1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.33:
3 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
5 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
6 are now caught and reported just like errors during macroexpansion.
7 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
8 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
9 * fixes and improvements related to Unicode and external formats:
10 ** bug fix: error handling and restart usage in the ucs-2 external format
12 ** there is now an implementation of the ucs-4 external format.
13 ** the utf-16 and utf-32 external formats are supported.
14 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
15 Willem Broekema; launchpad bug lp#489698).
16 * bug fix: some minor code rearrangements to reenable warning-free building
17 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
18 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
20 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
23 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
24 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
25 * improvement: support O_LARGEFILE access to files larger than 2GB on
26 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
27 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
28 definitions for methods specializing on the passed class itself.
29 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
30 definitions for methods specializing on the passed class itself, or on
32 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
33 about itself during the build (e.g. for M-? in Slime), if this flag is
34 enabled in customize-target-features.lisp. This will increase the core
35 size by about 5-6mb, though, so it's mostly interesting to SBCL
37 * new feature: various GENCGC tuning parameters have been experimentally
38 documented and exported from SB-EXT. See documentation for details.
39 * fixes and improvements related to Unicode and external formats:
40 ** the Unicode character database has been upgraded to the
41 Unicode 5.2 standard, giving names and properties to a number of new
42 characters, and providing a few extra characters with case
44 ** improvement: restarts for providing replacement input/output on coding
45 errors for fd-stream external formats.
46 ** improvement: where :<encoding> is a keyword corresponding to an
47 external format the system supports, it is now possible to specify
48 (:<encoding> :replacement <character>) as an external format which will
49 automatically substitute <character> on encoding or decoding errors for
50 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
52 ** improvement: the file streams underlying the standard streams (such as
53 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
54 which uses the replacement mechanism to handle encoding errors,
55 preventing various infinite error chains and unrecoverable I/O
57 ** minor incompatible change: the utf-8 external format now correctly
58 refuses to encode Lisp characters in the surrogate range (char-codes
59 between #xd800 and #xdfff).
60 ** fix a typo preventing conversion of strings into octet vectors
61 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
63 ** fix a bug in the octet multibyte handling of decoding errors and the
64 USE-VALUE restart. (launchpad bug lp#314939)
65 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
66 restart on fd-stream decoding errors.
67 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
68 error is near the end of file.
69 ** fix a double-error case in unibyte octet conversions, when the first
70 use of USE-VALUE is ignored.
71 ** fix bugs in handling of undefined code points in unibyte encodings.
72 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
73 streams after an UNREAD-CHAR.
74 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
75 object is allocated in a boxed region of dynamic space.
76 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
77 fork with multiple Lisp threads running is made, instead of going
78 ahead with unpredictable consequences. (reported by Leslie Polzer)
79 * bug fix: uses of slot accessors on specialized method parameters within
80 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
81 error while finalizing the class. This fix may cause classes with slot
82 accessors to be finalized later than previously. (reported by Lars Rune
83 Nøstdal; launchpad bug lp#473699)
84 * bug fix: restore buildability on the MIPS platform. (regression from
85 1.0.30.38, reported by Samium Gromoff)
86 * bug fix: inspecting closures is less likely to fail with a type error.
87 * bug fix: no timer starvation when setting the system clock back.
88 (launchpad bug lp#460283)
89 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
90 standard pprint dispatch table as specified by CLHS.
91 * bug fix: give CLISP a hint about a type declaration to enable it to build
92 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
94 * bug fix: correctly dump literal objects in defaulting forms of arglists.
95 (reported by Attila Lendvai; launchpad bug lp#310132)
96 * bug fix: distinguish in type specifiers between arrays that might be
97 complex and arrays that are definitely complex. (launchpad bug lp#309129)
98 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
99 type. (reported by Levente Mészáros; launchpad bug lp#485972)
100 * bug fix: setting the value of a symbol-macro within a method in the
101 presence of type declarations works properly again. (reported by Iban
102 Hatchondo; launchpad bug lp#485019)
104 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
105 * optimization: faster FIND and POSITION on strings of unknown element type
106 in high SPEED policies. (thanks to Karol Swietlicki)
107 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
109 * improvement: better error signalling for bogus parameter specializer names
110 in DEFMETHOD forms (reported by Pluijzer)
111 * bug fix: DELETE-FILE once again works on logical pathnames (regression
113 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
115 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
116 removes previous default initargs (reported by Lars Rune Nøstdal and
118 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
119 lambda-lists. (reported by Peter Seibel)
120 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
121 correctly when starting from an executable core without saved runtime
122 options (reported by Faré Rideau, thanks to Zach Beane)
123 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
124 an optimization note instead. (reported by Martin Cracauer)
125 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
126 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
127 new process group if it doesn't need to share stdin with the sbcl
128 process. (thanks to Leslie Polzer)
129 * bug fix: SATISFIES could be misoptimized to refer to a local function.
130 (reported by Stanislaw Halik)
132 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
133 * improvement: stack allocation is should now be possible in all nested
134 inlining cases: failure to stack allocate when equivalent code is manually
135 open coded is now considered a bug.
136 * improvements related to Unicode:
137 ** the Unicode character database has been upgraded to the
138 Unicode 5.1 standard, giving names and properties to a number of new
139 characters, and providing a few extra characters with case
141 ** the system now recognizes and produces names for Unicode Hangul
143 ** the EBCDIC-US external-format is now supported for octet operations
144 (as well as for stream operations).
145 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
146 be used to output toplevel compile-time effects into a separate .CFASL
148 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
149 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
151 * optimization: TRUNCATE on known single- and double-floats is upto 25%
153 * optimization: division of floating point numbers by constants uses
154 multiplication by reciprocal when an exact reciprocal exists.
155 * optimization: multiplication of single- and double-floats floats by
156 constant two has been optimized.
157 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
158 sufficient type information is available. (thanks to Leslie Polzer)
159 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
160 known structure objects are as efficient as defstruct generated accessors.
161 * optimization: unused vector creation can now be optimized away.
162 * improvement: ASDF systems can now depends on SB-INTROSPECT.
163 * improvement: a STYLE-WARNING is signalled when a generic function
164 clobbers an earlier FTYPE proclamation.
165 * improvement: the compiler is able to track the effective type of
166 generic function across method addition and removal even in the
167 absence of an explicit DEFGENERIC.
168 * improvement: DESCRIBE now reports on symbols naming undefined
169 but assumed or declared function as well.
170 * improvement: recompilation of systems using SB-GROVEL now works
171 (thanks to Leslie Polzer)
172 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
173 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
175 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
176 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
177 well as user defined declaration names.
178 ** VARIABLE-INFORMATION is now aware of alien variables.
179 * improvement: improved address space layout on OpenBSD (thanks to Josh
181 * improvement: pretty-printing of various Lisp forms has been improved
182 (thanks to Tobias Rittweiler)
183 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
184 unused were deleted in safe code. (reported by John Fremlin)
185 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
187 * bug fix: some out-of-line array predicates were missing (reported by
189 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
191 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
192 (thanks to Larry D'Anna)
193 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
194 as well. (reported by Josh Elsasser)
195 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
196 leading to possible attempts to delete the same file twice. See docstring
197 on DELETE-FILE for details. (reported by John Fremlin)
198 * bug fix: DELETE-FILE once again deletes the file named by the pathname
199 designator argument, rather than its truename. (reported by Luis
201 * bug fix: the low-level debugger had 32-bit assumptions and was missing
202 information about some array types. (thanks to Luis Oliveira)
203 * bug fix: moderately complex combinations of inline expansions could
204 be miscompiled if the result was declared to be dynamic extent.
205 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
206 form (+ <variable> <integer>) were miscompiled under certain
208 * bug fix: in some cases no compiler note about failure to stack allocate
209 was emitted, even if the objects were in fact heap allocated.
210 * bug fix: minor violation of "otherwise inaccessible" rule for stack
211 allocation could cause objects users might reasonably expect to
212 be heap allocated to be stack allocated.
213 * bug fix: DESCRIBE signalled an error for generic functions under
214 certain circumstances. (thanks to Leslie Polzer)
215 * bug fix: Fixed spelling of an error message.
217 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
218 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
219 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
220 of SB-THREAD:THREAD-ERROR-THREAD.
221 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
222 * new feature: docstrings for local and anonymous functions are no longer
223 discarded. (thanks to Leslie Polzer)
224 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
225 values in other threads.
226 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
227 about object allocation.
228 * optimization: division of a real float by a complex float is implemented
229 with a specialised code sequence.
230 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
231 keywords is an order of magnitude faster.
232 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
233 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
234 methods, and similarly for non-standard metaclass classes as long as there
235 are no methods additional on MAKE-INSTANCE.
236 * optimization: more efficient type-checks for FIXNUMs when the value
237 is known to be a signed word on x86 and x86-64.
238 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
239 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
241 * optimization: compiler is smarter about delegating argument type checks to
243 * optimization: several character functions are now compiled somewhat more
244 efficiently. (reported by Lynn Quam)
245 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
246 * improvement: less unsafe constant folding in floating point arithmetic,
247 especially for mixed complex/real -float operations.
248 * optimization: constant double and single floats are stored in native
249 unboxed format on x86[-64].
250 * optimization: smarter code for arithmetic operations with constant floats,
251 complex floats, or integers on x86[-64].
252 * optimization: smarter code for conjugate/multiplication of float complexes
253 and abs/negate of floats on x86-64.
254 * optimization: more efficient complex float and real float operations on
256 * improvement: complex float division is slightly more stable.
257 * improvement: DESCRIBE output has been reworked to be easier to read and
258 contains more pertinent information.
259 * improvement: failure to provide requested stack allocation compiler notes
260 provided in all cases (requested stack allocation not happening without a
261 note being issued is now considered a bug.)
262 * bug fix: SB-POSIX exports the documented types and functions
263 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
264 types. (reported by "abhi")
265 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
266 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
267 (thanks to Paul Khuong)
268 * bug fix: looping around HANDLER-CASE could silently consume stack space
269 on each iteration. (reported by "foobar")
270 * bug fix: better error signalling when calls to functions seeking elements
271 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
272 (reported by Tobias Rittweiler)
273 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
274 directory components now works as it used to. (various prolems reported by
275 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
276 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
277 (reported by Elliot Slaughter, thanks to Stas Boukarev)
278 * bug fix: bogus undefined variable warnings from fopcompiled references to
279 global variables. (thanks to Lars Rune Nøstdal)
280 * bug fix: foreign function names should now appear in backtraces on
281 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
282 * bug fix: SETF compiler macro documentation strings are not discarded
284 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
285 (thanks to Alex Plotnick)
286 * bug fix: SLEEP accepts large integer arguments, truncating them to
287 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
288 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
289 * bug fix: compiler notes for expensive slot type checks could be emitted
290 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
292 changes in sbcl-1.0.29 relative to 1.0.28:
293 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
294 https://bugs.launchpad.net/sbcl
295 Bugs can be reported directly there, or by sending email to
296 sbcl-bugs@lists.sourceforge.net
297 (no subscription required.)
298 * minor incompatible change: under weak type checking policy integer
299 types are weakened less aggressively.
300 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
301 allowed to return, which causes SBCL to quit with exit status 0. Previously
302 if the function returned with a small integer return value, that value
303 was accidentally reused as the exit status.
304 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
305 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
306 :HASH-FUNCTION argument. Refer to user manual for details.
307 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
309 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
310 microseconds since the Unix epoch on all platforms.
311 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
312 allows the compiler to safely elide boundedness checks for special
314 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
315 the symbol, prohibits both lexical and dynamic binding. This is mainly an
316 efficiency measure for threaded platforms, but also valueable in
318 * new feature: UNC pathnames are now understood by the system on Windows.
319 * optimization: the compiler uses a specialized version of FILL when the
320 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
322 * optimization: accesses to potentially non-simple arrays where element type
323 is known are 50% faster.
324 * optimization: compiler now generates faster array typechecking code.
325 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
327 * optimization: multidimensional array accesses in the absence of type
328 information regarding array rank are approximately 10% faster due to
329 open coding of ARRAY-RANK.
330 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
331 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
333 * optimization: result of call to VECTOR can now be stack allocated.
334 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
335 as long as the resulting array is one-dimensional and has a known
336 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
337 length of the list matches the known length of the vector does not
338 allocate the list as an intermediate step. Ditto for VECTOR and simple
340 * optimization: MAKE-ARRAY can now stack allocate in the presence of
341 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
342 known element type, and is known to be simple and one dimensional.
343 * improvement: SBCL now emits a compiler note where stack allocation was
344 requested but could not be provided (not in all cases, unfortunately)
345 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
346 * improvement: pretty-printing loop has been implemented properly. (thanks
347 to Tobias Rittweiler)
348 * documentation: CLOS slot typechecing policy has been documented.
349 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
350 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
351 pointer, leading to memory faults. (thanks to Bart Botta)
352 * bug fix: better floating point exception handling on x86/OpenBSD.
353 (thanks to Josh Elsasser)
354 * bug fix: exit status from QUIT when called under --script was lost
355 (reported by Hubert Kauker)
356 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
357 same implementation of FILL to initialize the array, even if a more
358 efficient one was available (reported by Stas Boukarev, thanks to
360 * bug fix: potential miscompilation of array stack allocation on x86 and
361 x86-64. (reported by Time Tossavainen)
362 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
363 result in their subforms being treated as top level forms. (reported by
365 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
366 than a machine register could leave garbage in the high bits of the
367 result register (bug 316325).
368 * bug fix: disable address space randomization Linux/x86-64 as well,
369 not just x86-64. (reported by Ken Olum)
370 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
371 generate incorrect code.
372 * bug fix: #201; type inference for CONS and ARRAY types could derive
373 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
374 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
375 could not be assigned to or bound with PROGV. (reported by Lorenz
377 * bug fix: the value of CL:- in the inspector was the previous expression
378 evaluated rather than the expression being evaluated.
379 * bug fix: constants can no longer be locally declared special.
380 * bug fix: signals delivered to threads started from foreign land (read:
381 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
382 thread by blocking all signals and resignalling.
383 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
384 from :INITFORM, if any.
386 changes in sbcl-1.0.28 relative to 1.0.27:
387 * a number of bugs in cross-compilation have been fixed, with the ultimate
388 result that building under (at least) clisp should be much more reliable.
389 * minor incompatible changes: echo-streams now propagate unread-char to the
390 underlying input stream, and no longer permit unreading more than one
392 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
393 frames, allowing for instance more reliable backtraces.
394 * improvement: the debugger REPL can now reference lexical variables
395 by name directly for code compiled with (DEBUG 3).
396 * improvement: errors from malformed declarations now have better source
397 paths associated with them. (thanks to Tobias Rittweiler)
398 * optimization: faster local calls on x86/x86-64
399 * bug fix: some error messages for out-of-bound array indexes confused the
400 index and the bound. (thanks to Stas Boukarev)
401 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
403 * bug fix: running regressions tests in shells without OSTYPE set now works.
404 (reported by Harald Hanche-Olsen)
405 * bug fix: more robust static space exhaustion signalling from
406 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
407 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
408 docstring away instead of storing it under names such as (LAMBDA (X)).
409 (reported by Leslie Polzer)
410 * bug fix: timers could go off in the wrong order, be delayed indefinitely
411 (thanks to Ole Arndt for the patch)
412 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
413 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
414 potentially-invalid effective methods in its cache.
415 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
416 instances as well (thanks to Paul Khuong)
417 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
418 SB-POSIX:WAITPID and their C equivalents.
419 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
421 changes in sbcl-1.0.27 relative to 1.0.26:
422 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
423 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
424 * improvement: the system either recovers from stack exhaustion or dies
425 properly as opposed to leaving the user uncertain of whether the handler
426 trampled on some random memory next to the stack or having to rely on
427 --lose-on-corruption (which is still a good idea to use in production
428 because stack exhaustion can happen in signal handlers which will likely
430 * bug fix: fix gc related interrupt handling bug on ppc (regression from
431 1.0.25.37, reported by Harald Hanche-Olsen)
432 * bug fix: work around signal delivery bug in darwin (regression from
433 1.0.25.44, reported by Sidney Markowitz)
434 * bug fix: fix ERROR leaking memory (reported by David Thompson)
436 changes in sbcl-1.0.26 relative to 1.0.25:
437 * incompatible change: an interruption (be it a function passed to
438 INTERRUPT-THREAD or a timer function) runs in an environment where
439 interrupts can be enabled. The interruption can use
440 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
441 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
442 potentially running out of stack. Keep in mind that in the absence
443 of WITHOUT-INTERRUPTS some potentially blocking operation such as
444 acquiring a lock can enable interrupts.
445 * incompatible change: GC-OFF and GC-ON are removed, as they were
446 always unsafe. Use WITHOUT-GCING instead.
447 * new feature: runtime option --disable-ldb
448 * new feature: runtime option --lose-on-corruption to die at the
449 slightest hint of possibly non-recoverable errors: running out of
450 memory, stack, alien stack, binding stack, encountering a memory
451 fault, etc. In the absence of --lose-on-corruption a warning is
453 * enhancement: detect binding stack exhaustion
454 * enhancement: detect alien stack exhaustion on x86/x86-64
455 * improvement: generally more stable and reliable interrupt handling
456 * improvement: there is a per thread interruption queue,
457 interruptions are executed in order of arrival
458 * improvement: a repeating timer reschedules itself when the it has
459 finished, but expiration times are spaced equally. If an
460 expiration time is in the past it will trigger after a short grace
461 period that may give a chance to other things to run.
462 * optimization: slightly faster gc on multithreaded builds
463 * optimization: faster WITHOUT-GCING
464 * bug fix: when JOIN-THREAD signals an error, do it when not holding
465 important locks so that the debugger/handler doesn't produce
466 recursive errors or deadlock.
467 * bug fix: real-time signals are not used anymore, so no more
468 hanging when the system wide real-time signal queue gets full.
469 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
471 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
472 * bug fix: fix random memory faults related to interrupts on alpha
473 * bug fix: fix random memory faults related to interrupts on sparc
474 * bug fix: fix deadlocks related to starting threads
475 * bug fix: fix deadlines on locks on futex platforms
476 * bug fix: restore errno in signal handlers
477 * bug fix: fix deadlocks related to hash tables
478 * bug fix: fix deadlocks in pcl
480 changes in sbcl-1.0.25 relative to 1.0.24:
481 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
482 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
483 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
484 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
485 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
486 better name and does not return values so stale on multiprocessor systems.
487 Also, HOLDING-MUTEX-P was added for about the only sane usage of
489 * improvement: unithread builds keep track of MUTEX-VALUE.
490 * improvement: reading from a TWO-WAY-STREAM does not touch the output
491 stream anymore making it thread safe to have a concurrent reader and
492 a writer, for instance, in a pipe.
493 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
494 arguments, making compiler-macros for SETF-functions able to inspect
495 their constant arguments.
496 * improvement: COMPILE-FILE reports times with millisecond accuracy
497 (thanks to Luis Oliveira)
498 * optimization: CHAR-CODE type derivation has been improved, making
499 TYPEP elimination on subtypes of CHARACTER work better. (reported
500 by Tobias Rittweiler, patch by Paul Khuong)
501 * bug fix: setting alien structure fields of type struct by value now
502 computes the right offset for the memory copy.
503 * bug fix: compilation problem involving inlined calls to aliens with
504 result type VOID. (reported by Ken Olum)
505 * bug fix: #235a; sequential inline expasion in different policies no
506 longer reuses the functional from the previous expansion site.
507 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
508 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
509 * bug fix: named ENUMs in multiply-referenced alien record types no longer
510 cause an implied type redefinition. Regression from 1.0.21.29.
511 * bug fix: sign-extension of small signed return values now works with C
512 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
513 * improvements to the Windows port:
514 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
515 Westfold, thanks to Rudi Schlatte)
517 changes in sbcl-1.0.24 relative to 1.0.23:
518 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
519 vector of a multidimensional SIMPLE-ARRAY.
520 * new feature: the system now signals a continuable error if standard
521 readtable modification is attempted. (thanks to Tobias Rittweiler)
522 * new feature: DIRECTORY has been extended with a non-standard keyword
523 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
524 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
525 has been robustified and documented better. (thanks to Robert Goldman)
526 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
527 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
528 * tradeoff: constant FORMAT control strings are now compiled unless
529 SPACE > SPEED (previously only when SPEED > SPACE.)
530 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
532 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
534 * bug fix: lexical type declarations are now correctly reported by
535 SB-CLTL2. (reported by Larry D'Anna)
536 * bug fix: STRING-TO-OCTETS did not handle :START properly when
537 using UTF-8 as external format. (thanks to Luis Oliveira)
538 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
539 are signalled correctly. (thanks to Stas Boukarev)
540 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
541 designator. (thanks to Tobias Rittweiler)
542 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
543 designator, and returns T instead of the function. (thanks to
545 * bug fix: direct superclasses of STANDARD-CLASS and
546 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
547 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
548 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
549 when no class with the named slot yet exists no longer causes a
550 compile-time style-warning.
551 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
552 in safe code. (reported by Didier Verna)
553 * bug fix: #430; nested structure constructors can stack allocate.
554 * bug fix: on some 64-bit platforms dynamic space size was truncated
555 to #xffff0000 bytes. (reported by Benjamin Lambert)
556 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
557 REPL unusable. (reported by Daniel Herring)
558 * bug fix: return values of READ-SEQUENCE did not take :START into
559 account on file streams, regressions since 1.0.12.22. (reported by
560 Thomas Russ, patch by Paul Khuong)
561 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
562 method specializer used to confuse permuation vector optimization.
563 * bug fix: system inserted bogus implicit type declarations for local
564 special variables in DEFMETHOD bodies.
565 * bug fix: #354; duplicated frames in backtraces due to
566 non-tail-call-optimized XEPs to functions with return type NIL
568 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
569 initializes structure object slots according to DEFSTRUCT initforms,
570 and DEFSTRUCT forms :INCLUDEind structure classes defined using
571 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
572 (reported by Bruno Haible and Stephen Wilson)
573 * bug fix: #395; fill-pointer output streams used now support
574 element-type BASE-CHAR as well.
575 * bug fix: compiler error when attempting to derive return value of
576 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
578 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
580 * bug fix: Mac OS X binaries should now be portable between Leopard
583 changes in sbcl-1.0.23 relative to 1.0.22:
584 * enhancement: when disassembling method functions, disassembly
585 for the associated fast function is also produced.
586 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
588 * optimization: printing with *PRINT-PRETTY* true is now more
589 efficient as long as the object being printed doesn't require
590 special handling by the pretty printer.
591 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
592 now interact correctly with type declarations.
593 * partial bug fix: PCL detects infinite recursion during wrapper
594 validation. (thanks to Attila Lendvai)
595 * bug fix: #426; nested function calls are inlined properly.
596 Previously if FOO was an inline function, in calls of the form
597 (FOO (FOO ...)) the outer call was not inlined.
598 * bug fix: long long arguments passed in registers to alien
599 functions on PowerPC Linux and NetBSD are now handled correctly
600 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
602 changes in sbcl-1.0.22 relative to 1.0.21:
603 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
604 for the shared object in the current directory, but passes the native
605 namestring of the designated pathname to the operation system's shared
606 object loading function as-it.
607 * minor incompatible change: --disable-debugger toplevel option now takes
608 effect before processing of initialization files and --eval or --load
610 * new feature: new commandline argument: --script, which supports
611 shebang lines. See documentation for details. (based on work by
613 * new feature: SAVE-LISP-AND-DIE can save current values of
614 --dynamic-space-size and --control-stack-size in the executable core,
615 causing it to skip normal runtime option processing. See documentation
616 for details. (thanks to Zach Beane)
617 * enhancement: inoccous calls to EVAL or generic functions dispatching
618 on subclasses of eg. STREAM no longer cause compiler notes to appear.
619 * enhancement: the system no longer resignals errors from --load and
620 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
621 associated with the original error to be lost. (thanks to Ariel
623 * enhancement: :DONT-SAVE keyword argument has been added to
624 LOAD-SHARED-OBJECT for controlling interaction with
625 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
626 the effects of an earlier LOAD-SHARED-OBJECT call.
627 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
628 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
629 provided. (reported by Cedric St-Jean)
630 * bug fix: circularity handling in the reader did not treat raw
631 structure slots correctly. (reported by Cedric St-Jean)
632 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
633 file descriptors when there were none.
634 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
635 was broken. (reporteed by Willem Broekema)
636 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
637 pathnames without a directory.
638 * bug fix: COMPILE-FILE did not warn about undefined variable
639 references at toplevel, and LOAD of the resulting fasl did
641 * bug fix: functions with non-required arguments used to end up with
642 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
643 * bug fix: redefining a function with non-required arguments didn't
644 update the system's knowledge about its call signature properly.
645 * bug fix: fixed #431; incompatible alien record type redefinitions
646 are detected and handled. (reported by Neil Haven)
647 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
648 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
650 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
651 after alien stack frames.
652 * bug fix: strings passed to mk{s,d}temp lacked null termination
654 changes in sbcl-1.0.21 relative to 1.0.20:
655 * new feature: the compiler is able to track the effective type of a
656 generic function across method addition and removal.
657 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
658 appropriately typed structure slots without locking.
659 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
660 information like those gathered by TIME using a programming-friendly
662 * new feature: TIME reports time taken even if the form performs a
663 non-local transfer of control.
664 * enhancement: reduced conservativism on GENCGC platforms: on
665 average 45% less pages pinned (measured from SBCL self build).
666 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
667 mutate constant symbols or violate declaimed type of the symbol.
668 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
669 sleeping on the semaphore in heavily contested semaphores.
670 * bug fix: semaphores and condition variables were not interrupt
672 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
673 owned by other threads anymore.
674 * bug fix: FIND on lists called KEY outside the specified
675 subsequence. (reported by budden)
676 * bug fix: LOG doesn't use single-float intermediate results when
677 given mixed integer and double-float arguments, leading to better
678 precision. (reported by Bob Felts)
679 * bug fix: LOG with base zero returned values of inconsistent type.
681 changes in sbcl-1.0.20 relative to 1.0.19:
682 * minor incompatible change: OPTIMIZE qualities
683 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
684 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
685 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
686 * documentation: some slot access efficiency guidelines have been
687 added to the user manual.
688 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
689 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
690 as ASSOC and MEMEBER.
691 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
692 to more efficient EQ-comparison versions more often.
693 * optimization: enhanced derivation of DOLIST iteration variable type
695 * optimization: constant folding of simple (LIST ...) forms as DOLIST
697 * optimization: runtime lookup of function definitions can be
698 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
699 * optimization: compiler is able to derive the return type of
700 (AREF (THE STRING X) Y) as being CHARACTER.
