
2264 * bug fix: signal handling and triggering gc do not conflict
2265 directly anymore, in particular a high frequency sb-sprof does
2266 not prevent gc from running
2267 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2268 approximation for timezone and DST information between the
2269 universal time epoch and the smallest negative 32-bit time_t.
2270 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2271 year 1899 so long as the encoded time is non-negative.
2272 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2273 the correct answer for non-constant rotate amounts when performing
2274 an inline 32-bit rotation.
2276 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2277 there is only one thread in the session
2278 ** bug fix: memory leak for streams created in one thread and
2279 written to in another
2280 ** bug fix: lockup when compiled with gcc4
2281 ** bug fix: race that allows the gc to be triggered when gc is
2283 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2284 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2286 ** bug fix: finished threads are gc'ed properly
2287 * fixed some bugs revealed by Paul Dietz' test suite:
2288 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2289 has arguments with a different set of applicable methods from
2290 the orignal arguments.
2291 ** The type error thrown by MAP now has a correct expected-type
2293 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2294 name a compiled function.
2295 ** Three MISC tests where a large bignum was improperly coerced to
2296 a float that couldn't represent that bignum during type
2297 derivation were fixed.
2298 ** SUBTYPEP can now handle types involving the negation of a
2299 list-form FUNCTION type.
2300 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2301 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2302 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2304 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2305 * New feature: Experimental support for bivalent streams: streams
2306 opened with :element-type :default now allow character and binary
2307 (unsigned-byte 8) I/O
2308 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2309 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2311 * bug fix: no more highly sporadic "couldn't check whether ~S is
2312 readable" when reading a stream and an interrupt hits in the middle
2313 of a select system call
2314 * compiler better recognizes complex arrays (reported by Eduardo
2316 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2317 for variables on SBCL built with linkage-tables. (reported by Luis
2319 * various error reporting improvements.
2320 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2321 (Thanks to Thiemo Seufer)
2322 * bug fix: nested reader invokations work correctly
2323 * bug fix: it is possible to have simultaneous references to foreign
2324 code and foreign data with the same name.
2326 ** added x86-64 support
2327 ** incompatible change: the threading api now works with thread
2328 objects instead of thread ids
2329 ** bug fix: threads are protected from signals and interruption when
2330 starting up or going down
2331 ** bug fix: a race where an exiting thread could lose its stack to gc
2332 ** bug fix: don't halt on infinite error in threads if possible
2333 ** fixed numerous gc deadlocks introduced in the pthread merge
2334 ** bug fix: fixed thread safety issues in read and print
2335 ** bug fix: debugger doesn't hang on session lock if interrupted at
2336 an inappropriate moment
2337 ** bug fix: run-program is now thread safe(r)
2338 ** bug fix: inner with-recursive-lock no longer releases the mutex
2339 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2340 * fixed some bugs revealed by Paul Dietz' test suite:
2341 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2342 EXPECTED-TYPE slots filled.
2343 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2345 ** MAKE-CONDITION accepts classes as type-designators.
2346 ** COMPILE may never return NIL.
2347 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2348 range before calling Unix time functions
2350 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2351 * numerous signal handling fixes to increase stability
2352 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2353 * minor incompatible change: we now correctly canonize default
2354 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2355 as per the MOP, rather than the historical (INITARG INITFUNCTION
2356 INITFORM). (reported by Bruno Haible)
2357 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2358 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2359 for more information.
2360 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2361 pathname is a directory pathname.
2362 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2363 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2365 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2366 :METACLASS argument in addition to a class name. (reported by
2367 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2368 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2369 (reported by Bruno Haible; patch from Gerd Moellmann)
2370 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2372 * bug fix: more cleanups to the floating point exception handling on
2373 x86-64 (thanks to James Knight)
2374 * bug fix: the compiler does not try to constant fold calls of
2375 COERCE and BIT-* functions when they return freshly constructed
2376 objects. (reported by Nikodemus Siivola and Paul Dietz)
2377 * optimization: improved the MIPS versions of generic (in the
2378 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2379 * optimization: direct conversion of (unsigned-byte 32) to floats on
2380 the PowerPC platform.
2381 * optimization: structure instances with raw slots now use less
2382 memory, and probably show better memory locality. (thanks to
2384 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2385 are treated as requests for stack allocation on the x86-64,
2386 Alpha-32, PPC and SPARC platforms.
2387 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2388 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2389 (thanks to Tony Martinez)
2391 ** gcing a dead thread can no longer lead to lockups
2392 ** threads block signals until they are set up properly
2393 ** errno is no longer shared by threads
2394 ** interrupt-thread restores the eflags register on x86
2395 ** fixed some lockups due to gc/thread interaction
2396 * fixed some bugs revealed by Paul Dietz' test suite:
2397 ** invalid dotted lists no longer raise a read error when
2398 *READ-SUPPRESS* is T
2399 ** signal an error if a symbol that names a declaration is used
2400 as the name of a type, or vice versa
2401 ** allow using the (declare (typespec var*)) abbreviation for
2402 (declare (type typespec var*)) with all type specifiers
2403 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2404 ** always return NIL from PPRINT-POP when OBJECT is NIL
2405 ** don't signal errors when pretty-printing malformed LABELS,
2406 FLET or MACROLET forms
2407 ** declarations in a DOLIST body are also in scope for the
2409 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2411 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2414 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2415 * fixed cross-compiler leakages that prevented building a 32-bit
2416 target with a 64-bit host compiler.
2417 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2418 opened with :IF-EXISTS :OVERWRITE.
2419 * fixed bug 281: error for an invalid qualifier in a short-form method
2420 combination method is not signalled until the faulty method is called.
2421 * bug fix: iteration variable type inferrer failed to deal with open
2422 intervals. (reported by Alan Shields)
2423 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2424 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2425 or more runtime options were provided to the sbcl binary.
2426 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2427 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2428 * optimization: there's now a fast-path for fixnum arguments in the
2429 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2430 * optimization: the code generated on x86-64 is more compact thanks
2431 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2432 * bug fix: floating-point exception handling now partly works on
2433 x86-64. (Thanks to James Knight)
2434 * improvement to the MIPS backend from Thiemo Seufer: C-style
2435 64-bit long long arguments and return values to alien functions
2437 * fixed some bugs revealed by Paul Dietz' test suite:
2438 ** the type-error signalled from WARN has a filled-in DATUM slot.
2439 ** the type-error required when a stream is not associated with
2440 a file has the stream as its datum.
2441 ** type-errors on single-floats on x86-64 no longer have
2442 :INVALID-OBJECT as the datum
2443 ** the type-errors signalled for invalid function names now have
2444 a correct expected type
2445 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2446 for structures defined with a :TYPE. Documentation strings for
2447 typed structures are no longer immediately discarded
2448 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2449 broadcast streams and synonym streams. FILE-LENGTH now also works
2450 on broadcast streams.
2452 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2453 * incompatible change: the --noprogrammer option, deprecated since
2454 version 0.7.5, has been removed. Please use the equivalent
2455 --disable-debugger option instead.
2456 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2458 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2459 inoperational for a while now) has been completely removed.
2460 * null lexical environments are now printed as #<NULL-LEXENV>,
2461 significantly reducing the amount of clutter in typical backtraces.
2462 * documentation on weak pointers, finalization, and after GC hooks
2463 has been added to the manual.
2464 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2465 as other specialized array types, is much faster. SUBSEQ and
2466 COPY-SEQ on such arrays have also been sped up.
2467 * optimization: EQL is now more efficient when at least other argument
2468 is known to be of type (OR FIXNUM (NOT NUMBER)).
2469 * fixed bug: the runtime is now less vulnerable to changes in the
2470 size of the SBCL object on OS X, and virtual memory is reserved for
2471 all spaces that need to be at a fixed address.
2472 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2473 * fixed bug: finalizers and after GC hooks that cause consing are now
2475 * fixed bug: compiler error messages and summaries are now printed to
2476 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2477 * fixed inference of the upper bound of an iteration variable.
2478 (reported by Rajat Datta).
2479 * fixed bug 211e: calling local functions with duplicated constant
2480 keyword argument no longer causes a bogus style warning about an
2482 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2483 ftype declaration to be disregarded. (reported by Dave Roberts)
2484 * fixed bug 373: caused by erronous compilation of references to alien
2485 variables in the runtime on ppc/darwin.
2486 * fixed bug 376: CONJUGATE type deriver.
2487 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2488 (reported by Baughn on #lisp)
2489 * a cross-compiler bug on non-x86 platforms has been identified and
2490 fixed. (thanks to Bruno Haible)
2491 * improvements to the MIPS runtime code for increased stability.
2492 (thanks to Thiemo Seufer)
2493 * increased the maximimum compact environment size to allow
2494 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2495 * partial workaround for bug 135: don't name the function we're
2496 calling for hairy cases of EVAL, so as not to accumulate one environment
2497 entry per call to EVAL. (reported by Kevin Reid)
2498 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2499 * optimization: type testing for non-vector arrays should be faster.
2500 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2502 * bug fix: setting 31st element of a bit vector to zero did not work
2504 * bug fix: redefining a class definition which failed due to a
2505 previous accessor / function clash now works (but see BUGS entry
2506 #380 for more problems in this area). (thanks to Zach Beane)
2507 * the long form of DEFINE-METHOD-COMBINATION disables method group
2508 checking when given a single method group with pattern *. (thanks
2510 * on x86 compiler supports stack allocation of results of simple
2511 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2512 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2513 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2514 * fixed some bugs related to Unicode integration:
2515 ** the restarts for recovering from input and output encoding
2516 errors only appear when there is in fact such an error to
2518 * fixed some bugs revealed by Paul Dietz' test suite:
2519 ** MISC.549 and similar: late transformation of unsafe type
2520 assertions into derived types caused unexpected code
2522 ** SCALE-FLOAT type deriver is less wrong.
2523 ** type derivers for EXP, LOG and similar functions compute result
2524 types for complex arguments better.
2525 ** (MISC.563) CONJUGATE type deriver works for very restricted
2527 ** out-of-line type testers for character strings are available.
2528 ** EQUAL compiler transform understands specialness of objects
2530 ** accessing double-floats stored on the stack now works on x86-64.
2531 ** debugger internals could sometimes create invalid lispobjs,
2532 resulting in GC crashes.
2533 ** MISC.548: type check weakening can convert required type into
2535 ** initialization forms for bindings are not in scope of free special
2538 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2539 * incompatible change: thread support for non-NPTL systems has
2540 been removed - locking is buggy and unreliable. A threaded
2541 SBCL build will now warn at startup and refuse to create
2542 new threads, unless futex support is detected in the kernel
2543 * incompatible change: the top level REPL now has only an ABORT
2544 restart associated with it, not TOPLEVEL and ABORT as it used to.
2545 TOP and TOPLEVEL are now available as debugger commands for
2546 returning to the top level.
2547 * incompatible change: forms evaluated in the REPL now use the
2548 global optimization policy.
2549 * incompatible change: user- and system-initialization files are
2550 no longer processed with LOAD, but by READ and EVAL; hence the
2551 global optimization policy, startup package, readtable, etc,
2553 * The .fasl file format number has been incremented because of
2554 various incompatible changes.
2555 * internal entry point details and argument counts no longer appear
2556 in backtraces unless explicitly requested by setting
2557 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2558 * built-in and standard functions no longer have names like "top
2559 level local call to FOO".
2560 * fixed bug 32: functions defined in non-null lexical environments
2561 now have more legible printed representation
2562 * fixed bug 33: functions defined in non-null lexical environemnts
2563 are now more amenable to inspection by INSPECT.
2564 * workaround for bug 354: XEPs no longer appear in backtraces unless
2565 explicitly requested.
2566 * fixed bug: receiving the signal which results from attempting to
2567 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2568 SIGBUS on other *BSDs) on architectures where the C stack is also
2569 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2570 to debugger confusion or wild instability. (reported by Cheuksan
2572 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2573 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2574 (reported by Lutz Euler)
2575 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2576 compiled code. (reported by Paul Dietz)
2577 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2578 specialized parameter to a method to something that is not TYPEP
2579 the specializer is now possible.
2580 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2581 face of package deletion.
2582 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2583 pathname. (reported by David Tolpin on c.l.l)
2584 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2585 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2586 STRING-STREAM and FILE-STREAM, respectively.
2587 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2588 than previously; in addition, its test suite is now run on build.
2589 (thanks to Luke Gorrie)
2590 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2591 * optimization: added a immediate representation for single-floats
2593 * fixed some bugs revealed by Paul Dietz' test suite:
2594 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2595 ** attempting to create a package with a colliding nickname causes
2596 correctable errors to be signalled.
2597 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2598 ** several x86-64 backend bugs related to sign-extension of immediate
2601 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2602 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2603 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2605 * fixed loading of multiply forward-referenced layouts.
2606 (thanks to Cheuksan Wang)
2607 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2608 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2609 (see COMPILE-FILE documentation for details of :PRINT option)
2610 * fixed bugs 19 and 317: fixed-format floating point printing is
2611 more accurate. This also fixes a bug reported by Adam Warner
2612 related to the ~@F format directive.
2613 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2615 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2616 dispatch table if the from-char is a dispatch macro character.
2617 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2618 by the wordsize no longer ignore the last word. (reported by Lutz
2620 * fixed bug in type checking of dynamic-extent variables. (reported
2622 * optimization: sequence traversal functions use their freedom to
2623 coerce function designators to functions.
2624 * optimization: code with many calls to CLOS methods specialized on
2625 CLOS classes has had redundant type checks removed.
2626 * optimization: added declarations to speed up operations that access
2627 the internal character database (for example STRING-UPCASE)
2628 * optimization: comparison operations between floats and sufficiently small
2629 fixnums no longer create extra rationals
2630 * fixed some bugs related to Unicode integration:
2631 ** portions of multibyte characters at the end of buffers for
2632 character-based file input are correctly transferred to the
2633 start of the buffer at the next read.
2634 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2635 passing it through to OPEN.
2636 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2637 argument given to internal calls to OPEN.
2638 ** the built-in comment readers (introduced by character sequences
2639 ";" and "#|") are more forgiving to encoding errors; they will
2640 STYLE-WARN and then attempt to resync the stream at a character
2641 boundary. (thanks to Teemu Kalvas)
2642 * fixed some bugs in the x86-64 port:
2643 ** Negative short int return values from c-calls are sign-extended
2645 ** The stack is aligned to 16-bytes for c-calls, as required by
2646 the ABI. (thanks to Cheuksan Wang)
2647 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2648 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2649 for enums. (thanks to Vincent Arkesteijn)
2650 ** Multiple small optimizations and bugfixes for floating point
2652 * fixed some bugs revealed by Paul Dietz' test suite:
2653 ** Space, Tab, Linefeed, Return and Page have the invalid
2654 secondary constituent character trait.
2655 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2657 ** WITH-INPUT-FROM-STRING should only update the index place on
2659 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2660 works more reliably.
2661 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2662 with one-dimensional arrays with fill pointers.
2663 ** TYPE-OF failed on a complex with an integer realpart and a
2665 ** compiler failure during type inference for the code of form
2666 (IF (EQL X (THE ...)) ...) (MISC.535).
2668 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2669 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2670 platform. The port supports 61-bit fixnums, large memory spaces
2671 and reloading shared object files.
2672 * enhancement: saving cores with foreign code loaded is now
2673 supported on ppc/Darwin in addition to the previously supported
2675 * enhancement: the statistical profiler now walks deeper into the
2676 call stack for more meaningful call-graphs and accrued time
2677 reports (x86/x86-64 only). It also now reports time spent in
2679 * enhancement: it is now possible to trace most individual methods
2680 of a generic function in addition to tracing the generic function
2682 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2684 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2685 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2686 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2687 produces an error. (thanks to Vincent Arkesteijn)
2688 * bug fix: NAMESTRING on pathnames with :WILD components in their
2689 directories works correctly. (thanks to Artem V. Andreev)
2690 * fixed bug 125: compiler preserves identity of closures. (reported
2692 * bug fixed: functions with &REST arguments sometimes failed with
2693 "Undefined function" when compiled with (DEBUG 3). (reported by
2695 * bug fix: overflow during compiling of setting element of a bit
2696 vector with constant index and value. (reported by Timmy Douglas)
2697 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2698 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2699 * on x86 compiler partially supports stack allocation of dynamic-extent
2701 * GO and RETURN-FROM do not check the extent of their exit points
2702 when compiled with SAFETY 0.