701 * optimization: CLRHASH on empty hash-tables no longer does pointless
702 work. (thanks to Alec Berryman)
703 * optimization: REPLACE deftransforms don't punt when :START1 or
705 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
706 no longer emits a code deletion note. (reported by Andrew
708 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
709 functions with an odd number of &OPTIONAL arguments, a &REST
710 argument, and one or more &KEY arguments at the call site.
711 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
712 arguments appeared at call sites as well.
713 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
714 READ-LINE &co can no longer read from them afterwards. (reported
716 * bug fix: fixed #427: unused local aliens no longer cause compiler
717 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
719 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
720 alien-stack leakage. (reported by Andy Hefner)
721 * bug fix: PROGV signals an error when an attempt to violate declared
722 type of a variable or bind a constant is made.
723 * bug fix: SET signals an error when an attempt to violate declared
724 type of a variable is made.
725 * bug fix: restart computation during the execution of a restart
726 test function no longer causes infinite recursion. (reported by
728 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
729 signals an error. (thanks to Pierre Mai)
730 * bug fix: EXPORT left symbol unexported in conflict situations.
731 (thanks to Michael Weber)
732 * bug fix: correctly handle name conflicts involving CL:NIL.
733 (thanks to Michael Weber)
734 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
735 conflicts arising from USEing package with conflicting symbols
736 correctly. (thanks to Michael Weber)
738 changes in sbcl-1.0.19 relative to 1.0.18:
739 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
740 warnings that go otherwise unhandled will be muffled if they are
741 of the type that's the value of this variable.
742 * optimization: stack allocation is slightly more efficient on x86
744 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
745 where a raw slot always is initialized using the initform whose
746 type is not know sufficiently well a compile-time are now compiled
747 correctly. (reported by John Morrison)
748 * bug fix: compiler no longer makes erronous assumptions in the
749 presence of non-foldable SATISFIES types.
750 * bug fix: stack analysis missed cleanups of dynamic-extent
751 arguments in non-let-converted calls to local functions.
752 * improvements to the Windows port:
753 ** adjusted address spaces for building on both Win32 and
754 Win64. (thanks for John Connors)
755 * fixed some bugs revealed by Paul Dietz' test suite:
756 ** interval arithmetic during type derivation used inexact integer
757 to single-float coercions.
758 ** arithmetic operations involving large integers and single
759 floats give the same results in compiled and interpreted code.
760 ** deriving the result type of COERCE no longer signals an error
761 if the derived type of the second argument is a MEMBER type
762 containing invalid type specifiers.
763 ** ADJOIN with constant NIL as second argument works correctly.
764 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
766 changes in sbcl-1.0.18 relative to 1.0.17:
767 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
768 profiles only the current thread.
769 * minor incompatible change: changes to SYMBOL-VALUE of constants
770 defined with DEFCONSTANT now signal an error.
771 * enhancement: SB-SPROF now has support for wallclock profiling,
772 and is also able to profile specific threads. REPORT output
773 has also additional sorting options.
774 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
776 * optimization: structure allocation has been improved
777 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
778 ** out of line constructors are ~10% faster.
779 ** inline constructors are ~15% faster.
780 ** inline constructors are capable of dynamic extent allocation
781 (generally on x86 and x86-64, in some cases on other platforms
783 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
785 * optimization: file compiler is now able to coalesce non-circular
786 lists, non-base strings, and bit-vectors. Additionally, constants
787 are never referenced through SYMBOL-VALUE at runtime.
788 * optimization: code defining methods on PRINT-OBJECT (and other
789 generic functions in the COMMON-LISP package) now loads faster.
790 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
792 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
793 fasl is now deleted. (reported by Attila Lendvai)
794 * bug fix: READ-LINE always returned NIL for the last line in files.
795 (reported by Yoshinori Tahara)
796 * bug fix: more accurate disassembly annotations of foreign function
797 calls. (thanks to Andy Hefner)
798 * bug fix: trimming non-simple strings and non-string string
799 designators when the there is nothing to trim works properly.
800 (thanks to James Knight)
801 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
802 (thanks to Travis Cross)
803 * fixed some bugs revealed by Paul Dietz' test suite:
804 ** NIL is a valid function name (regression at 1.0.13.38)
805 ** FILL on lists was missing its return value (regression at 1.0.12.27)
806 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
807 fill pointers properly (regression at 1.0.12.23)
808 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
809 displacement indices properly (regression at 1.0.12.23)
811 changes in sbcl-1.0.17 relative to 1.0.16:
812 * temporary regression: user code can no longer allocate closure
813 variable storage on stack, due to bug 419 without explicitly
814 requesting it. Please consult sbcl-devel for advice if you need to
815 use this feature in the meanwhile.
816 * new feature: runtime argument --control-stack-size can be used to
817 adjust thread default control stack size.
818 * enhancement: improved TIME output
819 ** all times are reported using the measured accuracy (milliseconds
820 for real and GC times, microseconds for everything else.)
821 ** processor cycle counts on x86 and x86-64.
822 ** interpreted forms are counted for both evaluator modes.
823 ** number of lambdas converted by the compiler is reported.
824 ** CPU percentage report (computed from real and total run time.)
825 ** more comprehensive run time reporting, using a condenced format
826 ** interperted form, lambda, and page fault counts are omitted
828 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
830 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
831 * optimization: two argument forms of LAST are upto ~10% faster
832 in normal SPEED policies.
833 * optimization: NCONC no longer needs to heap cons its &REST list
834 in normal SPEED policies.
835 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
837 * bug fix: UNION and NUNION work with :TEST-NOT once more,
838 regression since 1.0.9.1. (thanks to Eric Marsden)
839 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
840 since 1.0.15.36. (thanks to Paul Khuong)
841 * bug fix: LAST when always returned the whole list when given a bignum
842 as the second argument.
843 * bug fix: dynamic extent allocation of nested lists and vectors
844 could leak to otherwise accessible parts.
845 * bug fix: invalid optimization of heap-allocated alien variable
847 * bug fix: fasl header checking is less vulnerable to different
848 platform word lengths.
849 * bug fix: more correct assembler syntax for GNU binutils
850 2.18.50.0.4 support. (thanks to Marijn Schouten)
851 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
853 * bug fix: the fopcompiler can handle LOCALLY forms (with no
854 declarations) successfully. (reported by Attila Lendvai)
856 changes in sbcl-1.0.16 relative to 1.0.15:
857 * minor incompatible change: revert the changes to sb-posix's error
858 signaling added in 1.0.14.
859 * minor incompatible change: change PROBE-FILE back to returning
860 NIL whenever we can't get a truename, as was the case before 1.0.14.
861 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
862 inherits from ERROR instead of just CONDITION.
863 * new feature: SB-INTROSPECT can provide source locations for instances
864 as well. (thanks to Tobias Rittweiler)
865 * optimization: binding special variables now generates smaller code
866 on threaded platforms.
867 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
868 and cases where no :TEST is given but the compiler can infer that
869 the element to search is of type (OR FIXNUM (NOT NUMBER)).
870 * optimization: better LOGNOT on fixnums.
871 * optimization: modular arithmetic for a particular requested width
872 is implemented using a tagged representation unless a better
873 representation is available.
874 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
875 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
876 for wellformedness and returns a specialized vector. (reported by
877 Francois-Rene Rideau)
878 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
879 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
880 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
881 unscheduling timer at the same time as another timer fires.
882 * bug fix: don't reschedule timers for dead threads.
883 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
884 * bug fix: copying output from RUN-PROGRAM to a stream signalled
885 bogus errors if select() was interrupted.
886 * enhancement: add support for fcntl's struct flock to SB-POSIX.
888 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
889 * enhancement: cleaner backtraces for interactive interrupts, as
890 well as other cases where the interesting frames used to be
891 obscured by interrupt handling frames.
892 * enhancement: untracing a whole package using (UNTRACE "FOO") is
893 now supported, and tracing a whole package using (TRACE "FOO") now
894 traces SETF-functions as well.
895 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
896 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
897 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
898 * unreadably printed representation of hash-tables now includes
900 * bug fix: partially fixed #188: type propagation from assignments
901 is now more efficient.
902 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
903 'DOUBLE-FLOAT) are not flushable.
904 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
905 full address of the object, and none of the tag bits.
906 * bug fix: readably printing hash-tables now respects other printer
907 control variables. (reported by Cedric St-Jean)
908 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
909 keyword with MAKE-HASH-TABLE.
910 * bug fix: export SB-POSIX:MKSTEMP.
911 * bug fix: SORT was not interrupt safe.
912 * bug fix: XREF accounts for the last node of each basic-block as
914 * bug fix: MAKE-INSTANCE optimizations interacted badly with
915 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
916 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
917 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
918 create obsolete instances in the case of redefinition or
919 obsoletion of a superclass. (thanks to Andy Hefner)
920 * bug fix: Support for the Alpha architecture has been revived; it had
921 suffered somewhat from lack of maintenance since sbcl-1.0.
922 * improvements to the Windows port:
923 ** The system detects the codepage to use at startup, instead of
924 using the value from the saved core. (thanks to Kei Suzuki)
926 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
927 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
928 (see documentation for details.)
929 * revived support for OpenBSD (contributed by Josh Elsasser)
930 * partially fixed bug #108: ROOM no longer suffers from occasional
931 (AVER (SAP= CURRENT END)) failures.
932 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
933 bodies no longer cause a WARNING to be signalled. (reported by
935 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
936 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
937 the bug in TRUNCATE.) (reported by Michael Weber)
938 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
939 no samples. (reported by Andy Hefner)
940 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
941 no longer appear as (NIL ...) frames in backtraces.
942 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
943 is now actually exported from SB-EXT as documented. (reported by
945 * bug fix: sb-aclrepl now correctly understands how to inspect
946 single-floats on 64-bit platforms where single-floats are not boxed.
947 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
948 yet been finalized. (reported by Levente Meszaros)
949 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
950 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
951 SBCL-specific optimize qualities.
953 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
954 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
955 an executable in the search path, and does so in the child
956 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
957 been removed; it can be found in the CVS history, for somebody who
958 needs that search behavior (see the manual).
959 * minor incompatible change: compiler policy re. weakening type
960 checks has changed: now type checks are weakened only if SAFETY < 2
962 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
963 unparsing of directory pathnames as files. Analogously,
964 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
965 filename to parse into a directory pathname.
966 * enhancement: implicit generic function creation now signals a
967 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
968 users can bind handlers for and muffle around calls to LOAD.
969 * enhancement: RUN-PROGRAM allows unicode arguments and environments
970 to be used (using the default stream external format), and allows
971 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
972 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
973 strings and vectors whose element-type or simplicity is not fully
974 known at compile-time.
975 * optimization: STRING-TRIM and related functions no longer allocate
976 a new string when no trimming needs to be performed. These functions
977 are also faster than before when the input string has been declared
979 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
980 * optimization: READ-LINE is significantly faster for files containing
982 * optimization: non-open coded uses of character comparison operators
983 (e.g. char=) no longer cons when called with more than one parameter
984 on platforms supporting dynamic-extent allocation.
985 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
986 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
987 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
989 * bug fix: some sequence functions elided bounds checking when
991 * bug fix: too liberal weakening of union-type checks when SPEED >
993 * bug fix: more bogus fixnum declarations in ROOM implementation
996 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
997 * new feature: MAKE-HASH-TABLE now experimentally accepts a
998 :SYNCHRONIZED argument, which makes the hash-table safe for
999 concurrent accesses (but not iteration.) See also:
1000 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1001 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1002 * optimization: CONCATENATE on strings is an order of magnitude faster
1003 in code compiled with (> SPEED SPACE).
1004 * optimization: SUBSEQ is ~50% faster on lists.
1005 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1006 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1007 END is smaller then START.
1008 * bug fix: SB-PROFILE will no longer report extra consing for nested
1009 calls to profiled functions.
1010 * bug fix: ROOM implementation had bogus fixnum declarations which
1011 could cause type-errors when calling ROOM on large images.
1012 * bug fix: if file compilation is aborted, the partial fasl is now
1013 deleted, and COMPILE-FILE returns NIL as the primary value.
1014 * bug fix: number of thread safety issues relating to SBCL's internal
1015 hash-table usage have been fixed.
1016 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1017 be returned from its body when the values were being returned
1018 using unknown-values return convection and the W-P-O was wrapped
1019 inside an UNWIND-PROTECT.
1020 * bug fix: sb-posix should now compile again under Windows, enabling
1021 slime to work again.
1023 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1024 * incompatible change: hash-table accessor functions are no longer
1025 automatically protected by locks. Concurrent accesses on the same hash-table
1026 from multiple threads can give inconsistent results or even corrupt the
1027 hash-table completely. Multi-threaded applications should do their own
1028 locking at the correct granularity. In the current implementation it is
1029 still safe to have multiple readers access the same table, but it's not
1030 guaranteed that this property will be maintained in future releases.
1031 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1032 and will signal an error at runtime.
1033 * enhancement: SB-THREAD package now exports a semaphore interface.
1034 * enhancement: CONS can now stack-allocate on x86 and
1035 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1037 * enhancement: nested lists can now be stack allocated on
1038 platforms providing stack allocation support.
1039 * enhancement: dynamic-extent support has been extended to support
1040 cases where there are multiple possible sources for the stack
1042 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1043 if the mutex is uncontested on Linux.
1044 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1045 as specified by the CLHS. (thanks to Tobias Rittweiler)
1046 * bug fix: NaN comparison now works on x86-64.
1047 * bug fix: CLOSE :ABORT T on a stream with pending output now
1049 * bug fix: instances of non-standard metaclasses using standard
1050 instance structure protocol sometimes missed the slot type checks
1052 * bug fix: known functions can be inlined in high-debug code.
1054 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1055 * minor incompatible change: the MSI installer on Windows no longer
1056 associates .lisp and .fasl files with the installed SBCL.
1057 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1058 on Windows. (thanks to Luis Oliviera)
1059 * new feature: SB-CLTL2 contrib module now implements
1060 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1061 * optimization: scavenging weak pointers is now more efficient,
1062 requiring O(1) instead of O(N) per weak pointer to identify
1063 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1064 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1065 method bodies using constant slot-names when the first argument is
1066 a specializer parameter for the method.
1067 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1068 names, when the class of the instance is a direct instance
1069 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1070 as slow as the constant slot-name case.)
1071 * optimization: member type construction is now O(N) instead
1073 * optimization: UNION and NUNION are now O(N+M) for large
1074 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1076 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1077 now more readable in environments like Slime which display it.
1078 (thanks to Tobias C. Rittweiler)
1079 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1080 was able to infer the correct argument type for the object on which
1081 the CAS operation was being performed.
1082 * bug fix: copy propagation interfered with parallel assignment
1083 semantics in local calls. (reported by Paul Khuong)
1084 * bug fix: the signed modular fixnum shift compiled to wrong code on
1085 x86 and x86-64. (spotted by a slight modification to some of
1088 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1089 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1090 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1091 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1093 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1094 and generic function now print the object identity for anonymous
1095 instances. (thanks to Attila Lendvai)
1096 * enhancement: as an extension to MOP, SBCL now supports SETF of
1097 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1098 (thanks to Attila Lendvai)
1099 * enhancement: sb-cover has an alternate annotation mode which puts
1100 the form annotations at the car of the form, instead of annotating
1101 it completely. The benefit of this mode is that it shows better
1102 which forms were instrumented by the compiler.
1103 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1104 by printer control variables. (reported by Dan Corkill)
1105 * bug fix: system leaked memory when delayed output was performed by
1106 the OS in smaller chunks then expected. (thanks to David Smith)
1107 * bug fix: system leaked memory when file streams were not closed
1109 * bug fix: large objects written to slow streams that were modified
1110 after the write could end up with the modified state written to
1111 the underlying file descriptor.
1112 * bug fix: multiple threads operating in parallel on the same stream
1113 could cause buffer-overflows.
1114 * bug fix: source location information is stored correctly
1115 (broken since 1.0.6). This bug would generally show up as the
1116 Slime debugger highlighting the wrong form when the sldb-show-source
1118 * bug fix: Branch forms are again annotated as branches in the sb-cover
1120 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1121 on x86-64. (reported by Gregory Vanuxem)
1122 * bug fix: EQUALP could return wrong results for structures with raw
1123 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1124 word). (reported by Vjacheslav Fyodorov)
1125 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1128 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1129 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1130 atomic compare-and-swap operations on threaded platforms.
1131 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1132 allows assigning a global minimum value to optimization qualities
1133 (overriding proclamations and declarations).
1134 * enhancement: closed over variables can be stack-allocated on x86
1136 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1138 * optimization: slot definition lookup is now O(1). This speeds up
1139 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1140 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1141 * optimization: ASSOC and MEMBER can now be open-coded for all
1142 combinations of keyword arguments when second argument is constant
1143 and SPEED >= SPACE. In other cases a specialized version is
1145 * bug fix: using obsoleted structure instances with TYPEP and
1146 generic functions now signals a sensible error.
1147 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1148 (reported by Kristoffer Kvello)
1149 * bug fix: backtrace construction is now more careful when making
1150 lisp-objects from pointers on the stack, to avoid creating bogus
1151 objects that can be seen by the GC.
1152 * bug fix: defaulting of values in contexts expecting more than 7
1153 variables now works on x86-64. (reported by Christopher Laux)
1154 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1156 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1157 as the property-list of a symbol.
1158 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1159 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1160 situations requested, are once again file-compileable. (reported
1163 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1164 * MOP improvement: support for user-defined subclasses of
1165 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1166 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1167 the expansion of DEFMETHOD, is responsible for generating a form
1168 which creates a list of specializers when evaluated. Additional
1169 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1170 debugging and introspective support.
1171 * minor incompatible change: the (unsupported) spinlock interface
1172 has changed: free spinlock now has the value NIL, and a held spinlock
1173 has the owning thread as its value.
1174 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1175 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1177 * enhancement: name of a socket-stream is now "a socket" instead of
1178 "a constant string".
1179 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1180 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1181 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1182 for generic functions. (Thanks to Tobias C. Rittweiler)
1183 * optimization: bignum printing speed has been improved by 20-40%
1184 (depending on the bignum size.)
1185 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1187 * bug fix: the cache used by the CLOS to store precomputed effective
1188 methods, slot offsets, and constant return values is now thread and
1190 * bug fix: generic function dispatch function updating is now thread
1191 and interrupt safe (in the sense that the known issues have been
1193 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1194 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1195 * bug fix: asdf systems with dependencies to the SB-POSIX or
1196 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1197 * bug fix: interrupt safety of applicable method computation has been
1200 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1201 * new contrib: sb-cover, an experimental code coverage tool, is included
1202 as a contrib module.
1203 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1204 significantly faster.
1205 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1206 produce more efficient code on x86-64 if the compiler can determine
1207 that (+ INDEX <constant>) does not require a bounds check and FOO
1208 has an element type at least 8 bits wide.
1209 * enhancement: a new, experimental synchronous timeout facility is
1210 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1211 * enhancement: when a symbol name conflict error arises, the
1212 conflicting symbols are always printed with a package prefix.
1213 (thanks to Kevin Reid)
1214 * enhancement: stepping is now once again supported on the SPARC and
1215 MIPS platforms. (It is also now more likely to work on CheneyGC
1217 * enhancement: sb-sprof can now also track and report accurate call
1219 * bug fixes: the treatment of non-standard subclasses of
1220 SB-MOP:SPECIALIZER is more correct.
1221 * incompatible change: PURIFY no longer copies the data from the
1222 dynamic space into the static and read-only spaces on platforms
1223 that use the generational garbage collector
1224 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1226 * bug fix: GC race condition occasionally resulting in crashes with
1227 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1229 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1231 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1232 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1233 * bug fix: an asynchronous interrupt could previously leave the
1234 system running with GC inhibited.
1235 * bug fix: a DECLARE form evaluated at top-level now causes an error
1236 rather than silently (or verbosely) returning NIL.
1237 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1238 result in a error being signalled due to source form lookup errors.
1239 (reported by Peter Graves)
1241 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1242 * incompatible change: removed writer methods for host-ent-name,
1243 host-ent-addresses -- changing the values did not update the DNS
1245 * minor incompatible change: changed experimental JOIN-THREAD interface
1246 * documentation: the manual now lists reader and writer methods
1247 in class slot documentation sections. (thanks to Richard M Kreuter)
1248 * documentation: unwinding from asyncronous events has been
1249 documented as unsafe.
1250 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1251 in multithreaded application code.
1252 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1253 platforms. (thanks to James Anderson for the optimization hint)
1254 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1256 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1257 variants no longer cons.
1258 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1259 their NOT- variants no longer cons.
1260 * optimization: EQUAL hash tables no longer use SXHASH for objects
1261 of all data types, but instead use an EQL hash for types for which
1262 EQUAL is the same as EQL.
1263 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1264 are significantly faster.
1265 * optimization: new STRING-OUTPUT-STREAM implementation is much
1266 faster for large outputs and conses 30% less on average.
1267 * enhancement: XREF information is now collected for references made
1268 to global variables using SYMBOL-VALUE and a constant argument.
1269 * enhancement: SIGINT now causes a specific condition
1270 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1271 * bug fix: FILE-POSITION used to signal an error in some cases where
1272 ANSI requires it to return NIL.
1273 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1274 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1275 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1277 * bug fix: bad type declaration in the CLOS implementation has
1278 been fixed. (reported by James Anderson)
1279 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1280 (thanks to Lutz Euler)
1281 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1282 required by the CLHS. (reported by Samium Gromoff)
1283 * bug fix: dead unbound variable references now signal an error.
1284 (reported by Marco Monteiro)
1285 * bug fix: / with an unused value was being deleted in safe code.
1286 (thanks to Marco Monteiro and Kevin Reid)
1287 * bug fix: number of characters that can be written onto a single
1288 line in a file is unlimited.
1289 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1290 been fixed by inhibiting interrupts for when GC is disabled.
1291 * bug fix: some interrupt safety issues with GETHASH, (SETF
1292 GETHASH), CLRHASH and REMHASH have been fixed.
1293 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1294 specifier no longer causes infinite recursion.
1295 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1296 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1297 * bug fix: an &environment argument with the correct variable information
1298 is passed to macros that are expanded during byte compilation.
1299 (reported by Samium Gromoff)
1300 * bug fix: a logic error was causing the hash function for bignums to
1301 have excessive amounts of collisions. (reported by Faré Rideau)
1302 * bug fix: modifying the contents of an array could change the return
1303 value of SXHASH on that array, which is only allowed for strings
1304 and bit vectors. (bug introduced in 0.9.16)
1305 * improvement: the x86-64/darwin port now passes all tests (except
1306 for the debugger tests) but should still be considered
1307 experimental until this is fixed.
1308 * improvement: a style-warning is signaled for CASE (etc) clauses with
1309 duplicate keys. (thanks to Kevin Reid)
1310 * improvement: macos/x86 and macos/x86-64 now use mach exception
1311 handlers for dealing with illegal instructions (for trapping and
1312 error handling) and memory protection violations (for GC).
1314 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1315 * new platform: experimental support for x86-64/darwin (MacOS).
1316 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1317 and getnameinfo sockets functions are used instead of gethostbyaddr
1318 and gethostbyname, on platforms where the newer functions are available.
1319 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1321 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1322 don't print the startup banner, but behave as if --noinform was passed
1323 as a command line argument. (thanks to Kevin Reid)
1324 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1325 * optimization: code using alien values with undeclared types is much faster.
1326 * optimization: the compiler is now able to open code SEARCH in more cases.
1327 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1328 * bug fix: using standardized COMMON-LISP special variables as loop
1329 variables no longer signals bogus package lock violations. (reported
1331 * bug fix: declaring local loop variables to be of a range-limited type
1332 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1333 (reported by Andras Simon)
1334 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1335 bugs remain on x86-64.)
1336 * bug fix: the #= and ## reader macros now interact reasonably with
1337 funcallable instances.
1338 * bug fix: type-checks for function arguments were compiled using the
1339 compiler policy settings of the wrong lexical environment (in 1.0.2
1341 * bug fix: SHADOW accepts characters as string designators, as required
1342 by the spec (thanks to Eric Marsden)
1343 * bug fix: fixed GC safety issues when foreign functions are called with
1344 non-base strings as arguments
1345 * bug fix: more consistent error output for fatal-compiler-warnings, like
1347 * bug fix: functions with proper names were showing up as NIL in
1348 backtraces (reported by Edi Weitz, regression in 1.0.2)
1350 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1351 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1352 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1353 produce more efficient code on the x86 if the compiler can determine
1354 that (+ INDEX <constant>) does not require a bounds check and FOO
1355 has an element type at least 8 bits wide.
1356 * bug fix: references to undefined variables are handled the same way
1357 in toplevel forms as in the normal compiler
1358 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1360 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1361 (thanks to Jon Buller)
1362 * improvement: faster compilation times for complex functions
1363 * improvement: added readlink support to SB-POSIX (thanks to Richard
1366 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1367 * improvement: experimental support for mach exception handling on
1368 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1370 * improvement: support for GBK external format.
1371 (thanks to Chun Tian (binghe))
1372 * improvement: the debugger now displays variables that have been closed
1373 over, in code compiled with (DEBUG 2) or higher
1374 * improvement: support for executable cores on NetBSD (thanks to
1376 * new feature: added a RESTART-FRAME debugger command
1377 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1378 be used to provide an implementation for FILE-POSITION on Gray streams
1379 (thanks to Eric Marsden)
1380 * new feature: add syslog, openlog and closelog support to SB-POSIX
1381 (thanks to Richard Kreuter)
1382 * optimization: the function call overhead in code compiled with
1383 a high DEBUG optimization setting is significantly
1384 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1385 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1386 (thanks to Tony Martinez)
1387 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1388 evaluated once (reported by Kevin Reid)
1389 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1390 works again (thanks to Kilian Sprotte)
1391 * bug fix: an error is signaled for tagbodies with duplicate tags
1392 (thanks to Stephen Wilson)
1393 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1394 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1395 stack frames from alien callbacks.
1396 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1397 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1398 65k pages (thanks to David Woodhouse)
1399 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1401 changes in sbcl-1.0.1 relative to sbcl-1.0:
1402 * new platform: FreeBSD/x86-64, including support for threading.
1403 * new feature: the compiler stores cross-referencing information
1404 about function calls (who-calls), macroexpansion (who-macroexpands)
1405 and special variables (who-binds, who-sets, who-references) for code
1406 compiled with (< SPACE 3). This information is available through the
1407 sb-introspect contrib.