2703 * fixed some bugs related to Unicode integration:
2704 ** encoding and decoding errors are now much more robustly
2705 handled; it should now be possible to recover even from invalid
2706 input or output to the terminal. (thanks to Teemu Kalvas)
2707 ** provided a first cut at implementing STRING-TO-OCTETS and
2708 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2709 ** altered the SB-MD5 contributed module slightly, changing the
2710 interface just enough for it to be supportable for builds where
2711 lisp characters are not eight bits.
2712 * fixed some bugs revealed by Paul Dietz' test suite:
2713 ** the FORMATTER-generated functions for ~V[ conditionals require
2714 the correct number of arguments.
2715 ** READ-FROM-STRING returns the mandated second value when applied
2716 to displaced strings.
2717 ** the #\Rubout and #\Backspace characters are treated as invalid
2718 constituent characters by the tokenizer.
2720 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2721 * new feature: reloading changed shared object files with
2722 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2723 * new feature: references to foreign variables and functions
2724 can now be compiled and loaded before the corresponding shared
2725 object file is loaded, as long as the foreign definitions are
2726 available at runtime.
2727 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2728 supported. (thanks to Dan Debertin)
2729 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2730 just keywords. (thanks to Vincent Arkesteijn)
2731 * fixed bug #331: structure-class instances corresponding to
2732 DEFSTRUCT forms are now created eagerly.
2733 * fixed bug #345: backtraces from calls to undefined functions work
2734 on x86 as well. Related bug #61 is now also partially fixed on x86
2735 (backtraces from throws to unknown catch tags.)
2736 * bug fix: lambda-list parsing is now stricter vrt. order and number
2737 of lambda-list keywords.
2738 * bug fix: as specified by AMOP, an error is signalled if a
2739 class-option appears multiple times in a DEFCLASS form. (reported
2741 * bug fix: RANDOM can be compiled when the compiler derives the type
2742 of its numeric argument as a disjoint set of small integers.
2743 (reported by Paul Dietz)
2744 * bug fix: starting a core saved with shared objects loaded when
2745 those objects are not available no longer causes threaded SBCL to
2746 hang. (reported by Sean Ross)
2747 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2749 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2750 argument, which must always be NIL. (reported by Kalle Niemitalo)
2751 * bug fix: printing 1.0d+23 no longer results in an error.
2752 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2753 * bug fix: structure slot setters preserve evaluation order. (thanks
2755 * fixed some bugs related to Unicode integration:
2756 ** RUN-PROGRAM can allow its child to take input from a Lisp
2757 stream. (reported by Stefan Scholl)
2758 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2759 locales. (reported by Ken Causey)
2760 * fixed some bugs revealed by Paul Dietz' test suite:
2761 ** INCF, DECF and REMF evaluate their place form as specified in
2763 ** FORMATTER expands ~{ ~} iteration directives with V or #
2764 parameters correctly.
2765 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2766 consequent uses no arguments correctly.
2767 ** the system has a partial understanding of the (COMPLEX RATIO)
2769 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2772 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2773 * new feature: a build-time option (controlled by the :SB-UNICODE
2774 keyword feature, enabled by default) for building the system with
2775 support for the entire 21-bit character space defined by the
2777 * new feature: the system now has rudimentary external-format
2778 support; the primary user-visible change at this time is that
2779 characters with the high bit set (such as lower-case-e-acute) will
2780 print correctly to a terminal in a UTF-8 environment.
2781 * minor incompatible change: BASE-CHAR no longer names a class;
2782 however, CHARACTER continues to do so, as required by ANSI.
2783 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2784 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2785 has been moved to the SB-EXT package (temporarily re-exported from
2787 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2788 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2790 * fixed bug #351: better error-handlind and reporting for malformed
2792 * fixed bug #350: bignum-printing is now more memory-efficient,
2793 allowing printing of very large bignums, eg. (expt 2 10000000).
2794 (reported by Bruno Haible)
2795 * fixed bug #302: better primitive-type selection for intersection
2797 * fixed bug #308: non-graphic characters now all have names, as
2798 required. (reported by Bruno Haible)
2799 * bug fix: redefining a class with different superclasses now correctly
2800 removes it from the direct-subclasses of its previous superclasses.
2801 (reported by David Morse)
2802 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2803 new value now works. (reported by Bruno Haible)
2804 * bug fix: correct canonicalization of multiple non-standard slot
2805 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2806 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2807 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2808 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2809 now exists, an signals an error.
2810 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2811 during macroexpansion for non-iteration variables as well. (reported
2812 by Bruno Haible for CMUCL)
2813 * bug fix: Cyclic structures and unprintable objects in compiler
2814 messages no longer cause errors. (reported by Bruno Haible)
2815 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2816 and READ-FROM-STRING all now return a primary value of NIL if
2817 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2818 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2819 (reported by Bruno Haible for CMUCL)
2820 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2821 do not designate a condition. (reported by Bruno Haible for
2823 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2824 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2825 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2826 permitting resolution in favour of any of the conflicting symbols.
2827 (reported by Bruno Haible for CMUCL)
2828 * bug fix: EQUAL compiler optimizations is less aggressive on
2829 strings which can potentially compare true despite having distinct
2830 specialized array element types.
2831 * bug fix: unit enumerations can be defined without dividing by
2832 zero. (thanks to Vincent Arkesteijn)
2833 * FORMAT compile-time argument count checking has been enhanced.
2834 (report from Bruno Haible for CMUCL)
2835 * a partial workaround for the bug 262: the compiler does not try to
2836 inline-expand a local function doing RETURN-FROM from a deleted
2837 BLOCK. (thanks to Peter Denno for the bug report and to David
2838 Wragg for the simple test case)
2839 * fixed some bugs revealed by Paul Dietz' test suite:
2840 ** RENAME-PACKAGE allows all package designators as new package
2842 ** constraint propagation and lambda variable substitution are
2843 more cautious in dealing with partially deleted code.
2844 ** compiler sometimes lost reoptimization passes.
2845 ** CERROR, given a condition as condition designator, passes its
2846 remaining arguments to the continue format control without
2848 ** Case-altering FORMAT directives work correctly on non-ASCII
2850 ** The REAL type specifier handles bounds outside the floating
2851 point ranges without signalling FLOATING-POINT-OVERFLOW.
2852 ** Functions with IR1-transformations can create intercomponent
2853 references to global functions.
2854 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2856 ** FORMAT ~R treats a nil value for its first parameter correctly.
2858 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2859 * enhancement: saving cores with foreign code loaded is now
2860 supported on x86/NetBSD and sparc/Linux in addition to the previously
2861 supported platforms.
2862 * bug fix: on some platforms repeated installations caused multiple
2863 copies of HTML documentation to be installed -- should not happen
2864 any more. (reported by Stefan Scholl)
2865 * bug fix: parsing self-recursive alien record types multiple times
2866 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2867 original patch by Helmut Eller for CMUCL)
2868 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2869 (thanks to Richard Kreuter)
2870 * bug fix: defining classes whose accessors are methods on existing
2871 generic functions in other (locked) packages no longer signals
2872 bogus package lock violations. (reported by François-René Rideau)
2873 * bug fix: special variables as DEFMETHOD parameters no longer have
2874 associated bogus type declarations. (reported by David Wragg and
2876 * bug fix: read-write consistency on streams of element-type
2877 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2878 * bug fix: redefiniton of the only method of a generic function with
2879 no DEFGENERIC no longer emits a full WARNING. In addition,
2880 redefinition of generic functions with no DEFGENERIC to an
2881 incompatible lambda list now signals an error. (thanks to Zach
2883 * bug fix: DEFGENERIC now works even when there's a function of the
2884 same name in an enclosing lexical environment. (thanks to Zach
2886 * fixed compiler failure, caused by instrumenting code during
2887 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2888 * optimization: added loop analysis and improved register allocation
2889 * fixed some bugs revealed by Paul Dietz' test suite:
2890 ** POSITION on displaced vectors with non-zero displacement
2891 returns the right answer.
2892 ** (SIMPLE-STRING) is a valid type specifier for sequence
2894 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2896 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2897 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2899 ** The justification version of the FORMAT ~< directive treats
2900 non-zero minpad parameter correctly.
2902 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2903 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2904 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2905 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2906 the supported interface.
2907 * new feature: Single-stepping of code compiled with DEBUG 2 or
2908 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2909 * new feature: saving cores with foreign code loaded is now
2910 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2911 Timothy Moore's work for CMUCL)
2912 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2913 parameters to * instead of NIL if no initform is supplied.
2914 (reported by Johan Bockgaard)
2915 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2916 correctly when FUNCALL appears as the car of the form. Note:
2917 despite this FUNCALL forms are not currently subject to
2918 compiler-macro expansion. (port of Raymond Toy's fix for the
2919 same from CMUCL, reported by Johan Bockgaard)
2920 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2921 (thanks for Teemu Kalvas)
2922 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2923 case instead of terminating the iteration (thanks for Julian
2924 Squires, Sean Champ and Raymond Toy)
2925 * bug fix: incorrect expansion of defgeneric that caused a style
2926 warning. (thanks for Zach Beane)
2927 * bug fix: slot accessor effective method computation works properly
2928 for classes with multiple non-standard applicable methods on
2929 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2930 * on x86 compiler supports stack allocation of results of LIST and
2931 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2932 CMUCL implementation by Gerd Moellmann)
2933 * fixed some bugs revealed by Paul Dietz' test suite:
2934 ** FORMAT strings with both the ~<~:;~> form of the justification
2935 directive and pretty-printing directives cause an error.
2937 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2938 * incompatible change: the internal functions
2939 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2940 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2941 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2942 instead of the old functions.
2943 * new feature: on platforms where "dladdr" is available foreign
2944 function names now appear in backtraces. (based on Helmut Eller's
2946 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2947 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2949 * documentation: ASDF manual is now distributed with SBCL.
2950 * bug fix: SBCL can now load its contributed modules with REQUIRE
2951 even if the system-provided entries have been removed from
2952 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2954 * bug fix: backtraces involving undefined functions or assembly
2955 routines are more informative. (thanks to Brian Downing)
2956 * bug fix: mutually referent alien structures now work correctly.
2957 (reported by Rick Taube)
2958 * bug fix: structures defined by WITH-ALIEN can be referred to
2959 within other definitions in the same WITH-ALIEN.
2960 * bug fix: division operators (MOD, TRUNCATE and the like) with
2961 constant zero divisors and integer dividends no longer generate
2963 * bug fix: provide default methods for INPUT-STREAM-P and
2964 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2965 * bug fix: improve the detection and resolution of MOP metacycles
2966 (where effective method computation depends on the generic function
2967 having its effective method computed). (reported by Bruno Haible)
2968 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2969 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2970 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2971 * bug fix: pathnames with relative directory components can be
2972 represented relative to default pathnames.
2973 * optimization: in taking the GCD of bignums, reduce the two bignums
2974 to approximately the same size (using Euclid's algorithm) before
2975 applying the more sophisticated binary GCD. (thanks to Juho
2977 * optimization: COUNT on bitvectors now operates word-at-a-time.
2978 * optimization: ASH with a positive, but not necessarily constant,
2979 (leftwards) shift, when in a modular context, is compiled to a
2981 * fixed some bugs revealed by Paul Dietz' test suite:
2982 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2983 if the corresponding argument is NIL.
2984 ** FORMAT directives accepting a minpad parameter treat negative
2986 ** literal commas as character parameters to FORMAT directives are
2988 ** literal spaces directly after ~<Newline> directives within a
2989 format-logical-block (~:< ~@:>) do not induce :FILL-style
2990 conditional newlines.
2991 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2992 as specified: it no longer includes conditional newlines).
2993 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2995 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2996 :PER-LINE-PREFIX argument does not evaluate to a string.
2998 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2999 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
3000 the manual for details; for now, package locks can be disabled by
3001 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
3002 if no major problems are found then it is likely that they will be
3003 compiled in unconditionally.
3004 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
3005 unsupported operators on all platforms. To load a shared library
3006 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
3007 object file, link it into a shared library outside of SBCL and
3008 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
3010 * bug workaround: ROOM T and the GC were not working together
3011 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
3012 reporting facility aren't preserved. That reporting has been
3013 disabled, so now until and unless someone figures out how to make it
3014 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
3015 * minor incompatible change: as threatened around sbcl-0.8.0, the
3016 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
3017 The symbols are also exported from SB-PCL for backwards
3018 compatibility, but more so than before SB-PCL should be treated as
3019 an implementation-internal package.
3020 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
3022 * the SB-SPROF contrib now works on (most) non-x86 architectures.
3023 It is known as of this release not to work on the Alpha, however.
3024 * fixed bug #167: errors signalled due to illegal syntax in method
3025 bodies are now more legible.
3026 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
3027 designators and can hence be used with TYPEP.
3028 * fixed bug #333: CHECK-TYPE now ensures that the type error
3029 signalled, if any, has the right object to be accessed by
3030 TYPE-ERROR-DATUM. (reported by Tony Martinez)
3031 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
3032 5.1.2.3 for multiple-value place subforms. (reported by Kalle
3034 * fixed bug #334: programmatic addition of slots using specialized
3035 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
3036 and :ALLOCATION :CLASS slots.
3037 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
3038 float exponents. (rereported by Peter Seibel)
3039 * fixed a bug: #\Space (and other whitespace characters) are no
3040 longer considered to be macro characters in standard syntax by
3041 GET-MACRO-CHARACTER.
3042 * fixed bug: initialization of condition class metaobjects no longer
3043 causes an instance of the condition to be created. (reported by
3045 * fixed bug: it is now possible to have slots such that
3046 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
3047 description is neither :INSTANCE nor :CLASS.
3048 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
3049 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
3050 SLOT-VALUE-USING-CLASS), and no longer causes errors with
3051 non-standard SLOT-DEFINITION-ALLOCATION values.
3052 * fixed bugs: various race conditions handling exiting threads.
3053 CL-PPCRE's thread torture test now passes.
3054 * fixed arguably-a-bug: GC time proportional to number of threads in
3055 system even when most of them are idle
3056 * optimization: improved performance of BIT and SBIT on bit-vectors.
3057 * .fasl file incompatibility: The fasl file version number has
3058 been incremented because of changes associated with package locks.
3060 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
3061 * minor incompatible change: the system no longer provides
3062 optimization hints (conditions which are TYPEP
3063 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
3064 mode; these hints are still emitted when higher SPEED optimization
3066 * new contrib module: a sampling profiler (profiling by statistical
3067 sampling, rather than by instrumenting functions) is available as
3068 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
3069 * the behaviour of the standard function ED is now customizeable by
3070 third parties through a hook variable: see ED's documentation
3071 string for information on the protocol.
3072 * the compiler no longer emits efficiency notes for (FUNCALL X)
3073 when the type of X is uncertain under default optimization
3075 * fixed bug 276: mutating a binding of a specialized parameter to a
3076 method to something that is not TYPEP the specializer is now
3078 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
3079 exhibit the required behaviour on the x86 platform. (thanks to
3080 Peter van Eynde, Eric Marsden and Bruno Haible)
3081 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
3082 even for difficult arguments. (reported by Peter Graves)
3083 * fixed bug 141a: the backquote printer now descends quoted
3085 * fixed another bug in backquote printing: no more destructive
3086 modification of the form's list structure. (reported by Brian
3088 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
3089 move between its address being taken and the call to
3090 interrupt_thread, fixing a crashing race condition.
3091 * the SB-POSIX contrib implementation has been adjusted so that it
3092 no longer exhibits ridiculously poor performance when constructing
3093 instances corresponding to C structs.
3095 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
3096 * minor incompatible change: the sb-grovel contrib now treats C
3097 structures as alien (in the sense of SB-ALIEN) objects rather than
3098 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
3099 has implications for memory management of client code
3100 (sb-grovel-returned objects must now be manually managed) and for
3101 type safety (alien objects now have full types).
3102 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
3103 used to control emission of compiler diagnostics, rather than the
3104 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
3105 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
3106 quality should be considered deprecated.
3107 * install.sh now installs the user manual as well
3108 * (not quite a new documentable feature, but worth considering in
3109 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3110 beginnings of a semantically meaningful condition hierarchy is
3111 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3113 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3114 designator as the defaults argument.
3115 * fixed bug: Displaced arrays whose displaced-to array has become
3116 too small now cause ARRAY-DIMENSION to signal an error, providing
3117 for safer bounds-checking. (reported by Bruno Haible)
3118 * fixed bug: DEFCLASS slot definitions with identical :READER and
3119 :WRITER names now signal a reasonable error. (reported by Thomas
3121 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3123 * fixed bug: Invalid slot specification errors now print correctly.