1408 * new feature: users may subclass SEQUENCE, and have instances of
1409 these classes interoperate with standard Common Lisp functions if
1410 a number of methods are defined. (This feature is experimental
1411 and the interface subject to change based on feedback from SBCL
1412 users and the general community)
1413 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1414 x86/x86-64, rather than the previous fixed depth of 8
1415 * improvement: another pthread back-end of mutex "pthread-futex".
1416 * bug fix: non-ascii command-line arguments are processed correctly
1417 (thanks to Yaroslav Kavenchuk)
1418 * bug fix: non-required arguments were not passed correctly when a method
1419 defined using DEFMETHOD was called from a mop-generated method using
1420 CALL-NEXT-METHOD (reported by Pascal Costanza)
1421 * bug fix: recursion is now permitted in accessors through
1422 SLOT-UNBOUND. (reported by Pascal Costanza)
1423 * bug fix: an error was signaled at startup if the HOME environment
1424 variable was defined, but had an empty value (reported by Peter Van Eynde)
1425 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1426 signal a PROGRAM-ERROR, not a BUG.
1427 * bug fix: SB-POSIX:READDIR works when built with large file support.
1428 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1430 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1431 for better automatic indentation support. (thanks to Matt Pillsbury)
1432 * optimization: loading generic functions no longer takes O(n^2) time,
1433 proportional to the amount of methods in the generic function
1434 (reported by Todd Sabin and Jeremy Brown)
1435 * optimization: the FIND and POSITION family of sequence functions
1436 are significantly faster on arrays whose element types have been
1438 * improvements to the Windows port:
1439 ** Intermittent heap corruption problems have been fixed. (thanks
1440 to Alastair Bridgewater)
1441 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1443 ** Lisp is able to unwind foreign exception frames from alien
1444 callbacks. (thanks to Alastair Bridgewater)
1446 changes in sbcl-1.0 relative to sbcl-0.9.18:
1447 * improvement: experimental support for threading on FreeBSD/x86.
1448 (thanks to NIIMI Satoshi)
1449 * improvement: runtime option --dynamic-space-size can be used
1450 to set the size of the dynamic space reserved on startup.
1451 * improvement: floating point modes in effect are now saved in
1452 core, and restored on startup.
1453 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1454 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1455 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1456 threads are not running after *SAVE-HOOKS* have run.
1457 * improvement: writes to CLOS instance slots are type-checked in code
1458 compiled with (SAFETY 3)
1459 * improvement: floating-point exception handling on FreeBSD (thanks to
1461 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1462 (thanks to Zach Beane)
1463 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1465 * improvement: added support for the Shift-JIS external format.
1466 (contributed by NIIMI Satoshi)
1467 * improvement: callbacks are supported on Linux/PPC. (thanks to
1469 * bug fix: compiler bug triggered by a (non-standard) VALUES
1470 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1471 * bug fix: file compiler no longer confuses validated and already
1472 dumped structurres. (reported by Kaersten Poeck)
1473 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1474 fill-pointer signals a type-error as required. (thanks to
1476 * bug fix: disassemly of funcallable instances works.
1477 * bug fix: single stepping on PPC.
1478 * bug fix: fix thread-safety problems in the type system (generally
1479 manifesting as nonsensical errors like "STRING is a bad type specifier
1480 for sequences" or "The value 1 is not of type FIXNUM")
1481 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1482 (reported by Josip Gracin)
1483 * bug fix: an error is signaled for attempts to displace arrays with
1484 incompatible element types (thanks to Mario Mommer)
1485 * bug fix: more correct handling of wide characters in debug info
1486 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1487 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1488 and don't cause extra consing
1489 * optimization: MAP and MAP-INTO are significantly faster on vectors
1490 whose elements types have been declared.
1491 * Improvements to SB-SPROF:
1492 ** Support for allocation profiling
1493 ** Reduced profiling overhead, especially for long profiling runs
1494 * Improvements to the Windows port:
1495 ** floating point exceptions are now reported correctly.
1496 ** stack exhaustion detection works partially.
1497 ** more accurate GET-INTERNAL-REAL-TIME.
1498 ** better RUN-PROGRAM behaviour in case of errors in the child
1500 ** PROBE-FILE now simplifies pathnames correctly.
1501 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1502 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1503 1st 1970 (although time timezone data will be unreliable).
1504 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1506 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1507 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1508 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1510 * bug fix: two potential GC deadlocks affecting threaded builds.
1511 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1512 returns T (reported by Anton Kazennikov)
1513 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1514 with non-variable places
1515 * bug fix: remove a race condition in the setting of
1516 funcallable-instance functions, this should make threaded CLOS
1517 code more stable against memory faults.
1518 * bug fix: corruption of specials when unbinding is interrupted by an
1519 asynchronous unwind (reported by Hannu Koivisto)
1520 * improvement: the debugger will now also display local variables that
1521 are only used once, for code compiled with a DEBUG optimization quality
1524 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1525 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1526 * incompatible change: External-format support for FFI calls. The
1527 SB-ALIEN:C-STRING no longer implies an ASCII
1528 external-format. Instead the string is subject to external-format
1529 conversion. Additionally return values of type SB-ALIEN:C-STRING
1530 are by default of type (SIMPLE-ARRAY CHARACTER), not
1531 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1532 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1533 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1534 * incompatible change: SB-EXT package no longer contains the
1535 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1536 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1538 * incompatible change: the single-stepper is no longer available
1539 on Alpha, Mips and Sparc platforms.
1540 * minor incompatible change: the direct superclasses of
1541 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1542 not (STANDARD-OBJECT FUNCTION). This makes the
1543 class-precedence-lists of GENERIC-FUNCTION and
1544 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1546 * minor incompatible change: the default stream external format on
1547 non-unicode SBCL is no longer detected from the locale, but is always
1549 * new feature: Add a version of evaluator that uses an interpreter instead
1550 of the compiler. EVAL still uses the compiler by default, to switch it
1551 to use the interpreter, set the value of the variable
1552 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1553 * minor incompatible change: the single-stepper REPL has been merged
1554 with the normal debugger (see the "Stepping" heading of the debugger help
1555 for more details). The debugger command STEP will no longer switch
1556 to the single-stepper REPL.
1557 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1558 for a type now works.
1559 * bug fix: loading of default sysinit file works. (thanks to Leonid
1561 * bug fix: better detection of circularities in the file-compiler.
1562 (reported by Marco Monteiro)
1563 * bug fix: the CL pathname functions now work with files that have
1564 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1565 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1566 systems with Unix98 pty semantics.
1567 * bug fix: ASDF-INSTALL will now work with bsd tar.
1568 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1570 * bug fix: timers expiring in dead threads no longer cause a
1571 type-error (reported by Paul "Nonny Mouse").
1572 * bug fix: thanks to more lightweight single-stepper instrumentation,
1573 code compiled with (DEBUG 3) will compile and execute significantly faster,
1574 and will have more accurate type-inferencing than before
1575 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1576 whose bindings are modified
1577 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1578 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1579 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1580 CL:SOFTWARE-VERSION with proper version
1582 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1583 * feature: implemented the READER-METHOD-CLASS and
1584 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1585 as specified by AMOP.
1586 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1588 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1589 (thanks to Lutz Euler)
1590 * optimization: hashing of general arrays and vectors has been
1591 improved. (reported by Any Fingerhut)
1592 * enhancement: SB-INTROSPECT is now able to find definitions of
1593 profiled functions. (thanks to Troels Henriksen)
1594 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1595 as well, allowing compiler-macros for SETF-functions to expand.
1596 * enhancement: step-instrumentation no longer wraps known
1597 single-value functions with multiple-value context, allowing
1598 better type inference.
1599 * fixed bug #337: use of MAKE-METHOD in method combination now works
1600 even in the presence of user-defined method classes. (reported by
1601 Bruno Haible and Pascal Costanza)
1602 * fixed bug #339(c): if there are applicable methods not part of any
1603 long-form method-combination group, call INVALID-METHOD-ERROR.
1604 (reported by Bruno Haible)
1605 * fixed bug #361: the :FUNCTION initarg in the protocol for
1606 initialization of methods can now be used to override
1607 internally-produced optimized functions. (reported by Bruno
1609 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1610 system-provided lambda expression no longer cause warnings about
1611 unbound #:|pv-table| symbols.
1612 * bug fix: improved the handling of type declarations and the
1613 detection of violations for keyword arguments with non-constant
1615 * bug fix: potentially erronous calls to PATHNAME and
1616 MERGE-PATHNAMES were being flushed in some cases.
1617 (reported by Richard Kreuter)
1618 * bug fix: compiled calls to TYPEP were mishandling obsolete
1619 instances. (reported by James Bielman and Attila Lendvai)
1620 * bug fix: format strings with ~> without matching ~< no longer
1621 trigger an AVER, but signal an understandable error instead.
1622 (reported by Antonio Martinez)
1623 * bug fix: specifying an output-file without a pathname-type for
1624 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1625 of input-file instead of "fasl". (reported by Robert Dodier)
1626 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1627 (reported by James Y Knight).
1628 * bug fix: compiler-macros-function did not consider the environment
1629 argument for shadowing by local functions.
1630 * bug fix: compiler-macros expansion was inhibited by local INLINE
1632 * bug fix: inline expansions of known functions were subject to
1633 step-instrumentation in high debug policies, leading to problems
1634 with type-inference.
1635 * bug fix: compiler failed to differentiate between different CONS
1636 types in some cases.
1637 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1638 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1639 * bug fix: #368: incorrect use of expressed vs. upgraded array
1641 * bug fix: #306a: more precise unions of array types.
1642 * thread-safety improvements:
1643 ** CONDITION-WAIT could return early on Linux, if the thread was
1644 interrupted and subsequently continued with SIGCONT.
1645 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1647 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1648 * added support for the ucs-2 external format. (contributed by Ivan
1650 * minor incompatible change: pretty printing of objects of type
1651 (cons symbol) is, in the default pprint-dispatch-table, now
1652 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1654 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1655 called later in a class's lifetime, possibly as late as when the
1656 first instance of the class is created. Previously,
1657 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1658 class became finalizeable.
1659 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1660 for multibyte external-format streams. (thanks to "vbzoli")
1661 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1662 with :ALLOCATION :CLASS inherited from superclasses of the
1664 * fixed bug: anonymous classes can now be created using the :NAME
1665 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1666 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1667 * fixed bug: core-files saved with :EXECUTABLE T can again be
1668 executed when SBCL_HOME isn't set. (reported by James Knight)
1669 * fixed bug: toplevel LOCALLY forms with declarations could
1670 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1671 * fixed bug: printing from several different threads using different
1672 values of *print-case* could cause invalid output, due to
1673 some internal special variables of the printer not being bound
1674 thread-locally (reported by Max Mikhanosha)
1675 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1676 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1677 (reported by Pascal Costanza)
1678 * fixed bug: FINALIZE-INHERITANCE is called from
1679 REINITIALIZE-INSTANCE on classes when the class has previously
1680 been finalized, as required by AMOP.
1681 * minor code generation optimizations:
1682 ** better register allocation in CLOS dispatching functions
1683 ** overflow detection when coercing signed bytes to fixnums on x86-64
1684 ** is now implemented with one IMUL instruction instead of three shifts
1685 ** more efficient bit-vector access on x86 and x86-64
1686 ** more efficient access to raw structure slots on x86 and x86-64
1687 * fixed some bugs revealed by Paul Dietz' test suite:
1688 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1689 return its argument.
1691 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1692 * feature: thread support on Solaris/x86, and experimental thread support
1694 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1696 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1697 default method for SLOT-UNBOUND.
1698 * minor incompatible change: prevent the user from specializing the
1699 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1700 somewhat counter to the intent of the protocol, I (CSR) think, and
1701 additionally it just doesn't work in SBCL as currently
1702 implemented, thanks to optimizations (that are always valid for
1703 the other three SLOT-VALUEish functions, but not for the setter).
1704 * minor incompatible change: the :SB-LDB feature is now enabled by
1705 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1706 the low-level debugger.
1707 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1708 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1709 down to an SB-SYS:FD-STREAM, allowing direct communication in
1710 more cases. Temporary files and pipes are still used as a fallback
1712 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1713 on the non-reentrant C functions localtime(3) and gmtime(3).
1714 * bug fix: global optimization policy was not visible in LOCALLY and
1716 * bug fix: class objects can be used as specializers in methods.
1717 (reported by Pascal Costanza)
1718 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1719 with a type error. (reported by blitz_ on #lisp)
1720 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1721 failed with a type error. (reported by Pascal Bourguignon)
1722 * bug fix: merging pathnames against defaults with :DIRECTORY
1723 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1724 (reported by James Y Knight)
1725 * bug fix: saving large (>2GB) cores on x86-64 now works
1726 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1727 constant index and a (simple-array (signed-byte 32)) array
1728 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1729 error (patch by Robert J. Macomber)
1730 * feature: TIME now displays the amount of run-time spent in GC
1731 * bug fix: The debugger now does a better job of respecting
1732 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1733 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1734 workaround for bug 403.)
1735 * bug fix: floating point exception handling now works on FreeBSD
1736 (thanks to NIIMI Satoshi)
1737 * fixed some bugs revealed by Paul Dietz' test suite:
1738 ** MISC.641: LET-conversion were not supposed to work in late
1740 * improvements to the Win32/x86 port:
1741 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1742 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1744 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1745 * new feature: source path information is generated for macro-expansion
1746 errors for use in IDE's like Slime (thanks to Helmut Eller)
1747 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1748 * bug fix: compiler does not loop forever on an invalid type in
1750 * improvement: compilation of most CLOS applications is significantly
1752 * optimization: added a limited bytecode compiler for simple toplevel
1753 forms, speeding up compilation and FASL loading
1754 * bug fix: the statistical profiler now properly distinguishes anonymous
1757 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1758 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1759 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1760 into subclasses of the finalized class. Now user calls to
1761 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1762 system return before any subclasses are finalized.
1763 * minor incompatible change: The reader no longer ignores errors
1764 regarding non-existent packages in #+ and #- feature tests.
1765 * new feature: command line options --no-sysinit, --no-userinit to
1766 inhibit loading the corresponding init files
1767 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1768 for sending data through UDP sockets (thanks to François-René Rideau)
1769 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1770 error is signalled instead (thanks to François-René Rideau)
1771 * minor incompatible change: Error signalling behaviour of lexical
1772 operations violating package locks has changed slightly. Refer to
1773 documentation on package locks for details.
1774 * enhancement: EVAL can process IF-expressions without resorting to the
1776 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1777 constant-folded properly due to missing out-of-line functions.
1778 (reported by Mika Pihlajamäki)
1779 * bug fix: LISTEN sometimes returned T even in cases where no data was
1780 immediately available from the stream
1781 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1782 were reversed. (reported by Levente Mészáros)
1783 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1784 when given a vector displaced to an adjustable array. (reported
1786 * bug fix: derivation of float boundaries from numbers outside the
1787 appropriate float range (reported by John Wiseman)
1788 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1790 * bug fix: precomputing cache entries for generic functions with
1791 some subclasses of specializers as yet invalid does not attempt to
1792 fill a cache line with a negative offset. (reported by Levente
1794 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1795 allow condition class objects as arguments to DOCUMENTATION and
1796 (SETF DOCUMENTATION); only find and set documentation for
1797 structure names for the STRUCTURE doc-type. (suggested by Gary
1799 * improvements to the Win32/x86 port:
1800 ** (user-homedir-pathname) and default initialization file
1801 locations now know about the user's "Documents and Settings"
1802 directory (thanks to Yaroslav Kavenchuk)
1803 ** run-program is implemented (thanks to Mike Thomas)
1804 ** sockets support (thanks to Timothy Ritchey)
1805 ** better backtrace support (thanks to Alastair Bridgewater)
1806 ** sb-grovel supported
1807 ** asdf-install and sb-posix work somewhat
1808 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1809 * improvements to the Solaris/x86 port:
1810 ** works on Solaris 11/Solaris Express
1811 ** floating-point exception handling support
1812 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1813 * fixed some bugs revealed by Paul Dietz' test suite:
1814 ** REMOVE-METHOD returns its generic function argument even when
1815 the method is not one of the generic functions' methods.
1816 ** objects declared as MEMBER types can be admissible arguments to
1817 structure accessors.
1818 ** printing characters should simply be printed by the FORMAT ~:C
1820 ** compiler failure when compiling functions with hairy constant
1821 defaults for optional parameters.
1822 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1823 function, which is already optimized.
1825 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1826 * new platform: experimental support for SBCL x86/Darwin, including
1827 MacOS X 10.4.5 on Intel.
1828 * new feature: Unicode character names are now known to the system
1829 (through CHAR-NAME and NAME-CHAR).
1830 * new feature: the filesystem location of SBCL's core file is
1831 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1832 * minor incompatible change: the contrib modules SB-POSIX and
1833 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1834 this change is to make it easier to distribute
1835 location-independent binaries.
1836 * bug fix: as implied by AMOP, standardized classes no longer have
1837 slots named by external symbols of public packages. (reported by
1839 * optimization: calling functions via constant symbols -- as in
1840 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1841 via the function object as in (FUNCALL #'FOO).
1842 * optimization: on x86 and x86-64, the calling convention for the
1843 case of dealing with an unknown number of values has been altered
1844 to be friendlier to the prediction heuristics implemented,
1845 particularly with reference to CALL and RET pairing. (thanks to
1846 Alastair Bridgewater)
1847 * optimization: on x86, the code for access to thread-local symbol
1848 values and binding/unbinding of thread-local symbols is smaller.
1849 (thanks to Alastair Bridgewater)
1850 * enhancement: CONSTANTP is now able to determine constantness of
1851 more complex forms, including calls to constant-foldable standardized
1852 functions and some special forms beyond QUOTE.
1853 * fixed bug: occasional GC crashes on Solaris/x86
1854 * optimization: x86-64 supports stack allocation of results of simple
1855 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1856 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1857 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1858 (thanks to James Knight)
1859 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1860 by the standard (thanks to Stephen Compall)
1862 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1863 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1864 be used for bundling the runtime and the core file into one
1865 executable binary. This feature is not currently supported on all SBCL
1866 platforms. (thanks to James Bielman and NIIMI Satoshi)
1867 * new feature: a generational or ephemeral garbage collector is now
1868 the default on the PowerPC platform (both Linux and Darwin). The
1869 old Cheney (stop and copy) collector is a build-time option.
1870 (thanks to Cyrus Harmon, after Raymond Toy)
1871 * minor incompatible change: the method by which SBCL finds its
1872 contributed modules has changed; it no longer relies on symbolic
1873 links from an $SBCL_HOME/systems directory, but searches directly
1874 in the subdirectories of $SBCL_HOME.
1875 * enhancement: the dynamic heap size on the Linux/PPC platform is
1876 markedly larger, even using the older Cheney garbage collector.
1877 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1878 certain complicated string types. (reported by Gary King)
1879 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1880 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1881 * fixed bug: DOCUMENTATION on structures no longer signals an
1882 error if no structure type of the right name exists. (reported by
1884 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1885 applicable non-standard primary or :AROUND method on
1886 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1887 no longer cause unbound variable errors. (reported by Kevin Reid)
1888 * fixed bug: metacircle resolution in cases where methods have slots
1889 added before the slots from STANDARD-METHOD. (reported by Jean
1891 * fixed bug: the Power PC assembly code for calling into lisp
1892 sometimes computed its return address wrongly (depending on the
1893 15th bit of the address, fixed at link time). This used to
1894 manifest itself in a segmentation violation while building PCL.
1895 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1896 many others over the years)
1897 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1898 the stack pointer (treated as a signed integer) is negative.
1899 (thanks to Peter van Eynde)
1901 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1902 * new platform: experimental support for the Windows operating
1903 system has been added. (thanks to Alastair Bridgewater)
1904 * fixed several bugs in and robustified the PPC FFI (including
1905 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1906 * bug fix: allow non-simple string symbol names (reported by Paul
1908 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1910 * bug fix: saving a core corrupted callbacks on x86/x86-64
1911 * bug fix: closed a loophole in metacircularity detection and
1912 grounding in the PCL implementation of CLOS.
1913 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1914 (reported by tomppa on #lisp)
1915 * bug fix: clear the direction flag on Lisp -> C transitions, as
1916 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1917 (reported by Andrej Grozin and Hendrik Maryns)
1918 * optimization: major improvements to GC efficiency on GENCGC platforms
1919 * optimization: faster implementation of EQUAL
1920 * optimization: emit more efficient opcodes for some common
1921 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1922 * optimization: several other minor code-generation improvements on x86-64
1923 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1925 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1926 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1927 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1928 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1929 by ANSI as a generic function, and for consistency (SETF
1930 GENERIC-FUNCTION-NAME) is treated likewise.
1931 * fixed bug #233.b: make constraint propagation notice when a variable
1932 value is changed after it is referenced but before it is used
1933 * fixed bug #296: no more arbitrary behaviour from filenames with
1934 odd characters as --load arguments. (reported by Adam Warner)
1935 * fixed bug #390: :CHARACTER-SET pathname components now work as
1936 expected. (reported by Tim Daly Jr)
1937 * fixed bug #391: complicated :TYPE intersections in slot
1938 definitions no longer cause an error in PCL internals.
1939 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1940 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1941 returning the number of octets which would be written to the
1942 file-stream. (thanks to Robert J. Macomber)
1943 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1944 arguments to the internal stream functions. (thanks to David
1946 * bug fix: stack exhaustion is now handled on the x86/SunOS
1947 platform. (thanks to Daisuke Homma)
1948 * bug fix: more accurate ROOM results on GENCGC platforms
1949 * optimization: improved type inference for arithmetic-for
1950 index variables in LOOP
1951 * optimization: faster floating-point SQRT on x86-64
1952 * fixed some bugs revealed by Paul Dietz' test suite:
1953 ** DOCUMENTATION returns NIL instead of "" for method combinations
1954 that don't have a docstring
1956 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1957 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1958 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1959 therefore are not customizeable by user code (as seems to be at
1960 least permitted and maybe required by AMOP). As a consolation,
1961 however, the SBCL implementation of these functions now calls
1962 REINITIALIZE-INSTANCE as specified by AMOP.
1963 * bug fix: slot-definition documentation is propagated as per ANSI
1964 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1965 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1966 argument, as do eslotd initialization methods. (from Pascal
1967 Costanza's "Closer" project)
1968 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1969 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1971 * bug fix: it is now possible to have more than one subclass of
1972 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1973 (reported by Bruno Haible, Pascal Costanza and others)
1974 * bug fix: the dependent update protocol now works for generic
1975 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1976 and Pascal Costanza)
1977 * bug fix: condition-class instances corresponding to
1978 DEFINE-CONDITION forms are now created eagerly. (reported by
1979 Kalle Olavi Niemitalo on comp.lang.lisp)
1980 * bug fix: floating point printing is more accurate in some
1981 circumstances. (thanks to Simon Alexander)
1982 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1983 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1984 * bug fix: callbacks on OS X now preserve stack-alignment.
1985 (thanks to Cyrus Harmon)
1986 * enhancement: the x86-64 disassembler is much better at
1987 disassembling SSE instructions. (thanks to Lutz Euler)
1988 * enhancement: CHECK-TYPE on a variable now allows type-inference
1989 (thanks to Vincent Arkesteijn)
1990 * optimization: improved performance of EUC-JP external format.
1991 (thanks to NIIMI Satoshi)
1992 * optimization: performance improvements to IO on file streams of
1993 :ELEMENT-TYPE CHARACTER
1994 * optimization: much faster memory allocation on x86-64
1995 * optimization: faster garbage collections (latency and throughput)
1997 * optimization: faster bignum subtraction and division on x86 and x86-64
1998 (thanks to Lutz Euler)
1999 * optimization: more accurate type-inference for loops that use a
2000 floating point index variable or a negative step.
2002 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2003 * bug fix: add a workaround to SBCL looping infinitely at startup on
2004 Linux kernels with apparently buggy implementations of personality().
2005 (thanks to Svein Ove Aas)
2006 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2007 (thanks to David Lichteblau)
2008 * bug fix: local bindings shadow global symbol macros for
2009 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2010 * optimization: non-open coded uses of numeric comparison operators
2011 (e.g. >) no longer cons when called with more than one parameter
2012 on platforms supporting dynamic-extent allocation.
2013 * enhancement: saving cores with foreign code loaded is now supported
2014 on MIPS/Linux in addition to the previously supported platforms.
2015 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2016 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2017 * bug fix: division by zero in sb-sprof when no samples were collected
2018 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2019 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2020 (thanks to Alexander Kjeldaas)
2021 * new feature: ignoring the return values of destructive functions
2022 like NREVERSE signals a compile-time style-warning.
2023 (thanks to Kevin Reid)
2024 * new feature: the alignment of alien structure fields can be
2025 explicitly specified. (thanks to Cyrus Harmon)
2026 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2027 and dump core on SIGQUIT
2029 ** incompatible change: threads do not inherit values of specials
2030 from their parents (see manual)
2031 ** bug fix: threads stacks belonging to dead threads are freed by the
2032 next exiting thread, no need to gc to collect thread stacks anymore
2033 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2034 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2035 ** bug fix: dynamic variable and thread start related gc lossage
2036 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2038 * fixed some bugs revealed by Paul Dietz' test suite:
2039 ** SUBTYPEP is slightly more accurate on heinously complicated
2040 CONS types where some of the members have uncertain (in the
2041 NIL, NIL sense) type relationships to each other.
2043 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2044 no longer purified unless :PURIFY T is explicitly specified.
2045 ** Non-purified cores are significantly smaller than before
2047 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2048 * new feature: timers based on Zach Beane's excellent timer package
2049 * added support for the following external formats: koi8-u,
2050 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2051 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2052 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2053 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2054 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2055 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2056 * incompatible change: a threaded SBCL will no longer revert to
2057 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2058 * bug fix: interrupts are disabled until startup is complete; no
2059 more sigsegvs when receiving a signal to soon
2060 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2062 * bug fix: add a workaround for the memory randomization features in
2063 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2064 workaround will only be in effect on systems with the proc filesystem
2066 * bug fix: printing objects of type HASH-TABLE signals a
2067 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2069 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2070 have been running for over 50 days. (reported by Gilbert Baumann)
2071 * bug fix: the logic for getting names of functions gets less
2072 confused when confronded with alternate-metaclass
2073 funcallable-instances. (reported by Cyrus Harmon)
2074 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2075 consistent, even on internal alternate-metaclass objects.