3124 (thanks to Zach Beane)
3125 * fixed bug 320: Shared to local slot value transfers in class
3126 redefinitions now happen corrently from superclasses as
3127 well. (reported by Bruno Haible)
3128 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3130 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3131 improper lists correctly. (reported by Bruno Haible)
3132 * fixed bug 313: source-transform for <fun-name> was erroneously
3133 applied to a call of a value of a variable with name <fun-name>.
3134 (reported by Antonio Menezes Leitao)
3135 * fixed bug 307: The obsolete instance protocol ensures that
3136 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3137 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3138 remove dead unknown-values globs from the middle of the stack.
3139 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3140 easier to resume long computations after using *BREAK-ON-SIGNALS*
3141 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3142 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3143 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3144 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3145 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3146 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3148 * fixed a (fairly theoretical) bug in string printing: if
3149 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3150 does not have array-element-type equal to the most general string
3152 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3153 function name. (thanks to Nikodemus Siivola)
3154 * fixed bug: compile-time format string checker failed on
3155 non-closed ~{. (reported by Thomas F Burdick)
3156 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3157 don't warn on legal loop code involving more than one aggregate
3158 boolean. (thanks to Nikodemus Siivola)
3159 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3160 in keyword position to MAKE-PACKAGE was not being reported as
3162 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3163 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3165 * fixed bugs 280 and 312: the checking for multiple definitions in a
3166 file is less likely to become confused by uses of inline
3168 * fixed bug: the #S reader macro performs the keyword coercion
3169 specified for slot names. (reported by Kalle Niemitalo)
3170 * fixed bug: lambda lists may contain symbols whose names start with
3171 & but are not lambda-list-keywords; their occurrence triggers a
3173 * fixed bug 321: define-method-combination argument lambda lists do
3174 not cause invalid code to be generated when &optional or &aux
3175 variables have default values. (reported by Bruno Haible)
3176 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3177 have CLOS classes; this appears also to have fixed reported
3178 instability in stack exhaustion detection.
3179 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3180 MUFFLE-WARNING when no associated restart is present is now
3182 * optimization: rearranged the expansion of various defining macros
3183 so that each expands into only one top-level form in a
3184 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3186 * optimization: used a previously little-used slot in symbols to
3187 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3189 * fixed some bugs revealed by Paul Dietz' test suite:
3190 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3192 ** Fixed an optimization bug related to inheritance of initforms
3193 from local to shared slots.
3194 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3195 ** CAST optimizer forgot to flush argument derived type.
3196 ** print/read consistency on floats is now orders of magnitude
3197 more likely. (thanks also to Bruno Haible for a similar report
3199 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3200 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3201 for objects of type REAL. Make it so.
3202 ** SXHASH is specified (infelicitously) to respect similarity,
3203 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3204 it so. (thanks to Markus Ziegler)
3205 ** on the Alpha, the compiler succeeds in compiling functions
3206 returning a known number of arguments greater than 63.
3207 ** fixed handling of invalid NIL arguments in keyword position
3209 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3211 ** provided a readably-printable representation for RANDOM-STATE
3213 ** ensured that pathnames, where they have a namestring, always
3214 print using #P"..." syntax.
3216 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3217 * Support for the forthcoming 2.0 version of the NetBSD kernel
3218 running on x86 hardware has been added. (thanks to Perry
3219 E. Metzger most immediately, and others for their past work)
3220 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3221 current release version; SBCL's OpenBSD support had been broken
3222 since about the time of OpenBSD's switch to ELF binary format.)
3223 * [placeholder for DX summary]
3224 ** user code with &REST lists declared dynamic-extent, under high
3225 speed or space and low safety and debug optimization policy.
3226 * The manual has been converted to Texinfo format and the debugger
3227 chapter from the cmucl manual has been added.
3228 * A facility has been added to extract documentation strings from
3229 sbcl and store them as Texinfo-formatted snippets for inclusion in
3230 the manual (via Texinfo's @include directive)
3231 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3232 the test case to Dave Roberts)
3233 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3234 pointers. (reported by Sean Ross)
3235 * bug fix: PROFILE output is printed nicely even for large numerical
3236 values. (thanks to Zach Beane)
3237 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3238 greater than 32 handle EOF correctly.
3239 * bug fix: on X86 an immediate argument of the IMUL instruction is
3240 correctly printed in disassembly. (thanks to Lutz Euler)
3241 * bug fix: class slots in redefined classes preserve their old
3242 values. (thanks to Bruno Haible and Nikodemus Siivola)
3243 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3244 succeeds. (reported by Marco Baringer)
3245 * fixed some bugs revealed by Paul Dietz' test suite:
3246 ** READ-SEQUENCE now works on ECHO-STREAMs.
3247 ** RATIONALIZE works more according to its specification. (thanks
3250 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3251 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3252 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3253 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3254 should matter to you only if you rebind the printer control
3255 variables and then find you want different bindings in the
3256 debugger than in the ordinary execution of your program.)
3257 * The runtime build system has been tweaked to support building
3258 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3259 assembler and linker. (thanks to Nikodemus Siivola)
3260 * Unbound, undefined, undeclared variables now trigger full
3261 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3262 more often programmer error than deliberate exploitation of undefined
3264 * optimization: the hash algorithm for strings has changed to one
3265 that is less vulnerable to spurious collisions. (thanks to Juho
3267 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3268 needless bounds checking. (thanks to Juho Snellman)
3269 * optimization: implemented multiplication as a modular
3270 (UNSIGNED-BYTE 32) operation on the PPC backend.
3271 * fixed some bugs revealed by Paul Dietz' test suite:
3272 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3274 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3275 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3276 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3278 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3280 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3282 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3283 their output stream on EOF from read.
3284 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3285 have been read to end-of-file.
3286 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3288 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3289 description of determination of which consecutive characters
3291 ** Printing the "Space" character with escaping on now yields "#\\ ",
3292 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3293 ** Reading floating-point numbers with *READ-BASE* set to a number
3294 less than 10 works correctly.
3295 ** Reading floating-point numbers with *READ-BASE* set to a number
3296 more than 10 works correctly.
3297 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3298 the readtable currently in effect.
3300 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3301 * minor incompatible change: parsing of namestrings on a physical
3302 (Unix) host has changed; numbers after the final #\. in a
3303 namestring are no longer interpreted as a version field. This is
3304 intented to be largely invisible to the user, except that the
3305 meaning of the namestring "*.*.*" has changed: it now refers to a
3306 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3307 should usually be replaced by
3308 (make-pathname :name :wild :type :wild :version :wild)
3309 with the added benefit that this is more likely to be portable.
3310 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3311 option to OPEN now signals an error if the file being opened
3312 exists; this may have an impact on existing code.
3313 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3314 signals received, on the PowerPC platforms (both Linux and
3315 Darwin). (thanks to Pierre Mai for pointing out the location of
3317 * several fixes on OS X: The system now builds and runs cleanly on
3318 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3319 (thanks to Brian Mastenbrook)
3320 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3321 second-resolution: integer multiples of 1/3600 between -24 and 24.
3322 (thanks to Vincent Arkesteijn)
3323 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3324 type when called with 1 argument; PEEK-CHAR checked type of
3325 PEEK-TYPE only after having read first character from a
3326 stream. (reported by Peter Graves)
3327 * bug fix: the garbage collector now has much better locality
3328 behaviour, and in particular no longer treats all memory as being
3329 exhausted when many small objects point to each other in a deeply
3331 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3332 recognized as being TYPEP their class.
3333 * bug fix: the PUSHNEW documentation string has been corrected.
3334 (thanks to Vincent Arkesteijn)
3335 * bug fix: defaulting of the value for the last of an atypically
3336 large number of multiple values being bound was not being
3337 performed correctly on the Alpha or PPC platforms
3338 * optimization: implemented multiplication as a modular
3339 (UNSIGNED-BYTE 32) operation on the x86 backend.
3340 * optimization: SEARCH on simple-base-strings can now be open-coded.
3341 (see also contrib/compiler-extras.lisp for inspiration for
3342 teaching the compiler about the Boyer-Moore algorithm).
3343 * value, returned by MAX (and MIN) called with several EQUALP, but
3344 not EQL, arguments now does not depend on compiler settings.
3345 * fixed some bugs revealed by Paul Dietz' test suite:
3346 ** in stack analysis liveness information is propagated from
3347 non-local entry points.
3348 ** pathwise CAST removing failed when the CAST node did not start
3350 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3351 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3353 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3354 host is already defined.
3355 ** RENAME-FILE works on streams instead of signalling an internal
3357 ** PEEK-CHAR uses the current readtable when determining whether
3358 or not a character is whitespace.
3359 ** MERGE-PATHNAMES handles the case when the pathname does not
3360 specify a name while the default-pathname specifies a version.
3361 ** Pathnames now stand a better chance of respecting print/read
3363 ** Attempting to use standardized file system operators with a
3364 pathname with invalid :DIRECTORY components signals a
3366 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3368 ** DIRECTORY on logical pathnames is more correct.
3369 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3370 signal a TYPE-ERROR if their argument is not a stream.
3371 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3372 designator argument does not designate a stream.
3373 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3374 examining the synonym.
3375 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3377 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3378 element-type larger than ([UN]SIGNED-BYTE 32).
3380 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3381 * When built with the :SB-FUTEX feature, threaded builds now take
3382 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3383 for faster/more reliable mutex and condition variable support.
3384 * Incompatible change (but one you probably shouldn't have been using
3385 anyway): the interface and code for arbitrating between multiple
3386 threads in the same user session has been redesigned.
3387 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3388 error when called without an explicit environment argument.
3389 (thanks to Nikodemus Siivola)
3390 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3391 confused about their position. (thanks to Adam Warner and Gerd
3393 * bug fix: Pretty printing backquoted forms with unquotations in the
3394 argument list position of various code constructs such as LAMBDA
3395 now works correctly. (reported by Paul Dietz)
3396 * bug fix: Pretty printing unquotations no longer loses all
3397 stream position information.
3398 * optimization: performance of string output streams is now less
3399 poor for multiple small sequence writes.
3400 * optimization: performance of CSUBTYPEP in the presence of complex
3401 expressions involving CONS and NOT many times has been improved.
3402 (reported by Paul Dietz)
3403 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3405 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3406 streams for the REPL, for the debugger, and so on. (thanks to
3408 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3409 * fixed some bugs revealed by Paul Dietz' test suite:
3410 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3411 is ignored if it is too small, rather than propagating through
3412 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3413 ** extremely complex negations of CONS types were not being
3414 sufficiently canonicalized, leading to inconsistencies in
3416 ** VALUES tranformer lost derived type.
3418 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3419 * fixed a bootstrapping bug: the build process no longer assumes
3420 that the various BOOLE-related constants have the same value in
3421 host and target lisps. (noted by Paul Dietz' test suite on an
3422 SBCL binary built from CLISP)
3423 * The system can now be dynamically linked on the MIPS platform,
3424 which enables dynamic loading of foreign code from Lisp. (thanks
3425 to Ralf Baechle for discussions on the MIPS ABI)
3426 * The system now records debugging information for its own source
3427 files in a filesystem-position-independent manner, relative to
3428 the "SYS" logical host.
3429 * fixed a compiler bug: MV-LET convertion did not check references
3430 to the "max args" entry point. (reported by Brian Downing)
3431 * tweaked disassembly notes to be less confident about proclaiming
3432 some instruction as an LRA. (thanks to Brian Downing)
3433 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3434 now each have their own history, command character, and other
3435 characteristics. (thanks to David Lichteblau)
3436 * fixed some bugs revealed by Paul Dietz' test suite:
3437 ** compiler failure in compiling LOGAND expressions including a
3439 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3440 shift greater than 32.
3441 ** FLUSH-DEST did not mark blocks for type check regeneration.
3442 ** HANDLER-CASE failed to accept declarations in handler clauses
3443 in some circumstances.
3445 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3446 * New code in contrib/sb-introspect (still probably not entirely
3447 stable yet) provides some support for smart Lisp development
3448 environments like SLIME.
3449 * The conditions signalled for errors occurring when loading .fasl
3450 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3451 in a way which should help ASDF recover gracefully.
3452 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3453 stuff has been cleaned up. If you code contrib/ stuff, this might
3454 affect you, and you can look at contrib/README, contrib/STANDARDS,
3455 and/or the 0.8.4.27 diff to check.
3456 * In full calls the compiler now does not generate checks for declared
3457 argument types for all arguments.
3458 * various threading fixes
3459 ** and some experimental patches which didn't make it into
3460 the main tree for this release, but which are shipped in
3461 contrib/experimental-thread.patch as a possible fix for some
3462 failures (deadlock, spinning...) in GC-intensive multithreaded
3464 * fixed PPC build problem (source code incompatibility of different
3465 library versions): added offsetof() hackery which attempts to divine
3466 where glibc maintainers put uc_mcontext today
3467 * fixed bug 282: compiler does not trust type assertions while passing
3468 arguments to a full call.
3469 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3470 &OPTIONAL VALUES type parameter.
3471 * fix bug 214: algorithm for noting rejected templates is now more
3472 similar to that of template seletion. (also reported by rydis on
3474 * fixed bug 141b: printing backquoted information readably and prettily
3475 inserts a space where necessary.
3476 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3477 (CHARACTER 10) are now reported as errors, rather than propagated
3478 as unknown types. (reported by piso on #lisp)
3479 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3480 with values NIL and :ERROR. (thanks to Milan Zamazal)
3481 * fixed bug 191c: CLOS now does proper keyword argument checking as
3482 described in CLHS 7.6.5 and 7.6.5.1.
3483 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3484 longer raise an error; further, using a list as a for-as-arithmetic
3485 counter now raises a meaningful error.
3486 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3487 now understood by sequence creation functions such as MAKE-SEQUENCE
3489 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3490 a STRING-INPUT-STREAM.
3491 * compiler enhancement: SIGNUM is now better able to derive the type
3493 * type declarations inside WITH-SLOTS are checked. (reported by
3495 * fixed some bugs revealed by Paul Dietz' test suite:
3496 ** incorrect optimization of TRUNCATE for a positive first
3497 argument and negative second.
3498 ** compiler failure in let-convertion during flushing dead code.
3499 ** compiler failure while deriving type of TRUNCATE on an
3500 interval, containing 0.
3501 ** ASH of a negative bignum by a negative bignum count now returns
3503 ** intersection of CONS types now canonicalizes properly, fixing
3504 inconsistencies in SUBTYPEP.
3506 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3507 * incompatible change: The --disable-debugger command line
3508 option now clobbers the debugger at a more fundamental
3509 level, by redefining #'INVOKE-DEBUGGER instead of by
3510 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3511 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3512 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3513 enter the debugger REPL and then suffer recursive errors
3514 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3515 while under the new system, BREAK in --disable-debugger mode
3516 terminates the system just as an unhandled error would.
3517 * fixed compiler performance when processing loops with a step >1;
3518 * bug fix: DOCUMENTATION now retrieves generic function
3519 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3520 support has been systematized, and now supports the methods
3521 specified by ANSI, along with a default method and a method for
3522 slot documentation. (reported by Nathan Froyd)
3523 * bug fix: effective methods associated with a generic function are
3524 no longer cached over a change of that generic function's method
3525 combination. (reported by Andreas Fuchs)
3526 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3527 names a non-existent directory. (thanks to Andreas Fuchs)
3528 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3529 the PPC platform, lead to stack corruption; this has been fixed.
3530 (reported by Rainer Joswig)
3531 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3532 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3533 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3534 arithmetic in general, is now much more reliable on the PPC
3536 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3537 platform now returns the right answer.
3538 * optimization: restored some effective method precomputation in
3539 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3540 precomputation is now tunable.
3541 * optimization: compiler-internal data structure use has been
3542 reviewed, and changes have been made that should improve the
3543 performance of the compiler by about 20%.
3544 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3545 simple-base-strings and simple-bit-vectors is improved.
3546 * optimization: the optimization of 32-bit logical and arithmetic
3547 functions introduced in version 0.8.3 on the x86 has been
3548 implemented on the mips, ppc and sparc platforms; an
3549 implementation of the same facility, but for 64-bit arithmetic,
3550 has been added for the alpha.
3551 * microoptimization: the compiler is better able to make use of the
3552 x86 LEA instruction for multiplication by constants.
3553 * bug fix: in some situations compiler did not report usage of
3554 generic arithmetic in (SPEED 3) policy.
3555 * bug 145b fix: compiler used wrong type specifier while converting
3556 MEMBER-types to numeric.
3557 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3559 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3560 index argument now works properly on the Alpha platform.