2076 * bug fix: flush closure information collected by physical
2077 environment analysis prepass before the main pass. (bug reported
2079 * bug fix: compiler pack phase does not modify a hashtable, which is
2080 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2081 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2082 may be expanded. (reported by Alexander <malishev> on c.l.l)
2083 * bug fix: no more style-warnings when compiling code using LAST
2084 under high-SPEED low-SPACE settings. (reported by David Wallin
2086 * fixed some incompatibilities between SBCL's MOP and the MOP
2088 ** the METAOBJECT class is now implemented;
2089 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2090 STANDARD-OBJECT, as required;
2091 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2092 now compatible (as required by VALIDATE-SUPERCLASS); there
2093 remains a consistency requirement over the presence or absence
2094 of the FUNCTION class in the superclass list of finalized
2095 classes; see the manual for more details;
2096 ** the :AROUND method for COMPUTE-SLOTS on
2097 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2098 requested slot ordering.
2100 ** bug fix: parent thread now can be gc'ed even with a live
2102 ** bug fix: binding a special with PROGV to no value is not
2103 the same as that symbol not having been bound (thanks to
2105 ** bug fix: binding specials is thread safe (thanks to
2107 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2108 and SB-SPROF do not die with 'no handler for signal XX in
2109 interrupt_handle_now(..)' anymore
2110 ** bug fix: WITH-TIMEOUT works with multiple running threads
2111 * fixed some bugs revealed by Paul Dietz' test suite:
2112 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2113 the :method-class keyword argument.
2115 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2116 * new port: the Solaris operating system on x86 processors is now
2117 mostly supported, though some rough edges in the environment
2118 remain. (thanks to Daisuke Homma)
2119 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2120 space for its heap. (thanks to Thiemo Seufer)
2121 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2122 closures. (thanks to Thiemo Seufer)
2123 * minor incompatible change: eof selects abort in the debugger.
2124 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2125 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2127 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2128 deprecated, and will go away in a future revision of SBCL.
2129 * minor incompatible change: GC-ON and GC-OFF are no longer
2130 implemented with a counter, it does not matter how many times gc
2131 is switched on or off
2132 * bug fix: discriminating functions for generic function classes
2133 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2134 make invalid assumptions about method precedence order. (reported
2136 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2137 platforms. (thanks to Thiemo Seufer)
2138 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2139 * bug fix: correct pathname printing with printer escaping is on.
2140 (thanks to Kevin Reid)
2141 * bug fix: complex VOP definitions in "user-space" no longer trigger
2142 package locks. (reported by Zach Beane)
2143 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2144 for funcallable-instance functions, and is no different from
2146 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2147 characters in the current readtable. (reported by Nicholas Neuss)
2148 * bug fix: SUBTYPEP on various CONS types returns more a more
2149 accurate acknowledgment of its certainty.
2150 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2151 lists in some cases. This partially fixes bug 384.
2152 * flush all standard streams before prompting in the REPL and the
2154 * bug fix: signal handling and triggering gc do not conflict
2155 directly anymore, in particular a high frequency sb-sprof does
2156 not prevent gc from running
2157 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2158 approximation for timezone and DST information between the
2159 universal time epoch and the smallest negative 32-bit time_t.
2160 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2161 year 1899 so long as the encoded time is non-negative.
2162 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2163 the correct answer for non-constant rotate amounts when performing
2164 an inline 32-bit rotation.
2166 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2167 there is only one thread in the session
2168 ** bug fix: memory leak for streams created in one thread and
2169 written to in another
2170 ** bug fix: lockup when compiled with gcc4
2171 ** bug fix: race that allows the gc to be triggered when gc is
2173 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2174 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2176 ** bug fix: finished threads are gc'ed properly
2177 * fixed some bugs revealed by Paul Dietz' test suite:
2178 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2179 has arguments with a different set of applicable methods from
2180 the orignal arguments.
2181 ** The type error thrown by MAP now has a correct expected-type
2183 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2184 name a compiled function.
2185 ** Three MISC tests where a large bignum was improperly coerced to
2186 a float that couldn't represent that bignum during type
2187 derivation were fixed.
2188 ** SUBTYPEP can now handle types involving the negation of a
2189 list-form FUNCTION type.
2190 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2191 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2192 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2194 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2195 * New feature: Experimental support for bivalent streams: streams
2196 opened with :element-type :default now allow character and binary
2197 (unsigned-byte 8) I/O
2198 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2199 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2201 * bug fix: no more highly sporadic "couldn't check whether ~S is
2202 readable" when reading a stream and an interrupt hits in the middle
2203 of a select system call
2204 * compiler better recognizes complex arrays (reported by Eduardo
2206 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2207 for variables on SBCL built with linkage-tables. (reported by Luis
2209 * various error reporting improvements.
2210 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2211 (Thanks to Thiemo Seufer)
2212 * bug fix: nested reader invokations work correctly
2213 * bug fix: it is possible to have simultaneous references to foreign
2214 code and foreign data with the same name.
2216 ** added x86-64 support
2217 ** incompatible change: the threading api now works with thread
2218 objects instead of thread ids
2219 ** bug fix: threads are protected from signals and interruption when
2220 starting up or going down
2221 ** bug fix: a race where an exiting thread could lose its stack to gc
2222 ** bug fix: don't halt on infinite error in threads if possible
2223 ** fixed numerous gc deadlocks introduced in the pthread merge
2224 ** bug fix: fixed thread safety issues in read and print
2225 ** bug fix: debugger doesn't hang on session lock if interrupted at
2226 an inappropriate moment
2227 ** bug fix: run-program is now thread safe(r)
2228 ** bug fix: inner with-recursive-lock no longer releases the mutex
2229 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2230 * fixed some bugs revealed by Paul Dietz' test suite:
2231 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2232 EXPECTED-TYPE slots filled.
2233 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2235 ** MAKE-CONDITION accepts classes as type-designators.
2236 ** COMPILE may never return NIL.
2237 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2238 range before calling Unix time functions
2240 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2241 * numerous signal handling fixes to increase stability
2242 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2243 * minor incompatible change: we now correctly canonize default
2244 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2245 as per the MOP, rather than the historical (INITARG INITFUNCTION
2246 INITFORM). (reported by Bruno Haible)
2247 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2248 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2249 for more information.
2250 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2251 pathname is a directory pathname.
2252 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2253 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2255 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2256 :METACLASS argument in addition to a class name. (reported by
2257 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2258 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2259 (reported by Bruno Haible; patch from Gerd Moellmann)
2260 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2262 * bug fix: more cleanups to the floating point exception handling on
2263 x86-64 (thanks to James Knight)
2264 * bug fix: the compiler does not try to constant fold calls of
2265 COERCE and BIT-* functions when they return freshly constructed
2266 objects. (reported by Nikodemus Siivola and Paul Dietz)
2267 * optimization: improved the MIPS versions of generic (in the
2268 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2269 * optimization: direct conversion of (unsigned-byte 32) to floats on
2270 the PowerPC platform.
2271 * optimization: structure instances with raw slots now use less
2272 memory, and probably show better memory locality. (thanks to
2274 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2275 are treated as requests for stack allocation on the x86-64,
2276 Alpha-32, PPC and SPARC platforms.
2277 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2278 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2279 (thanks to Tony Martinez)
2281 ** gcing a dead thread can no longer lead to lockups
2282 ** threads block signals until they are set up properly
2283 ** errno is no longer shared by threads
2284 ** interrupt-thread restores the eflags register on x86
2285 ** fixed some lockups due to gc/thread interaction
2286 * fixed some bugs revealed by Paul Dietz' test suite:
2287 ** invalid dotted lists no longer raise a read error when
2288 *READ-SUPPRESS* is T
2289 ** signal an error if a symbol that names a declaration is used
2290 as the name of a type, or vice versa
2291 ** allow using the (declare (typespec var*)) abbreviation for
2292 (declare (type typespec var*)) with all type specifiers
2293 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2294 ** always return NIL from PPRINT-POP when OBJECT is NIL
2295 ** don't signal errors when pretty-printing malformed LABELS,
2296 FLET or MACROLET forms
2297 ** declarations in a DOLIST body are also in scope for the
2299 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2301 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2304 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2305 * fixed cross-compiler leakages that prevented building a 32-bit
2306 target with a 64-bit host compiler.
2307 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2308 opened with :IF-EXISTS :OVERWRITE.
2309 * fixed bug 281: error for an invalid qualifier in a short-form method
2310 combination method is not signalled until the faulty method is called.
2311 * bug fix: iteration variable type inferrer failed to deal with open
2312 intervals. (reported by Alan Shields)
2313 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2314 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2315 or more runtime options were provided to the sbcl binary.
2316 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2317 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2318 * optimization: there's now a fast-path for fixnum arguments in the
2319 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2320 * optimization: the code generated on x86-64 is more compact thanks
2321 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2322 * bug fix: floating-point exception handling now partly works on
2323 x86-64. (Thanks to James Knight)
2324 * improvement to the MIPS backend from Thiemo Seufer: C-style
2325 64-bit long long arguments and return values to alien functions
2327 * fixed some bugs revealed by Paul Dietz' test suite:
2328 ** the type-error signalled from WARN has a filled-in DATUM slot.
2329 ** the type-error required when a stream is not associated with
2330 a file has the stream as its datum.
2331 ** type-errors on single-floats on x86-64 no longer have
2332 :INVALID-OBJECT as the datum
2333 ** the type-errors signalled for invalid function names now have
2334 a correct expected type
2335 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2336 for structures defined with a :TYPE. Documentation strings for
2337 typed structures are no longer immediately discarded
2338 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2339 broadcast streams and synonym streams. FILE-LENGTH now also works
2340 on broadcast streams.
2342 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2343 * incompatible change: the --noprogrammer option, deprecated since
2344 version 0.7.5, has been removed. Please use the equivalent
2345 --disable-debugger option instead.
2346 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2348 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2349 inoperational for a while now) has been completely removed.
2350 * null lexical environments are now printed as #<NULL-LEXENV>,
2351 significantly reducing the amount of clutter in typical backtraces.
2352 * documentation on weak pointers, finalization, and after GC hooks
2353 has been added to the manual.
2354 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2355 as other specialized array types, is much faster. SUBSEQ and
2356 COPY-SEQ on such arrays have also been sped up.
2357 * optimization: EQL is now more efficient when at least other argument
2358 is known to be of type (OR FIXNUM (NOT NUMBER)).
2359 * fixed bug: the runtime is now less vulnerable to changes in the
2360 size of the SBCL object on OS X, and virtual memory is reserved for
2361 all spaces that need to be at a fixed address.
2362 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2363 * fixed bug: finalizers and after GC hooks that cause consing are now
2365 * fixed bug: compiler error messages and summaries are now printed to
2366 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2367 * fixed inference of the upper bound of an iteration variable.
2368 (reported by Rajat Datta).
2369 * fixed bug 211e: calling local functions with duplicated constant
2370 keyword argument no longer causes a bogus style warning about an
2372 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2373 ftype declaration to be disregarded. (reported by Dave Roberts)
2374 * fixed bug 373: caused by erronous compilation of references to alien
2375 variables in the runtime on ppc/darwin.
2376 * fixed bug 376: CONJUGATE type deriver.
2377 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2378 (reported by Baughn on #lisp)
2379 * a cross-compiler bug on non-x86 platforms has been identified and
2380 fixed. (thanks to Bruno Haible)
2381 * improvements to the MIPS runtime code for increased stability.
2382 (thanks to Thiemo Seufer)
2383 * increased the maximimum compact environment size to allow
2384 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2385 * partial workaround for bug 135: don't name the function we're
2386 calling for hairy cases of EVAL, so as not to accumulate one environment
2387 entry per call to EVAL. (reported by Kevin Reid)
2388 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2389 * optimization: type testing for non-vector arrays should be faster.
2390 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2392 * bug fix: setting 31st element of a bit vector to zero did not work
2394 * bug fix: redefining a class definition which failed due to a
2395 previous accessor / function clash now works (but see BUGS entry
2396 #380 for more problems in this area). (thanks to Zach Beane)
2397 * the long form of DEFINE-METHOD-COMBINATION disables method group
2398 checking when given a single method group with pattern *. (thanks
2400 * on x86 compiler supports stack allocation of results of simple
2401 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2402 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2403 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2404 * fixed some bugs related to Unicode integration:
2405 ** the restarts for recovering from input and output encoding
2406 errors only appear when there is in fact such an error to
2408 * fixed some bugs revealed by Paul Dietz' test suite:
2409 ** MISC.549 and similar: late transformation of unsafe type
2410 assertions into derived types caused unexpected code
2412 ** SCALE-FLOAT type deriver is less wrong.
2413 ** type derivers for EXP, LOG and similar functions compute result
2414 types for complex arguments better.
2415 ** (MISC.563) CONJUGATE type deriver works for very restricted
2417 ** out-of-line type testers for character strings are available.
2418 ** EQUAL compiler transform understands specialness of objects
2420 ** accessing double-floats stored on the stack now works on x86-64.
2421 ** debugger internals could sometimes create invalid lispobjs,
2422 resulting in GC crashes.
2423 ** MISC.548: type check weakening can convert required type into
2425 ** initialization forms for bindings are not in scope of free special
2428 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2429 * incompatible change: thread support for non-NPTL systems has
2430 been removed - locking is buggy and unreliable. A threaded
2431 SBCL build will now warn at startup and refuse to create
2432 new threads, unless futex support is detected in the kernel
2433 * incompatible change: the top level REPL now has only an ABORT
2434 restart associated with it, not TOPLEVEL and ABORT as it used to.
2435 TOP and TOPLEVEL are now available as debugger commands for
2436 returning to the top level.
2437 * incompatible change: forms evaluated in the REPL now use the
2438 global optimization policy.
2439 * incompatible change: user- and system-initialization files are
2440 no longer processed with LOAD, but by READ and EVAL; hence the
2441 global optimization policy, startup package, readtable, etc,
2443 * The .fasl file format number has been incremented because of
2444 various incompatible changes.
2445 * internal entry point details and argument counts no longer appear
2446 in backtraces unless explicitly requested by setting
2447 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2448 * built-in and standard functions no longer have names like "top
2449 level local call to FOO".
2450 * fixed bug 32: functions defined in non-null lexical environments
2451 now have more legible printed representation
2452 * fixed bug 33: functions defined in non-null lexical environemnts
2453 are now more amenable to inspection by INSPECT.
2454 * workaround for bug 354: XEPs no longer appear in backtraces unless
2455 explicitly requested.
2456 * fixed bug: receiving the signal which results from attempting to
2457 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2458 SIGBUS on other *BSDs) on architectures where the C stack is also
2459 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2460 to debugger confusion or wild instability. (reported by Cheuksan
2462 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2463 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2464 (reported by Lutz Euler)
2465 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2466 compiled code. (reported by Paul Dietz)
2467 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2468 specialized parameter to a method to something that is not TYPEP
2469 the specializer is now possible.
2470 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2471 face of package deletion.
2472 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2473 pathname. (reported by David Tolpin on c.l.l)
2474 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2475 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2476 STRING-STREAM and FILE-STREAM, respectively.
2477 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2478 than previously; in addition, its test suite is now run on build.
2479 (thanks to Luke Gorrie)
2480 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2481 * optimization: added a immediate representation for single-floats
2483 * fixed some bugs revealed by Paul Dietz' test suite:
2484 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2485 ** attempting to create a package with a colliding nickname causes
2486 correctable errors to be signalled.
2487 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2488 ** several x86-64 backend bugs related to sign-extension of immediate
2491 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2492 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2493 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2495 * fixed loading of multiply forward-referenced layouts.
2496 (thanks to Cheuksan Wang)
2497 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2498 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2499 (see COMPILE-FILE documentation for details of :PRINT option)
2500 * fixed bugs 19 and 317: fixed-format floating point printing is
2501 more accurate. This also fixes a bug reported by Adam Warner
2502 related to the ~@F format directive.
2503 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2505 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2506 dispatch table if the from-char is a dispatch macro character.
2507 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2508 by the wordsize no longer ignore the last word. (reported by Lutz
2510 * fixed bug in type checking of dynamic-extent variables. (reported
2512 * optimization: sequence traversal functions use their freedom to
2513 coerce function designators to functions.
2514 * optimization: code with many calls to CLOS methods specialized on
2515 CLOS classes has had redundant type checks removed.
2516 * optimization: added declarations to speed up operations that access
2517 the internal character database (for example STRING-UPCASE)
2518 * optimization: comparison operations between floats and sufficiently small
2519 fixnums no longer create extra rationals
2520 * fixed some bugs related to Unicode integration:
2521 ** portions of multibyte characters at the end of buffers for
2522 character-based file input are correctly transferred to the
2523 start of the buffer at the next read.
2524 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2525 passing it through to OPEN.
2526 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2527 argument given to internal calls to OPEN.
2528 ** the built-in comment readers (introduced by character sequences
2529 ";" and "#|") are more forgiving to encoding errors; they will
2530 STYLE-WARN and then attempt to resync the stream at a character
2531 boundary. (thanks to Teemu Kalvas)
2532 * fixed some bugs in the x86-64 port:
2533 ** Negative short int return values from c-calls are sign-extended
2535 ** The stack is aligned to 16-bytes for c-calls, as required by
2536 the ABI. (thanks to Cheuksan Wang)
2537 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2538 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2539 for enums. (thanks to Vincent Arkesteijn)
2540 ** Multiple small optimizations and bugfixes for floating point
2542 * fixed some bugs revealed by Paul Dietz' test suite:
2543 ** Space, Tab, Linefeed, Return and Page have the invalid
2544 secondary constituent character trait.
2545 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2547 ** WITH-INPUT-FROM-STRING should only update the index place on
2549 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2550 works more reliably.
2551 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2552 with one-dimensional arrays with fill pointers.
2553 ** TYPE-OF failed on a complex with an integer realpart and a
2555 ** compiler failure during type inference for the code of form
2556 (IF (EQL X (THE ...)) ...) (MISC.535).
2558 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2559 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2560 platform. The port supports 61-bit fixnums, large memory spaces
2561 and reloading shared object files.
2562 * enhancement: saving cores with foreign code loaded is now
2563 supported on ppc/Darwin in addition to the previously supported
2565 * enhancement: the statistical profiler now walks deeper into the
2566 call stack for more meaningful call-graphs and accrued time
2567 reports (x86/x86-64 only). It also now reports time spent in
2569 * enhancement: it is now possible to trace most individual methods
2570 of a generic function in addition to tracing the generic function
2572 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2574 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2575 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2576 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2577 produces an error. (thanks to Vincent Arkesteijn)
2578 * bug fix: NAMESTRING on pathnames with :WILD components in their
2579 directories works correctly. (thanks to Artem V. Andreev)
2580 * fixed bug 125: compiler preserves identity of closures. (reported
2582 * bug fixed: functions with &REST arguments sometimes failed with
2583 "Undefined function" when compiled with (DEBUG 3). (reported by
2585 * bug fix: overflow during compiling of setting element of a bit
2586 vector with constant index and value. (reported by Timmy Douglas)
2587 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2588 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2589 * on x86 compiler partially supports stack allocation of dynamic-extent
2591 * GO and RETURN-FROM do not check the extent of their exit points
2592 when compiled with SAFETY 0.
2593 * fixed some bugs related to Unicode integration:
2594 ** encoding and decoding errors are now much more robustly
2595 handled; it should now be possible to recover even from invalid
2596 input or output to the terminal. (thanks to Teemu Kalvas)
2597 ** provided a first cut at implementing STRING-TO-OCTETS and
2598 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2599 ** altered the SB-MD5 contributed module slightly, changing the
2600 interface just enough for it to be supportable for builds where
2601 lisp characters are not eight bits.
2602 * fixed some bugs revealed by Paul Dietz' test suite:
2603 ** the FORMATTER-generated functions for ~V[ conditionals require
2604 the correct number of arguments.
2605 ** READ-FROM-STRING returns the mandated second value when applied
2606 to displaced strings.
2607 ** the #\Rubout and #\Backspace characters are treated as invalid
2608 constituent characters by the tokenizer.
2610 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2611 * new feature: reloading changed shared object files with
2612 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2613 * new feature: references to foreign variables and functions
2614 can now be compiled and loaded before the corresponding shared
2615 object file is loaded, as long as the foreign definitions are
2616 available at runtime.
2617 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2618 supported. (thanks to Dan Debertin)
2619 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2620 just keywords. (thanks to Vincent Arkesteijn)
2621 * fixed bug #331: structure-class instances corresponding to
2622 DEFSTRUCT forms are now created eagerly.
2623 * fixed bug #345: backtraces from calls to undefined functions work
2624 on x86 as well. Related bug #61 is now also partially fixed on x86
2625 (backtraces from throws to unknown catch tags.)
2626 * bug fix: lambda-list parsing is now stricter vrt. order and number
2627 of lambda-list keywords.
2628 * bug fix: as specified by AMOP, an error is signalled if a
2629 class-option appears multiple times in a DEFCLASS form. (reported
2631 * bug fix: RANDOM can be compiled when the compiler derives the type
2632 of its numeric argument as a disjoint set of small integers.
2633 (reported by Paul Dietz)
2634 * bug fix: starting a core saved with shared objects loaded when
2635 those objects are not available no longer causes threaded SBCL to
2636 hang. (reported by Sean Ross)
2637 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2639 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2640 argument, which must always be NIL. (reported by Kalle Niemitalo)
2641 * bug fix: printing 1.0d+23 no longer results in an error.
2642 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2643 * bug fix: structure slot setters preserve evaluation order. (thanks
2645 * fixed some bugs related to Unicode integration:
2646 ** RUN-PROGRAM can allow its child to take input from a Lisp
2647 stream. (reported by Stefan Scholl)
2648 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2649 locales. (reported by Ken Causey)
2650 * fixed some bugs revealed by Paul Dietz' test suite:
2651 ** INCF, DECF and REMF evaluate their place form as specified in
2653 ** FORMATTER expands ~{ ~} iteration directives with V or #
2654 parameters correctly.
2655 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2656 consequent uses no arguments correctly.
2657 ** the system has a partial understanding of the (COMPLEX RATIO)
2659 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2662 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2663 * new feature: a build-time option (controlled by the :SB-UNICODE
2664 keyword feature, enabled by default) for building the system with
2665 support for the entire 21-bit character space defined by the
2667 * new feature: the system now has rudimentary external-format
2668 support; the primary user-visible change at this time is that
2669 characters with the high bit set (such as lower-case-e-acute) will
2670 print correctly to a terminal in a UTF-8 environment.
2671 * minor incompatible change: BASE-CHAR no longer names a class;
2672 however, CHARACTER continues to do so, as required by ANSI.
2673 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2674 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2675 has been moved to the SB-EXT package (temporarily re-exported from
2677 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2678 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2680 * fixed bug #351: better error-handlind and reporting for malformed
2682 * fixed bug #350: bignum-printing is now more memory-efficient,
2683 allowing printing of very large bignums, eg. (expt 2 10000000).
2684 (reported by Bruno Haible)
2685 * fixed bug #302: better primitive-type selection for intersection
2687 * fixed bug #308: non-graphic characters now all have names, as
2688 required. (reported by Bruno Haible)
2689 * bug fix: redefining a class with different superclasses now correctly
2690 removes it from the direct-subclasses of its previous superclasses.
2691 (reported by David Morse)
2692 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2693 new value now works. (reported by Bruno Haible)
2694 * bug fix: correct canonicalization of multiple non-standard slot
2695 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2696 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2697 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2698 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2699 now exists, an signals an error.
2700 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2701 during macroexpansion for non-iteration variables as well. (reported
2702 by Bruno Haible for CMUCL)
2703 * bug fix: Cyclic structures and unprintable objects in compiler
2704 messages no longer cause errors. (reported by Bruno Haible)
2705 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2706 and READ-FROM-STRING all now return a primary value of NIL if
2707 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2708 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2709 (reported by Bruno Haible for CMUCL)
2710 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2711 do not designate a condition. (reported by Bruno Haible for
2713 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2714 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2715 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2716 permitting resolution in favour of any of the conflicting symbols.
2717 (reported by Bruno Haible for CMUCL)
2718 * bug fix: EQUAL compiler optimizations is less aggressive on
2719 strings which can potentially compare true despite having distinct
2720 specialized array element types.
2721 * bug fix: unit enumerations can be defined without dividing by
2722 zero. (thanks to Vincent Arkesteijn)
2723 * FORMAT compile-time argument count checking has been enhanced.
2724 (report from Bruno Haible for CMUCL)
2725 * a partial workaround for the bug 262: the compiler does not try to
2726 inline-expand a local function doing RETURN-FROM from a deleted
2727 BLOCK. (thanks to Peter Denno for the bug report and to David
2728 Wragg for the simple test case)
2729 * fixed some bugs revealed by Paul Dietz' test suite:
2730 ** RENAME-PACKAGE allows all package designators as new package
2732 ** constraint propagation and lambda variable substitution are
2733 more cautious in dealing with partially deleted code.
2734 ** compiler sometimes lost reoptimization passes.
2735 ** CERROR, given a condition as condition designator, passes its
2736 remaining arguments to the continue format control without
2738 ** Case-altering FORMAT directives work correctly on non-ASCII
2740 ** The REAL type specifier handles bounds outside the floating
2741 point ranges without signalling FLOATING-POINT-OVERFLOW.
2742 ** Functions with IR1-transformations can create intercomponent
2743 references to global functions.
2744 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2746 ** FORMAT ~R treats a nil value for its first parameter correctly.
2748 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2749 * enhancement: saving cores with foreign code loaded is now
2750 supported on x86/NetBSD and sparc/Linux in addition to the previously
2751 supported platforms.
2752 * bug fix: on some platforms repeated installations caused multiple
2753 copies of HTML documentation to be installed -- should not happen
2754 any more. (reported by Stefan Scholl)
2755 * bug fix: parsing self-recursive alien record types multiple times
2756 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2757 original patch by Helmut Eller for CMUCL)
2758 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2759 (thanks to Richard Kreuter)
2760 * bug fix: defining classes whose accessors are methods on existing
2761 generic functions in other (locked) packages no longer signals
2762 bogus package lock violations. (reported by François-René Rideau)
2763 * bug fix: special variables as DEFMETHOD parameters no longer have
2764 associated bogus type declarations. (reported by David Wragg and
2766 * bug fix: read-write consistency on streams of element-type
2767 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2768 * bug fix: redefiniton of the only method of a generic function with
2769 no DEFGENERIC no longer emits a full WARNING. In addition,
2770 redefinition of generic functions with no DEFGENERIC to an
2771 incompatible lambda list now signals an error. (thanks to Zach
2773 * bug fix: DEFGENERIC now works even when there's a function of the
2774 same name in an enclosing lexical environment. (thanks to Zach
2776 * fixed compiler failure, caused by instrumenting code during
2777 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2778 * optimization: added loop analysis and improved register allocation
2779 * fixed some bugs revealed by Paul Dietz' test suite:
2780 ** POSITION on displaced vectors with non-zero displacement
2781 returns the right answer.
2782 ** (SIMPLE-STRING) is a valid type specifier for sequence
2784 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2786 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2787 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2789 ** The justification version of the FORMAT ~< directive treats
2790 non-zero minpad parameter correctly.
2792 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2793 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2794 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2795 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2796 the supported interface.
2797 * new feature: Single-stepping of code compiled with DEBUG 2 or
2798 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2799 * new feature: saving cores with foreign code loaded is now
2800 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2801 Timothy Moore's work for CMUCL)
2802 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2803 parameters to * instead of NIL if no initform is supplied.
2804 (reported by Johan Bockgaard)
2805 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2806 correctly when FUNCALL appears as the car of the form. Note:
2807 despite this FUNCALL forms are not currently subject to
2808 compiler-macro expansion. (port of Raymond Toy's fix for the
2809 same from CMUCL, reported by Johan Bockgaard)
2810 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2811 (thanks for Teemu Kalvas)
2812 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2813 case instead of terminating the iteration (thanks for Julian
2814 Squires, Sean Champ and Raymond Toy)
2815 * bug fix: incorrect expansion of defgeneric that caused a style
2816 warning. (thanks for Zach Beane)
2817 * bug fix: slot accessor effective method computation works properly
2818 for classes with multiple non-standard applicable methods on
2819 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2820 * on x86 compiler supports stack allocation of results of LIST and
2821 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2822 CMUCL implementation by Gerd Moellmann)
2823 * fixed some bugs revealed by Paul Dietz' test suite:
2824 ** FORMAT strings with both the ~<~:;~> form of the justification
2825 directive and pretty-printing directives cause an error.
2827 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2828 * incompatible change: the internal functions
2829 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2830 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2831 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2832 instead of the old functions.
2833 * new feature: on platforms where "dladdr" is available foreign
2834 function names now appear in backtraces. (based on Helmut Eller's
2836 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2837 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2839 * documentation: ASDF manual is now distributed with SBCL.
2840 * bug fix: SBCL can now load its contributed modules with REQUIRE
2841 even if the system-provided entries have been removed from
2842 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2844 * bug fix: backtraces involving undefined functions or assembly
2845 routines are more informative. (thanks to Brian Downing)
2846 * bug fix: mutually referent alien structures now work correctly.
2847 (reported by Rick Taube)
2848 * bug fix: structures defined by WITH-ALIEN can be referred to
2849 within other definitions in the same WITH-ALIEN.
2850 * bug fix: division operators (MOD, TRUNCATE and the like) with
2851 constant zero divisors and integer dividends no longer generate
2853 * bug fix: provide default methods for INPUT-STREAM-P and
2854 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2855 * bug fix: improve the detection and resolution of MOP metacycles
2856 (where effective method computation depends on the generic function
2857 having its effective method computed). (reported by Bruno Haible)
2858 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2859 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2860 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2861 * bug fix: pathnames with relative directory components can be
2862 represented relative to default pathnames.
2863 * optimization: in taking the GCD of bignums, reduce the two bignums
2864 to approximately the same size (using Euclid's algorithm) before
2865 applying the more sophisticated binary GCD. (thanks to Juho
2867 * optimization: COUNT on bitvectors now operates word-at-a-time.
2868 * optimization: ASH with a positive, but not necessarily constant,
2869 (leftwards) shift, when in a modular context, is compiled to a
2871 * fixed some bugs revealed by Paul Dietz' test suite:
2872 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2873 if the corresponding argument is NIL.
2874 ** FORMAT directives accepting a minpad parameter treat negative
2876 ** literal commas as character parameters to FORMAT directives are
2878 ** literal spaces directly after ~<Newline> directives within a
2879 format-logical-block (~:< ~@:>) do not induce :FILL-style
2880 conditional newlines.
2881 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2882 as specified: it no longer includes conditional newlines).
2883 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2885 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2886 :PER-LINE-PREFIX argument does not evaluate to a string.
2888 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2889 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2890 the manual for details; for now, package locks can be disabled by
2891 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2892 if no major problems are found then it is likely that they will be
2893 compiled in unconditionally.
2894 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2895 unsupported operators on all platforms. To load a shared library
2896 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2897 object file, link it into a shared library outside of SBCL and
2898 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2900 * bug workaround: ROOM T and the GC were not working together
2901 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2902 reporting facility aren't preserved. That reporting has been
2903 disabled, so now until and unless someone figures out how to make it
2904 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2905 * minor incompatible change: as threatened around sbcl-0.8.0, the
2906 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2907 The symbols are also exported from SB-PCL for backwards
2908 compatibility, but more so than before SB-PCL should be treated as
2909 an implementation-internal package.
2910 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2912 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2913 It is known as of this release not to work on the Alpha, however.
2914 * fixed bug #167: errors signalled due to illegal syntax in method
2915 bodies are now more legible.
2916 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2917 designators and can hence be used with TYPEP.
2918 * fixed bug #333: CHECK-TYPE now ensures that the type error
2919 signalled, if any, has the right object to be accessed by
2920 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2921 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2922 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2924 * fixed bug #334: programmatic addition of slots using specialized
2925 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2926 and :ALLOCATION :CLASS slots.
2927 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2928 float exponents. (rereported by Peter Seibel)
2929 * fixed a bug: #\Space (and other whitespace characters) are no
2930 longer considered to be macro characters in standard syntax by
2931 GET-MACRO-CHARACTER.
2932 * fixed bug: initialization of condition class metaobjects no longer
2933 causes an instance of the condition to be created. (reported by
2935 * fixed bug: it is now possible to have slots such that
2936 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2937 description is neither :INSTANCE nor :CLASS.
2938 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2939 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2940 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2941 non-standard SLOT-DEFINITION-ALLOCATION values.
2942 * fixed bugs: various race conditions handling exiting threads.
2943 CL-PPCRE's thread torture test now passes.
2944 * fixed arguably-a-bug: GC time proportional to number of threads in
2945 system even when most of them are idle
2946 * optimization: improved performance of BIT and SBIT on bit-vectors.
2947 * .fasl file incompatibility: The fasl file version number has
2948 been incremented because of changes associated with package locks.
2950 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2951 * minor incompatible change: the system no longer provides
2952 optimization hints (conditions which are TYPEP
2953 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2954 mode; these hints are still emitted when higher SPEED optimization
2956 * new contrib module: a sampling profiler (profiling by statistical
2957 sampling, rather than by instrumenting functions) is available as
2958 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2959 * the behaviour of the standard function ED is now customizeable by
2960 third parties through a hook variable: see ED's documentation
2961 string for information on the protocol.
2962 * the compiler no longer emits efficiency notes for (FUNCALL X)
2963 when the type of X is uncertain under default optimization
2965 * fixed bug 276: mutating a binding of a specialized parameter to a
2966 method to something that is not TYPEP the specializer is now
2968 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2969 exhibit the required behaviour on the x86 platform. (thanks to
2970 Peter van Eynde, Eric Marsden and Bruno Haible)
2971 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2972 even for difficult arguments. (reported by Peter Graves)
2973 * fixed bug 141a: the backquote printer now descends quoted
2975 * fixed another bug in backquote printing: no more destructive
2976 modification of the form's list structure. (reported by Brian
2978 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2979 move between its address being taken and the call to
2980 interrupt_thread, fixing a crashing race condition.
2981 * the SB-POSIX contrib implementation has been adjusted so that it
2982 no longer exhibits ridiculously poor performance when constructing
2983 instances corresponding to C structs.
2985 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2986 * minor incompatible change: the sb-grovel contrib now treats C
2987 structures as alien (in the sense of SB-ALIEN) objects rather than
2988 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2989 has implications for memory management of client code
2990 (sb-grovel-returned objects must now be manually managed) and for
2991 type safety (alien objects now have full types).
2992 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2993 used to control emission of compiler diagnostics, rather than the
2994 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2995 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2996 quality should be considered deprecated.
2997 * install.sh now installs the user manual as well
2998 * (not quite a new documentable feature, but worth considering in
2999 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3000 beginnings of a semantically meaningful condition hierarchy is
3001 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3003 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3004 designator as the defaults argument.
3005 * fixed bug: Displaced arrays whose displaced-to array has become
3006 too small now cause ARRAY-DIMENSION to signal an error, providing
3007 for safer bounds-checking. (reported by Bruno Haible)
3008 * fixed bug: DEFCLASS slot definitions with identical :READER and
3009 :WRITER names now signal a reasonable error. (reported by Thomas
3011 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3013 * fixed bug: Invalid slot specification errors now print correctly.
3014 (thanks to Zach Beane)
3015 * fixed bug 320: Shared to local slot value transfers in class
3016 redefinitions now happen corrently from superclasses as
3017 well. (reported by Bruno Haible)
3018 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3020 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3021 improper lists correctly. (reported by Bruno Haible)
3022 * fixed bug 313: source-transform for <fun-name> was erroneously
3023 applied to a call of a value of a variable with name <fun-name>.
3024 (reported by Antonio Menezes Leitao)
3025 * fixed bug 307: The obsolete instance protocol ensures that
3026 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3027 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3028 remove dead unknown-values globs from the middle of the stack.
3029 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3030 easier to resume long computations after using *BREAK-ON-SIGNALS*
3031 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3032 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3033 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3034 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3035 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3036 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3038 * fixed a (fairly theoretical) bug in string printing: if
3039 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3040 does not have array-element-type equal to the most general string
3042 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3043 function name. (thanks to Nikodemus Siivola)
3044 * fixed bug: compile-time format string checker failed on
3045 non-closed ~{. (reported by Thomas F Burdick)
3046 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3047 don't warn on legal loop code involving more than one aggregate
3048 boolean. (thanks to Nikodemus Siivola)
3049 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3050 in keyword position to MAKE-PACKAGE was not being reported as
3052 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3053 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3055 * fixed bugs 280 and 312: the checking for multiple definitions in a
3056 file is less likely to become confused by uses of inline
3058 * fixed bug: the #S reader macro performs the keyword coercion
3059 specified for slot names. (reported by Kalle Niemitalo)
3060 * fixed bug: lambda lists may contain symbols whose names start with
3061 & but are not lambda-list-keywords; their occurrence triggers a
3063 * fixed bug 321: define-method-combination argument lambda lists do
3064 not cause invalid code to be generated when &optional or &aux
3065 variables have default values. (reported by Bruno Haible)
3066 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3067 have CLOS classes; this appears also to have fixed reported
3068 instability in stack exhaustion detection.
3069 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3070 MUFFLE-WARNING when no associated restart is present is now
3072 * optimization: rearranged the expansion of various defining macros
3073 so that each expands into only one top-level form in a
3074 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3076 * optimization: used a previously little-used slot in symbols to
3077 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3079 * fixed some bugs revealed by Paul Dietz' test suite:
3080 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3082 ** Fixed an optimization bug related to inheritance of initforms
3083 from local to shared slots.
3084 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3085 ** CAST optimizer forgot to flush argument derived type.
3086 ** print/read consistency on floats is now orders of magnitude
3087 more likely. (thanks also to Bruno Haible for a similar report
3089 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3090 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3091 for objects of type REAL. Make it so.
3092 ** SXHASH is specified (infelicitously) to respect similarity,
3093 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3094 it so. (thanks to Markus Ziegler)
3095 ** on the Alpha, the compiler succeeds in compiling functions
3096 returning a known number of arguments greater than 63.
3097 ** fixed handling of invalid NIL arguments in keyword position
3099 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3101 ** provided a readably-printable representation for RANDOM-STATE
3103 ** ensured that pathnames, where they have a namestring, always
3104 print using #P"..." syntax.
3106 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3107 * Support for the forthcoming 2.0 version of the NetBSD kernel
3108 running on x86 hardware has been added. (thanks to Perry
3109 E. Metzger most immediately, and others for their past work)
3110 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3111 current release version; SBCL's OpenBSD support had been broken
3112 since about the time of OpenBSD's switch to ELF binary format.)
3113 * [placeholder for DX summary]
3114 ** user code with &REST lists declared dynamic-extent, under high
3115 speed or space and low safety and debug optimization policy.
3116 * The manual has been converted to Texinfo format and the debugger
3117 chapter from the cmucl manual has been added.
3118 * A facility has been added to extract documentation strings from
3119 sbcl and store them as Texinfo-formatted snippets for inclusion in
3120 the manual (via Texinfo's @include directive)
3121 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3122 the test case to Dave Roberts)
3123 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3124 pointers. (reported by Sean Ross)
3125 * bug fix: PROFILE output is printed nicely even for large numerical
3126 values. (thanks to Zach Beane)
3127 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3128 greater than 32 handle EOF correctly.
3129 * bug fix: on X86 an immediate argument of the IMUL instruction is
3130 correctly printed in disassembly. (thanks to Lutz Euler)
3131 * bug fix: class slots in redefined classes preserve their old
3132 values. (thanks to Bruno Haible and Nikodemus Siivola)
3133 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3134 succeeds. (reported by Marco Baringer)
3135 * fixed some bugs revealed by Paul Dietz' test suite:
3136 ** READ-SEQUENCE now works on ECHO-STREAMs.
3137 ** RATIONALIZE works more according to its specification. (thanks
3140 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3141 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3142 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3143 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3144 should matter to you only if you rebind the printer control
3145 variables and then find you want different bindings in the
3146 debugger than in the ordinary execution of your program.)
3147 * The runtime build system has been tweaked to support building
3148 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3149 assembler and linker. (thanks to Nikodemus Siivola)
3150 * Unbound, undefined, undeclared variables now trigger full
3151 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3152 more often programmer error than deliberate exploitation of undefined
3154 * optimization: the hash algorithm for strings has changed to one
3155 that is less vulnerable to spurious collisions. (thanks to Juho
3157 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3158 needless bounds checking. (thanks to Juho Snellman)
3159 * optimization: implemented multiplication as a modular
3160 (UNSIGNED-BYTE 32) operation on the PPC backend.
3161 * fixed some bugs revealed by Paul Dietz' test suite:
3162 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3164 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3165 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3166 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3168 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3170 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3172 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3173 their output stream on EOF from read.
3174 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3175 have been read to end-of-file.
3176 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3178 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3179 description of determination of which consecutive characters
3181 ** Printing the "Space" character with escaping on now yields "#\\ ",
3182 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3183 ** Reading floating-point numbers with *READ-BASE* set to a number
3184 less than 10 works correctly.
3185 ** Reading floating-point numbers with *READ-BASE* set to a number
3186 more than 10 works correctly.
3187 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3188 the readtable currently in effect.
3190 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3191 * minor incompatible change: parsing of namestrings on a physical
3192 (Unix) host has changed; numbers after the final #\. in a
3193 namestring are no longer interpreted as a version field. This is
3194 intented to be largely invisible to the user, except that the
3195 meaning of the namestring "*.*.*" has changed: it now refers to a
3196 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3197 should usually be replaced by
3198 (make-pathname :name :wild :type :wild :version :wild)
3199 with the added benefit that this is more likely to be portable.
3200 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3201 option to OPEN now signals an error if the file being opened
3202 exists; this may have an impact on existing code.
3203 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3204 signals received, on the PowerPC platforms (both Linux and
3205 Darwin). (thanks to Pierre Mai for pointing out the location of
3207 * several fixes on OS X: The system now builds and runs cleanly on
3208 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3209 (thanks to Brian Mastenbrook)
3210 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3211 second-resolution: integer multiples of 1/3600 between -24 and 24.
3212 (thanks to Vincent Arkesteijn)
3213 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3214 type when called with 1 argument; PEEK-CHAR checked type of
3215 PEEK-TYPE only after having read first character from a
3216 stream. (reported by Peter Graves)
3217 * bug fix: the garbage collector now has much better locality
3218 behaviour, and in particular no longer treats all memory as being
3219 exhausted when many small objects point to each other in a deeply
3221 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3222 recognized as being TYPEP their class.
3223 * bug fix: the PUSHNEW documentation string has been corrected.
3224 (thanks to Vincent Arkesteijn)
3225 * bug fix: defaulting of the value for the last of an atypically
3226 large number of multiple values being bound was not being
3227 performed correctly on the Alpha or PPC platforms
3228 * optimization: implemented multiplication as a modular
3229 (UNSIGNED-BYTE 32) operation on the x86 backend.
3230 * optimization: SEARCH on simple-base-strings can now be open-coded.
3231 (see also contrib/compiler-extras.lisp for inspiration for
3232 teaching the compiler about the Boyer-Moore algorithm).
3233 * value, returned by MAX (and MIN) called with several EQUALP, but
3234 not EQL, arguments now does not depend on compiler settings.
3235 * fixed some bugs revealed by Paul Dietz' test suite:
3236 ** in stack analysis liveness information is propagated from
3237 non-local entry points.
3238 ** pathwise CAST removing failed when the CAST node did not start
3240 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3241 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3243 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3244 host is already defined.
3245 ** RENAME-FILE works on streams instead of signalling an internal
3247 ** PEEK-CHAR uses the current readtable when determining whether
3248 or not a character is whitespace.
3249 ** MERGE-PATHNAMES handles the case when the pathname does not
3250 specify a name while the default-pathname specifies a version.
3251 ** Pathnames now stand a better chance of respecting print/read
3253 ** Attempting to use standardized file system operators with a
3254 pathname with invalid :DIRECTORY components signals a
3256 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3258 ** DIRECTORY on logical pathnames is more correct.
3259 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3260 signal a TYPE-ERROR if their argument is not a stream.
3261 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3262 designator argument does not designate a stream.
3263 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3264 examining the synonym.
3265 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3267 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3268 element-type larger than ([UN]SIGNED-BYTE 32).
3270 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3271 * When built with the :SB-FUTEX feature, threaded builds now take
3272 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3273 for faster/more reliable mutex and condition variable support.
3274 * Incompatible change (but one you probably shouldn't have been using
3275 anyway): the interface and code for arbitrating between multiple
3276 threads in the same user session has been redesigned.
3277 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3278 error when called without an explicit environment argument.
3279 (thanks to Nikodemus Siivola)
3280 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3281 confused about their position. (thanks to Adam Warner and Gerd
3283 * bug fix: Pretty printing backquoted forms with unquotations in the
3284 argument list position of various code constructs such as LAMBDA
3285 now works correctly. (reported by Paul Dietz)
3286 * bug fix: Pretty printing unquotations no longer loses all
3287 stream position information.
3288 * optimization: performance of string output streams is now less
3289 poor for multiple small sequence writes.
3290 * optimization: performance of CSUBTYPEP in the presence of complex
3291 expressions involving CONS and NOT many times has been improved.
3292 (reported by Paul Dietz)
3293 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3295 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3296 streams for the REPL, for the debugger, and so on. (thanks to
3298 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3299 * fixed some bugs revealed by Paul Dietz' test suite:
3300 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3301 is ignored if it is too small, rather than propagating through
3302 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3303 ** extremely complex negations of CONS types were not being
3304 sufficiently canonicalized, leading to inconsistencies in
3306 ** VALUES tranformer lost derived type.
3308 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3309 * fixed a bootstrapping bug: the build process no longer assumes
3310 that the various BOOLE-related constants have the same value in
3311 host and target lisps. (noted by Paul Dietz' test suite on an
3312 SBCL binary built from CLISP)
3313 * The system can now be dynamically linked on the MIPS platform,
3314 which enables dynamic loading of foreign code from Lisp. (thanks
3315 to Ralf Baechle for discussions on the MIPS ABI)
3316 * The system now records debugging information for its own source
3317 files in a filesystem-position-independent manner, relative to
3318 the "SYS" logical host.
3319 * fixed a compiler bug: MV-LET convertion did not check references
3320 to the "max args" entry point. (reported by Brian Downing)
3321 * tweaked disassembly notes to be less confident about proclaiming
3322 some instruction as an LRA. (thanks to Brian Downing)
3323 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3324 now each have their own history, command character, and other
3325 characteristics. (thanks to David Lichteblau)
3326 * fixed some bugs revealed by Paul Dietz' test suite:
3327 ** compiler failure in compiling LOGAND expressions including a
3329 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3330 shift greater than 32.
3331 ** FLUSH-DEST did not mark blocks for type check regeneration.
3332 ** HANDLER-CASE failed to accept declarations in handler clauses
3333 in some circumstances.
3335 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3336 * New code in contrib/sb-introspect (still probably not entirely
3337 stable yet) provides some support for smart Lisp development
3338 environments like SLIME.
3339 * The conditions signalled for errors occurring when loading .fasl
3340 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3341 in a way which should help ASDF recover gracefully.
3342 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3343 stuff has been cleaned up. If you code contrib/ stuff, this might
3344 affect you, and you can look at contrib/README, contrib/STANDARDS,
3345 and/or the 0.8.4.27 diff to check.
3346 * In full calls the compiler now does not generate checks for declared
3347 argument types for all arguments.
3348 * various threading fixes
3349 ** and some experimental patches which didn't make it into
3350 the main tree for this release, but which are shipped in
3351 contrib/experimental-thread.patch as a possible fix for some
3352 failures (deadlock, spinning...) in GC-intensive multithreaded
3354 * fixed PPC build problem (source code incompatibility of different
3355 library versions): added offsetof() hackery which attempts to divine
3356 where glibc maintainers put uc_mcontext today
3357 * fixed bug 282: compiler does not trust type assertions while passing
3358 arguments to a full call.
3359 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3360 &OPTIONAL VALUES type parameter.
3361 * fix bug 214: algorithm for noting rejected templates is now more
3362 similar to that of template seletion. (also reported by rydis on
3364 * fixed bug 141b: printing backquoted information readably and prettily
3365 inserts a space where necessary.
3366 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3367 (CHARACTER 10) are now reported as errors, rather than propagated
3368 as unknown types. (reported by piso on #lisp)
3369 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3370 with values NIL and :ERROR. (thanks to Milan Zamazal)
3371 * fixed bug 191c: CLOS now does proper keyword argument checking as
3372 described in CLHS 7.6.5 and 7.6.5.1.
3373 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3374 longer raise an error; further, using a list as a for-as-arithmetic
3375 counter now raises a meaningful error.
3376 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3377 now understood by sequence creation functions such as MAKE-SEQUENCE
3379 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3380 a STRING-INPUT-STREAM.
3381 * compiler enhancement: SIGNUM is now better able to derive the type
3383 * type declarations inside WITH-SLOTS are checked. (reported by
3385 * fixed some bugs revealed by Paul Dietz' test suite:
3386 ** incorrect optimization of TRUNCATE for a positive first
3387 argument and negative second.
3388 ** compiler failure in let-convertion during flushing dead code.
3389 ** compiler failure while deriving type of TRUNCATE on an
3390 interval, containing 0.
3391 ** ASH of a negative bignum by a negative bignum count now returns
3393 ** intersection of CONS types now canonicalizes properly, fixing
3394 inconsistencies in SUBTYPEP.
3396 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3397 * incompatible change: The --disable-debugger command line
3398 option now clobbers the debugger at a more fundamental
3399 level, by redefining #'INVOKE-DEBUGGER instead of by
3400 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3401 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3402 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3403 enter the debugger REPL and then suffer recursive errors
3404 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3405 while under the new system, BREAK in --disable-debugger mode
3406 terminates the system just as an unhandled error would.
3407 * fixed compiler performance when processing loops with a step >1;
3408 * bug fix: DOCUMENTATION now retrieves generic function
3409 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3410 support has been systematized, and now supports the methods
3411 specified by ANSI, along with a default method and a method for
3412 slot documentation. (reported by Nathan Froyd)
3413 * bug fix: effective methods associated with a generic function are
3414 no longer cached over a change of that generic function's method
3415 combination. (reported by Andreas Fuchs)
3416 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3417 names a non-existent directory. (thanks to Andreas Fuchs)
3418 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3419 the PPC platform, lead to stack corruption; this has been fixed.
3420 (reported by Rainer Joswig)
3421 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3422 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3423 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3424 arithmetic in general, is now much more reliable on the PPC
3426 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3427 platform now returns the right answer.
3428 * optimization: restored some effective method precomputation in
3429 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3430 precomputation is now tunable.
3431 * optimization: compiler-internal data structure use has been
3432 reviewed, and changes have been made that should improve the
3433 performance of the compiler by about 20%.
3434 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3435 simple-base-strings and simple-bit-vectors is improved.
3436 * optimization: the optimization of 32-bit logical and arithmetic
3437 functions introduced in version 0.8.3 on the x86 has been
3438 implemented on the mips, ppc and sparc platforms; an
3439 implementation of the same facility, but for 64-bit arithmetic,
3440 has been added for the alpha.
3441 * microoptimization: the compiler is better able to make use of the
3442 x86 LEA instruction for multiplication by constants.
3443 * bug fix: in some situations compiler did not report usage of
3444 generic arithmetic in (SPEED 3) policy.
3445 * bug 145b fix: compiler used wrong type specifier while converting
3446 MEMBER-types to numeric.
3447 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3449 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3450 index argument now works properly on the Alpha platform.
3451 * bug fix: floating point exception treatment on the Alpha platform
3453 * bug fix: FILE-POSITION works much better on string input and
3454 output streams. (thanks to Nikodemus Siivola)
3455 * bug fix: many threading/garbage collection symptoms sorted.
3456 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3457 might be pseudo-atomic.
3458 * internal change: Stopping for GC is now done with signals not
3459 ptrace. GC is now done in whichever thread wanted it, instead of
3461 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3463 * bug fix: result form in DO is not contained in the implicit
3465 * incompatible change: ICR structure is changed; the value part of
3466 CONTINUATION is now called LVAR; corresponding functions are
3467 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3468 * added type deriver for ISQRT (thanks to Robert E. Brown).
3469 * bug fix: better support for loading from the command line when an
3470 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3472 * fixed some bugs revealed by Paul Dietz' test suite:
3473 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3474 ** ROUND and FROUND now give the right answer when given very
3475 small float arguments.
3476 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3478 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3479 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3480 ** LCM should return a non-negative integer.
3481 ** PARSE-INTEGER returned the index of a terminator instead of the
3482 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3483 ** PARSE-INTEGER returned an incorrect index being applied to a
3485 ** LCM with two arguments of 0 returns 0 rather than signalling
3487 ** unsigned addition of a 32-bit constant with the high bit set no
3488 longer causes an internal compiler error on the x86.
3489 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3490 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3491 with negative last argument.
3492 ** byte specifiers with zero size and position no longer cause
3493 an error during type derivation.
3494 ** bignum multiplication on the Alpha platform now returns the
3496 * porting: The system now builds on SuSE AMD64, although it still
3497 generates a 32-bit binary.
3498 * .fasl file incompatibility: The fasl file version number has
3499 been incremented (because of the changes to internal compiler
3500 data structures referred to above).