3561 * bug fix: floating point exception treatment on the Alpha platform
3563 * bug fix: FILE-POSITION works much better on string input and
3564 output streams. (thanks to Nikodemus Siivola)
3565 * bug fix: many threading/garbage collection symptoms sorted.
3566 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3567 might be pseudo-atomic.
3568 * internal change: Stopping for GC is now done with signals not
3569 ptrace. GC is now done in whichever thread wanted it, instead of
3571 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3573 * bug fix: result form in DO is not contained in the implicit
3575 * incompatible change: ICR structure is changed; the value part of
3576 CONTINUATION is now called LVAR; corresponding functions are
3577 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3578 * added type deriver for ISQRT (thanks to Robert E. Brown).
3579 * bug fix: better support for loading from the command line when an
3580 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3582 * fixed some bugs revealed by Paul Dietz' test suite:
3583 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3584 ** ROUND and FROUND now give the right answer when given very
3585 small float arguments.
3586 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3588 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3589 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3590 ** LCM should return a non-negative integer.
3591 ** PARSE-INTEGER returned the index of a terminator instead of the
3592 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3593 ** PARSE-INTEGER returned an incorrect index being applied to a
3595 ** LCM with two arguments of 0 returns 0 rather than signalling
3597 ** unsigned addition of a 32-bit constant with the high bit set no
3598 longer causes an internal compiler error on the x86.
3599 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3600 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3601 with negative last argument.
3602 ** byte specifiers with zero size and position no longer cause
3603 an error during type derivation.
3604 ** bignum multiplication on the Alpha platform now returns the
3606 * porting: The system now builds on SuSE AMD64, although it still
3607 generates a 32-bit binary.
3608 * .fasl file incompatibility: The fasl file version number has
3609 been incremented (because of the changes to internal compiler
3610 data structures referred to above).
3612 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3613 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3614 more accurately, on the Darwin kernel running on PowerPC hardware.
3615 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3616 * Compiler code deletion notes now signal a condition of type
3617 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3618 an associated MUFFLE-WARNING restart.
3619 * The compiler now performs limited argument count validation of
3620 constant format strings in FORMAT, and where appropriate in ERROR,
3621 CERROR and WARN. (thanks to Gerd Moellmann)
3622 * New ASDF-INSTALL contrib can be used for automatic download and
3623 installation of third-party Lisp code from CCLAN or other sites
3625 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3626 which forces another thread to execute a function supplied by the
3628 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3629 now accept and act upon their :ELEMENT-TYPE keyword argument.
3630 (reported by Martin Atzmueller, Edi Weitz)
3631 * bug fix: FILE-POSITION now accepts position designators up to
3632 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3633 is the greater. (thanks to Patrik Nordebo)
3634 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3635 Kalle Olavi Niemitalo)
3636 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3637 printable. (reported by Eric Marsden)
3638 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3639 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3640 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3641 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3642 (thanks to Stig Sandoe)
3643 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3644 UNREAD-CHAR now correctly includes the unread character in the
3645 target sequence. (thanks to Gerd Moellmann)
3646 * bug fix in threaded builds: the system can now be suspended and
3647 resumed by shell job control with minimal disruption.
3648 * bug fixes in times and timezones >2038AD
3649 * better handling of "where is GNU make?" problem in build scripts
3650 (thanks to Nikodemus Siivola)
3651 * new optimization: inside a named function any reference to a
3652 function with the same name is considered to be a self-reference;
3653 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3654 optimization quality.
3655 * new optimization on x86: logical functions and + now have
3656 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3657 used when the result is truncated to 32 bits.
3658 * VALUES declaration is partially enabled.
3659 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3660 * fixed some bugs revealed by Paul Dietz' test suite:
3661 ** The system now obeys the constraint imposed by
3662 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3663 types form a lattice under type intersection.
3664 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3665 ** ASSOC now ignores NIL elements in an alist.
3666 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3667 and (1+ MOST-POSITIVE-FIXNUM) answers.
3668 ** The addition of a method with invalid qualifiers to a generic
3669 function does not cause an error to be signalled immediately;
3670 a warning is signalled, and the error is generated only on
3671 calling the generic function.
3672 * changed .fasl file version number, in part to add type codes for
3673 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3674 obscure ANSI requirements
3676 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3677 * fixed bug 148: failure to inline-expand a local function left
3678 garbage, confusing the compiler.
3679 * fixed bugs 3cd: structure slot readers perform type check if the
3680 slot can have an invalid value (i.e. it is either not initialized
3681 or can be written with a less specific slot writer).
3682 * bug fix: the compiler now traps array references to elements off
3683 the end of an array; previously, the bounds checking in some
3684 circumstances could go off-by-one.
3685 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3687 * type declarations for array element types now obey the description
3688 on the CLHS page "Declaration TYPE", as per discussions on
3689 sbcl-help around 2003-05-08. This means that a declaration
3690 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3691 declaration, all references to BAR will be asserted or assumed
3692 (with THE, so dependent on compiler policy) to involve objects of
3693 type FOO. Note that no such declaration is implied in
3694 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3695 * declared types of functions from the "Conditions"
3696 chapter. (reported by Paul Dietz)
3697 * bug fix: CERROR accepts a function as its first argument.
3698 * bug fix: NTH an NTHCDR accept a bignum as index
3699 arguments. (reported by Adam Warner)
3700 * optimization: character compare routines now optimize comparing
3701 against a constant character. (reported by Gilbert Baumann)
3702 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3703 argument now works properly on the MIPS platform.
3704 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3705 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3706 declared to have a functional type, does not cause an error
3708 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3709 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3710 * fixed bug: if last continuation of a deleted block has a
3711 destination, this destination should be deleted too. (reported by
3713 * fixed a bug in the bootstrap process: the host compiler's values
3714 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3715 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3716 test case from Patrik Nordebo)
3717 * improved the ability of the disassembler on the PPC platform to
3718 provide helpful disassembly notes.
3719 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3720 the class in more cases than previously.
3721 * bug fix: FILE-POSITION now understands :START and :END for
3722 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3723 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3724 8) data. (thanks to David Lichteblau)
3725 * bug fix: it is possible to add a method to a generic function
3726 without lambda list.
3727 * bug fix: reader failed to signal END-OF-FILE inside an
3728 object representation. (reported by Nikodemus Siivola)
3729 * fixed some bugs revealed by Paul Dietz' test suite:
3730 ** LAST and [N]BUTLAST should accept a bignum.
3731 ** condition slot accessors are methods.
3732 ** (VECTOR NIL) is a subtype of STRING.
3734 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3735 * minor incompatible change: some nonsensical specialized lambda
3736 lists (used in DEFMETHOD) which were previously ignored now signal
3738 * minor incompatible change: the system is now aware of the types of
3739 variables in the COMMON-LISP package, and will signal errors for
3740 most violations of these type constraints (where previously they
3741 were silently accepted).
3742 * minor incompatible change: COMPILE-FILE now uses the freedom
3743 afforded (ANSI 3.2.2.3) to use derived function types for
3744 functions defined in the same file. This also permits the system
3745 to warn on static type mismatches and function
3746 redefinition. (Currently it does not work with high DEBUG level.)
3747 * minor incompatible change: VALUES declaration is disabled.
3748 * When issuing notes, the compiler now signals a condition of type
3749 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3750 restart for use in user handlers. It is expected that the
3751 COMPILER-NOTE condition will eventually become a condition
3752 supertype to a hierarchy of note types, which will then be
3753 handleable in a similar fashion. However, at the moment, no such
3754 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3755 but it's an implementation detail, not a classification for the
3757 * Changes in type checking closed the following bugs:
3758 ** type checking of unused values (192b, 194d, 203);
3759 ** template selection based on unsafe type assertions (192c, 236);
3760 ** type checking in branches (194bc).
3761 * A short form of VALUES type specifier has ANSI meaning (it has
3762 increased the number of situations when SBCL cannot perform type
3764 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3765 keywords or constants is permissible.
3766 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3767 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3768 classes. (thanks to Antonio Martinez)
3769 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3770 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3771 argument to a pretty-print stream. Instead, it leaves any such
3772 operations to DESCRIBE-OBJECT methods.
3773 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3775 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3776 lambda lists. (thanks to David Lichteblau)
3777 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3778 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3779 (thanks to Gerd Moellmann)
3780 * fixed bug 246: increased compilation speed of long
3781 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3783 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3785 * DEFCONSTANT now throws a condition of type
3786 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3787 constant to a non-EQL value; CONTINUE and ABORT restarts
3788 respectively change and preserve the value.
3789 * fixed bug 63: The code walker, part of the implementation of CLOS,
3790 is now better at handling symbol macros.
3791 * bug fix: There is no longer an internal implementation type named
3792 CL:LENGTH. (reported by Raymond Toy)
3793 * bug fix: In macro-like defining macros/special operators the
3794 implicit block does not enclose the lambda list.
3795 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3796 enclosed in parentheses) are not suitable as type specifiers, and
3797 their use properly signals an error now.
3798 * bug fix: An argument count mismatch for a type specifier in code
3799 being compiled no longer causes an unhandled error at compile
3800 time, but signals a compile-time warning.
3801 * fixed simple vector readable printing
3802 * bug fix: DESCRIBE takes more care over whether the class
3803 precedence list slot of a class is bound before accessing it.
3804 (reported by Markus Krummenacker)
3805 * bug fix: FORMATTER can successfully compile pretty-printer format
3806 strings which use variants of the ~* directive inside.
3807 * bug fix: SEARCH now applies its TEST predicate to the elements of
3808 the arguments in the correct order. (thanks to Wolfhard Buss)
3809 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3810 in type inference. (thanks to Robert E. Brown)
3811 * bug fix: Reading in symbols with an explicit package name of ""
3812 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3813 * fixed some bugs revealed by Paul Dietz' test suite:
3814 ** NIL is now allowed as a structure slot name.
3815 ** Arbitrary numbers, not just REALs, are allowed in certain
3816 circumstances in LOOP for-as-arithmetic clauses.
3817 ** Multiple class redefinitions before slot access no longer
3818 causes a type error.
3819 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3820 association between the name and a class.
3821 ** Generic functions with non-standard method-combination and over
3822 five methods all of which return constants no longer return NIL
3823 after the first few invocations. (thanks to Gerd Moellmann)
3824 ** CALL-NEXT-METHOD with no arguments now passes the original
3825 values of the arguments, even in the presence of assignment.
3826 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3828 ** DISASSEMBLE works with closures and funcallable instances.
3829 ** ADD-METHOD now returns the generic function, not the new method.
3830 ** FIND-METHOD signals an error if the lengths of the specializers
3831 is incompatible with the generic function, even if the ERRORP
3833 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3834 which its argument is a member.
3835 ** DEFCLASS only redefines the class named by its class-name
3836 argument if that name is the proper name of the class;
3837 otherwise, it creates a new class.
3838 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3839 of the UNBOUND-SLOT condition to the name of the slot.
3840 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3841 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3842 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3843 SLOT-MAKUNBOUND in the specified fashion.
3845 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3846 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3847 cross-compilation host. As a consequence, we can now bootstrap our
3848 way up to SBCL starting with a bare gcc toolchain and human-readable
3849 source code (first the source to CLISP, then the source to SBCL).
3850 * A contributed module containing a partial implementation of the
3851 simple-streams interface has been included. (thanks to Rudi
3853 * A contributed module implementing the RFC1321 Message Digest
3854 Algorithm, known as MD5, has been included.
3855 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3856 no longer has any effect, as the code controlled by this feature
3857 has been deleted. (As far as we know, no-one has ever built using
3858 this feature, and its semantics were confused in any case).
3859 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3860 work on conditions (as required by the ANSI specification),
3861 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3862 expected behaviour on conditions. Users should note, however,
3863 that such behaviour is not required by the ANSI specification,
3864 and so use of this behaviour may render their code unportable.
3865 * fixed some bugs revealed by Paul Dietz' test suite:
3866 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3868 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3869 ** SXHASH on condition objects no longer returns NIL.
3870 ** :ALLOCATION :CLASS slots are better treated; their values are
3871 updated on class redefinition, and initforms inherited from
3872 superclasses are applied.
3873 ** REMOVE-METHOD returns its generic function argument even when
3874 no method was removed.
3875 ** SHARED-INITIALIZE now initializes the values of the requested
3876 slots, including those with :ALLOCATION :CLASS.
3877 ** ALLOCATE-INSTANCE now works on structure classes defined via
3878 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3880 ** SLOT-EXISTS-P now works on conditions, as well as structures
3882 ** MAKE-LOAD-FORM now has the required methods on
3883 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3884 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3885 keyword, and now implements the SLOT-NAMES argument.
3886 ** methods with &OPTIONAL arguments no longer allow too many
3887 arguments to be passed in the call without error.
3888 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3889 option is consistent with the required arguments of the generic
3890 function lambda list.
3891 * bug fix: REQUIRE accepts a string designator. (Thanks to
3893 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3894 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3895 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3896 * bug fix: The compiler now checks for duplicated variables in macro
3898 * bug fix: SETQ on globals returns the correct value.
3899 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3900 not a COMPILER-ERROR (followed by some other strange error on
3901 choosing the CONTINUE restart).
3902 * bug fix: make.sh and friends are now more consistent in the way that
3903 they look for GNU "make".
3905 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3906 * experimental native threads support (on x86 Linux >=2.4 only).
3907 This is not compiled in by default: you need to add :SB-THREAD to
3908 the target features. See the "Beyond ANSI" chapter of the manual
3910 * fix for longstanding nonANSIism: The old distinction between
3911 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3912 The return value from CL:FIND-CLASS is now a CLOS class, and
3913 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3914 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3915 * An interface to the MetaObject Protocol, as described in Kiczales,
3916 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3917 MIT Press, 1991, is available from the SB-MOP package.
3918 * incompatible change: the SB-PCL package should now be considered
3919 a private implementation detail, and no longer a semi-private MOP
3921 * minor incompatible change: due to rearrangement for threads, the
3922 control stack and binding stack are now allocated at arbitrary
3923 addresses instead of being hardcoded per-port. Users affected by
3924 this probably have to be doing advanced things with shared
3925 libraries, and will know who they are.
3926 * minor incompatible change: Previously, all --eval forms used were
3927 processed with READ before any of them were processed with EVAL.
3928 Now each --eval form is processed with both READ and EVAL before
3929 the next --eval form is processed. (Thus package operations like
3930 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3931 work as the user might reasonably expect.)
3932 * minor incompatible change: *STANDARD-INPUT* is now only an
3933 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3935 * minor incompatible change: Y-OR-N-P is now character-oriented,
3936 not line oriented. Also, YES-OR-NO-P now works without errors.
3937 (thanks to Antonio Martinez)
3938 * sb-aclrepl module improvements: an integrated inspector, added
3939 repl features, and a bug fix to :trace command.
3940 * Known functions, which cannot be open coded by the backend, are
3941 considered to be able to check types of their arguments. (fixing
3942 a bug report by Nathan J. Froyd)
3943 * fixed a bug in computing method discriminating functions: It is
3944 now possible to define methods specialized on classes which have
3945 forward-referenced superclasses. (thanks to Gerd Moellmann)
3946 * fixed evaluation order in optional entries (reported by Gilbert
3948 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3949 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3950 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3951 required-by-AMOP NAME argument, as well as CLASS and
3952 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3953 * fixed bug 20: DEFMETHOD can define methods using names that are
3954 not the proper names of classes to designate class specializers.
3955 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3956 Unix file descriptors, instead of blowing up. (thanks to Antonio
3958 * Garbage collection refactoring: user-visible change is that a
3959 call to the GC function during WITHOUT-GCING will not do garbage
3960 collection until the end of the WITHOUT-GCING. If you were doing
3961 this you were probably losing anyway.
3962 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3963 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3964 TYPEP the latter but not the former.
3965 * The compiler issues a full WARNING for calls to undefined functions
3966 with names from the CL package.
3967 * MAP-INTO for a vector destination is open coded. (reported by
3968 Brian Downing on c.l.l)
3969 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3970 documentation string.
3971 * fixed some bugs revealed by Paul Dietz' test suite:
3972 ** COPY-ALIST now signals an error if its argument is a dotted
3974 ** Condition slots are now accessed more correctly in the presence
3975 of multiple initargs for a given slot.
3976 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3977 exclude restarts of the same name associated with a different
3979 ** DEFCLASS of forward-referenced classes with another
3980 forward-referenced class in the superclasses list no longer
3982 ** Condition slots are now initialized once each, not multiple
3983 times. (thanks to Gerd Moellmann)
3984 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3986 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3987 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3988 arguments contain duplicated elements.
3989 ** RESTART-CASE understands local macros.
3990 ** RESTART-CASE associates exactly its own restarts with a condition.
3991 ** ENDP in safe mode checks its argument to be of type LIST.
3992 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3993 in question is unbound.
3994 ** Optimized MAKE-INSTANCE functions no longer cause internal
3995 assertion failures in the presence of duplicate initargs.
3996 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3997 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3998 propagated even in the ctor optimized implementation of
4000 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
4002 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
4003 * a better implementation of SXHASH on (simple) bit vectors,
4004 measured both in execution speed and in distribution of results
4005 over the positive fixnums, has been installed. Likewise, a better
4006 implementation of EQUAL for simple bit vectors is now available.
4007 * fixed CEILING optimization for a divisor of form 2^k.
4008 * fixed bug 240 (emitting extra style warnings "using the lexical
4009 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
4010 by Antonio Martinez)
4011 * fixed SXHASH, giving different results for NIL depending on type
4012 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
4013 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
4014 variable. (found by Rolf Wester)
4015 * SBCL does not ignore type declarations for special
4016 variables. (reported by rif on c.l.l 2003-03-05)
4017 * some bug fixes in contrib/sb-aclrepl/
4018 * fixed some bugs revealed by Paul Dietz' test suite:
4019 ** a bug in the CONS type specifier, whereby the CAR and CDR
4020 types got intertwined, has been fixed;
4021 ** the type system is now able to reason about the interaction
4022 between INTEGER and RATIO types more completely;
4023 ** APPEND, [N]REVERSE and NRECONC check that those their
4024 arguments, which must be proper lists, are really so;
4025 ** An array specialized to be unable to hold elements has been
4026 implemented, as required -- yes, really -- by ANSI;
4027 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
4028 on malformed property lists;
4030 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
4031 * incompatible packaging change: in line with Unix convention,
4032 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
4033 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
4034 * REQUIRE and PROVIDE are now optionally capable of doing something
4035 useful. See the documentation string for REQUIRE.
4036 * infrastructure for a managed SBCL contrib system: contributed
4037 modules in this release include:
4038 ** the ASDF system definition facility;
4039 ** an interface to the BSD Sockets API;
4040 ** an ACL-like convenience interface to the repl;
4041 (thanks to Kevin Rosenberg)
4042 ** an implementation of ROTATE-BYTE, with efficient implementation
4044 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
4045 gives a closer approximation to the right answer than previously.
4046 (thanks to Raymond Toy)
4047 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
4048 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
4049 optional environment arguments, as required by ANSI.
4050 * fixed bugs in other functions taking environment objects, allowing
4051 calls with an explicit NIL environment argument to be compiled
4053 * fixed bug 228: primary return values from
4054 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
4055 COMPILE or FUNCTION.
4056 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
4057 :TYPE will no longer signal errors on innocuous objects.
4058 * fixed bug 231b: SETQ is better at respecting type declarations in
4059 the lexical environment.
4060 * fixed a bug in DEFCLASS: classes named by symbols with no or
4061 unprintable packages can now be defined.
4062 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
4063 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
4064 * fixed compiler failure related to checking types of functions.
4065 (reported by Robert E. Brown)
4066 * the compiler is now much more consistent in its error-checking
4067 treatment of bounding index arguments to sequence functions: in
4068 (SAFETY 3) code, errors will be signalled in almost all cases if
4069 invalid sequence bounding indices are passed to functions defined
4070 by ANSI to operate on sequences.
4071 * fixed a bug in the build procedure: documentation of SBCL-specific
4072 packages is now preserved and available in the final Lisp image.
4073 * lifted FDEFINITION lookup out of loops in the implementation of
4074 many list operations. (thanks to Robert E. Brown)
4075 * fixed a bug in the reader: the #n# reader macro now works for
4076 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
4077 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
4078 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
4079 E. Brown and Tony Martinez respectively)
4080 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
4081 SLOT-BOUNDP: in particular, it is now able to optimize them much
4082 better, and is now not vulnerable to having packages renamed.
4083 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
4084 always called when a slot is not present in an instance. (thanks
4086 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
4087 longer signal an error on unreading a character following EOF on
4088 the previous constituent stream. (thanks to Tony Martinez)
4089 * fixed some bugs revealed by Paul Dietz' test suite:
4090 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
4091 not just nonnegative fixnums;
4092 ** the logical bit-array operators such as BIT-AND now accept an
4093 explicit NIL for their "opt-arg" argument (to indicate a
4094 freshly-consed result bit-array);
4095 ** ELT now signals an error on an invalid sequence index in safe
4097 ** the type system is now cleverer about negations of numeric
4098 types, and consequently understands the BIGNUM and RATIO types
4100 ** the type system is now cleverer about the interaction between
4101 INTEGER and RATIO types: while bugs still remain, many more
4102 cases are accurately computed;
4103 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
4104 if it is in the last clause;
4105 ** CONSTANTLY now correctly returns a side-effect-free function in
4107 ** DECLARE is no longer treated as a special-operator; in
4108 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
4109 * incremented fasl file version number due to the change in the
4110 DEFSTRUCT-SLOT-DESCRIPTION structure.
4112 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4113 * minor incompatible change: code processed by the "interpreter" or
4114 EVAL now has a compilation optimization policy of (DEBUG 2)
4115 (changed from (DEBUG 1)) to improve debuggability of interactive
4116 development, and to allow the use of the debug RETURN command in
4118 * an experimental implementation of the RETURN command for the
4119 debugger has been included. (thanks to Frederik Kuivinen)
4120 * fixed bug 62: constraints were not propagated into a loop.
4121 * fixed bug in embedded calls of SORT (reported and investigated by
4123 * fixed some bugs revealed by Paul Dietz' test suite:
4124 ** printing and reading of arrays with some dimensions having
4125 length 0 (thanks to Gerd Moellmann);
4126 ** BOA constructor with &AUX argument without a default value does
4127 not cause a type error;
4128 ** CONSTANTP now returns true for all self-evaluating objects.
4130 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4131 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4132 accessors that are related by inheritance, as specified in the
4133 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4134 Valtteri Vuorikoski)
4135 * The compiler is now able to inline functions that were defined in
4136 a complex lexical environment (e.g. inside a MACROLET).
4137 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4139 * Support for the upcoming FreeBSD-5.0 release has been included.
4140 (thanks to Dag-Erling Smorgrav)
4141 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4142 effect when it is not in a toplevel context.
4143 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4144 stand a better chance of being correct. (thanks to Gerd
4146 * fixed bug in COERCE, which now signals an error on coercing a
4147 rational to a bounded real type which excludes the expected
4149 * The compiler is now able to derive types more accurately from the
4150 COERCE and COMPILE functions.
4151 * fixed bug 223: functional binding is considered to be constant
4152 only for symbols in the CL package.
4153 * fixed bug 231: SETQ did not check the type of a variable being set
4154 (reported by Robert E. Brown)
4155 * A new optimization for MAKE-INSTANCE has been included, fixing
4156 various bugs (including relating to :ALLOCATION :CLASS slots and
4157 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4159 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4160 ** As required by ANSI, LOOP now disallows anonymous collection
4161 clauses such as COLLECT I in conjunction with aggregate boolean
4162 clauses such as THEREIS (= I 1);
4163 ** LOOP now signals an error when any variable is reused in the
4164 same loop (including the potentially useful construct analogous
4165 to WITH A = 1 WITH A = (1+ A);
4166 ** IT is only a special loop symbol within the first clause of a
4167 conditional loop clause;
4168 ** LOOP with a typed iteration variable over a hashtable now
4169 signals a type error iff it should.
4170 * fixed some other bugs revealed by Paul Dietz' test suite:
4171 ** FILE-STREAM now names the class previously known as FD-STREAM;
4172 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4173 argument) no longer signals an error;
4174 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4175 of the slot symbol, rather than using the current package
4176 ((:CONC-NAME "") continues to intern the slot's name in the
4178 * incremented fasl file version number, because of the incompatible
4179 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4180 of the new implementation of DEFINE-COMPILER-MACRO.
4182 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4183 * Support for building SBCL for MIPS platforms running in
4184 little-endian mode has now been checked in, and basic
4185 functionality on said platforms verified.
4186 * minor incompatible change: PCL now records the pathname of a file
4187 in which methods and the like are defined, rather than its
4189 * minor incompatible change: TRUENAME now considers the truename of
4190 a file naming a directory to be the pathname with :DIRECTORY
4191 component indicating that directory.
4192 * minor incompatible change: a NAMED clause in the extended form of
4193 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4194 reason for the previous behaviour is unclear.
4195 * more systematization and improvement of CLOS and MOP conformance
4196 in PCL (thanks to Gerd Moellman and Pierre Mai):
4197 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4199 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4200 primary methods with no specializers;
4201 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4203 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4204 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4205 FORWARD-REFERENCED-CLASSes; error reporting on
4206 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4208 ** SXHASH on CLOS instances now uses a slot internal to the
4209 instance to return different numbers on distinct instances,
4210 while preserving the same return value through invocations of
4212 ** DEFMETHOD signals errors when methods with longer incongruent
4213 lambda lists are added to generic functions;
4214 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4215 CLASS, as specified in AMOP;
4216 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4217 on the order returned by the primary method for classes of
4218 class STANDARD-CLASS;
4219 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4220 * fixed some bugs shown by Paul Dietz' test suite:
4221 ** DOLIST puts its body in TAGBODY;
4222 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4224 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4225 value producing form;
4226 ** if more variables are given to PROGV than values, extra
4227 variables are bound and made to have no value;
4228 ** NSUBSTITUTE on list arguments gets the right answer with
4230 ** ELT signals an error of type TYPE-ERROR when the index argument
4231 is not a valid sequence index;
4232 ** LOOP signals (at macroexpansion time) an error of type
4233 PROGRAM-ERROR when duplicate variable names are found;
4234 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4235 ** FUNCALL of special-operators now cause an error of type
4237 ** PSETQ now works as required in the presence of side-effecting
4238 symbol-macro places;
4239 ** NCONC accepts any object as its last argument;
4240 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4242 ** loop-for-as-package does not require a package to be explicitely
4244 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4246 * fixed bug 166: compiler preserves "there is a way to go"
4247 invariant when deleting code.
4248 * fixed bug 172: macro lambda lists with required arguments after
4249 &REST arguments now cause an error to be signalled. (thanks to
4251 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4252 bug. (thanks to Matthew Danish)
4253 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4255 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4256 arguments are lexically rebound. (thanks to Gerd Moellmann and
4258 * fixed bug 194: error messages are now more informative when there
4259 is no primary method applicable in a call to a generic
4260 function. (thanks to Gerd Moellmann)
4261 * fixed bug in command line argument checking (thanks to Julian
4263 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4264 (thanks to Matthew Danish)
4265 * incremented fasl file version number, because of the
4266 SXHASH-related changes in the layout of CLOS data structures
4268 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4269 * minor incompatible change: The runtime (the Unix executable named
4270 "sbcl") is now much pickier about the .core files it will load.
4271 Essentially it now requires .core files to descend from the same
4272 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4273 as the runtime does. (The intent is to prevent the crashes which
4274 can occur, and which can even be reported as mysterious failures,
4275 when people patch the sources or change the build parameters
4276 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4277 sbcl and .core files.)
4278 * fixed bug: VALUES-LIST is no longer optimized away.
4279 * fixed bug 142: The FFI conversion of C string values to Lisp
4280 string values no longer conses excessively. (thanks to Nathan
4281 Froyd porting Raymond Toy's fix to CMU CL)
4282 * began to systematize and improve MOP conformance in PCL (thanks to
4283 Nathan Froyd, Gerd Moellman and Pierre Mai):
4284 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4286 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4287 ** FINALIZE-INHERITANCE is now called on class finalization;
4288 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4289 argument precedence order.
4290 * fixed bug 202: The compiler no longer fails on functions whose
4291 derived types contradict their declared type.
4292 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4293 so it can be non-toplevel.
4294 * The fasl file version number has changed (because of the new
4295 implementation of DEFMACRO).
4296 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4297 safe code, that any length requirement by their type-specifier
4298 argument is valid. The exceptions to this are described in bug
4300 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4301 functions now check the types of their inputs as required by ANSI.
4302 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4303 introduced symbol is DECLAREd to be SPECIAL.
4304 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4305 * fixed bug: :COUNT argument to sequence functions may be negative
4306 * fixed bug: body of DO-SYMBOLS may contain declarations
4307 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4308 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4309 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4310 symbol macro only once
4311 * fixed printing of call frame when argument list is unavailable
4312 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4313 * compiler no longer signals WARNING on unknown keyword
4316 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4317 * A beta-quality port to the mips architecture running Linux,
4318 based on the old CMUCL backend, has been made. It has been tested
4319 on a big-endian kernel, and works sufficiently well to be able to
4320 rebuild itself; it has not been tested in little-endian mode.
4321 * fixed an inconsistency between gencgc.c and purify.c which made
4322 dumping/loading .core files unreliable
4323 * fixed bug 120a: The compiler now deals correctly with IFs where
4324 the consequent is the same as the alternative, instead of
4325 misderiving the return type. (thanks to Alexey Dejneka)
4326 * fixed bug 113: Logical pathnames are now dumpable (the logical
4327 host is resolved at load-time, throwing an error if it is not
4329 * fixed bug 174: FORMAT's error message is slightly clearer when a
4330 non-printing character is used in a format directive.
4331 * fixed several bugs in compiler checking of type declarations, i.e.
4332 violations of the Python "declarations are assertions" principle
4333 (thanks to Alexey Dejneka)
4334 * fixed several bugs in PCL's error checking (thanks to Gerd
4336 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4338 * fixed bug in compilation of functions as first class values
4339 (thanks to Antonio Martinez-Shotton)
4340 * The compiler's handling TYPE-ERRORs which it can prove will
4341 inevitably happen at runtime has been cleaned up and corrected
4342 in several ways. (thanks to Alexey Dejneka)
4343 * improved argument type checking for various basic arithmetic
4344 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4345 lavished on them in the past that they can be compiled in many
4346 ways in different special cases
4347 * fixed bug 181: compiler checks validity of user supplied type
4349 * cleaned up code flushing in optimization: Function calls which
4350 should signal errors for safety purposes (e.g. which ANSI says
4351 should signal errors when their arguments are of incorrect type)
4352 are no longer optimized away.
4353 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4354 * incremented fasl file version number, because changes in the
4355 implementation of sequence functions like COERCE caused
4356 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4357 in old inline expansions) to become undefined. (Actually these
4358 changes were later undone, so we might very well be binary
4359 compatible with 0.7.7 after all, but leaving the version number
4360 incremented seemed like the simplest and most conservative
4363 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4364 * An alpha-quality port to the parisc architecture running Linux,
4365 based on the old CMUCL backend, has been made. This, even more so
4366 than the other backends, should be considered still a work in
4367 progress; known problems include that the Linux kernel in 64-bit
4368 mode does not propagate the correct sigcontext structure to
4369 userspace, and consequently SBCL on a parisc64 kernel will not
4371 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4372 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4373 suppressed.) Also now that the compiler is looking at declarations
4374 in the environment, it checks optimization declarations as well,
4375 and suppresses inlining when (> DEBUG SPEED).
4376 * More fixes have been made to treatment of floating point exception
4377 treatment and other Unix signals. In particular, floating point
4378 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4379 * The detection and handling of control stack exhaustion (infinite
4380 or very deeply nested recursion) has changed. Stack exhaustion
4381 detection is now done by write-protecting pages at the OS level
4382 and applies at all optimization settings; when found, a
4383 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4384 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4385 be caught using IGNORE-ERRORS.
4386 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4387 correctly when transforming recursive LABELS functions to LETs.
4388 (thanks to Alexey Dejneka)
4389 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4390 that are names of constants or global variables.
4391 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4392 alien routines with docstrings.
4393 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4394 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4396 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4397 (thanks to Eric Marsden)
4398 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4399 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4400 * Bug fix: The SPARC backend can now compile functions involving
4401 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4402 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4403 to a functions expecting a PATHNAME-DESIGNATOR.
4404 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4405 lambda lists. (thanks to Alexey Dejneka)
4406 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4407 functions so that EXPORTing the name of the class doesn't cause
4408 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4409 to Antonio Martinez for reporting this)
4410 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4411 constant-folded without failing an assertion. (thanks to Einar
4412 Floystad Dorum for reporting this)
4413 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4414 types are now dealt with more correctly by the compiler.