3502 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3503 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3504 more accurately, on the Darwin kernel running on PowerPC hardware.
3505 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3506 * Compiler code deletion notes now signal a condition of type
3507 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3508 an associated MUFFLE-WARNING restart.
3509 * The compiler now performs limited argument count validation of
3510 constant format strings in FORMAT, and where appropriate in ERROR,
3511 CERROR and WARN. (thanks to Gerd Moellmann)
3512 * New ASDF-INSTALL contrib can be used for automatic download and
3513 installation of third-party Lisp code from CCLAN or other sites
3515 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3516 which forces another thread to execute a function supplied by the
3518 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3519 now accept and act upon their :ELEMENT-TYPE keyword argument.
3520 (reported by Martin Atzmueller, Edi Weitz)
3521 * bug fix: FILE-POSITION now accepts position designators up to
3522 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3523 is the greater. (thanks to Patrik Nordebo)
3524 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3525 Kalle Olavi Niemitalo)
3526 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3527 printable. (reported by Eric Marsden)
3528 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3529 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3530 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3531 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3532 (thanks to Stig Sandoe)
3533 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3534 UNREAD-CHAR now correctly includes the unread character in the
3535 target sequence. (thanks to Gerd Moellmann)
3536 * bug fix in threaded builds: the system can now be suspended and
3537 resumed by shell job control with minimal disruption.
3538 * bug fixes in times and timezones >2038AD
3539 * better handling of "where is GNU make?" problem in build scripts
3540 (thanks to Nikodemus Siivola)
3541 * new optimization: inside a named function any reference to a
3542 function with the same name is considered to be a self-reference;
3543 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3544 optimization quality.
3545 * new optimization on x86: logical functions and + now have
3546 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3547 used when the result is truncated to 32 bits.
3548 * VALUES declaration is partially enabled.
3549 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3550 * fixed some bugs revealed by Paul Dietz' test suite:
3551 ** The system now obeys the constraint imposed by
3552 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3553 types form a lattice under type intersection.
3554 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3555 ** ASSOC now ignores NIL elements in an alist.
3556 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3557 and (1+ MOST-POSITIVE-FIXNUM) answers.
3558 ** The addition of a method with invalid qualifiers to a generic
3559 function does not cause an error to be signalled immediately;
3560 a warning is signalled, and the error is generated only on
3561 calling the generic function.
3562 * changed .fasl file version number, in part to add type codes for
3563 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3564 obscure ANSI requirements
3566 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3567 * fixed bug 148: failure to inline-expand a local function left
3568 garbage, confusing the compiler.
3569 * fixed bugs 3cd: structure slot readers perform type check if the
3570 slot can have an invalid value (i.e. it is either not initialized
3571 or can be written with a less specific slot writer).
3572 * bug fix: the compiler now traps array references to elements off
3573 the end of an array; previously, the bounds checking in some
3574 circumstances could go off-by-one.
3575 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3577 * type declarations for array element types now obey the description
3578 on the CLHS page "Declaration TYPE", as per discussions on
3579 sbcl-help around 2003-05-08. This means that a declaration
3580 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3581 declaration, all references to BAR will be asserted or assumed
3582 (with THE, so dependent on compiler policy) to involve objects of
3583 type FOO. Note that no such declaration is implied in
3584 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3585 * declared types of functions from the "Conditions"
3586 chapter. (reported by Paul Dietz)
3587 * bug fix: CERROR accepts a function as its first argument.
3588 * bug fix: NTH an NTHCDR accept a bignum as index
3589 arguments. (reported by Adam Warner)
3590 * optimization: character compare routines now optimize comparing
3591 against a constant character. (reported by Gilbert Baumann)
3592 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3593 argument now works properly on the MIPS platform.
3594 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3595 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3596 declared to have a functional type, does not cause an error
3598 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3599 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3600 * fixed bug: if last continuation of a deleted block has a
3601 destination, this destination should be deleted too. (reported by
3603 * fixed a bug in the bootstrap process: the host compiler's values
3604 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3605 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3606 test case from Patrik Nordebo)
3607 * improved the ability of the disassembler on the PPC platform to
3608 provide helpful disassembly notes.
3609 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3610 the class in more cases than previously.
3611 * bug fix: FILE-POSITION now understands :START and :END for
3612 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3613 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3614 8) data. (thanks to David Lichteblau)
3615 * bug fix: it is possible to add a method to a generic function
3616 without lambda list.
3617 * bug fix: reader failed to signal END-OF-FILE inside an
3618 object representation. (reported by Nikodemus Siivola)
3619 * fixed some bugs revealed by Paul Dietz' test suite:
3620 ** LAST and [N]BUTLAST should accept a bignum.
3621 ** condition slot accessors are methods.
3622 ** (VECTOR NIL) is a subtype of STRING.
3624 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3625 * minor incompatible change: some nonsensical specialized lambda
3626 lists (used in DEFMETHOD) which were previously ignored now signal
3628 * minor incompatible change: the system is now aware of the types of
3629 variables in the COMMON-LISP package, and will signal errors for
3630 most violations of these type constraints (where previously they
3631 were silently accepted).
3632 * minor incompatible change: COMPILE-FILE now uses the freedom
3633 afforded (ANSI 3.2.2.3) to use derived function types for
3634 functions defined in the same file. This also permits the system
3635 to warn on static type mismatches and function
3636 redefinition. (Currently it does not work with high DEBUG level.)
3637 * minor incompatible change: VALUES declaration is disabled.
3638 * When issuing notes, the compiler now signals a condition of type
3639 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3640 restart for use in user handlers. It is expected that the
3641 COMPILER-NOTE condition will eventually become a condition
3642 supertype to a hierarchy of note types, which will then be
3643 handleable in a similar fashion. However, at the moment, no such
3644 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3645 but it's an implementation detail, not a classification for the
3647 * Changes in type checking closed the following bugs:
3648 ** type checking of unused values (192b, 194d, 203);
3649 ** template selection based on unsafe type assertions (192c, 236);
3650 ** type checking in branches (194bc).
3651 * A short form of VALUES type specifier has ANSI meaning (it has
3652 increased the number of situations when SBCL cannot perform type
3654 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3655 keywords or constants is permissible.
3656 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3657 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3658 classes. (thanks to Antonio Martinez)
3659 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3660 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3661 argument to a pretty-print stream. Instead, it leaves any such
3662 operations to DESCRIBE-OBJECT methods.
3663 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3665 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3666 lambda lists. (thanks to David Lichteblau)
3667 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3668 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3669 (thanks to Gerd Moellmann)
3670 * fixed bug 246: increased compilation speed of long
3671 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3673 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3675 * DEFCONSTANT now throws a condition of type
3676 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3677 constant to a non-EQL value; CONTINUE and ABORT restarts
3678 respectively change and preserve the value.
3679 * fixed bug 63: The code walker, part of the implementation of CLOS,
3680 is now better at handling symbol macros.
3681 * bug fix: There is no longer an internal implementation type named
3682 CL:LENGTH. (reported by Raymond Toy)
3683 * bug fix: In macro-like defining macros/special operators the
3684 implicit block does not enclose the lambda list.
3685 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3686 enclosed in parentheses) are not suitable as type specifiers, and
3687 their use properly signals an error now.
3688 * bug fix: An argument count mismatch for a type specifier in code
3689 being compiled no longer causes an unhandled error at compile
3690 time, but signals a compile-time warning.
3691 * fixed simple vector readable printing
3692 * bug fix: DESCRIBE takes more care over whether the class
3693 precedence list slot of a class is bound before accessing it.
3694 (reported by Markus Krummenacker)
3695 * bug fix: FORMATTER can successfully compile pretty-printer format
3696 strings which use variants of the ~* directive inside.
3697 * bug fix: SEARCH now applies its TEST predicate to the elements of
3698 the arguments in the correct order. (thanks to Wolfhard Buss)
3699 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3700 in type inference. (thanks to Robert E. Brown)
3701 * bug fix: Reading in symbols with an explicit package name of ""
3702 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3703 * fixed some bugs revealed by Paul Dietz' test suite:
3704 ** NIL is now allowed as a structure slot name.
3705 ** Arbitrary numbers, not just REALs, are allowed in certain
3706 circumstances in LOOP for-as-arithmetic clauses.
3707 ** Multiple class redefinitions before slot access no longer
3708 causes a type error.
3709 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3710 association between the name and a class.
3711 ** Generic functions with non-standard method-combination and over
3712 five methods all of which return constants no longer return NIL
3713 after the first few invocations. (thanks to Gerd Moellmann)
3714 ** CALL-NEXT-METHOD with no arguments now passes the original
3715 values of the arguments, even in the presence of assignment.
3716 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3718 ** DISASSEMBLE works with closures and funcallable instances.
3719 ** ADD-METHOD now returns the generic function, not the new method.
3720 ** FIND-METHOD signals an error if the lengths of the specializers
3721 is incompatible with the generic function, even if the ERRORP
3723 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3724 which its argument is a member.
3725 ** DEFCLASS only redefines the class named by its class-name
3726 argument if that name is the proper name of the class;
3727 otherwise, it creates a new class.
3728 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3729 of the UNBOUND-SLOT condition to the name of the slot.
3730 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3731 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3732 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3733 SLOT-MAKUNBOUND in the specified fashion.
3735 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3736 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3737 cross-compilation host. As a consequence, we can now bootstrap our
3738 way up to SBCL starting with a bare gcc toolchain and human-readable
3739 source code (first the source to CLISP, then the source to SBCL).
3740 * A contributed module containing a partial implementation of the
3741 simple-streams interface has been included. (thanks to Rudi
3743 * A contributed module implementing the RFC1321 Message Digest
3744 Algorithm, known as MD5, has been included.
3745 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3746 no longer has any effect, as the code controlled by this feature
3747 has been deleted. (As far as we know, no-one has ever built using
3748 this feature, and its semantics were confused in any case).
3749 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3750 work on conditions (as required by the ANSI specification),
3751 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3752 expected behaviour on conditions. Users should note, however,
3753 that such behaviour is not required by the ANSI specification,
3754 and so use of this behaviour may render their code unportable.
3755 * fixed some bugs revealed by Paul Dietz' test suite:
3756 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3758 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3759 ** SXHASH on condition objects no longer returns NIL.
3760 ** :ALLOCATION :CLASS slots are better treated; their values are
3761 updated on class redefinition, and initforms inherited from
3762 superclasses are applied.
3763 ** REMOVE-METHOD returns its generic function argument even when
3764 no method was removed.
3765 ** SHARED-INITIALIZE now initializes the values of the requested
3766 slots, including those with :ALLOCATION :CLASS.
3767 ** ALLOCATE-INSTANCE now works on structure classes defined via
3768 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3770 ** SLOT-EXISTS-P now works on conditions, as well as structures
3772 ** MAKE-LOAD-FORM now has the required methods on
3773 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3774 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3775 keyword, and now implements the SLOT-NAMES argument.
3776 ** methods with &OPTIONAL arguments no longer allow too many
3777 arguments to be passed in the call without error.
3778 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3779 option is consistent with the required arguments of the generic
3780 function lambda list.
3781 * bug fix: REQUIRE accepts a string designator. (Thanks to
3783 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3784 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3785 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3786 * bug fix: The compiler now checks for duplicated variables in macro
3788 * bug fix: SETQ on globals returns the correct value.
3789 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3790 not a COMPILER-ERROR (followed by some other strange error on
3791 choosing the CONTINUE restart).
3792 * bug fix: make.sh and friends are now more consistent in the way that
3793 they look for GNU "make".
3795 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3796 * experimental native threads support (on x86 Linux >=2.4 only).
3797 This is not compiled in by default: you need to add :SB-THREAD to
3798 the target features. See the "Beyond ANSI" chapter of the manual
3800 * fix for longstanding nonANSIism: The old distinction between
3801 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3802 The return value from CL:FIND-CLASS is now a CLOS class, and
3803 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3804 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3805 * An interface to the MetaObject Protocol, as described in Kiczales,
3806 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3807 MIT Press, 1991, is available from the SB-MOP package.
3808 * incompatible change: the SB-PCL package should now be considered
3809 a private implementation detail, and no longer a semi-private MOP
3811 * minor incompatible change: due to rearrangement for threads, the
3812 control stack and binding stack are now allocated at arbitrary
3813 addresses instead of being hardcoded per-port. Users affected by
3814 this probably have to be doing advanced things with shared
3815 libraries, and will know who they are.
3816 * minor incompatible change: Previously, all --eval forms used were
3817 processed with READ before any of them were processed with EVAL.
3818 Now each --eval form is processed with both READ and EVAL before
3819 the next --eval form is processed. (Thus package operations like
3820 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3821 work as the user might reasonably expect.)
3822 * minor incompatible change: *STANDARD-INPUT* is now only an
3823 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3825 * minor incompatible change: Y-OR-N-P is now character-oriented,
3826 not line oriented. Also, YES-OR-NO-P now works without errors.
3827 (thanks to Antonio Martinez)
3828 * sb-aclrepl module improvements: an integrated inspector, added
3829 repl features, and a bug fix to :trace command.
3830 * Known functions, which cannot be open coded by the backend, are
3831 considered to be able to check types of their arguments. (fixing
3832 a bug report by Nathan J. Froyd)
3833 * fixed a bug in computing method discriminating functions: It is
3834 now possible to define methods specialized on classes which have
3835 forward-referenced superclasses. (thanks to Gerd Moellmann)
3836 * fixed evaluation order in optional entries (reported by Gilbert
3838 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3839 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3840 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3841 required-by-AMOP NAME argument, as well as CLASS and
3842 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3843 * fixed bug 20: DEFMETHOD can define methods using names that are
3844 not the proper names of classes to designate class specializers.
3845 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3846 Unix file descriptors, instead of blowing up. (thanks to Antonio
3848 * Garbage collection refactoring: user-visible change is that a
3849 call to the GC function during WITHOUT-GCING will not do garbage
3850 collection until the end of the WITHOUT-GCING. If you were doing
3851 this you were probably losing anyway.
3852 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3853 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3854 TYPEP the latter but not the former.
3855 * The compiler issues a full WARNING for calls to undefined functions
3856 with names from the CL package.
3857 * MAP-INTO for a vector destination is open coded. (reported by
3858 Brian Downing on c.l.l)
3859 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3860 documentation string.
3861 * fixed some bugs revealed by Paul Dietz' test suite:
3862 ** COPY-ALIST now signals an error if its argument is a dotted
3864 ** Condition slots are now accessed more correctly in the presence
3865 of multiple initargs for a given slot.
3866 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3867 exclude restarts of the same name associated with a different
3869 ** DEFCLASS of forward-referenced classes with another
3870 forward-referenced class in the superclasses list no longer
3872 ** Condition slots are now initialized once each, not multiple
3873 times. (thanks to Gerd Moellmann)
3874 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3876 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3877 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3878 arguments contain duplicated elements.
3879 ** RESTART-CASE understands local macros.
3880 ** RESTART-CASE associates exactly its own restarts with a condition.
3881 ** ENDP in safe mode checks its argument to be of type LIST.
3882 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3883 in question is unbound.
3884 ** Optimized MAKE-INSTANCE functions no longer cause internal
3885 assertion failures in the presence of duplicate initargs.
3886 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3887 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3888 propagated even in the ctor optimized implementation of
3890 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3892 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3893 * a better implementation of SXHASH on (simple) bit vectors,
3894 measured both in execution speed and in distribution of results
3895 over the positive fixnums, has been installed. Likewise, a better
3896 implementation of EQUAL for simple bit vectors is now available.
3897 * fixed CEILING optimization for a divisor of form 2^k.
3898 * fixed bug 240 (emitting extra style warnings "using the lexical
3899 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3900 by Antonio Martinez)
3901 * fixed SXHASH, giving different results for NIL depending on type
3902 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3903 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3904 variable. (found by Rolf Wester)
3905 * SBCL does not ignore type declarations for special
3906 variables. (reported by rif on c.l.l 2003-03-05)
3907 * some bug fixes in contrib/sb-aclrepl/
3908 * fixed some bugs revealed by Paul Dietz' test suite:
3909 ** a bug in the CONS type specifier, whereby the CAR and CDR
3910 types got intertwined, has been fixed;
3911 ** the type system is now able to reason about the interaction
3912 between INTEGER and RATIO types more completely;
3913 ** APPEND, [N]REVERSE and NRECONC check that those their
3914 arguments, which must be proper lists, are really so;
3915 ** An array specialized to be unable to hold elements has been
3916 implemented, as required -- yes, really -- by ANSI;
3917 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3918 on malformed property lists;
3920 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3921 * incompatible packaging change: in line with Unix convention,
3922 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3923 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3924 * REQUIRE and PROVIDE are now optionally capable of doing something
3925 useful. See the documentation string for REQUIRE.
3926 * infrastructure for a managed SBCL contrib system: contributed
3927 modules in this release include:
3928 ** the ASDF system definition facility;
3929 ** an interface to the BSD Sockets API;
3930 ** an ACL-like convenience interface to the repl;
3931 (thanks to Kevin Rosenberg)
3932 ** an implementation of ROTATE-BYTE, with efficient implementation
3934 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3935 gives a closer approximation to the right answer than previously.
3936 (thanks to Raymond Toy)
3937 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3938 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3939 optional environment arguments, as required by ANSI.
3940 * fixed bugs in other functions taking environment objects, allowing
3941 calls with an explicit NIL environment argument to be compiled
3943 * fixed bug 228: primary return values from
3944 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3945 COMPILE or FUNCTION.
3946 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3947 :TYPE will no longer signal errors on innocuous objects.
3948 * fixed bug 231b: SETQ is better at respecting type declarations in
3949 the lexical environment.
3950 * fixed a bug in DEFCLASS: classes named by symbols with no or
3951 unprintable packages can now be defined.
3952 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3953 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3954 * fixed compiler failure related to checking types of functions.
3955 (reported by Robert E. Brown)
3956 * the compiler is now much more consistent in its error-checking
3957 treatment of bounding index arguments to sequence functions: in
3958 (SAFETY 3) code, errors will be signalled in almost all cases if
3959 invalid sequence bounding indices are passed to functions defined
3960 by ANSI to operate on sequences.
3961 * fixed a bug in the build procedure: documentation of SBCL-specific
3962 packages is now preserved and available in the final Lisp image.
3963 * lifted FDEFINITION lookup out of loops in the implementation of
3964 many list operations. (thanks to Robert E. Brown)
3965 * fixed a bug in the reader: the #n# reader macro now works for
3966 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3967 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3968 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3969 E. Brown and Tony Martinez respectively)
3970 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3971 SLOT-BOUNDP: in particular, it is now able to optimize them much
3972 better, and is now not vulnerable to having packages renamed.
3973 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3974 always called when a slot is not present in an instance. (thanks
3976 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3977 longer signal an error on unreading a character following EOF on
3978 the previous constituent stream. (thanks to Tony Martinez)
3979 * fixed some bugs revealed by Paul Dietz' test suite:
3980 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3981 not just nonnegative fixnums;
3982 ** the logical bit-array operators such as BIT-AND now accept an
3983 explicit NIL for their "opt-arg" argument (to indicate a
3984 freshly-consed result bit-array);
3985 ** ELT now signals an error on an invalid sequence index in safe
3987 ** the type system is now cleverer about negations of numeric
3988 types, and consequently understands the BIGNUM and RATIO types
3990 ** the type system is now cleverer about the interaction between
3991 INTEGER and RATIO types: while bugs still remain, many more
3992 cases are accurately computed;
3993 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3994 if it is in the last clause;
3995 ** CONSTANTLY now correctly returns a side-effect-free function in
3997 ** DECLARE is no longer treated as a special-operator; in
3998 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3999 * incremented fasl file version number due to the change in the
4000 DEFSTRUCT-SLOT-DESCRIPTION structure.
4002 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4003 * minor incompatible change: code processed by the "interpreter" or
4004 EVAL now has a compilation optimization policy of (DEBUG 2)
4005 (changed from (DEBUG 1)) to improve debuggability of interactive
4006 development, and to allow the use of the debug RETURN command in
4008 * an experimental implementation of the RETURN command for the
4009 debugger has been included. (thanks to Frederik Kuivinen)
4010 * fixed bug 62: constraints were not propagated into a loop.
4011 * fixed bug in embedded calls of SORT (reported and investigated by
4013 * fixed some bugs revealed by Paul Dietz' test suite:
4014 ** printing and reading of arrays with some dimensions having
4015 length 0 (thanks to Gerd Moellmann);
4016 ** BOA constructor with &AUX argument without a default value does
4017 not cause a type error;
4018 ** CONSTANTP now returns true for all self-evaluating objects.
4020 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4021 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4022 accessors that are related by inheritance, as specified in the
4023 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4024 Valtteri Vuorikoski)
4025 * The compiler is now able to inline functions that were defined in
4026 a complex lexical environment (e.g. inside a MACROLET).
4027 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4029 * Support for the upcoming FreeBSD-5.0 release has been included.
4030 (thanks to Dag-Erling Smorgrav)
4031 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4032 effect when it is not in a toplevel context.
4033 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4034 stand a better chance of being correct. (thanks to Gerd
4036 * fixed bug in COERCE, which now signals an error on coercing a
4037 rational to a bounded real type which excludes the expected
4039 * The compiler is now able to derive types more accurately from the
4040 COERCE and COMPILE functions.
4041 * fixed bug 223: functional binding is considered to be constant
4042 only for symbols in the CL package.
4043 * fixed bug 231: SETQ did not check the type of a variable being set
4044 (reported by Robert E. Brown)
4045 * A new optimization for MAKE-INSTANCE has been included, fixing
4046 various bugs (including relating to :ALLOCATION :CLASS slots and
4047 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4049 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4050 ** As required by ANSI, LOOP now disallows anonymous collection
4051 clauses such as COLLECT I in conjunction with aggregate boolean
4052 clauses such as THEREIS (= I 1);
4053 ** LOOP now signals an error when any variable is reused in the
4054 same loop (including the potentially useful construct analogous
4055 to WITH A = 1 WITH A = (1+ A);
4056 ** IT is only a special loop symbol within the first clause of a
4057 conditional loop clause;
4058 ** LOOP with a typed iteration variable over a hashtable now
4059 signals a type error iff it should.
4060 * fixed some other bugs revealed by Paul Dietz' test suite:
4061 ** FILE-STREAM now names the class previously known as FD-STREAM;
4062 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4063 argument) no longer signals an error;
4064 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4065 of the slot symbol, rather than using the current package
4066 ((:CONC-NAME "") continues to intern the slot's name in the
4068 * incremented fasl file version number, because of the incompatible
4069 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4070 of the new implementation of DEFINE-COMPILER-MACRO.
4072 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4073 * Support for building SBCL for MIPS platforms running in
4074 little-endian mode has now been checked in, and basic
4075 functionality on said platforms verified.
4076 * minor incompatible change: PCL now records the pathname of a file
4077 in which methods and the like are defined, rather than its
4079 * minor incompatible change: TRUENAME now considers the truename of
4080 a file naming a directory to be the pathname with :DIRECTORY
4081 component indicating that directory.
4082 * minor incompatible change: a NAMED clause in the extended form of
4083 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4084 reason for the previous behaviour is unclear.
4085 * more systematization and improvement of CLOS and MOP conformance
4086 in PCL (thanks to Gerd Moellman and Pierre Mai):
4087 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4089 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4090 primary methods with no specializers;
4091 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4093 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4094 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4095 FORWARD-REFERENCED-CLASSes; error reporting on
4096 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4098 ** SXHASH on CLOS instances now uses a slot internal to the
4099 instance to return different numbers on distinct instances,
4100 while preserving the same return value through invocations of
4102 ** DEFMETHOD signals errors when methods with longer incongruent
4103 lambda lists are added to generic functions;
4104 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4105 CLASS, as specified in AMOP;
4106 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4107 on the order returned by the primary method for classes of
4108 class STANDARD-CLASS;
4109 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4110 * fixed some bugs shown by Paul Dietz' test suite:
4111 ** DOLIST puts its body in TAGBODY;
4112 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4114 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4115 value producing form;
4116 ** if more variables are given to PROGV than values, extra
4117 variables are bound and made to have no value;
4118 ** NSUBSTITUTE on list arguments gets the right answer with
4120 ** ELT signals an error of type TYPE-ERROR when the index argument
4121 is not a valid sequence index;
4122 ** LOOP signals (at macroexpansion time) an error of type
4123 PROGRAM-ERROR when duplicate variable names are found;
4124 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4125 ** FUNCALL of special-operators now cause an error of type
4127 ** PSETQ now works as required in the presence of side-effecting
4128 symbol-macro places;
4129 ** NCONC accepts any object as its last argument;
4130 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4132 ** loop-for-as-package does not require a package to be explicitely
4134 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4136 * fixed bug 166: compiler preserves "there is a way to go"
4137 invariant when deleting code.
4138 * fixed bug 172: macro lambda lists with required arguments after
4139 &REST arguments now cause an error to be signalled. (thanks to
4141 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4142 bug. (thanks to Matthew Danish)
4143 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4145 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4146 arguments are lexically rebound. (thanks to Gerd Moellmann and
4148 * fixed bug 194: error messages are now more informative when there
4149 is no primary method applicable in a call to a generic
4150 function. (thanks to Gerd Moellmann)
4151 * fixed bug in command line argument checking (thanks to Julian
4153 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4154 (thanks to Matthew Danish)
4155 * incremented fasl file version number, because of the
4156 SXHASH-related changes in the layout of CLOS data structures
4158 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4159 * minor incompatible change: The runtime (the Unix executable named
4160 "sbcl") is now much pickier about the .core files it will load.
4161 Essentially it now requires .core files to descend from the same
4162 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4163 as the runtime does. (The intent is to prevent the crashes which
4164 can occur, and which can even be reported as mysterious failures,
4165 when people patch the sources or change the build parameters
4166 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4167 sbcl and .core files.)
4168 * fixed bug: VALUES-LIST is no longer optimized away.
4169 * fixed bug 142: The FFI conversion of C string values to Lisp
4170 string values no longer conses excessively. (thanks to Nathan
4171 Froyd porting Raymond Toy's fix to CMU CL)
4172 * began to systematize and improve MOP conformance in PCL (thanks to
4173 Nathan Froyd, Gerd Moellman and Pierre Mai):
4174 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4176 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4177 ** FINALIZE-INHERITANCE is now called on class finalization;
4178 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4179 argument precedence order.
4180 * fixed bug 202: The compiler no longer fails on functions whose
4181 derived types contradict their declared type.
4182 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4183 so it can be non-toplevel.
4184 * The fasl file version number has changed (because of the new
4185 implementation of DEFMACRO).