4415 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4416 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4417 behaviour analogous to RENAME-FILE. This puts its behaviour more
4418 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4420 * The fasl file version number has changed again. (because of the
4421 bug fix involving the names of PCL MAKE-INSTANCE functions)
4423 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4424 * bug fix: Floating point exceptions are treated much more
4425 consistently on the x86/Linux and PPC/Linux platforms.
4426 * Array initialization with :INITIAL-ELEMENT is now much faster for
4427 cases when the compiler cannot open code the array creation, but
4428 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4429 array accesses have also seen a speed increase.
4430 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4431 specified. (This was at the root of some bad interactions between
4432 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4434 * bug fix: Internal error arguments for undefined functions are now
4435 computed correctly on the PPC/Linux platform.
4436 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4437 Raymond Toy's patch for CMU CL)
4438 * Support for the Solaris 9 operating environment has been included
4439 (thanks to Daniel Merritt)
4440 * A very ugly but hopefully complete draft of the missing FFI chapter
4441 of the manual has been created by reformatting the corresponding
4442 CMU CL manual chapter into (currently very ugly and incoherent)
4443 DocBook and bringing it up to date for SBCL behavior. Thus, the
4444 manual is now essentially complete, at least by my extreme
4445 once-and-only-once standards, whereby it's acceptable to refer to
4446 the doc strings of SB-EXT functions as the primary documentation.
4447 * The fasl file version number has changed again, due to cleanup of
4448 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4449 is no longer a static symbol.)
4451 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4452 * SBCL now builds with OpenMCL (version 0.12) as the
4453 cross-compilation host; also, more progress has been made toward
4454 bootstrapping under CLISP.
4455 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4457 * bug 158 fixed: The compiler can now deal with integer loop
4458 increments different from 1; fixing this turned out also to fix
4460 * bug 169 fixed: no more bogus warnings about using lexical bindings
4461 despite the presence of perfectly good SPECIAL declarations (thanks
4462 to David Lichteblau)
4463 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4464 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4465 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4467 * bug 180 fixed: Method combination specifications no longer ignore
4468 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4469 * bug fix: Structure type predicate functions now check their argument
4470 count as they should.
4471 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4472 correctly. (thanks to Pierre Mai)
4473 * minor incompatible change: The --noprogrammer option is deprecated
4474 in favor of the new --disable-debugger option, which is very similar.
4475 (The major difference is that it takes effect at a slightly different
4476 time at startup, causing handling of errors in --sysinit and
4477 --userinit files will be affected differently.) The
4478 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4479 been added to allow this functionality to be controlled from ordinary
4480 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4481 maintainers, who might want to run non-interactive scripts to
4482 build SBCL cores which will later be used interactively.)
4483 * minor incompatible change: The LOAD function no longer, when given
4484 a wild pathname to load, loads all files matching that pathname.
4485 Instead, an error of type FILE-ERROR is signalled.
4487 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4488 * bug 147 fixed: The compiler preserves its block link/count
4489 invariants more correctly now so that it doesn't crash. (thanks
4491 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4493 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4494 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4495 no-value-for-this-character as NIL (as specified by ANSI).
4496 * HOST-NAMESTRING on physical pathnames now returns a string that is
4497 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4498 (thanks to Christophe Rhodes)
4499 * The Alpha port handles icache flushing more correctly. (thanks to
4501 * More progress has been made toward bootstrapping under CLISP. (thanks
4502 to Christophe Rhodes)
4503 * The fasl file format has changed again, because dynamic loading
4504 on OpenBSD (which has non-ELF object files) motivated some cleanups
4505 in the way that foreign symbols are transformed and passed around.
4506 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4507 is no longer treated as whitespace by the reader, but instead as
4508 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4509 |A
\7fB|, instead of A as it used to.
4511 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4512 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4513 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4514 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4515 this as of 0.7.1.45, but was left out of the previous news section
4516 (thanks to Dan Barlow)
4517 * SBCL now runs on the Solaris operating system on SPARC architectures
4518 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4519 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4520 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4521 * SPARC backend cleanups, allowing builds of cores optimized for V8
4522 and V9 SPARCS, and also emission of code targeted to a particular
4523 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4525 * SBCL is closer to bootstrapping under CLISP, thanks to various
4526 fixes by Christophe Rhodes.
4527 * The fasl file format has changed again, to allow the compiler's
4528 INFO database to support symbol macros.
4529 * The user manual (in doc/) is formatted into HTML more nicely.
4530 (thanks to coreythomas)
4531 * The system is smarter about SUBTYPEP relationships, especially
4532 those involving NOT types (including types such as ATOM which are
4533 represented internally using NOT types). Thus SUBTYPEP is less
4534 likely to return (VALUES NIL NIL) in general, and in particular
4535 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4537 * The fasl file format has changed again, because the internal
4538 representation of types now includes a new slot to support the new
4539 SUBTYPEP-of-NOT-types logic.
4540 * (not a change in the main branch of SBCL, but a related prototype
4541 which can hopefully be merged into the main branch of SBCL in the
4542 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4543 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4544 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4545 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4546 dispatch macro character combinations. (thanks to Alexey Dejneka)
4547 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4548 that superficially look like logical namestrings correctly.
4549 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4551 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4552 default supplied-p) for &optional and &key arguments. (thanks to
4555 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4556 * incompatible change: The compiler is now less aggressive about
4557 tail call optimization, doing it only when (> SPACE DEBUG) or
4558 (> SPEED DEBUG). (This is an incompatible change because there are
4559 programs which relied on the old CMU-CL-style behavior to optimize
4560 away their unbounded recursion which will now die of stack overflow.)
4561 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4562 for non-GENCGC systems has been increased to 20M (since that
4563 seems much closer to the likely performance optimum for modern
4564 systems than the old 4M value was)
4565 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4566 and *DEBUG-PRINT-LEVEL*
4567 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4568 of CMU CL's support for SPARC, and various endianness and other
4569 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4570 * new syntactic sugar for the Unix command line: --load foo.bar is now
4571 an alternate notation for --eval '(load "foo.bar")'.
4573 ** The system now detects stack overflow and handles it gracefully,
4574 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4575 optimization settings. (This is a good thing in general, and
4576 its introduction in this version should be particularly timely
4577 for anyone whose code fails because of suppression of tail
4579 ** The system now hunts for the C variable "environ" in a more
4580 devious way, to avoid segfaults when the C library version differs
4581 between compile time and run time. (thanks to Christophe Rhodes)
4582 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4583 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4584 ** The compiler no longer issues bogus style warnings for undefined
4585 classes in the same source file as the DEFCLASSes which defined
4586 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4588 ** fixes in CONDITION class precedence list for undefined function
4589 errors (thanks to Alexei Dejneka)
4590 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4591 correctly. (thanks to Dan Barlow)
4592 ** portability fixes aiming at bootstrapping under CLISP (thanks
4593 to Dave McDonald and Christophe Rhodes)
4594 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4595 ** fixes in type translation and and type inference (thanks to
4597 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4598 ** various fixes in the new ports (thanks to Dan Barlow)
4599 * several changes related to debugging:
4600 ** suppression of tail recursion, as noted above
4601 ** stack overflow detection, as noted above
4602 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4603 is now the default. (For some time encapsulation has been more
4604 reliable than the breakpoint-based :ENCAPSULATE NIL
4605 implementation, at least on X86 systems; and I just noticed that
4606 encapsulation also seems closer to the spirit of the ANSI
4609 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4611 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4612 up properly again. (There was a packaging bug in 0.7.0 which
4613 left their definitions in SB-SYS::LOAD-FOREIGN and
4614 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4615 vital for most things which interface to C-level interfaces,
4616 like extensions working with sockets or databases or
4617 Perl-compatible regexes or whatever, and the need to fix
4618 this bug is the main reason that 0.7.1 was released so
4620 ** DEFGENERIC is now choosier about the methods it redefines, so that
4621 reLOADing a previously-LOADed file containing DEFGENERICs does
4622 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4623 cycle now works as it should. (thanks to Alexey Dejneka)
4624 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4625 fixed by Christophe Rhodes. (He actually submitted this patch
4626 months ago, and I delayed until after 0.7.0.)
4627 ** Bug 111 (internal compiler confusion about runtime checks on
4628 FUNCTION types) was fixed by Alexey Dejneka.
4629 * Some internal cleanups (getting rid of variables which aren't
4630 needed now that the byte interpreter is gone) caused the fasl
4631 file format number to change again.
4633 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4634 * major incompatible change: The default fasl file extension, i.e. the
4635 default extension for files produced by COMPILE-FILE, has changed
4636 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4638 ** There are many changes in the implementation of the compiler.
4639 SBCL is now essentially a compiler-only implementation of ANSI
4640 Common Lisp. EVAL still "interprets" a few special cases, but
4641 almost all the interesting cases are handled by creating
4642 a LAMBDA expression, calling COMPILE on it, then calling
4643 FUNCALL on the result.
4644 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4645 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4646 Since the code is newer, there might still be some new bugs
4647 (though not as many as before Martin Atzmueller's fixes:-). But
4648 the new code is substantially simpler and clearer, and hopefully
4649 any remaining bugs will be simpler, less fundamental, and more
4650 fixable then the bugs in the old code.
4651 ** The revised compiler is still a little unsteady on its feet.
4653 *** The debugging information it produces (particularly the names
4654 of FUNCTION objects) is sometimes much less useful than what
4655 the old compiler produced.
4656 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4657 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4658 within a MACROLET) has been temporarily weakened.
4659 ** There are new compiler optimizations for various functions:
4660 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4661 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4662 *** the math functions TRUNCATE, FLOOR, and CEILING
4663 *** the function-of-all-trades COERCE
4664 Mostly these should be transparent, but there's one
4665 potentially-annoying problem (bug 117): when the compiler
4666 inline-expands a function and does type analysis on the result,
4667 it can create control paths which have type mismatches, and
4668 when it can't prove that those control paths aren't taken,
4669 it will issue WARNINGs about the type mismatches. This is
4670 a particular problem in practice for the new sequence functions.
4671 It's not clear how this should be fixed, and for now, a
4672 workaround is given in the entry for 117 in the BUGS file.
4673 ** (Because of the interaction between the two previous items --
4674 occasional inlining problems and new inline expansions -- some
4675 of the new sequence function optimizations won't really kick in
4676 completely until debugging information, and then inlining, are
4677 straightened out in some future version.)
4678 * minor incompatible changes:
4679 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4680 DIRECTORY no longer implicitly promotes NIL slots of its
4681 pathname argument to :WILD. In particular, when you ask for the
4682 contents of a directory (which you used to be able to do without
4683 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4684 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4685 ** changes in behavior that ANSI explicitly defines to be
4686 implementation dependent:
4687 *** The new compiler-only implementation still conforms with ANSI,
4688 but acts a little different than before. Besides the obvious
4689 changes in performance tradeoffs (that the cost per form passed
4690 to EVAL has gone up, and the cost per form executed by EVAL
4691 has gone down), the behavior of the system changes a little
4692 because there are no longer any interpreted function objects.
4693 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4694 e.g. doing COMPILE on the output of interactive DEFUN is
4696 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4698 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4699 has changed from (:CL) to NIL.
4700 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4701 e.g. #\Newline and #\Tab, don't have names specified in the
4702 ANSI Common Lisp standard, is now based on their ASCII symbolic
4703 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4704 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4705 are no longer used for output.
4706 ** changes in internal implementation constants:
4707 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4708 4 million. (If your application spends a lot of time GCing and
4709 you have a lot of RAM, you might want to experiment with
4710 increasing it even more.)
4711 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4712 However, almost all old code should still continue to work without
4713 immediate update, as SB-C-CALL is now a (deprecated) nickname
4715 ** Old operator names in the style DEF-FOO are now deprecated in
4716 favor of new corresponding names DEFINE-FOO, for consistency with
4717 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4718 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4719 internal symbols, but a few supported extensions like
4720 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4721 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4722 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4723 etc. as you get deeper into recursive calls to the debugger
4724 command loop, instead of the old "5]", "5]]", "5]]]"
4725 sequence. (I was motivated to do this when squabbles between
4726 ILISP and SBCL left me very deeply nested in the debugger. In the
4727 short term, this change will probably provoke more ILISP/SBCL
4728 squabbles, but hopefully it will be an improvement in the long run.)
4729 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4730 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4731 function, since declaiming return types involving aliens is
4732 (1) annoyingly messy to do by hand and (2) vital to efficient
4733 compilation of code which calls such functions.
4734 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4735 longer reexported by the SB-EXT package. They're solely useful
4736 for alien code, so it seems more logical that you should get
4737 them from the SB-ALIEN package, not in SB-EXT.
4738 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4739 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4740 features. Instead, the code that they used to control is always
4741 built into the system.
4742 * many other bug fixes
4743 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4744 advantage of the new EVAL-WHEN stuff and to clean them up in
4745 general, and they are now more ANSI-compliant in a number of
4746 ways. Martin Atzmueller is responsible for a lot of this.
4747 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4749 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4750 *** correct ERROR type for various file operations
4751 *** some fixes for Lisp streams
4752 *** DEFMETHOD syntax checking
4753 *** changing old weird representation of debug information as
4754 strings (which, among their other deficiencies, don't transform
4755 correctly when you rename packages, and don't change their
4756 print representation when you change things like *PACKAGE*
4757 and *PRINT-LENGTH*) to symbols and lists of symbols
4758 He also made several improvements and fixed several bugs in DESCRIBE.
4759 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4761 *** misbehavior of WRITE-STRING/WRITE-LINE
4762 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4763 and several other LOOP problems as well
4764 *** DIRECTORY when similar filenames are present
4765 *** DEFGENERIC with :METHOD options
4766 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4767 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4768 *** argument ordering in FIND with :TEST option
4769 *** mishandled package designator argument in APROPOS-LIST
4770 *** various problems in the backquote readmacro
4771 *** a bug in APROPOS
4772 *** probably some others that I'm not describing very well here,
4773 since the CVS log documents them by reference to sbcl-devel
4774 messages, and the SourceForge archives aren't working well.:-(
4775 ** Dan Barlow improved the Alpha port (and is making progress on the
4776 PPC port, for those of you who think different).
4777 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4778 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4779 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4780 continues to work on the SPARC port (for those of you in a position
4781 to look down upon our little PC-compatible boxes from a great height).
4782 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4783 and uses that copy, rather than the current dynamic value, when
4784 it's trying to decide whether to truncate output. Thus e.g.
4785 (let ((*print-lines* 50))
4786 (pprint-logical-block (stream nil)
4788 (let ((*print-lines* 8))
4789 (print (aref possiblybigthings i) stream)))))
4790 should now truncate the logical block only at 50 lines, instead of
4791 often truncating it at 8 lines, as it did before.
4792 * The doc/cmucl/ directory, containing old CMU CL documentation
4793 from the time of the fork, is no longer part of the base system.
4794 SourceForge has shut down its anonymous FTP service, and with it
4795 my original plan for distributing the old CMU CL documentation
4796 there. For now, if you need these files you can download an old
4797 SBCL source release and extract them from it.
4798 * The fasl file version number changed again, for dozens of reasons,
4799 some of which are apparent above.
4801 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4802 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4803 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4804 patch, so that DISASSEMBLE output is much nicer.
4805 * The code in the SB-PROFILE package now seems reasonably stable.
4806 I still haven't decided what the final interface should look like
4807 (I'd like PROFILE to interact cleanly with TRACE, since both
4808 facilities use function encapsulation) but if you have a need
4809 for profiling now, you can probably use it successfully with
4810 the current CMU-CL-style interface.
4811 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4812 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4813 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4814 link now returns the dangling link itself, and for similar
4815 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4816 link itself. (In these cases the old code signalled an error and
4817 looped endlessly, respectively.) Thus, DIRECTORY now works even
4818 in the presence of dangling and cyclic symbolic links.
4819 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4820 is now a supported extension again, since the consensus on
4821 sbcl-devel was that it can be useful for ordinary development
4822 work, not just for debugging SBCL itself.
4823 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4824 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4825 that currently-defined functions might be redefined later with
4826 different return types.
4827 * Hash tables can be printed readably, as inspired by CMU CL code
4828 of Eric Marsden and SBCL code of Martin Atzmueller.
4829 * better error handling in CLOS method combination, thanks to
4830 Martin Atzmueller porting Pierre Mai's CMU CL patches
4831 * more overflow fixes for >16Mbyte I/O buffers
4832 * A bug in READ has been fixed, so that now a single Ctrl-D
4833 character suffices to cause end-of-file on character streams.