4186 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4187 safe code, that any length requirement by their type-specifier
4188 argument is valid. The exceptions to this are described in bug
4190 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4191 functions now check the types of their inputs as required by ANSI.
4192 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4193 introduced symbol is DECLAREd to be SPECIAL.
4194 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4195 * fixed bug: :COUNT argument to sequence functions may be negative
4196 * fixed bug: body of DO-SYMBOLS may contain declarations
4197 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4198 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4199 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4200 symbol macro only once
4201 * fixed printing of call frame when argument list is unavailable
4202 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4203 * compiler no longer signals WARNING on unknown keyword
4206 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4207 * A beta-quality port to the mips architecture running Linux,
4208 based on the old CMUCL backend, has been made. It has been tested
4209 on a big-endian kernel, and works sufficiently well to be able to
4210 rebuild itself; it has not been tested in little-endian mode.
4211 * fixed an inconsistency between gencgc.c and purify.c which made
4212 dumping/loading .core files unreliable
4213 * fixed bug 120a: The compiler now deals correctly with IFs where
4214 the consequent is the same as the alternative, instead of
4215 misderiving the return type. (thanks to Alexey Dejneka)
4216 * fixed bug 113: Logical pathnames are now dumpable (the logical
4217 host is resolved at load-time, throwing an error if it is not
4219 * fixed bug 174: FORMAT's error message is slightly clearer when a
4220 non-printing character is used in a format directive.
4221 * fixed several bugs in compiler checking of type declarations, i.e.
4222 violations of the Python "declarations are assertions" principle
4223 (thanks to Alexey Dejneka)
4224 * fixed several bugs in PCL's error checking (thanks to Gerd
4226 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4228 * fixed bug in compilation of functions as first class values
4229 (thanks to Antonio Martinez-Shotton)
4230 * The compiler's handling TYPE-ERRORs which it can prove will
4231 inevitably happen at runtime has been cleaned up and corrected
4232 in several ways. (thanks to Alexey Dejneka)
4233 * improved argument type checking for various basic arithmetic
4234 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4235 lavished on them in the past that they can be compiled in many
4236 ways in different special cases
4237 * fixed bug 181: compiler checks validity of user supplied type
4239 * cleaned up code flushing in optimization: Function calls which
4240 should signal errors for safety purposes (e.g. which ANSI says
4241 should signal errors when their arguments are of incorrect type)
4242 are no longer optimized away.
4243 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4244 * incremented fasl file version number, because changes in the
4245 implementation of sequence functions like COERCE caused
4246 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4247 in old inline expansions) to become undefined. (Actually these
4248 changes were later undone, so we might very well be binary
4249 compatible with 0.7.7 after all, but leaving the version number
4250 incremented seemed like the simplest and most conservative
4253 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4254 * An alpha-quality port to the parisc architecture running Linux,
4255 based on the old CMUCL backend, has been made. This, even more so
4256 than the other backends, should be considered still a work in
4257 progress; known problems include that the Linux kernel in 64-bit
4258 mode does not propagate the correct sigcontext structure to
4259 userspace, and consequently SBCL on a parisc64 kernel will not
4261 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4262 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4263 suppressed.) Also now that the compiler is looking at declarations
4264 in the environment, it checks optimization declarations as well,
4265 and suppresses inlining when (> DEBUG SPEED).
4266 * More fixes have been made to treatment of floating point exception
4267 treatment and other Unix signals. In particular, floating point
4268 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4269 * The detection and handling of control stack exhaustion (infinite
4270 or very deeply nested recursion) has changed. Stack exhaustion
4271 detection is now done by write-protecting pages at the OS level
4272 and applies at all optimization settings; when found, a
4273 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4274 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4275 be caught using IGNORE-ERRORS.
4276 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4277 correctly when transforming recursive LABELS functions to LETs.
4278 (thanks to Alexey Dejneka)
4279 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4280 that are names of constants or global variables.
4281 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4282 alien routines with docstrings.
4283 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4284 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4286 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4287 (thanks to Eric Marsden)
4288 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4289 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4290 * Bug fix: The SPARC backend can now compile functions involving
4291 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4292 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4293 to a functions expecting a PATHNAME-DESIGNATOR.
4294 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4295 lambda lists. (thanks to Alexey Dejneka)
4296 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4297 functions so that EXPORTing the name of the class doesn't cause
4298 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4299 to Antonio Martinez for reporting this)
4300 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4301 constant-folded without failing an assertion. (thanks to Einar
4302 Floystad Dorum for reporting this)
4303 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4304 types are now dealt with more correctly by the compiler.
4305 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4306 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4307 behaviour analogous to RENAME-FILE. This puts its behaviour more
4308 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4310 * The fasl file version number has changed again. (because of the
4311 bug fix involving the names of PCL MAKE-INSTANCE functions)
4313 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4314 * bug fix: Floating point exceptions are treated much more
4315 consistently on the x86/Linux and PPC/Linux platforms.
4316 * Array initialization with :INITIAL-ELEMENT is now much faster for
4317 cases when the compiler cannot open code the array creation, but
4318 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4319 array accesses have also seen a speed increase.
4320 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4321 specified. (This was at the root of some bad interactions between
4322 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4324 * bug fix: Internal error arguments for undefined functions are now
4325 computed correctly on the PPC/Linux platform.
4326 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4327 Raymond Toy's patch for CMU CL)
4328 * Support for the Solaris 9 operating environment has been included
4329 (thanks to Daniel Merritt)
4330 * A very ugly but hopefully complete draft of the missing FFI chapter
4331 of the manual has been created by reformatting the corresponding
4332 CMU CL manual chapter into (currently very ugly and incoherent)
4333 DocBook and bringing it up to date for SBCL behavior. Thus, the
4334 manual is now essentially complete, at least by my extreme
4335 once-and-only-once standards, whereby it's acceptable to refer to
4336 the doc strings of SB-EXT functions as the primary documentation.
4337 * The fasl file version number has changed again, due to cleanup of
4338 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4339 is no longer a static symbol.)
4341 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4342 * SBCL now builds with OpenMCL (version 0.12) as the
4343 cross-compilation host; also, more progress has been made toward
4344 bootstrapping under CLISP.
4345 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4347 * bug 158 fixed: The compiler can now deal with integer loop
4348 increments different from 1; fixing this turned out also to fix
4350 * bug 169 fixed: no more bogus warnings about using lexical bindings
4351 despite the presence of perfectly good SPECIAL declarations (thanks
4352 to David Lichteblau)
4353 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4354 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4355 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4357 * bug 180 fixed: Method combination specifications no longer ignore
4358 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4359 * bug fix: Structure type predicate functions now check their argument
4360 count as they should.
4361 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4362 correctly. (thanks to Pierre Mai)
4363 * minor incompatible change: The --noprogrammer option is deprecated
4364 in favor of the new --disable-debugger option, which is very similar.
4365 (The major difference is that it takes effect at a slightly different
4366 time at startup, causing handling of errors in --sysinit and
4367 --userinit files will be affected differently.) The
4368 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4369 been added to allow this functionality to be controlled from ordinary
4370 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4371 maintainers, who might want to run non-interactive scripts to
4372 build SBCL cores which will later be used interactively.)
4373 * minor incompatible change: The LOAD function no longer, when given
4374 a wild pathname to load, loads all files matching that pathname.
4375 Instead, an error of type FILE-ERROR is signalled.
4377 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4378 * bug 147 fixed: The compiler preserves its block link/count
4379 invariants more correctly now so that it doesn't crash. (thanks
4381 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4383 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4384 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4385 no-value-for-this-character as NIL (as specified by ANSI).
4386 * HOST-NAMESTRING on physical pathnames now returns a string that is
4387 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4388 (thanks to Christophe Rhodes)
4389 * The Alpha port handles icache flushing more correctly. (thanks to
4391 * More progress has been made toward bootstrapping under CLISP. (thanks
4392 to Christophe Rhodes)
4393 * The fasl file format has changed again, because dynamic loading
4394 on OpenBSD (which has non-ELF object files) motivated some cleanups
4395 in the way that foreign symbols are transformed and passed around.
4396 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4397 is no longer treated as whitespace by the reader, but instead as
4398 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4399 |A
\7fB|, instead of A as it used to.
4401 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4402 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4403 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4404 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4405 this as of 0.7.1.45, but was left out of the previous news section
4406 (thanks to Dan Barlow)
4407 * SBCL now runs on the Solaris operating system on SPARC architectures
4408 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4409 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4410 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4411 * SPARC backend cleanups, allowing builds of cores optimized for V8
4412 and V9 SPARCS, and also emission of code targeted to a particular
4413 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4415 * SBCL is closer to bootstrapping under CLISP, thanks to various
4416 fixes by Christophe Rhodes.
4417 * The fasl file format has changed again, to allow the compiler's
4418 INFO database to support symbol macros.
4419 * The user manual (in doc/) is formatted into HTML more nicely.
4420 (thanks to coreythomas)
4421 * The system is smarter about SUBTYPEP relationships, especially
4422 those involving NOT types (including types such as ATOM which are
4423 represented internally using NOT types). Thus SUBTYPEP is less
4424 likely to return (VALUES NIL NIL) in general, and in particular
4425 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4427 * The fasl file format has changed again, because the internal
4428 representation of types now includes a new slot to support the new
4429 SUBTYPEP-of-NOT-types logic.
4430 * (not a change in the main branch of SBCL, but a related prototype
4431 which can hopefully be merged into the main branch of SBCL in the
4432 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4433 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4434 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4435 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4436 dispatch macro character combinations. (thanks to Alexey Dejneka)
4437 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4438 that superficially look like logical namestrings correctly.
4439 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4441 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4442 default supplied-p) for &optional and &key arguments. (thanks to
4445 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4446 * incompatible change: The compiler is now less aggressive about
4447 tail call optimization, doing it only when (> SPACE DEBUG) or
4448 (> SPEED DEBUG). (This is an incompatible change because there are
4449 programs which relied on the old CMU-CL-style behavior to optimize
4450 away their unbounded recursion which will now die of stack overflow.)
4451 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4452 for non-GENCGC systems has been increased to 20M (since that
4453 seems much closer to the likely performance optimum for modern
4454 systems than the old 4M value was)
4455 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4456 and *DEBUG-PRINT-LEVEL*
4457 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4458 of CMU CL's support for SPARC, and various endianness and other
4459 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4460 * new syntactic sugar for the Unix command line: --load foo.bar is now
4461 an alternate notation for --eval '(load "foo.bar")'.
4463 ** The system now detects stack overflow and handles it gracefully,
4464 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4465 optimization settings. (This is a good thing in general, and
4466 its introduction in this version should be particularly timely
4467 for anyone whose code fails because of suppression of tail
4469 ** The system now hunts for the C variable "environ" in a more
4470 devious way, to avoid segfaults when the C library version differs
4471 between compile time and run time. (thanks to Christophe Rhodes)
4472 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4473 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4474 ** The compiler no longer issues bogus style warnings for undefined
4475 classes in the same source file as the DEFCLASSes which defined
4476 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4478 ** fixes in CONDITION class precedence list for undefined function
4479 errors (thanks to Alexei Dejneka)
4480 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4481 correctly. (thanks to Dan Barlow)
4482 ** portability fixes aiming at bootstrapping under CLISP (thanks
4483 to Dave McDonald and Christophe Rhodes)
4484 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4485 ** fixes in type translation and and type inference (thanks to
4487 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4488 ** various fixes in the new ports (thanks to Dan Barlow)
4489 * several changes related to debugging:
4490 ** suppression of tail recursion, as noted above
4491 ** stack overflow detection, as noted above
4492 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4493 is now the default. (For some time encapsulation has been more
4494 reliable than the breakpoint-based :ENCAPSULATE NIL
4495 implementation, at least on X86 systems; and I just noticed that
4496 encapsulation also seems closer to the spirit of the ANSI
4499 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4501 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4502 up properly again. (There was a packaging bug in 0.7.0 which
4503 left their definitions in SB-SYS::LOAD-FOREIGN and
4504 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4505 vital for most things which interface to C-level interfaces,
4506 like extensions working with sockets or databases or
4507 Perl-compatible regexes or whatever, and the need to fix
4508 this bug is the main reason that 0.7.1 was released so
4510 ** DEFGENERIC is now choosier about the methods it redefines, so that
4511 reLOADing a previously-LOADed file containing DEFGENERICs does
4512 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4513 cycle now works as it should. (thanks to Alexey Dejneka)
4514 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4515 fixed by Christophe Rhodes. (He actually submitted this patch
4516 months ago, and I delayed until after 0.7.0.)
4517 ** Bug 111 (internal compiler confusion about runtime checks on
4518 FUNCTION types) was fixed by Alexey Dejneka.
4519 * Some internal cleanups (getting rid of variables which aren't
4520 needed now that the byte interpreter is gone) caused the fasl
4521 file format number to change again.
4523 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4524 * major incompatible change: The default fasl file extension, i.e. the
4525 default extension for files produced by COMPILE-FILE, has changed
4526 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4528 ** There are many changes in the implementation of the compiler.
4529 SBCL is now essentially a compiler-only implementation of ANSI
4530 Common Lisp. EVAL still "interprets" a few special cases, but
4531 almost all the interesting cases are handled by creating
4532 a LAMBDA expression, calling COMPILE on it, then calling
4533 FUNCALL on the result.
4534 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4535 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4536 Since the code is newer, there might still be some new bugs
4537 (though not as many as before Martin Atzmueller's fixes:-). But
4538 the new code is substantially simpler and clearer, and hopefully
4539 any remaining bugs will be simpler, less fundamental, and more
4540 fixable then the bugs in the old code.
4541 ** The revised compiler is still a little unsteady on its feet.
4543 *** The debugging information it produces (particularly the names
4544 of FUNCTION objects) is sometimes much less useful than what
4545 the old compiler produced.
4546 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4547 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4548 within a MACROLET) has been temporarily weakened.
4549 ** There are new compiler optimizations for various functions:
4550 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4551 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4552 *** the math functions TRUNCATE, FLOOR, and CEILING
4553 *** the function-of-all-trades COERCE
4554 Mostly these should be transparent, but there's one
4555 potentially-annoying problem (bug 117): when the compiler
4556 inline-expands a function and does type analysis on the result,
4557 it can create control paths which have type mismatches, and
4558 when it can't prove that those control paths aren't taken,
4559 it will issue WARNINGs about the type mismatches. This is
4560 a particular problem in practice for the new sequence functions.
4561 It's not clear how this should be fixed, and for now, a
4562 workaround is given in the entry for 117 in the BUGS file.
4563 ** (Because of the interaction between the two previous items --
4564 occasional inlining problems and new inline expansions -- some
4565 of the new sequence function optimizations won't really kick in
4566 completely until debugging information, and then inlining, are
4567 straightened out in some future version.)
4568 * minor incompatible changes:
4569 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4570 DIRECTORY no longer implicitly promotes NIL slots of its
4571 pathname argument to :WILD. In particular, when you ask for the
4572 contents of a directory (which you used to be able to do without
4573 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4574 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4575 ** changes in behavior that ANSI explicitly defines to be
4576 implementation dependent:
4577 *** The new compiler-only implementation still conforms with ANSI,
4578 but acts a little different than before. Besides the obvious
4579 changes in performance tradeoffs (that the cost per form passed
4580 to EVAL has gone up, and the cost per form executed by EVAL
4581 has gone down), the behavior of the system changes a little
4582 because there are no longer any interpreted function objects.
4583 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4584 e.g. doing COMPILE on the output of interactive DEFUN is
4586 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4588 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4589 has changed from (:CL) to NIL.
4590 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4591 e.g. #\Newline and #\Tab, don't have names specified in the
4592 ANSI Common Lisp standard, is now based on their ASCII symbolic
4593 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4594 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4595 are no longer used for output.
4596 ** changes in internal implementation constants:
4597 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4598 4 million. (If your application spends a lot of time GCing and
4599 you have a lot of RAM, you might want to experiment with
4600 increasing it even more.)
4601 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4602 However, almost all old code should still continue to work without
4603 immediate update, as SB-C-CALL is now a (deprecated) nickname
4605 ** Old operator names in the style DEF-FOO are now deprecated in
4606 favor of new corresponding names DEFINE-FOO, for consistency with
4607 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4608 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4609 internal symbols, but a few supported extensions like
4610 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4611 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4612 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4613 etc. as you get deeper into recursive calls to the debugger
4614 command loop, instead of the old "5]", "5]]", "5]]]"
4615 sequence. (I was motivated to do this when squabbles between
4616 ILISP and SBCL left me very deeply nested in the debugger. In the
4617 short term, this change will probably provoke more ILISP/SBCL
4618 squabbles, but hopefully it will be an improvement in the long run.)
4619 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4620 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4621 function, since declaiming return types involving aliens is
4622 (1) annoyingly messy to do by hand and (2) vital to efficient
4623 compilation of code which calls such functions.
4624 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4625 longer reexported by the SB-EXT package. They're solely useful
4626 for alien code, so it seems more logical that you should get
4627 them from the SB-ALIEN package, not in SB-EXT.
4628 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4629 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4630 features. Instead, the code that they used to control is always
4631 built into the system.
4632 * many other bug fixes
4633 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4634 advantage of the new EVAL-WHEN stuff and to clean them up in
4635 general, and they are now more ANSI-compliant in a number of
4636 ways. Martin Atzmueller is responsible for a lot of this.
4637 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4639 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4640 *** correct ERROR type for various file operations
4641 *** some fixes for Lisp streams
4642 *** DEFMETHOD syntax checking
4643 *** changing old weird representation of debug information as
4644 strings (which, among their other deficiencies, don't transform
4645 correctly when you rename packages, and don't change their
4646 print representation when you change things like *PACKAGE*
4647 and *PRINT-LENGTH*) to symbols and lists of symbols
4648 He also made several improvements and fixed several bugs in DESCRIBE.
4649 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4651 *** misbehavior of WRITE-STRING/WRITE-LINE
4652 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4653 and several other LOOP problems as well
4654 *** DIRECTORY when similar filenames are present
4655 *** DEFGENERIC with :METHOD options
4656 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4657 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4658 *** argument ordering in FIND with :TEST option
4659 *** mishandled package designator argument in APROPOS-LIST
4660 *** various problems in the backquote readmacro
4661 *** a bug in APROPOS
4662 *** probably some others that I'm not describing very well here,
4663 since the CVS log documents them by reference to sbcl-devel
4664 messages, and the SourceForge archives aren't working well.:-(
4665 ** Dan Barlow improved the Alpha port (and is making progress on the
4666 PPC port, for those of you who think different).
4667 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4668 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4669 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4670 continues to work on the SPARC port (for those of you in a position
4671 to look down upon our little PC-compatible boxes from a great height).
4672 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4673 and uses that copy, rather than the current dynamic value, when
4674 it's trying to decide whether to truncate output. Thus e.g.
4675 (let ((*print-lines* 50))
4676 (pprint-logical-block (stream nil)
4678 (let ((*print-lines* 8))
4679 (print (aref possiblybigthings i) stream)))))
4680 should now truncate the logical block only at 50 lines, instead of
4681 often truncating it at 8 lines, as it did before.
4682 * The doc/cmucl/ directory, containing old CMU CL documentation
4683 from the time of the fork, is no longer part of the base system.
4684 SourceForge has shut down its anonymous FTP service, and with it
4685 my original plan for distributing the old CMU CL documentation
4686 there. For now, if you need these files you can download an old
4687 SBCL source release and extract them from it.
4688 * The fasl file version number changed again, for dozens of reasons,
4689 some of which are apparent above.
4691 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4692 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4693 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4694 patch, so that DISASSEMBLE output is much nicer.
4695 * The code in the SB-PROFILE package now seems reasonably stable.
4696 I still haven't decided what the final interface should look like
4697 (I'd like PROFILE to interact cleanly with TRACE, since both
4698 facilities use function encapsulation) but if you have a need
4699 for profiling now, you can probably use it successfully with
4700 the current CMU-CL-style interface.
4701 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4702 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4703 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4704 link now returns the dangling link itself, and for similar
4705 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4706 link itself. (In these cases the old code signalled an error and
4707 looped endlessly, respectively.) Thus, DIRECTORY now works even
4708 in the presence of dangling and cyclic symbolic links.
4709 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4710 is now a supported extension again, since the consensus on
4711 sbcl-devel was that it can be useful for ordinary development
4712 work, not just for debugging SBCL itself.
4713 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4714 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4715 that currently-defined functions might be redefined later with
4716 different return types.
4717 * Hash tables can be printed readably, as inspired by CMU CL code
4718 of Eric Marsden and SBCL code of Martin Atzmueller.
4719 * better error handling in CLOS method combination, thanks to
4720 Martin Atzmueller porting Pierre Mai's CMU CL patches
4721 * more overflow fixes for >16Mbyte I/O buffers
4722 * A bug in READ has been fixed, so that now a single Ctrl-D
4723 character suffices to cause end-of-file on character streams.
4724 In particular, now you only need one Ctrl-D at the command
4725 line (not two) to exit SBCL.
4726 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4728 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4729 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4730 does the right thing.
4731 * The implementation of some type tests, especially for CONDITION
4732 types, is now tidier and maybe faster, due to CMU CL code
4733 originally by Douglas Crosher, ported by Martin Atzmueller.
4734 * Some math functions have been fixed, and there are new
4735 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4736 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4737 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4738 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4739 likely to be built into sbcl-0.7.0.)
4740 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4741 CL, ported to SBCL by Martin Atzmueller)
4742 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4743 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4744 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4745 * There's a new slam.sh hack to shorten the edit/compile/debug
4746 cycle for low-level changes to SBCL itself, and a new
4747 :SB-AFTER-XC-CORE target feature to control the generation of
4748 the after-xc.core file needed by slam.sh.
4749 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4750 COMPILE-FILE is no longer supported, so that now every function
4751 gets an entry point, so that block compilation looks a little
4752 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4753 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4754 returns the number of bytes consed since the system started,
4755 rather than the number consed since the first time the function
4756 was called. (The new definition parallels ANSI functions like
4757 CL:GET-INTERNAL-RUN-TIME.)
4758 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4759 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4760 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4761 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4762 * Fasl file version numbers are now independent of the target CPU,
4763 since historically most system changes which required version
4764 number changes have affected all CPUs equally. Similarly,
4765 the byte fasl file version is now equal to the ordinary
4768 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4769 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4770 is no longer recognized. I apologize for this, because it was
4771 listed in SB-EXT as a supported extension, but I found that
4772 its existing behavior was poorly specified, as well as incorrectly
4773 specified, and it looked like too much of a mess to straighten it
4774 out. I have enough on my hands trying to get ANSI stuff to work..
4775 * many patches ported from CMU CL by Martin Atzmueller, with
4776 half a dozen bug fixes in pretty-printing and the debugger, and
4777 half a dozen others elsewhere
4778 * fixed bug 13: Floating point infinities are now supported again.
4779 They might still be a little bit flaky, but thanks to bug reports
4780 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4781 as flaky as they were.
4782 * The --noprogrammer command line option is now supported. (Its
4783 behavior is slightly different in detail from what the old man
4784 page claimed it would do, but it's still appropriate under the
4785 same circumstances that the man page talks about.)
4786 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4787 are now supported, and enabled by default. Thus, the compiler can
4788 handle many floating point and complex operations much less
4789 inefficiently. (Thus e.g. you can implement a complex FFT
4791 * The compiler now detects type mismatches between DECLAIM FTYPE
4792 and DEFUN better, and implements CHECK-TYPE more correctly, and
4793 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4794 thanks to patches from Martin Atzmueller.
4795 * various fixes to make the cross-compiler more portable to
4796 ANSI-conforming-but-different cross-compilation hosts (notably
4797 Lispworks for Windows, following bug reports from Arthur Lemmens)
4798 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4799 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4800 fixed thanks to Pierre Mai's CMU CL patches.
4801 * fixes to keep the system from overflowing internal counters when
4802 it tries to use i/o buffers larger than 16M bytes
4803 * fixed bug 45a: Various internal functions required to support
4804 complex special functions have been merged from CMU CL sources.
4805 (When I was first setting up SBCL, I misunderstood a compile-time
4806 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4807 * improved support for type intersection and union, fixing bug 12
4808 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4809 more obscure bugs as well
4810 * some steps toward byte-compiling non-performance-critical
4811 parts of the system, courtesy of patches from Martin Atzmueller
4812 * Christophe Rhodes has made some debian packages of sbcl at
4813 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4814 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4815 stable, but are nonetheless usable. When he's ready, I'd be happy
4816 to add them to the SourceForge "File Releases" section. (And if
4817 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4818 * new fasl file format version number (because of changes in
4819 internal representation of (OR ..) types to accommodate the new
4820 support for (AND ..) types, among other things)
4822 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4823 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4825 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4827 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4828 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4829 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4830 * DESCRIBE now works on structure objects again.
4831 * Most function call argument type mismatches are now handled as
4832 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4833 know whether the function will be redefined before the call is
4834 executed. (The compiler could flag local calls with full WARNINGs,
4835 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4836 it doesn't keep track of enough information to know whether calls
4837 are local in this sense.)
4838 * Compiler output is now more verbose, with messages truncated
4839 later than before. (There should be some supported way for users
4840 to override the default verbosity, but I haven't decided how to
4841 provide it yet, so this behavior is still controlled by the internal
4842 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4843 src/compiler/ir1util.lisp.)
4844 * Fasl file format version numbers have increased again, because
4845 support for the Gray streams extension changes the layout of the
4846 system's STREAM objects.
4847 * The Gray subclassable streams extension now works, thanks to a
4848 patch from Martin Atzmueller.
4849 * The full LOAD-FOREIGN extension (not just the primitive
4850 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4851 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4852 but like most other programs, it defaults to copying the Unix
4853 environment from the original process instead of starting the
4854 new process in an empty environment.
4855 * Extensions which manipulate the Unix environment now support
4856 an :ENVIRONMENT keyword option which doesn't smash case or
4857 do other bad things. The CMU-CL-style :ENV option is retained
4858 for porting convenience.
4859 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4860 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4862 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4864 * A patch from Martin Atzmueller seems to have solved the SIGINT
4865 problem, and as far as we know, signal-handling now works cleanly.
4866 (If you find any new bugs, please report them!)
4867 * The system no longer defaults Lisp source file names to types
4868 ".l", ".cl", or ".lsp", but only to ".lisp".
4869 * The compiler no longer uses special default file extensions for
4870 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4871 seems to expect a single default extension for all compiled code,
4872 and there's no compelling reason to try to stretch the standard
4873 to allow two different extensions.) Instead, byte-compiled files
4874 default to the same extension as native-compiled files.
4875 * Fasl file format version numbers have increased again, because
4876 a rearrangement of internal implementation packages made some
4877 dumped symbols in old fasl files unreadable in new cores.
4878 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4879 many fewer weird special cases.
4880 * Bug #17 (differing COMPILE-FILE behavior between logical and
4881 physical pathnames) has been fixed, and some related misbehavior too,
4882 thanks to a patch from Martin Atzmueller.
4883 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4884 by Tim Moore, ported to SBCL by Martin Atzmueller.
4885 * Martin Atzmueller fixed several filesystem-related problems,
4886 including bug #36, in part by porting CMU CL patches, which were
4887 written in part by Paul Werkowski.
4888 * More compiler warnings in src/runtime/ are gone, thanks to
4889 more patches from Martin Atzmueller.
4890 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4893 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4895 * DESCRIBE now works on CONDITION objects.
4896 * The debugger now handles errors which arise when trying to print
4897 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4899 * The build system now uses an additional file, customize-target-features.lisp,
4900 to allow local modifications to the target *FEATURES* list. (The point of
4901 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4902 debugging features enabled, without having to worry about propagating it
4903 into everyone's system when I do a "cvs update".) When no
4904 customize-target-features.lisp file exists, the target *FEATURES* list
4905 should be constructed the same way as before.
4906 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4907 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4908 warn about behavior which is undefined under the ANSI spec. Note: This
4909 is specified by ANSI, but it's not very popular with programmers.
4910 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4911 macro in the SBCL sources for an example of a workaround which you
4912 might use to make portable ANSI-standard code which does what you want.
4913 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4914 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4915 when it's used as a non-toplevel form. (This is required in order
4916 to implement the DEFCONSTANT-EQX macro.)
4917 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4918 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4919 funny things, due to bugs in the implementation of EVAL-WHEN
4920 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4921 non-toplevel DEFCONSTANTs are uncommon.)
4922 * The core file version number and fasl file version number have been
4923 incremented, because the old noncompliant DEFCONSTANT behavior involved
4924 calling functions which no longer exist, and because I also took the
4925 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4926 * fixed bug 1 (error handling before read-eval-print loop starts), and
4927 redid debugger restarts and related debugger commands somewhat while
4929 ** The QUIT debugger command is gone, since it did something
4930 rather different than the SB-EXT:QUIT command, and since it never
4931 worked properly outside the main toplevel read/eval/print loop.
4932 Invoking the new TOPLEVEL restart provides the same functionality.
4933 ** The GO debugger command is also gone, since you can just invoke
4934 the CONTINUE restart directly instead.
4935 ** The TOP debugger command is also gone, since it's redundant with the
4936 FRAME 0 command, and since it interfered with abbreviations for the
4938 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4940 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4941 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4942 some porting work by Martin Atzmueller)
4943 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4944 with a boatload of patches which he ported to SBCL. Now that those
4946 ** The system tries to make sure that its low-priority messages
4947 are prefixed by semicolons, to help people who like to use
4948 syntax highlighting in their ILISP buffer. (This patch
4949 was originally due to Raymond Toy.)
4950 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4951 patches originally written by Raymond Toy.
4952 ** The compiler understands coercion between single-value and
4953 multiple-VALUES type expressions better, getting rid of some very
4954 weird behavior, thanks to patches originally by Robert MacLachlan
4955 and Douglas Crosher.
4956 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4957 lambda lists, thanks to a patch originally by Pierre Mai.
4958 ** The system no longer bogusly warns about "abbreviated type
4960 ** The compiler gets less confused by inlining and RETURN-FROM,
4961 thanks to some patches originally by Tim Moore.
4962 ** The system no longer hangs when dumping circular lists to fasl
4963 files, thanks to a patch originally from Douglas Crosher.
4964 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4965 undefined function error.
4966 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4967 in structure slot initforms) for now, documented workaround instead:-|
4968 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4969 * fixed bug 5: added stubs for various Gray stream functions called
4970 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4971 installed, at least appropriate type errors are generated
4972 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4973 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4974 consistently in DEFMETHOD forms.
4975 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4976 it doesn't seem to affect SBCL after all
4977 * The C runtime system now builds with better optimization and many
4978 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4980 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4982 * The system is now under CVS at SourceForge (instead of the
4983 CVS repository on my home machine).
4984 * The new signal handling code has been tweaked to treat register
4985 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4986 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4987 downstream, e.g. in debug-int.lisp, has implicit dependencies
4988 on the unsignedness of integer representation of machine words,
4989 and that caused the system to bomb out with infinite regress
4990 when trying to recover from type errors involving signed values,
4991 e.g. (BUTLAST '(1 2 3) -1).)
4992 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4993 (This was one of the bugs Peter Van Eynde reported back in July.)
4994 * The system now uses code inspired by Colin Walters' O(N)
4995 implementation of MAP (from the cmucl-imp@cons.org mailing
4996 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4997 inline the MAP operation, and there is more than one
4998 sequence argument to the MAP call (so that it can't just
4999 do ETYPECASE once and for all based on the type of the
5000 single sequence argument). (The old non-inline implementation
5001 of the general M-argument sequence-of-length-N case required
5002 O(M*N*N) time when any of the sequence arguments were LISTs.)
5003 * The QUIT :UNIX-CODE keyword argument has been renamed to
5004 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5005 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5007 * Raymond Wiker's patches to port dynamic loading from Linux to
5008 FreeBSD have been added.
5009 * The BUGS file is now more nearly up to date, thanks in large part
5010 to Martin Atzmueller's review of it.
5011 * The debugger now flushes standard output streams before it begins
5012 its output ("debugger invoked" and so forth).
5013 * The core version number and fasl file version number have both
5014 been incremented, because of incompatible changes in the layout
5016 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5017 used to not be called for a saved Lisp image.)
5018 * Martin Atzmueller's version of a patch to fix a compiler crash,
5019 as posted on sbcl-devel 13 September 2000, has been installed.
5020 * Instead of installing Martin Atzmueller's patch for the
5021 compiler transform for SUBSEQ, I deleted the compiler transform,
5022 and transforms for some similar consing operations.
5023 * A bug in signal handling which kept TRACE from working on OpenBSD
5025 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5026 away by constant folding
5027 * The system now defines its address space constants in one place
5028 (in the Lisp sources), and propagates them automatically elsewhere
5029 (through GENESIS and the sbcl.h file). Therefore, patching the
5030 address map is less unnecessarily tedious and error-prone. The
5031 Lisp names of address space constants have also been systematized.
5032 * CVS tags like dollar-Header-dollar have been removed from
5033 the sources, because they have never saved me trouble and
5034 they've been source of trouble working with patches and other
5035 diff-related operations.
5036 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5037 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5039 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5041 * The system has been ported to OpenBSD.
5042 * The system now compiles with a simple "sh make.sh" on the systems
5043 that it's supported on. I.e., now you no longer need to tweak
5044 text in the target-features.lisp-expr and symlinks in src/runtime/
5045 by hand, the make.sh takes care of it for you.
5046 * The system is no longer so grossly inefficient when compiling code
5047 involving vectors implemented as general (not simple) vectors (VECTOR T),
5048 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5049 which dares to use the various sequence functions on non-simple
5050 vectors, takes less of a performance hit.
5051 * There is now a primitive type predicate VECTOR-T-P
5052 to test for the (VECTOR T) type, so that e.g.
5053 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5054 can now be compiled with some semblance of efficiency. (The old code
5055 turned the type declaration into a full call to %TYPEP at runtime!)
5056 * AREF on (VECTOR T) is still not fast, since it's still compiled
5057 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5058 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5059 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5060 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5061 without making full calls to SUBTYPEP at runtime.
5062 (Some analogous efficiency issues for non-simple vectors specialized to
5063 element types other than T, or for non-simple multidimensional arrays,
5064 have not been addressed. They could almost certainly be handled the
5065 same way if anyone is motivated to do so.)
5066 * The changes in array handling break binary compatibility, so
5067 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5068 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5069 instead of (VALUES T T).
5070 * By following the instructions that Dan Barlow posted to sbcl-devel
5071 on 2 July 2000, I was able to enable primitive dynamic object
5072 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5073 functionality is not implemented (since it calls ld to resolve
5074 library references automatically, requiring RUN-PROGRAM for its
5075 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5076 try to resolve library references) is now supported.
5077 * The system now flushes the standard output streams when it terminates,
5078 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5079 them at several other probably-convenient times, e.g. in each pass of
5080 the toplevel read-eval-print loop, and after evaluating a form given
5081 as an "--eval" command-line option. (These changes were motivated by a
5082 discussion of stream flushing issues on cmucl-imp in August 2000.)
5083 * The source transform for TYPEP of array types no longer assumes
5084 that an array whose element type is a not-yet-defined type
5085 is implemented as an array of T, but instead punts, so that the
5086 type will be interpreted at runtime.
5087 * There is now some support for cross-compiling in make.sh: each of
5088 the phases of make.sh has its own script. (This should be transparent
5089 to people doing ordinary, non-cross-compile builds.)
5090 * Since my laptop doesn't have hundreds of megabytes of memory like
5091 my desktop machine, I became more motivated to do some items on
5092 my to-do list in order to reduce the size of the system a little:
5093 ** Arrange for various needed-only-at-cold-init things to be
5094 uninterned after cold init. To support this, those things have
5095 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5096 symbols with such names are now uninterned after cold init).
5097 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5098 fasl files for cold load.
5099 ** Remove the old compiler structure pooling code (which used to
5100 be conditional on the target feature :SB-ALLOC) completely.
5101 ** Redo the representation of some data in cold init to be more compact.
5102 (I also looked into supporting byte compiled code at bootstrap time,
5103 which would probably reduce the size of the system a lot, but that
5104 looked too complicated, so I punted for now.)
5105 * The maximum signal nesting depth in the src/runtime/ support code has
5106 been reduced from 4096 to 256. (I don't know any reason for the very
5107 large old value. If the new smaller value turns out to break something,
5108 I'll probably just bump it back up.)
5109 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5111 * Many, many bugs reported by Peter Van Eynde have been added to
5112 the BUGS list; some have even been fixed.
5113 * While enabling dynamic object file loading, I tried to make the
5114 code easier to understand, renaming various functions and variables
5115 with less ambiguous names, and changing some function calling
5116 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5117 * While trying to figure out how to do the OpenBSD port, I tried to
5118 clean up some of the code in src/runtime/. In particular, I dropped
5119 support for non-POSIX signal handling, added various comments,
5120 tweaked the code to reduce the number of compilation warnings, and
5121 renamed some files to increase consistency.
5122 * To support the new automatic configuration functionality in make.sh,
5123 the source file target-features.lisp-expr has been replaced with the
5124 source file base-target-features.lisp-expr and the machine-generated
5125 file local-target-features.lisp-expr.
5126 * fixed a stupid quoting error in make.sh so that using CMU CL
5127 "lisp -batch" as cross-compilation host works again
5129 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5131 * DESCRIBE no longer tries to call itself recursively to describe
5132 bound/fbound values, so that it no longer fails on symbols which are
5133 bound to themselves (like keywords, T, and NIL).
5134 * DESCRIBE now works on generic functions.
5135 * The printer now prints less-screwed-up representations of closures
5136 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5137 * A private symbol is used instead of the :EMPTY keyword previously
5138 used to mark empty slots in hash tables. Thus
5139 (DEFVAR *HT* (MAKE-HASH-TABLE))
5140 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5141 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5142 now does what ANSI says that it should. (You can still get
5143 similar noncompliant behavior if bang on the hash table
5144 implementation with all the symbols you get back from
5145 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5146 This breaks binary compatibility, since tests for equality to
5147 :EMPTY are wired into things like the macroexpansion of
5148 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5150 * There's now a minimal placeholder implementation for CL:STEP,
5151 as required by ANSI.
5152 * An obscure bug in the interaction of the normal compiler, the byte
5153 compiler, inlining, and structure predicates has been patched
5154 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5155 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5157 * Missing ordinary arguments in a macro call are now detected even
5158 when the macro lambda list contains &KEY or &REST.
5159 * The debugger no longer complains about encountering the top of the
5160 stack when you type "FRAME 0" to explicitly instruct it to go to
5161 the top of the stack. And it now prints the frame you request even
5162 if it's the current frame (instead of saying "You are here.").
5163 * As specified by ANSI, the system now always prints keywords
5164 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5166 * The default initial SIZE of HASH-TABLEs is now smaller.
5167 * Type information from CLOS class dispatch is now propagated
5168 into DEFMETHOD bodies, so that e.g.
5169 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5171 is now basically equivalent to
5172 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5173 (DECLARE (TYPE SINGLE-FLOAT X))
5175 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5176 operation, without having to do run-time type dispatch.
5177 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5178 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5179 * Since I don't seem to be making big file reorganizations very often
5180 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5181 is overflowing my ability to conveniently back them up), I've finally
5182 checked the system into CVS. (The CVS repository is on my home system,
5183 not at SourceForge -- putting it on SourceForge might come later.)
5184 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5185 high-level GC-NOTIFY-FOO functions send their output. (There's
5186 still very little control of where low-level verbose GC functions
5187 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5188 less than it used to -- the GC-NOTIFY-FOO functions are now under
5189 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5190 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5191 in only one place in the source code, and propagates it automatically
5192 everywhere that it's needed. Thus e.g. when I bump the version from
5193 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5194 * The C source files now include boilerplate legalese and documentation
5195 at the head of each file (just as the Lisp source files already did).
5196 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5197 to his page will be replaced with a link to his new CLiki service.
5199 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5201 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5202 * The build process now looks for GNU make under the default name "gmake",
5203 instead of "make" as it used to. If GNU make is not available as "gmake"
5204 on your system, you can change this default behavior by setting the
5205 GNUMAKE environment variable.
5206 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5207 can build without error under CMU CL.
5209 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5211 * There is now a partial SBCL user manual (with some new text and some
5212 text cribbed from the CMU CL manual).
5213 * The beginnings of a profiler have been added (starting with the
5214 CMU CL profiler and simplifying and cleaning up). Eventually the
5215 main interface should be through the TRACE macro, but for now,
5216 it's still accessed through vaguely CMU-CL-style functions and macros
5217 exported from the package SB-PROFILE.
5218 * Some problems left over from porting CMU CL to the new
5219 cross-compilation bootstrap process have been cleaned up:
5220 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5221 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5222 part when getting rid of the compiler *BACKEND* stuff.)
5223 ** The value of *NULL-TYPE* was screwed up, because it was
5224 being initialized before the type system knew the final
5225 definition of the 'NULL type. This screwed up several key
5226 optimizations in the compiler, causing inefficiency in all sorts
5227 of places. (I found it because I wanted to understand why
5228 GET-INTERNAL-RUN-TIME was consing.)
5229 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5230 PROCLAIM FTYPE information. Unfortunately this broke binary
5231 compatibility again, since now the forms output by DEFGENERIC
5232 to refer to functions which didn't exist in 0.6.3.
5233 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5234 can use the new (as of 0.6.3) transform for SOME into MAP into
5236 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5237 values to (INTEGER 0 1000000), so that the time code will no longer
5238 occasionally get blown up by Linux returning 1000000 microseconds
5239 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5240 its output conform to the ANSI spec. (Alas, this makes its output
5241 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5242 * A full call to MAP NIL with a single sequence argument no longer conses.
5243 * fixes to problems pointed out by Martin Atzmueller:
5244 * The manual page no longer talks about multiprocessing as though
5245 it were currently supported.
5246 * The ILISP support patches have been removed from the distribution,
5247 because as of version 5.10.1, ILISP now supports SBCL without us
5248 having to maintain patches.
5249 * added a modified version of Raymond Toy's recent CMU CL patch for
5250 EQUALP comparison of HASH-TABLE
5252 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5254 * The system still can't cross-compile itself with
5255 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5256 entails), but at least it can compile more of itself that way
5257 than it used to be able to, and various buglets which were uncovered
5258 by trying to cross-compile itself that way have now been fixed.
5259 * This release breaks binary compatibility again. This time
5260 at least I've incremented the FASL file format version to 2, so that the
5261 problem can be detected reliably instead of just causing weird errors.
5262 * various new style warnings:
5263 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5264 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5265 ** using the lexical binding of a variable named in the *FOO* style
5266 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5267 as specified by ANSI.
5268 * *RANDOM-STATE* is no longer automatically initialized from
5269 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5270 default behavior of the system is to repeat its behavior every time
5271 it's run. If you'd like to change this behavior, you can always
5272 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5273 old convention there was no comparably easy way to get the system to
5274 repeat its behavior every time it was run.
5275 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5276 FTYPE declarations has been removed, in favor of their ANSI
5277 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5278 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5279 the types of their sequence arguments can be determined at compile time.
5280 This is done through a new open code expansion for MAP which eliminates
5281 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5282 argument types can be determined at compile time.
5283 * The optimizer now transforms COERCE into an identity operation when it
5284 can prove that the coerced object is already of the correct type. (This
5285 can be a win for machine generated code, including the output of other
5286 optimization transforms, such as the MAP transform above.)
5287 * Credit information has been moved from source file headers into CREDITS.
5288 * Source file headers have been made more standard.
5289 * The CASE macro now compiles without complaining even when it has
5292 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5294 * (Note that the way that the PCL macroexpansions were rewritten
5295 to accommodate the change in DEFGENERIC below breaks binary
5296 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5297 not run under sbcl-0.6.2. Once we get out of alpha releases,
5298 i.e. hit release 1.0.0, we'll probably try to maintain binary
5299 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5300 and sbcl-1.4.4. Until then, however, it might be fairly common
5301 for maintenance releases to break binary compatibility.)
5302 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5304 * The reporting of unhandled signals has been changed to print some
5305 explanatory text as well as the report form. (Previously only
5306 the report form was printed.)
5307 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5308 it defines, so that the compiler no longer issues undefined function
5309 warnings for compiled-but-not-yet-loaded generic functions.
5310 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5311 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5312 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5313 section of the ANSI spec are supported.
5314 * The "" nickname for the "KEYWORD" package has been removed.
5315 The reader still handles symbol tokens which begin with a package marker
5316 as keywords, but it doesn't expose its mechanism for doing so in the
5317 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5318 * The system now issues STYLE-WARNINGs for contradictory TYPE
5319 proclamations. (Warnings for contradictory FTYPE proclamations would
5320 be nice too, but those can't be done usefully unless the type system
5321 is made smarter about FUNCTION types.)
5322 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5323 systematized, so that "*target-*.lisp is supposed to exist only on the
5324 target and imply that there's a related file which exists on the
5325 host, and *host-*.lisp is supposed to exist only on the host and imply
5326 that there's a related file which exists on the target. This involves a
5327 lot of renaming. Hopefully the acute confusion caused by the renaming
5328 will be justified by the reduction in chronic confusion..
5329 ** runtime-type.lisp -> early-target-type.lisp
5330 ** target-type.lisp -> late-target-type.lisp
5331 ** early-host-format.lisp -> early-format.lisp
5332 ** late-host-format.lisp -> late-format.lisp
5333 ** host-error.lisp -> misc-error.lisp
5334 ** early-error.lisp -> early-target-error.lisp
5335 ** late-error.lisp -> late-target-error.lisp
5336 ** host-defboot.lisp -> early-defboot.lisp
5337 ** code/misc.lisp -> code/target-misc.lisp
5338 ** code/host-misc.lisp -> code/misc.lisp
5339 ** code/numbers.lisp -> code/target-numbers.lisp
5340 ** code/early-numbers.lisp -> numbers.lisp
5341 ** early-host-type.lisp -> early-type.lisp
5342 ** late-host-type.lisp -> late-type.lisp
5343 ** host-typep.lisp -> typep.lisp
5344 ** load.lisp -> target-load.lisp
5345 ** host-load.lisp -> load.lisp
5346 ** host-disassem.lisp -> disassem.lisp
5347 ** host-insts.lisp -> insts.lisp
5348 ** byte-comp.lisp -> target-byte-comp.lisp
5349 ** host-byte-comp.lisp -> byte-comp.lisp
5350 ** host-signal.lisp -> signal.lisp
5351 ** host-defstruct.lisp -> defstruct.lisp
5352 ** late-target-type.lisp -> deftypes-for-target.lisp
5353 Furthermore, several other previously target-only files foo.lisp (e.g.
5354 hash-table.lisp and random.lisp) have been split into a target-and-host
5355 foo.lisp file and a target-only target-foo.lisp file, with their key type
5356 definitions in the target-and-host part, so that the cross-compiler will
5357 know more about target types.
5358 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5359 gone away. In their place are various *BACKEND-FOO* variables
5360 corresponding to the slots of the old structure.
5361 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5362 into the target SBCL has been fixed.
5363 * The chill.lisp system for loading cold code into a running SBCL
5365 * Support for the CMU CL "scavenger hook" extension has been removed.
5366 (It was undocumented and unused in the CMU CL sources that SBCL was
5367 derived from, and stale in sbcl-0.6.1.)
5368 * Various errors in the cross-compiler type system were detected
5369 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5370 (enabling various consistency checks). Many of them were fixed,
5371 but some hard problems remain, so the compiler is back to
5372 running without *TYPE-SYSTEM-INITIALIZED* for now.
5373 * As part of the cross-compiler type system cleanup, I implemented
5374 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5375 * I have started adding UNCROSS calls throughout the type system
5376 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5377 on cross-compiler input files will be able to go away, and various
5379 * CONSTANTP now returns true for quoted forms (as explicitly required
5382 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5384 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5385 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5386 (MAKE-LIST -1). In the longer term, it ought to become true
5387 that declarations are assertions even at SAFETY 1. For now, it's not
5388 quite true even at SAFETY 3, but it's at least more nearly true..
5389 (Note that this change seems to increases the size of the system by
5390 O(5%) and to decrease the speed of the compiler by 20% or more.)
5391 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5392 for the problem of printing dozens of lines of distracting information
5393 about low-level system machinery as part of the top stack frame
5394 on entry to the debugger when an undefined function was called.
5395 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5396 is not reset to COMMON-LISP-USER.
5397 * Compilation of stuff related to dyncount.lisp has been made conditional
5398 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5399 smaller. The various dyncount-related symbols have been moved into
5400 a new "SB-DYNCOUNT" package.
5401 * tty-inspect.lisp has been renamed to inspect.lisp.
5402 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5403 feature has gone away. (When we eventually port to other flavors of
5404 libc and/or Unix, we'll try to make the differences between flavors
5405 invisible at the user level.)
5406 * Various other *FEATURES* tags, and/or their associated conditionals,
5407 have been removed if obsolescent, or given better documentation, or
5408 sometimes given more-mnemonic names.
5410 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5412 * tidied up "make.sh" script
5413 * tidied up system directory structure
5414 * better "clean.sh" behavior
5415 * added doc/FOR-CMUCL-DEVELOPERS
5416 * many many small tweaks to output format, e.g. removing possibly-confusing
5417 trailing #\. character in DESCRIBE-INSTANCE
5418 * (EQUALP #\A 'A) no longer signals an error.
5419 * new hashing code, including EQUALP hashing
5420 * tidied up Lisp initialization and toplevel
5421 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5422 * command line argument processing
5423 * added POSIX-GETENV function to deal with Unix-ish environment variables
5424 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5425 terminating SBCL on EOF
5426 * non-verbose GC by default
5427 * There is no more "sbcl" shell script; the sbcl file is now the C
5428 runtime executable (just like CMU CL).
5429 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5431 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5432 symbol and package information as Lisp native symbol and package objects
5433 instead of strings naming symbols and strings naming packages. This way,
5434 whenever packages are renamed (as in warm init), debug information is
5435 transformed along with everything else.
5436 * tweaked the optimization policy declarations which control the building
5437 of SBCL itself. Now, among other things, the system no longer saves
5438 source location debugging information. (This helps two problems at once
5439 by reducing SBCL size and by keeping SBCL from trying to look for its
5440 sources -- which may not exist -- when reporting errors.)
5441 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5442 debugging and testing purposes
5443 * cleaned up printing, making the printer call PRINT-OBJECT for
5444 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5445 methods, giving nearly-ANSI behavior
5446 * converted almost all special variables to use *FOO* naming convention
5447 * deleted PARSE-TIME functionality, since it can be done portably
5448 * moved some files out of cold init into warm init
5449 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5451 * regularized formatting of source files
5452 * added an install.sh script
5453 * fixed ridiculous memory usage of cross-compiler by making
5454 compiler/alloc.lisp not try to do pooling unless it can hook
5455 itself into the GC of the cross-compilation host. Now the system
5456 builds nicely on my old laptop.
5457 * added :SB-ALLOC in target-features.lisp-expr
5458 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5459 was not implemented)
5460 * re-did condition handling and note reporting in the compiler. Notes
5461 are no longer handled by signalling conditions. Style warnings
5462 and warnings are handled more correctly and reported in such a way
5463 that it's easy to find one or the other in your output (so that you
5464 can e.g. figure out which of many problems caused COMPILE-FILE to
5466 * changed the severity of several compiler warnings from full WARNING
5467 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5468 compiler note reporting so that it doesn't use the condition system
5469 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5470 COMPILE-FILE command)
5471 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5472 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5473 no longer supported, since their implementation was deeply intertwingled
5474 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5475 * removed bogus "support" for compiler macros named (SETF FOO), and
5476 removed the compiler macro for SETF INFO (but only after making a fool
5477 of myself on the cmucl-imp mailing list by posting a bogus patch for
5478 DEFINE-COMPILER-MACRO..)
5479 * Compiled files containing forms which have side effects on the Lisp
5480 reader (such as DEFPACKAGE forms) are now handled more correctly.
5481 (Compiler queuing of top level lambdas has been suppressed by setting
5482 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5483 * deleted various currently-unused source files, e.g. gengc.lisp. They
5484 may be added back at some point e.g. when porting to other architectures,
5485 but until they are it's distracting to distribute them and to try to
5487 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5488 there were so many of them they're just distractions, and UNCROSS is
5489 known to be able to handle the current sources
5490 * moved PROFILE functionality into TRACE, so that it will be clear
5491 how the wrapping and unwrapping of functions when you profile them
5492 interacts with the wrapping and unwrapping of functions when you
5493 trace them. (Actually, the functionality isn't there yet, but at least
5494 the interface specification is there. Hopefully, the functionality will
5495 arrive with some future maintenance release.)
5496 * removed host-oops.lisp
5497 * changed signature of QUIT function to allow UNIX-CODE argument
5498 * fixed READ-SEQUENCE bug
5499 * tweaked verbose GC output so that it looks more like the progress
5500 output that ANSI specifies for functions like LOAD
5501 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5502 * added <http://sbcl.sourceforge.com> to the banner information printed by