4834 In particular, now you only need one Ctrl-D at the command
4835 line (not two) to exit SBCL.
4836 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4838 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4839 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4840 does the right thing.
4841 * The implementation of some type tests, especially for CONDITION
4842 types, is now tidier and maybe faster, due to CMU CL code
4843 originally by Douglas Crosher, ported by Martin Atzmueller.
4844 * Some math functions have been fixed, and there are new
4845 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4846 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4847 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4848 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4849 likely to be built into sbcl-0.7.0.)
4850 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4851 CL, ported to SBCL by Martin Atzmueller)
4852 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4853 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4854 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4855 * There's a new slam.sh hack to shorten the edit/compile/debug
4856 cycle for low-level changes to SBCL itself, and a new
4857 :SB-AFTER-XC-CORE target feature to control the generation of
4858 the after-xc.core file needed by slam.sh.
4859 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4860 COMPILE-FILE is no longer supported, so that now every function
4861 gets an entry point, so that block compilation looks a little
4862 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4863 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4864 returns the number of bytes consed since the system started,
4865 rather than the number consed since the first time the function
4866 was called. (The new definition parallels ANSI functions like
4867 CL:GET-INTERNAL-RUN-TIME.)
4868 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4869 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4870 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4871 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4872 * Fasl file version numbers are now independent of the target CPU,
4873 since historically most system changes which required version
4874 number changes have affected all CPUs equally. Similarly,
4875 the byte fasl file version is now equal to the ordinary
4878 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4879 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4880 is no longer recognized. I apologize for this, because it was
4881 listed in SB-EXT as a supported extension, but I found that
4882 its existing behavior was poorly specified, as well as incorrectly
4883 specified, and it looked like too much of a mess to straighten it
4884 out. I have enough on my hands trying to get ANSI stuff to work..
4885 * many patches ported from CMU CL by Martin Atzmueller, with
4886 half a dozen bug fixes in pretty-printing and the debugger, and
4887 half a dozen others elsewhere
4888 * fixed bug 13: Floating point infinities are now supported again.
4889 They might still be a little bit flaky, but thanks to bug reports
4890 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4891 as flaky as they were.
4892 * The --noprogrammer command line option is now supported. (Its
4893 behavior is slightly different in detail from what the old man
4894 page claimed it would do, but it's still appropriate under the
4895 same circumstances that the man page talks about.)
4896 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4897 are now supported, and enabled by default. Thus, the compiler can
4898 handle many floating point and complex operations much less
4899 inefficiently. (Thus e.g. you can implement a complex FFT
4901 * The compiler now detects type mismatches between DECLAIM FTYPE
4902 and DEFUN better, and implements CHECK-TYPE more correctly, and
4903 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4904 thanks to patches from Martin Atzmueller.
4905 * various fixes to make the cross-compiler more portable to
4906 ANSI-conforming-but-different cross-compilation hosts (notably
4907 Lispworks for Windows, following bug reports from Arthur Lemmens)
4908 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4909 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4910 fixed thanks to Pierre Mai's CMU CL patches.
4911 * fixes to keep the system from overflowing internal counters when
4912 it tries to use i/o buffers larger than 16M bytes
4913 * fixed bug 45a: Various internal functions required to support
4914 complex special functions have been merged from CMU CL sources.
4915 (When I was first setting up SBCL, I misunderstood a compile-time
4916 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4917 * improved support for type intersection and union, fixing bug 12
4918 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4919 more obscure bugs as well
4920 * some steps toward byte-compiling non-performance-critical
4921 parts of the system, courtesy of patches from Martin Atzmueller
4922 * Christophe Rhodes has made some debian packages of sbcl at
4923 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4924 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4925 stable, but are nonetheless usable. When he's ready, I'd be happy
4926 to add them to the SourceForge "File Releases" section. (And if
4927 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4928 * new fasl file format version number (because of changes in
4929 internal representation of (OR ..) types to accommodate the new
4930 support for (AND ..) types, among other things)
4932 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4933 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4935 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4937 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4938 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4939 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4940 * DESCRIBE now works on structure objects again.
4941 * Most function call argument type mismatches are now handled as
4942 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4943 know whether the function will be redefined before the call is
4944 executed. (The compiler could flag local calls with full WARNINGs,
4945 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4946 it doesn't keep track of enough information to know whether calls
4947 are local in this sense.)
4948 * Compiler output is now more verbose, with messages truncated
4949 later than before. (There should be some supported way for users
4950 to override the default verbosity, but I haven't decided how to
4951 provide it yet, so this behavior is still controlled by the internal
4952 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4953 src/compiler/ir1util.lisp.)
4954 * Fasl file format version numbers have increased again, because
4955 support for the Gray streams extension changes the layout of the
4956 system's STREAM objects.
4957 * The Gray subclassable streams extension now works, thanks to a
4958 patch from Martin Atzmueller.
4959 * The full LOAD-FOREIGN extension (not just the primitive
4960 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4961 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4962 but like most other programs, it defaults to copying the Unix
4963 environment from the original process instead of starting the
4964 new process in an empty environment.
4965 * Extensions which manipulate the Unix environment now support
4966 an :ENVIRONMENT keyword option which doesn't smash case or
4967 do other bad things. The CMU-CL-style :ENV option is retained
4968 for porting convenience.
4969 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4970 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4972 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4974 * A patch from Martin Atzmueller seems to have solved the SIGINT
4975 problem, and as far as we know, signal-handling now works cleanly.
4976 (If you find any new bugs, please report them!)
4977 * The system no longer defaults Lisp source file names to types
4978 ".l", ".cl", or ".lsp", but only to ".lisp".
4979 * The compiler no longer uses special default file extensions for
4980 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4981 seems to expect a single default extension for all compiled code,
4982 and there's no compelling reason to try to stretch the standard
4983 to allow two different extensions.) Instead, byte-compiled files
4984 default to the same extension as native-compiled files.
4985 * Fasl file format version numbers have increased again, because
4986 a rearrangement of internal implementation packages made some
4987 dumped symbols in old fasl files unreadable in new cores.
4988 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4989 many fewer weird special cases.
4990 * Bug #17 (differing COMPILE-FILE behavior between logical and
4991 physical pathnames) has been fixed, and some related misbehavior too,
4992 thanks to a patch from Martin Atzmueller.
4993 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4994 by Tim Moore, ported to SBCL by Martin Atzmueller.
4995 * Martin Atzmueller fixed several filesystem-related problems,
4996 including bug #36, in part by porting CMU CL patches, which were
4997 written in part by Paul Werkowski.
4998 * More compiler warnings in src/runtime/ are gone, thanks to
4999 more patches from Martin Atzmueller.
5000 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
5003 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
5005 * DESCRIBE now works on CONDITION objects.
5006 * The debugger now handles errors which arise when trying to print
5007 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
5009 * The build system now uses an additional file, customize-target-features.lisp,
5010 to allow local modifications to the target *FEATURES* list. (The point of
5011 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
5012 debugging features enabled, without having to worry about propagating it
5013 into everyone's system when I do a "cvs update".) When no
5014 customize-target-features.lisp file exists, the target *FEATURES* list
5015 should be constructed the same way as before.
5016 * fixed bugs in DEFCONSTANT ANSI-compatibility:
5017 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
5018 warn about behavior which is undefined under the ANSI spec. Note: This
5019 is specified by ANSI, but it's not very popular with programmers.
5020 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
5021 macro in the SBCL sources for an example of a workaround which you
5022 might use to make portable ANSI-standard code which does what you want.
5023 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
5024 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
5025 when it's used as a non-toplevel form. (This is required in order
5026 to implement the DEFCONSTANT-EQX macro.)
5027 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
5028 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
5029 funny things, due to bugs in the implementation of EVAL-WHEN
5030 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
5031 non-toplevel DEFCONSTANTs are uncommon.)
5032 * The core file version number and fasl file version number have been
5033 incremented, because the old noncompliant DEFCONSTANT behavior involved
5034 calling functions which no longer exist, and because I also took the
5035 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
5036 * fixed bug 1 (error handling before read-eval-print loop starts), and
5037 redid debugger restarts and related debugger commands somewhat while
5039 ** The QUIT debugger command is gone, since it did something
5040 rather different than the SB-EXT:QUIT command, and since it never
5041 worked properly outside the main toplevel read/eval/print loop.
5042 Invoking the new TOPLEVEL restart provides the same functionality.
5043 ** The GO debugger command is also gone, since you can just invoke
5044 the CONTINUE restart directly instead.
5045 ** The TOP debugger command is also gone, since it's redundant with the
5046 FRAME 0 command, and since it interfered with abbreviations for the
5048 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
5050 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
5051 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
5052 some porting work by Martin Atzmueller)
5053 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
5054 with a boatload of patches which he ported to SBCL. Now that those
5056 ** The system tries to make sure that its low-priority messages
5057 are prefixed by semicolons, to help people who like to use
5058 syntax highlighting in their ILISP buffer. (This patch
5059 was originally due to Raymond Toy.)
5060 ** The system now optimizes INTEGER-LENGTH better, thanks to more
5061 patches originally written by Raymond Toy.
5062 ** The compiler understands coercion between single-value and
5063 multiple-VALUES type expressions better, getting rid of some very
5064 weird behavior, thanks to patches originally by Robert MacLachlan
5065 and Douglas Crosher.
5066 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
5067 lambda lists, thanks to a patch originally by Pierre Mai.
5068 ** The system no longer bogusly warns about "abbreviated type
5070 ** The compiler gets less confused by inlining and RETURN-FROM,
5071 thanks to some patches originally by Tim Moore.
5072 ** The system no longer hangs when dumping circular lists to fasl
5073 files, thanks to a patch originally from Douglas Crosher.
5074 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
5075 undefined function error.
5076 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
5077 in structure slot initforms) for now, documented workaround instead:-|
5078 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
5079 * fixed bug 5: added stubs for various Gray stream functions called
5080 in the not-a-CL:STREAM case, so that even when Gray streams aren't
5081 installed, at least appropriate type errors are generated
5082 * fixed bug 8: better reporting of various PROGRAM-ERRORs
5083 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
5084 consistently in DEFMETHOD forms.
5085 * removed bug 21 from BUGS, since Martin Atzmueller points out that
5086 it doesn't seem to affect SBCL after all
5087 * The C runtime system now builds with better optimization and many
5088 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
5090 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
5092 * The system is now under CVS at SourceForge (instead of the
5093 CVS repository on my home machine).
5094 * The new signal handling code has been tweaked to treat register
5095 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
5096 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
5097 downstream, e.g. in debug-int.lisp, has implicit dependencies
5098 on the unsignedness of integer representation of machine words,
5099 and that caused the system to bomb out with infinite regress
5100 when trying to recover from type errors involving signed values,
5101 e.g. (BUTLAST '(1 2 3) -1).)
5102 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
5103 (This was one of the bugs Peter Van Eynde reported back in July.)
5104 * The system now uses code inspired by Colin Walters' O(N)
5105 implementation of MAP (from the cmucl-imp@cons.org mailing
5106 list, 2 September 2000) when it can't use a DEFTRANSFORM to
5107 inline the MAP operation, and there is more than one
5108 sequence argument to the MAP call (so that it can't just
5109 do ETYPECASE once and for all based on the type of the
5110 single sequence argument). (The old non-inline implementation
5111 of the general M-argument sequence-of-length-N case required
5112 O(M*N*N) time when any of the sequence arguments were LISTs.)
5113 * The QUIT :UNIX-CODE keyword argument has been renamed to
5114 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5115 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5117 * Raymond Wiker's patches to port dynamic loading from Linux to
5118 FreeBSD have been added.
5119 * The BUGS file is now more nearly up to date, thanks in large part
5120 to Martin Atzmueller's review of it.
5121 * The debugger now flushes standard output streams before it begins
5122 its output ("debugger invoked" and so forth).
5123 * The core version number and fasl file version number have both
5124 been incremented, because of incompatible changes in the layout
5126 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5127 used to not be called for a saved Lisp image.)
5128 * Martin Atzmueller's version of a patch to fix a compiler crash,
5129 as posted on sbcl-devel 13 September 2000, has been installed.
5130 * Instead of installing Martin Atzmueller's patch for the
5131 compiler transform for SUBSEQ, I deleted the compiler transform,
5132 and transforms for some similar consing operations.
5133 * A bug in signal handling which kept TRACE from working on OpenBSD
5135 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5136 away by constant folding
5137 * The system now defines its address space constants in one place
5138 (in the Lisp sources), and propagates them automatically elsewhere
5139 (through GENESIS and the sbcl.h file). Therefore, patching the
5140 address map is less unnecessarily tedious and error-prone. The
5141 Lisp names of address space constants have also been systematized.
5142 * CVS tags like dollar-Header-dollar have been removed from
5143 the sources, because they have never saved me trouble and
5144 they've been source of trouble working with patches and other
5145 diff-related operations.
5146 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5147 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5149 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5151 * The system has been ported to OpenBSD.
5152 * The system now compiles with a simple "sh make.sh" on the systems
5153 that it's supported on. I.e., now you no longer need to tweak
5154 text in the target-features.lisp-expr and symlinks in src/runtime/
5155 by hand, the make.sh takes care of it for you.
5156 * The system is no longer so grossly inefficient when compiling code
5157 involving vectors implemented as general (not simple) vectors (VECTOR T),
5158 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5159 which dares to use the various sequence functions on non-simple
5160 vectors, takes less of a performance hit.
5161 * There is now a primitive type predicate VECTOR-T-P
5162 to test for the (VECTOR T) type, so that e.g.
5163 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5164 can now be compiled with some semblance of efficiency. (The old code
5165 turned the type declaration into a full call to %TYPEP at runtime!)
5166 * AREF on (VECTOR T) is still not fast, since it's still compiled
5167 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5168 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5169 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5170 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5171 without making full calls to SUBTYPEP at runtime.
5172 (Some analogous efficiency issues for non-simple vectors specialized to
5173 element types other than T, or for non-simple multidimensional arrays,
5174 have not been addressed. They could almost certainly be handled the
5175 same way if anyone is motivated to do so.)
5176 * The changes in array handling break binary compatibility, so
5177 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5178 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5179 instead of (VALUES T T).
5180 * By following the instructions that Dan Barlow posted to sbcl-devel
5181 on 2 July 2000, I was able to enable primitive dynamic object
5182 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5183 functionality is not implemented (since it calls ld to resolve
5184 library references automatically, requiring RUN-PROGRAM for its
5185 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5186 try to resolve library references) is now supported.
5187 * The system now flushes the standard output streams when it terminates,
5188 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5189 them at several other probably-convenient times, e.g. in each pass of
5190 the toplevel read-eval-print loop, and after evaluating a form given
5191 as an "--eval" command-line option. (These changes were motivated by a
5192 discussion of stream flushing issues on cmucl-imp in August 2000.)
5193 * The source transform for TYPEP of array types no longer assumes
5194 that an array whose element type is a not-yet-defined type
5195 is implemented as an array of T, but instead punts, so that the
5196 type will be interpreted at runtime.
5197 * There is now some support for cross-compiling in make.sh: each of
5198 the phases of make.sh has its own script. (This should be transparent
5199 to people doing ordinary, non-cross-compile builds.)
5200 * Since my laptop doesn't have hundreds of megabytes of memory like
5201 my desktop machine, I became more motivated to do some items on
5202 my to-do list in order to reduce the size of the system a little:
5203 ** Arrange for various needed-only-at-cold-init things to be
5204 uninterned after cold init. To support this, those things have
5205 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5206 symbols with such names are now uninterned after cold init).
5207 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5208 fasl files for cold load.
5209 ** Remove the old compiler structure pooling code (which used to
5210 be conditional on the target feature :SB-ALLOC) completely.
5211 ** Redo the representation of some data in cold init to be more compact.
5212 (I also looked into supporting byte compiled code at bootstrap time,
5213 which would probably reduce the size of the system a lot, but that
5214 looked too complicated, so I punted for now.)
5215 * The maximum signal nesting depth in the src/runtime/ support code has
5216 been reduced from 4096 to 256. (I don't know any reason for the very
5217 large old value. If the new smaller value turns out to break something,
5218 I'll probably just bump it back up.)
5219 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5221 * Many, many bugs reported by Peter Van Eynde have been added to
5222 the BUGS list; some have even been fixed.
5223 * While enabling dynamic object file loading, I tried to make the
5224 code easier to understand, renaming various functions and variables
5225 with less ambiguous names, and changing some function calling
5226 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5227 * While trying to figure out how to do the OpenBSD port, I tried to
5228 clean up some of the code in src/runtime/. In particular, I dropped
5229 support for non-POSIX signal handling, added various comments,
5230 tweaked the code to reduce the number of compilation warnings, and
5231 renamed some files to increase consistency.
5232 * To support the new automatic configuration functionality in make.sh,
5233 the source file target-features.lisp-expr has been replaced with the
5234 source file base-target-features.lisp-expr and the machine-generated
5235 file local-target-features.lisp-expr.
5236 * fixed a stupid quoting error in make.sh so that using CMU CL
5237 "lisp -batch" as cross-compilation host works again
5239 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5241 * DESCRIBE no longer tries to call itself recursively to describe
5242 bound/fbound values, so that it no longer fails on symbols which are
5243 bound to themselves (like keywords, T, and NIL).
5244 * DESCRIBE now works on generic functions.
5245 * The printer now prints less-screwed-up representations of closures
5246 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5247 * A private symbol is used instead of the :EMPTY keyword previously
5248 used to mark empty slots in hash tables. Thus
5249 (DEFVAR *HT* (MAKE-HASH-TABLE))
5250 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5251 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5252 now does what ANSI says that it should. (You can still get
5253 similar noncompliant behavior if bang on the hash table
5254 implementation with all the symbols you get back from
5255 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5256 This breaks binary compatibility, since tests for equality to
5257 :EMPTY are wired into things like the macroexpansion of
5258 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5260 * There's now a minimal placeholder implementation for CL:STEP,
5261 as required by ANSI.
5262 * An obscure bug in the interaction of the normal compiler, the byte
5263 compiler, inlining, and structure predicates has been patched
5264 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5265 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5267 * Missing ordinary arguments in a macro call are now detected even
5268 when the macro lambda list contains &KEY or &REST.
5269 * The debugger no longer complains about encountering the top of the
5270 stack when you type "FRAME 0" to explicitly instruct it to go to
5271 the top of the stack. And it now prints the frame you request even
5272 if it's the current frame (instead of saying "You are here.").
5273 * As specified by ANSI, the system now always prints keywords
5274 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5276 * The default initial SIZE of HASH-TABLEs is now smaller.
5277 * Type information from CLOS class dispatch is now propagated
5278 into DEFMETHOD bodies, so that e.g.
5279 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5281 is now basically equivalent to
5282 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5283 (DECLARE (TYPE SINGLE-FLOAT X))
5285 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5286 operation, without having to do run-time type dispatch.
5287 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5288 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5289 * Since I don't seem to be making big file reorganizations very often
5290 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5291 is overflowing my ability to conveniently back them up), I've finally
5292 checked the system into CVS. (The CVS repository is on my home system,
5293 not at SourceForge -- putting it on SourceForge might come later.)
5294 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5295 high-level GC-NOTIFY-FOO functions send their output. (There's
5296 still very little control of where low-level verbose GC functions
5297 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5298 less than it used to -- the GC-NOTIFY-FOO functions are now under
5299 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5300 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5301 in only one place in the source code, and propagates it automatically
5302 everywhere that it's needed. Thus e.g. when I bump the version from
5303 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5304 * The C source files now include boilerplate legalese and documentation
5305 at the head of each file (just as the Lisp source files already did).
5306 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5307 to his page will be replaced with a link to his new CLiki service.
5309 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5311 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5312 * The build process now looks for GNU make under the default name "gmake",
5313 instead of "make" as it used to. If GNU make is not available as "gmake"
5314 on your system, you can change this default behavior by setting the
5315 GNUMAKE environment variable.
5316 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5317 can build without error under CMU CL.
5319 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5321 * There is now a partial SBCL user manual (with some new text and some
5322 text cribbed from the CMU CL manual).
5323 * The beginnings of a profiler have been added (starting with the
5324 CMU CL profiler and simplifying and cleaning up). Eventually the
5325 main interface should be through the TRACE macro, but for now,
5326 it's still accessed through vaguely CMU-CL-style functions and macros
5327 exported from the package SB-PROFILE.
5328 * Some problems left over from porting CMU CL to the new
5329 cross-compilation bootstrap process have been cleaned up:
5330 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5331 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5332 part when getting rid of the compiler *BACKEND* stuff.)
5333 ** The value of *NULL-TYPE* was screwed up, because it was
5334 being initialized before the type system knew the final
5335 definition of the 'NULL type. This screwed up several key
5336 optimizations in the compiler, causing inefficiency in all sorts
5337 of places. (I found it because I wanted to understand why
5338 GET-INTERNAL-RUN-TIME was consing.)
5339 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5340 PROCLAIM FTYPE information. Unfortunately this broke binary
5341 compatibility again, since now the forms output by DEFGENERIC
5342 to refer to functions which didn't exist in 0.6.3.
5343 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5344 can use the new (as of 0.6.3) transform for SOME into MAP into
5346 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5347 values to (INTEGER 0 1000000), so that the time code will no longer
5348 occasionally get blown up by Linux returning 1000000 microseconds
5349 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5350 its output conform to the ANSI spec. (Alas, this makes its output
5351 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5352 * A full call to MAP NIL with a single sequence argument no longer conses.
5353 * fixes to problems pointed out by Martin Atzmueller:
5354 * The manual page no longer talks about multiprocessing as though
5355 it were currently supported.
5356 * The ILISP support patches have been removed from the distribution,
5357 because as of version 5.10.1, ILISP now supports SBCL without us
5358 having to maintain patches.
5359 * added a modified version of Raymond Toy's recent CMU CL patch for
5360 EQUALP comparison of HASH-TABLE
5362 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5364 * The system still can't cross-compile itself with
5365 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5366 entails), but at least it can compile more of itself that way
5367 than it used to be able to, and various buglets which were uncovered
5368 by trying to cross-compile itself that way have now been fixed.
5369 * This release breaks binary compatibility again. This time
5370 at least I've incremented the FASL file format version to 2, so that the
5371 problem can be detected reliably instead of just causing weird errors.
5372 * various new style warnings:
5373 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5374 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5375 ** using the lexical binding of a variable named in the *FOO* style
5376 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5377 as specified by ANSI.
5378 * *RANDOM-STATE* is no longer automatically initialized from
5379 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5380 default behavior of the system is to repeat its behavior every time
5381 it's run. If you'd like to change this behavior, you can always
5382 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5383 old convention there was no comparably easy way to get the system to
5384 repeat its behavior every time it was run.
5385 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5386 FTYPE declarations has been removed, in favor of their ANSI
5387 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5388 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5389 the types of their sequence arguments can be determined at compile time.
5390 This is done through a new open code expansion for MAP which eliminates
5391 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5392 argument types can be determined at compile time.
5393 * The optimizer now transforms COERCE into an identity operation when it
5394 can prove that the coerced object is already of the correct type. (This
5395 can be a win for machine generated code, including the output of other
5396 optimization transforms, such as the MAP transform above.)
5397 * Credit information has been moved from source file headers into CREDITS.
5398 * Source file headers have been made more standard.
5399 * The CASE macro now compiles without complaining even when it has
5402 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5404 * (Note that the way that the PCL macroexpansions were rewritten
5405 to accommodate the change in DEFGENERIC below breaks binary
5406 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5407 not run under sbcl-0.6.2. Once we get out of alpha releases,
5408 i.e. hit release 1.0.0, we'll probably try to maintain binary
5409 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5410 and sbcl-1.4.4. Until then, however, it might be fairly common
5411 for maintenance releases to break binary compatibility.)
5412 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5414 * The reporting of unhandled signals has been changed to print some
5415 explanatory text as well as the report form. (Previously only
5416 the report form was printed.)
5417 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5418 it defines, so that the compiler no longer issues undefined function
5419 warnings for compiled-but-not-yet-loaded generic functions.
5420 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5421 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5422 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5423 section of the ANSI spec are supported.
5424 * The "" nickname for the "KEYWORD" package has been removed.
5425 The reader still handles symbol tokens which begin with a package marker
5426 as keywords, but it doesn't expose its mechanism for doing so in the
5427 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5428 * The system now issues STYLE-WARNINGs for contradictory TYPE
5429 proclamations. (Warnings for contradictory FTYPE proclamations would
5430 be nice too, but those can't be done usefully unless the type system
5431 is made smarter about FUNCTION types.)
5432 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5433 systematized, so that "*target-*.lisp is supposed to exist only on the
5434 target and imply that there's a related file which exists on the
5435 host, and *host-*.lisp is supposed to exist only on the host and imply
5436 that there's a related file which exists on the target. This involves a
5437 lot of renaming. Hopefully the acute confusion caused by the renaming
5438 will be justified by the reduction in chronic confusion..
5439 ** runtime-type.lisp -> early-target-type.lisp
5440 ** target-type.lisp -> late-target-type.lisp
5441 ** early-host-format.lisp -> early-format.lisp
5442 ** late-host-format.lisp -> late-format.lisp
5443 ** host-error.lisp -> misc-error.lisp
5444 ** early-error.lisp -> early-target-error.lisp
5445 ** late-error.lisp -> late-target-error.lisp
5446 ** host-defboot.lisp -> early-defboot.lisp
5447 ** code/misc.lisp -> code/target-misc.lisp
5448 ** code/host-misc.lisp -> code/misc.lisp
5449 ** code/numbers.lisp -> code/target-numbers.lisp
5450 ** code/early-numbers.lisp -> numbers.lisp
5451 ** early-host-type.lisp -> early-type.lisp
5452 ** late-host-type.lisp -> late-type.lisp
5453 ** host-typep.lisp -> typep.lisp
5454 ** load.lisp -> target-load.lisp
5455 ** host-load.lisp -> load.lisp
5456 ** host-disassem.lisp -> disassem.lisp
5457 ** host-insts.lisp -> insts.lisp
5458 ** byte-comp.lisp -> target-byte-comp.lisp
5459 ** host-byte-comp.lisp -> byte-comp.lisp
5460 ** host-signal.lisp -> signal.lisp
5461 ** host-defstruct.lisp -> defstruct.lisp
5462 ** late-target-type.lisp -> deftypes-for-target.lisp
5463 Furthermore, several other previously target-only files foo.lisp (e.g.
5464 hash-table.lisp and random.lisp) have been split into a target-and-host
5465 foo.lisp file and a target-only target-foo.lisp file, with their key type
5466 definitions in the target-and-host part, so that the cross-compiler will
5467 know more about target types.
5468 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5469 gone away. In their place are various *BACKEND-FOO* variables
5470 corresponding to the slots of the old structure.
5471 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5472 into the target SBCL has been fixed.
5473 * The chill.lisp system for loading cold code into a running SBCL
5475 * Support for the CMU CL "scavenger hook" extension has been removed.
5476 (It was undocumented and unused in the CMU CL sources that SBCL was
5477 derived from, and stale in sbcl-0.6.1.)
5478 * Various errors in the cross-compiler type system were detected
5479 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5480 (enabling various consistency checks). Many of them were fixed,
5481 but some hard problems remain, so the compiler is back to
5482 running without *TYPE-SYSTEM-INITIALIZED* for now.
5483 * As part of the cross-compiler type system cleanup, I implemented
5484 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5485 * I have started adding UNCROSS calls throughout the type system
5486 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5487 on cross-compiler input files will be able to go away, and various
5489 * CONSTANTP now returns true for quoted forms (as explicitly required
5492 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5494 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5495 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5496 (MAKE-LIST -1). In the longer term, it ought to become true
5497 that declarations are assertions even at SAFETY 1. For now, it's not
5498 quite true even at SAFETY 3, but it's at least more nearly true..
5499 (Note that this change seems to increases the size of the system by
5500 O(5%) and to decrease the speed of the compiler by 20% or more.)
5501 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5502 for the problem of printing dozens of lines of distracting information
5503 about low-level system machinery as part of the top stack frame
5504 on entry to the debugger when an undefined function was called.
5505 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5506 is not reset to COMMON-LISP-USER.
5507 * Compilation of stuff related to dyncount.lisp has been made conditional
5508 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5509 smaller. The various dyncount-related symbols have been moved into
5510 a new "SB-DYNCOUNT" package.
5511 * tty-inspect.lisp has been renamed to inspect.lisp.
5512 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5513 feature has gone away. (When we eventually port to other flavors of
5514 libc and/or Unix, we'll try to make the differences between flavors
5515 invisible at the user level.)
5516 * Various other *FEATURES* tags, and/or their associated conditionals,
5517 have been removed if obsolescent, or given better documentation, or
5518 sometimes given more-mnemonic names.
5520 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5522 * tidied up "make.sh" script
5523 * tidied up system directory structure
5524 * better "clean.sh" behavior
5525 * added doc/FOR-CMUCL-DEVELOPERS
5526 * many many small tweaks to output format, e.g. removing possibly-confusing
5527 trailing #\. character in DESCRIBE-INSTANCE
5528 * (EQUALP #\A 'A) no longer signals an error.
5529 * new hashing code, including EQUALP hashing
5530 * tidied up Lisp initialization and toplevel
5531 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5532 * command line argument processing
5533 * added POSIX-GETENV function to deal with Unix-ish environment variables
5534 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5535 terminating SBCL on EOF
5536 * non-verbose GC by default
5537 * There is no more "sbcl" shell script; the sbcl file is now the C
5538 runtime executable (just like CMU CL).
5539 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5541 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5542 symbol and package information as Lisp native symbol and package objects
5543 instead of strings naming symbols and strings naming packages. This way,
5544 whenever packages are renamed (as in warm init), debug information is
5545 transformed along with everything else.
5546 * tweaked the optimization policy declarations which control the building
5547 of SBCL itself. Now, among other things, the system no longer saves
5548 source location debugging information. (This helps two problems at once
5549 by reducing SBCL size and by keeping SBCL from trying to look for its
5550 sources -- which may not exist -- when reporting errors.)
5551 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5552 debugging and testing purposes
5553 * cleaned up printing, making the printer call PRINT-OBJECT for
5554 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5555 methods, giving nearly-ANSI behavior
5556 * converted almost all special variables to use *FOO* naming convention
5557 * deleted PARSE-TIME functionality, since it can be done portably
5558 * moved some files out of cold init into warm init
5559 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5561 * regularized formatting of source files
5562 * added an install.sh script
5563 * fixed ridiculous memory usage of cross-compiler by making
5564 compiler/alloc.lisp not try to do pooling unless it can hook
5565 itself into the GC of the cross-compilation host. Now the system
5566 builds nicely on my old laptop.
5567 * added :SB-ALLOC in target-features.lisp-expr
5568 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5569 was not implemented)
5570 * re-did condition handling and note reporting in the compiler. Notes
5571 are no longer handled by signalling conditions. Style warnings
5572 and warnings are handled more correctly and reported in such a way
5573 that it's easy to find one or the other in your output (so that you
5574 can e.g. figure out which of many problems caused COMPILE-FILE to
5576 * changed the severity of several compiler warnings from full WARNING
5577 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5578 compiler note reporting so that it doesn't use the condition system
5579 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5580 COMPILE-FILE command)
5581 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5582 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5583 no longer supported, since their implementation was deeply intertwingled
5584 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5585 * removed bogus "support" for compiler macros named (SETF FOO), and
5586 removed the compiler macro for SETF INFO (but only after making a fool
5587 of myself on the cmucl-imp mailing list by posting a bogus patch for
5588 DEFINE-COMPILER-MACRO..)
5589 * Compiled files containing forms which have side effects on the Lisp
5590 reader (such as DEFPACKAGE forms) are now handled more correctly.
5591 (Compiler queuing of top level lambdas has been suppressed by setting
5592 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5593 * deleted various currently-unused source files, e.g. gengc.lisp. They
5594 may be added back at some point e.g. when porting to other architectures,
5595 but until they are it's distracting to distribute them and to try to
5597 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5598 there were so many of them they're just distractions, and UNCROSS is
5599 known to be able to handle the current sources
5600 * moved PROFILE functionality into TRACE, so that it will be clear
5601 how the wrapping and unwrapping of functions when you profile them
5602 interacts with the wrapping and unwrapping of functions when you
5603 trace them. (Actually, the functionality isn't there yet, but at least
5604 the interface specification is there. Hopefully, the functionality will
5605 arrive with some future maintenance release.)
5606 * removed host-oops.lisp
5607 * changed signature of QUIT function to allow UNIX-CODE argument
5608 * fixed READ-SEQUENCE bug
5609 * tweaked verbose GC output so that it looks more like the progress
5610 output that ANSI specifies for functions like LOAD
5611 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5612 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob