1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.36:
3 * enhancement: Backtrace from THROW to uncaught tag on x86oids now shows
4 stack frame thrown from.
5 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) take advantage of
6 constraint propgation, allowing better compilation eg. when used to
7 access structures with WITH-SLOTS. (lp#520366)
8 * optimization: the compiler is now more aware of the type of the underlying
9 storage vector for multidimensional simple arrays resulting in better code
10 for accessing such arrays.
11 * optimization: passing NIL as the environment argument to TYPEP no longer
12 inhibits optimizing it. (lp#309788)
13 * bug fix: Fix compiler error involving MAKE-ARRAY and IF forms
14 in :INITIAL-CONTENTS. (lp#523612)
15 * bug fix: FUNCTION-LAMBDA-EXPRESSION lost declarations from interpreted
16 functions. (lp#524707)
17 * bug fix: bogus style warnings from certain (SETF SLOT-VALUE) and
18 WITH-SLOTS usages during compilation.
19 * bug fix: SB-C::CLASS-INFO now prints correctly (lp#514762)
21 changes in sbcl-1.0.36 relative to sbcl-1.0.35:
22 * new feature: SB-EXT:TYPEXPAND-1, SB-EXT:TYPEXPAND, and
23 SB-EXT:TYPEXPAND-ALL behave exactly like their MACROEXPAND counterparts
24 but work on type specifiers.
25 * new feature: SB-EXT:DEFINED-TYPE-NAME-P returns whether a symbol is known
26 to name a type specifier.
27 * new feature: SB-EXT:VALID-TYPE-SPECIFIER-P returns whether a given type
28 specifier is valid where "valid" basically means "would be accepted as
29 second argument of TYPEP".
30 * new feature: SB-INTROSPECT:FUNCTION-TYPE takes a function-designator and
31 returns the function's declared, or derived FTYPE.
32 * new feature: SB-POSIX now supports accessing the d_ino member of
33 dirent structures. (Thanks to Philipp Marek and Pierre THEIRRY)
34 * new feature: The function SB-EXT:SEED-RANDOM-STATE has been added to
35 provide for seeding a RANDOM-STATE object with user-provided data or
36 from the operating system's PRNG. Also, (MAKE-RANDOM-STATE T) will
37 attempt to initialize the returned state from the operating system's
38 PRNG where possible. (Thanks to Fare Rideau; launchpad bug lp#310116)
39 * bug fix: Fix SB-SIMPLE-STREAMS:READ-VECTOR to correctly set the
40 FILE-POSITION of the stream being read from. (launchpad bug lp#491087)
41 * bug fix: Fix grammar and style issues for the docstrings of
42 printer-related variables and functions. (Thanks to mon_key; launchpad
44 * bug fix: Fix compilation on chenygc platforms. Thanks to Larry Valkama and
46 * bug fix: SB-THREAD:CONDITION-WAIT sometimes signaled a deadline twice
47 in a row even though a handler defered the deadline long into the
49 * bug fix: A deadline handler was run without interrupts enabled for a
50 deadline signaled within SB-THREAD:CONDITION-WAIT. That could result
51 in infinitely spinning, non-killable threads.
52 * bug fix: Backtrace from internal-errors on x86-64 os x was truncated
53 before reaching the erring stack frame.
54 * bug fix: Fix type derivation for EXPT when raising a fixnum to a
55 real power. (launchpad bug lp#525949)
56 * bug fix: Fix SB-EXT:GENERATION-* accessors for generations > 0 on
57 GENCGC platforms. (launchpad bug lp#529014)
58 * bug fix: More robust checks for invalid DEFMETHOD argument specializers.
59 (launchpad bug lp#525916)
60 * bug fix: Fix building on Darwin when sysctl is not in the user's PATH.
61 (Thanks to Robert Goldman)
63 changes in sbcl-1.0.35 relative to sbcl-1.0.34:
64 * optimization: ROUND with a single single-float or double-float argument
65 is properly inlined when possible.
66 * optimization: Slightly better code is generated for integer<->float
67 conversions and for single-float<->double-float conversions on x86-64.
68 * optimization: SB-ROTATE-BYTE:ROTATE-BYTE now generates more efficient
69 code for 32-bit and 64-bit rotations on x86-64.
70 * bug fix: The install script changes the ownership of directories as well
71 as files for contrib modules using asdf. (thanks to Eugene Ossintsev;
72 launchpad bug lp#508485)
73 * bug fix: TRUNCATE with a single single-float or double-float argument is
74 properly inlined when possible. (launchpad bug lp#489388)
75 * bug fix: Passing a rotation count of zero to SB-ROTATE-BYTE:ROTATE-BYTE
76 no longer causes a compiler error on x86 and ppc.
77 * bug fix: GET-MACRO-CHARACTER bogusly computed its second return value
78 always relative to *READTABLE* rather than the passed argument.
80 changes in sbcl-1.0.34 relative to sbcl-1.0.33:
81 * minor incompatible change: threading support is now enabled by default
83 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
85 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
86 are now caught and reported just like errors during macroexpansion.
87 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
88 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
89 * enhancement: ASDF systems can now depends on SB-QUEUE.
90 * fixes and improvements related to Unicode and external formats:
91 ** bug fix: error handling and restart usage in the ucs-2 external format
93 ** there is now an implementation of the ucs-4 external format.
94 ** the utf-16 and utf-32 external formats are supported.
95 * bug fix: SB-POSIX wrapper for putenv no longer tries to put lisp strings
96 in the environment. setenv() and unsetenv() are also provided. (reported by
97 Fare Rideau; launchpad bug lp#460455)
98 * bug fix: LOAD of both .fasl and .FASL type files now forces fasl-style
99 loading. This ensures sensible errors for .FASL files from other
100 implementations on case-insensitive filesystems. (reported by Willem
101 Broekema; launchpad bug lp#489417)
102 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
103 Willem Broekema; launchpad bug lp#489698).
104 * bug fix: some minor code rearrangements to reenable warning-free building
105 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
106 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
108 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
111 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
112 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
113 * improvement: support O_LARGEFILE access to files larger than 2GB on
114 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
115 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
116 definitions for methods specializing on the passed class itself.
117 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
118 definitions for methods specializing on the passed class itself, or on
120 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
121 about itself during the build (e.g. for M-? in Slime), if this flag is
122 enabled in customize-target-features.lisp. This will increase the core
123 size by about 5-6mb, though, so it's mostly interesting to SBCL
125 * new feature: various GENCGC tuning parameters have been experimentally
126 documented and exported from SB-EXT. See documentation for details.
127 * fixes and improvements related to Unicode and external formats:
128 ** the Unicode character database has been upgraded to the
129 Unicode 5.2 standard, giving names and properties to a number of new
130 characters, and providing a few extra characters with case
132 ** improvement: restarts for providing replacement input/output on coding
133 errors for fd-stream external formats.
134 ** improvement: where :<encoding> is a keyword corresponding to an
135 external format the system supports, it is now possible to specify
136 (:<encoding> :replacement <character>) as an external format which will
137 automatically substitute <character> on encoding or decoding errors for
138 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
140 ** improvement: the file streams underlying the standard streams (such as
141 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
142 which uses the replacement mechanism to handle encoding errors,
143 preventing various infinite error chains and unrecoverable I/O
145 ** minor incompatible change: the utf-8 external format now correctly
146 refuses to encode Lisp characters in the surrogate range (char-codes
147 between #xd800 and #xdfff).
148 ** fix a typo preventing conversion of strings into octet vectors
149 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
151 ** fix a bug in the octet multibyte handling of decoding errors and the
152 USE-VALUE restart. (launchpad bug lp#314939)
153 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
154 restart on fd-stream decoding errors.
155 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
156 error is near the end of file.
157 ** fix a double-error case in unibyte octet conversions, when the first
158 use of USE-VALUE is ignored.
159 ** fix bugs in handling of undefined code points in unibyte encodings.
160 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
161 streams after an UNREAD-CHAR.
162 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
163 object is allocated in a boxed region of dynamic space.
164 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
165 fork with multiple Lisp threads running is made, instead of going
166 ahead with unpredictable consequences. (reported by Leslie Polzer)
167 * bug fix: uses of slot accessors on specialized method parameters within
168 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
169 error while finalizing the class. This fix may cause classes with slot
170 accessors to be finalized later than previously. (reported by Lars Rune
171 Nøstdal; launchpad bug lp#473699)
172 * bug fix: restore buildability on the MIPS platform. (regression from
173 1.0.30.38, reported by Samium Gromoff)
174 * bug fix: inspecting closures is less likely to fail with a type error.
175 * bug fix: no timer starvation when setting the system clock back.
176 (launchpad bug lp#460283)
177 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
178 standard pprint dispatch table as specified by CLHS.
179 * bug fix: give CLISP a hint about a type declaration to enable it to build
180 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
182 * bug fix: correctly dump literal objects in defaulting forms of arglists.
183 (reported by Attila Lendvai; launchpad bug lp#310132)
184 * bug fix: distinguish in type specifiers between arrays that might be
185 complex and arrays that are definitely complex. (launchpad bug lp#309129)
186 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
187 type. (reported by Levente Mészáros; launchpad bug lp#485972)
188 * bug fix: setting the value of a symbol-macro within a method in the
189 presence of type declarations works properly again. (reported by Iban
190 Hatchondo; launchpad bug lp#485019)
192 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
193 * optimization: faster FIND and POSITION on strings of unknown element type
194 in high SPEED policies. (thanks to Karol Swietlicki)
195 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
197 * improvement: better error signalling for bogus parameter specializer names
198 in DEFMETHOD forms (reported by Pluijzer)
199 * bug fix: DELETE-FILE once again works on logical pathnames (regression
201 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
203 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
204 removes previous default initargs (reported by Lars Rune Nøstdal and
206 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
207 lambda-lists. (reported by Peter Seibel)
208 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
209 correctly when starting from an executable core without saved runtime
210 options (reported by Faré Rideau, thanks to Zach Beane)
211 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
212 an optimization note instead. (reported by Martin Cracauer)
213 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
214 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
215 new process group if it doesn't need to share stdin with the sbcl
216 process. (thanks to Leslie Polzer)
217 * bug fix: SATISFIES could be misoptimized to refer to a local function.
218 (reported by Stanislaw Halik)
220 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
221 * improvement: stack allocation is should now be possible in all nested
222 inlining cases: failure to stack allocate when equivalent code is manually
223 open coded is now considered a bug.
224 * improvements related to Unicode:
225 ** the Unicode character database has been upgraded to the
226 Unicode 5.1 standard, giving names and properties to a number of new
227 characters, and providing a few extra characters with case
229 ** the system now recognizes and produces names for Unicode Hangul
231 ** the EBCDIC-US external-format is now supported for octet operations
232 (as well as for stream operations).
233 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
234 be used to output toplevel compile-time effects into a separate .CFASL
236 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
237 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
239 * optimization: TRUNCATE on known single- and double-floats is upto 25%
241 * optimization: division of floating point numbers by constants uses
242 multiplication by reciprocal when an exact reciprocal exists.
243 * optimization: multiplication of single- and double-floats floats by
244 constant two has been optimized.
245 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
246 sufficient type information is available. (thanks to Leslie Polzer)
247 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
248 known structure objects are as efficient as defstruct generated accessors.
249 * optimization: unused vector creation can now be optimized away.
250 * improvement: ASDF systems can now depends on SB-INTROSPECT.
251 * improvement: a STYLE-WARNING is signalled when a generic function
252 clobbers an earlier FTYPE proclamation.
253 * improvement: the compiler is able to track the effective type of
254 generic function across method addition and removal even in the
255 absence of an explicit DEFGENERIC.
256 * improvement: DESCRIBE now reports on symbols naming undefined
257 but assumed or declared function as well.
258 * improvement: recompilation of systems using SB-GROVEL now works
259 (thanks to Leslie Polzer)
260 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
261 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
263 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
264 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
265 well as user defined declaration names.
266 ** VARIABLE-INFORMATION is now aware of alien variables.
267 * improvement: improved address space layout on OpenBSD (thanks to Josh
269 * improvement: pretty-printing of various Lisp forms has been improved
270 (thanks to Tobias Rittweiler)
271 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
272 unused were deleted in safe code. (reported by John Fremlin)
273 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
275 * bug fix: some out-of-line array predicates were missing (reported by
277 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
279 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
280 (thanks to Larry D'Anna)
281 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
282 as well. (reported by Josh Elsasser)
283 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
284 leading to possible attempts to delete the same file twice. See docstring
285 on DELETE-FILE for details. (reported by John Fremlin)
286 * bug fix: DELETE-FILE once again deletes the file named by the pathname
287 designator argument, rather than its truename. (reported by Luis
289 * bug fix: the low-level debugger had 32-bit assumptions and was missing
290 information about some array types. (thanks to Luis Oliveira)
291 * bug fix: moderately complex combinations of inline expansions could
292 be miscompiled if the result was declared to be dynamic extent.
293 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
294 form (+ <variable> <integer>) were miscompiled under certain
296 * bug fix: in some cases no compiler note about failure to stack allocate
297 was emitted, even if the objects were in fact heap allocated.
298 * bug fix: minor violation of "otherwise inaccessible" rule for stack
299 allocation could cause objects users might reasonably expect to
300 be heap allocated to be stack allocated.
301 * bug fix: DESCRIBE signalled an error for generic functions under
302 certain circumstances. (thanks to Leslie Polzer)
303 * bug fix: Fixed spelling of an error message.
305 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
306 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
307 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
308 of SB-THREAD:THREAD-ERROR-THREAD.
309 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
310 * new feature: docstrings for local and anonymous functions are no longer
311 discarded. (thanks to Leslie Polzer)
312 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
313 values in other threads.
314 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
315 about object allocation.
316 * optimization: division of a real float by a complex float is implemented
317 with a specialised code sequence.
318 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
319 keywords is an order of magnitude faster.
320 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
321 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
322 methods, and similarly for non-standard metaclass classes as long as there
323 are no methods additional on MAKE-INSTANCE.
324 * optimization: more efficient type-checks for FIXNUMs when the value
325 is known to be a signed word on x86 and x86-64.
326 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
327 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
329 * optimization: compiler is smarter about delegating argument type checks to
331 * optimization: several character functions are now compiled somewhat more
332 efficiently. (reported by Lynn Quam)
333 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
334 * improvement: less unsafe constant folding in floating point arithmetic,
335 especially for mixed complex/real -float operations.
336 * optimization: constant double and single floats are stored in native
337 unboxed format on x86[-64].
338 * optimization: smarter code for arithmetic operations with constant floats,
339 complex floats, or integers on x86[-64].
340 * optimization: smarter code for conjugate/multiplication of float complexes
341 and abs/negate of floats on x86-64.
342 * optimization: more efficient complex float and real float operations on
344 * improvement: complex float division is slightly more stable.
345 * improvement: DESCRIBE output has been reworked to be easier to read and
346 contains more pertinent information.
347 * improvement: failure to provide requested stack allocation compiler notes
348 provided in all cases (requested stack allocation not happening without a
349 note being issued is now considered a bug.)
350 * bug fix: SB-POSIX exports the documented types and functions
351 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
352 types. (reported by "abhi")
353 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
354 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
355 (thanks to Paul Khuong)
356 * bug fix: looping around HANDLER-CASE could silently consume stack space
357 on each iteration. (reported by "foobar")
358 * bug fix: better error signalling when calls to functions seeking elements
359 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
360 (reported by Tobias Rittweiler)
361 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
362 directory components now works as it used to. (various prolems reported by
363 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
364 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
365 (reported by Elliot Slaughter, thanks to Stas Boukarev)
366 * bug fix: bogus undefined variable warnings from fopcompiled references to
367 global variables. (thanks to Lars Rune Nøstdal)
368 * bug fix: foreign function names should now appear in backtraces on
369 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
370 * bug fix: SETF compiler macro documentation strings are not discarded
372 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
373 (thanks to Alex Plotnick)
374 * bug fix: SLEEP accepts large integer arguments, truncating them to
375 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
376 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
377 * bug fix: compiler notes for expensive slot type checks could be emitted
378 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
380 changes in sbcl-1.0.29 relative to 1.0.28:
381 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
382 https://bugs.launchpad.net/sbcl
383 Bugs can be reported directly there, or by sending email to
384 sbcl-bugs@lists.sourceforge.net
385 (no subscription required.)
386 * minor incompatible change: under weak type checking policy integer
387 types are weakened less aggressively.
388 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
389 allowed to return, which causes SBCL to quit with exit status 0. Previously
390 if the function returned with a small integer return value, that value
391 was accidentally reused as the exit status.
392 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
393 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
394 :HASH-FUNCTION argument. Refer to user manual for details.
395 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
397 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
398 microseconds since the Unix epoch on all platforms.
399 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
400 allows the compiler to safely elide boundedness checks for special
402 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
403 the symbol, prohibits both lexical and dynamic binding. This is mainly an
404 efficiency measure for threaded platforms, but also valueable in
406 * new feature: UNC pathnames are now understood by the system on Windows.
407 * optimization: the compiler uses a specialized version of FILL when the
408 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
410 * optimization: accesses to potentially non-simple arrays where element type
411 is known are 50% faster.
412 * optimization: compiler now generates faster array typechecking code.
413 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
415 * optimization: multidimensional array accesses in the absence of type
416 information regarding array rank are approximately 10% faster due to
417 open coding of ARRAY-RANK.
418 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
419 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
421 * optimization: result of call to VECTOR can now be stack allocated.
422 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
423 as long as the resulting array is one-dimensional and has a known
424 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
425 length of the list matches the known length of the vector does not
426 allocate the list as an intermediate step. Ditto for VECTOR and simple
428 * optimization: MAKE-ARRAY can now stack allocate in the presence of
429 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
430 known element type, and is known to be simple and one dimensional.
431 * improvement: SBCL now emits a compiler note where stack allocation was
432 requested but could not be provided (not in all cases, unfortunately)
433 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
434 * improvement: pretty-printing loop has been implemented properly. (thanks
435 to Tobias Rittweiler)
436 * documentation: CLOS slot typechecing policy has been documented.
437 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
438 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
439 pointer, leading to memory faults. (thanks to Bart Botta)
440 * bug fix: better floating point exception handling on x86/OpenBSD.
441 (thanks to Josh Elsasser)
442 * bug fix: exit status from QUIT when called under --script was lost
443 (reported by Hubert Kauker)
444 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
445 same implementation of FILL to initialize the array, even if a more
446 efficient one was available (reported by Stas Boukarev, thanks to
448 * bug fix: potential miscompilation of array stack allocation on x86 and
449 x86-64. (reported by Time Tossavainen)
450 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
451 result in their subforms being treated as top level forms. (reported by
453 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
454 than a machine register could leave garbage in the high bits of the
455 result register (bug 316325).
456 * bug fix: disable address space randomization Linux/x86-64 as well,
457 not just x86-64. (reported by Ken Olum)
458 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
459 generate incorrect code.
460 * bug fix: #201; type inference for CONS and ARRAY types could derive
461 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
462 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
463 could not be assigned to or bound with PROGV. (reported by Lorenz
465 * bug fix: the value of CL:- in the inspector was the previous expression
466 evaluated rather than the expression being evaluated.
467 * bug fix: constants can no longer be locally declared special.
468 * bug fix: signals delivered to threads started from foreign land (read:
469 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
470 thread by blocking all signals and resignalling.
471 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
472 from :INITFORM, if any.
474 changes in sbcl-1.0.28 relative to 1.0.27:
475 * a number of bugs in cross-compilation have been fixed, with the ultimate
476 result that building under (at least) clisp should be much more reliable.
477 * minor incompatible changes: echo-streams now propagate unread-char to the
478 underlying input stream, and no longer permit unreading more than one
480 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
481 frames, allowing for instance more reliable backtraces.
482 * improvement: the debugger REPL can now reference lexical variables
483 by name directly for code compiled with (DEBUG 3).
484 * improvement: errors from malformed declarations now have better source
485 paths associated with them. (thanks to Tobias Rittweiler)
486 * optimization: faster local calls on x86/x86-64
487 * bug fix: some error messages for out-of-bound array indexes confused the
488 index and the bound. (thanks to Stas Boukarev)
489 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
491 * bug fix: running regressions tests in shells without OSTYPE set now works.
492 (reported by Harald Hanche-Olsen)
493 * bug fix: more robust static space exhaustion signalling from
494 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
495 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
496 docstring away instead of storing it under names such as (LAMBDA (X)).
497 (reported by Leslie Polzer)
498 * bug fix: timers could go off in the wrong order, be delayed indefinitely
499 (thanks to Ole Arndt for the patch)
500 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
501 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
502 potentially-invalid effective methods in its cache.
503 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
504 instances as well (thanks to Paul Khuong)
505 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
506 SB-POSIX:WAITPID and their C equivalents.
507 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
509 changes in sbcl-1.0.27 relative to 1.0.26:
510 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
511 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
512 * improvement: the system either recovers from stack exhaustion or dies
513 properly as opposed to leaving the user uncertain of whether the handler
514 trampled on some random memory next to the stack or having to rely on
515 --lose-on-corruption (which is still a good idea to use in production
516 because stack exhaustion can happen in signal handlers which will likely
518 * bug fix: fix gc related interrupt handling bug on ppc (regression from
519 1.0.25.37, reported by Harald Hanche-Olsen)
520 * bug fix: work around signal delivery bug in darwin (regression from
521 1.0.25.44, reported by Sidney Markowitz)
522 * bug fix: fix ERROR leaking memory (reported by David Thompson)
524 changes in sbcl-1.0.26 relative to 1.0.25:
525 * incompatible change: an interruption (be it a function passed to
526 INTERRUPT-THREAD or a timer function) runs in an environment where
527 interrupts can be enabled. The interruption can use
528 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
529 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
530 potentially running out of stack. Keep in mind that in the absence
531 of WITHOUT-INTERRUPTS some potentially blocking operation such as
532 acquiring a lock can enable interrupts.
533 * incompatible change: GC-OFF and GC-ON are removed, as they were
534 always unsafe. Use WITHOUT-GCING instead.
535 * new feature: runtime option --disable-ldb
536 * new feature: runtime option --lose-on-corruption to die at the
537 slightest hint of possibly non-recoverable errors: running out of
538 memory, stack, alien stack, binding stack, encountering a memory
539 fault, etc. In the absence of --lose-on-corruption a warning is
541 * enhancement: detect binding stack exhaustion
542 * enhancement: detect alien stack exhaustion on x86/x86-64
543 * improvement: generally more stable and reliable interrupt handling
544 * improvement: there is a per thread interruption queue,
545 interruptions are executed in order of arrival
546 * improvement: a repeating timer reschedules itself when the it has
547 finished, but expiration times are spaced equally. If an
548 expiration time is in the past it will trigger after a short grace
549 period that may give a chance to other things to run.
550 * optimization: slightly faster gc on multithreaded builds
551 * optimization: faster WITHOUT-GCING
552 * bug fix: when JOIN-THREAD signals an error, do it when not holding
553 important locks so that the debugger/handler doesn't produce
554 recursive errors or deadlock.
555 * bug fix: real-time signals are not used anymore, so no more
556 hanging when the system wide real-time signal queue gets full.
557 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
559 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
560 * bug fix: fix random memory faults related to interrupts on alpha
561 * bug fix: fix random memory faults related to interrupts on sparc
562 * bug fix: fix deadlocks related to starting threads
563 * bug fix: fix deadlines on locks on futex platforms
564 * bug fix: restore errno in signal handlers
565 * bug fix: fix deadlocks related to hash tables
566 * bug fix: fix deadlocks in pcl
568 changes in sbcl-1.0.25 relative to 1.0.24:
569 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
570 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
571 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
572 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
573 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
574 better name and does not return values so stale on multiprocessor systems.
575 Also, HOLDING-MUTEX-P was added for about the only sane usage of
577 * improvement: unithread builds keep track of MUTEX-VALUE.
578 * improvement: reading from a TWO-WAY-STREAM does not touch the output
579 stream anymore making it thread safe to have a concurrent reader and
580 a writer, for instance, in a pipe.
581 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
582 arguments, making compiler-macros for SETF-functions able to inspect
583 their constant arguments.
584 * improvement: COMPILE-FILE reports times with millisecond accuracy
585 (thanks to Luis Oliveira)
586 * optimization: CHAR-CODE type derivation has been improved, making
587 TYPEP elimination on subtypes of CHARACTER work better. (reported
588 by Tobias Rittweiler, patch by Paul Khuong)
589 * bug fix: setting alien structure fields of type struct by value now
590 computes the right offset for the memory copy.
591 * bug fix: compilation problem involving inlined calls to aliens with
592 result type VOID. (reported by Ken Olum)
593 * bug fix: #235a; sequential inline expasion in different policies no
594 longer reuses the functional from the previous expansion site.
595 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
596 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
597 * bug fix: named ENUMs in multiply-referenced alien record types no longer
598 cause an implied type redefinition. Regression from 1.0.21.29.
599 * bug fix: sign-extension of small signed return values now works with C
600 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
601 * improvements to the Windows port:
602 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
603 Westfold, thanks to Rudi Schlatte)
605 changes in sbcl-1.0.24 relative to 1.0.23:
606 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
607 vector of a multidimensional SIMPLE-ARRAY.
608 * new feature: the system now signals a continuable error if standard
609 readtable modification is attempted. (thanks to Tobias Rittweiler)
610 * new feature: DIRECTORY has been extended with a non-standard keyword
611 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
612 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
613 has been robustified and documented better. (thanks to Robert Goldman)
614 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
615 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
616 * tradeoff: constant FORMAT control strings are now compiled unless
617 SPACE > SPEED (previously only when SPEED > SPACE.)
618 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
620 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
622 * bug fix: lexical type declarations are now correctly reported by
623 SB-CLTL2. (reported by Larry D'Anna)
624 * bug fix: STRING-TO-OCTETS did not handle :START properly when
625 using UTF-8 as external format. (thanks to Luis Oliveira)
626 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
627 are signalled correctly. (thanks to Stas Boukarev)
628 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
629 designator. (thanks to Tobias Rittweiler)
630 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
631 designator, and returns T instead of the function. (thanks to
633 * bug fix: direct superclasses of STANDARD-CLASS and
634 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
635 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
636 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
637 when no class with the named slot yet exists no longer causes a
638 compile-time style-warning.
639 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
640 in safe code. (reported by Didier Verna)
641 * bug fix: #430; nested structure constructors can stack allocate.
642 * bug fix: on some 64-bit platforms dynamic space size was truncated
643 to #xffff0000 bytes. (reported by Benjamin Lambert)
644 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
645 REPL unusable. (reported by Daniel Herring)
646 * bug fix: return values of READ-SEQUENCE did not take :START into
647 account on file streams, regressions since 1.0.12.22. (reported by
648 Thomas Russ, patch by Paul Khuong)
649 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
650 method specializer used to confuse permuation vector optimization.
651 * bug fix: system inserted bogus implicit type declarations for local
652 special variables in DEFMETHOD bodies.
653 * bug fix: #354; duplicated frames in backtraces due to
654 non-tail-call-optimized XEPs to functions with return type NIL
656 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
657 initializes structure object slots according to DEFSTRUCT initforms,
658 and DEFSTRUCT forms :INCLUDEind structure classes defined using
659 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
660 (reported by Bruno Haible and Stephen Wilson)
661 * bug fix: #395; fill-pointer output streams used now support
662 element-type BASE-CHAR as well.
663 * bug fix: compiler error when attempting to derive return value of
664 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
666 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
668 * bug fix: Mac OS X binaries should now be portable between Leopard
671 changes in sbcl-1.0.23 relative to 1.0.22:
672 * enhancement: when disassembling method functions, disassembly
673 for the associated fast function is also produced.
674 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
676 * optimization: printing with *PRINT-PRETTY* true is now more
677 efficient as long as the object being printed doesn't require
678 special handling by the pretty printer.
679 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
680 now interact correctly with type declarations.
681 * partial bug fix: PCL detects infinite recursion during wrapper
682 validation. (thanks to Attila Lendvai)
683 * bug fix: #426; nested function calls are inlined properly.
684 Previously if FOO was an inline function, in calls of the form
685 (FOO (FOO ...)) the outer call was not inlined.
686 * bug fix: long long arguments passed in registers to alien
687 functions on PowerPC Linux and NetBSD are now handled correctly
688 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
690 changes in sbcl-1.0.22 relative to 1.0.21:
691 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
692 for the shared object in the current directory, but passes the native
693 namestring of the designated pathname to the operation system's shared
694 object loading function as-it.
695 * minor incompatible change: --disable-debugger toplevel option now takes
696 effect before processing of initialization files and --eval or --load
698 * new feature: new commandline argument: --script, which supports
699 shebang lines. See documentation for details. (based on work by
701 * new feature: SAVE-LISP-AND-DIE can save current values of
702 --dynamic-space-size and --control-stack-size in the executable core,
703 causing it to skip normal runtime option processing. See documentation
704 for details. (thanks to Zach Beane)
705 * enhancement: inoccous calls to EVAL or generic functions dispatching
706 on subclasses of eg. STREAM no longer cause compiler notes to appear.
707 * enhancement: the system no longer resignals errors from --load and
708 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
709 associated with the original error to be lost. (thanks to Ariel
711 * enhancement: :DONT-SAVE keyword argument has been added to
712 LOAD-SHARED-OBJECT for controlling interaction with
713 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
714 the effects of an earlier LOAD-SHARED-OBJECT call.
715 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
716 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
717 provided. (reported by Cedric St-Jean)
718 * bug fix: circularity handling in the reader did not treat raw
719 structure slots correctly. (reported by Cedric St-Jean)
720 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
721 file descriptors when there were none.
722 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
723 was broken. (reporteed by Willem Broekema)
724 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
725 pathnames without a directory.
726 * bug fix: COMPILE-FILE did not warn about undefined variable
727 references at toplevel, and LOAD of the resulting fasl did
729 * bug fix: functions with non-required arguments used to end up with
730 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
731 * bug fix: redefining a function with non-required arguments didn't
732 update the system's knowledge about its call signature properly.
733 * bug fix: fixed #431; incompatible alien record type redefinitions
734 are detected and handled. (reported by Neil Haven)
735 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
736 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
738 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
739 after alien stack frames.
740 * bug fix: strings passed to mk{s,d}temp lacked null termination
742 changes in sbcl-1.0.21 relative to 1.0.20:
743 * new feature: the compiler is able to track the effective type of a
744 generic function across method addition and removal.
745 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
746 appropriately typed structure slots without locking.
747 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
748 information like those gathered by TIME using a programming-friendly
750 * new feature: TIME reports time taken even if the form performs a
751 non-local transfer of control.
752 * enhancement: reduced conservativism on GENCGC platforms: on
753 average 45% less pages pinned (measured from SBCL self build).
754 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
755 mutate constant symbols or violate declaimed type of the symbol.
756 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
757 sleeping on the semaphore in heavily contested semaphores.
758 * bug fix: semaphores and condition variables were not interrupt
760 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
761 owned by other threads anymore.
762 * bug fix: FIND on lists called KEY outside the specified
763 subsequence. (reported by budden)
764 * bug fix: LOG doesn't use single-float intermediate results when
765 given mixed integer and double-float arguments, leading to better
766 precision. (reported by Bob Felts)
767 * bug fix: LOG with base zero returned values of inconsistent type.
769 changes in sbcl-1.0.20 relative to 1.0.19:
770 * minor incompatible change: OPTIMIZE qualities
771 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
772 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
773 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
774 * documentation: some slot access efficiency guidelines have been
775 added to the user manual.
776 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
777 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
778 as ASSOC and MEMEBER.
779 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
780 to more efficient EQ-comparison versions more often.
781 * optimization: enhanced derivation of DOLIST iteration variable type
783 * optimization: constant folding of simple (LIST ...) forms as DOLIST
785 * optimization: runtime lookup of function definitions can be
786 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
787 * optimization: compiler is able to derive the return type of
788 (AREF (THE STRING X) Y) as being CHARACTER.
789 * optimization: CLRHASH on empty hash-tables no longer does pointless
790 work. (thanks to Alec Berryman)
791 * optimization: REPLACE deftransforms don't punt when :START1 or
793 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
794 no longer emits a code deletion note. (reported by Andrew
796 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
797 functions with an odd number of &OPTIONAL arguments, a &REST
798 argument, and one or more &KEY arguments at the call site.
799 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
800 arguments appeared at call sites as well.
801 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
802 READ-LINE &co can no longer read from them afterwards. (reported
804 * bug fix: fixed #427: unused local aliens no longer cause compiler
805 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
807 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
808 alien-stack leakage. (reported by Andy Hefner)
809 * bug fix: PROGV signals an error when an attempt to violate declared
810 type of a variable or bind a constant is made.
811 * bug fix: SET signals an error when an attempt to violate declared
812 type of a variable is made.
813 * bug fix: restart computation during the execution of a restart
814 test function no longer causes infinite recursion. (reported by
816 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
817 signals an error. (thanks to Pierre Mai)
818 * bug fix: EXPORT left symbol unexported in conflict situations.
819 (thanks to Michael Weber)
820 * bug fix: correctly handle name conflicts involving CL:NIL.
821 (thanks to Michael Weber)
822 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
823 conflicts arising from USEing package with conflicting symbols
824 correctly. (thanks to Michael Weber)
826 changes in sbcl-1.0.19 relative to 1.0.18:
827 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
828 warnings that go otherwise unhandled will be muffled if they are
829 of the type that's the value of this variable.
830 * optimization: stack allocation is slightly more efficient on x86
832 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
833 where a raw slot always is initialized using the initform whose
834 type is not know sufficiently well a compile-time are now compiled
835 correctly. (reported by John Morrison)
836 * bug fix: compiler no longer makes erronous assumptions in the
837 presence of non-foldable SATISFIES types.
838 * bug fix: stack analysis missed cleanups of dynamic-extent
839 arguments in non-let-converted calls to local functions.
840 * improvements to the Windows port:
841 ** adjusted address spaces for building on both Win32 and
842 Win64. (thanks for John Connors)
843 * fixed some bugs revealed by Paul Dietz' test suite:
844 ** interval arithmetic during type derivation used inexact integer
845 to single-float coercions.
846 ** arithmetic operations involving large integers and single
847 floats give the same results in compiled and interpreted code.
848 ** deriving the result type of COERCE no longer signals an error
849 if the derived type of the second argument is a MEMBER type
850 containing invalid type specifiers.
851 ** ADJOIN with constant NIL as second argument works correctly.
852 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
854 changes in sbcl-1.0.18 relative to 1.0.17:
855 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
856 profiles only the current thread.
857 * minor incompatible change: changes to SYMBOL-VALUE of constants
858 defined with DEFCONSTANT now signal an error.
859 * enhancement: SB-SPROF now has support for wallclock profiling,
860 and is also able to profile specific threads. REPORT output
861 has also additional sorting options.
862 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
864 * optimization: structure allocation has been improved
865 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
866 ** out of line constructors are ~10% faster.
867 ** inline constructors are ~15% faster.
868 ** inline constructors are capable of dynamic extent allocation
869 (generally on x86 and x86-64, in some cases on other platforms
871 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
873 * optimization: file compiler is now able to coalesce non-circular
874 lists, non-base strings, and bit-vectors. Additionally, constants
875 are never referenced through SYMBOL-VALUE at runtime.
876 * optimization: code defining methods on PRINT-OBJECT (and other
877 generic functions in the COMMON-LISP package) now loads faster.
878 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
880 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
881 fasl is now deleted. (reported by Attila Lendvai)
882 * bug fix: READ-LINE always returned NIL for the last line in files.
883 (reported by Yoshinori Tahara)
884 * bug fix: more accurate disassembly annotations of foreign function
885 calls. (thanks to Andy Hefner)
886 * bug fix: trimming non-simple strings and non-string string
887 designators when the there is nothing to trim works properly.
888 (thanks to James Knight)
889 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
890 (thanks to Travis Cross)
891 * fixed some bugs revealed by Paul Dietz' test suite:
892 ** NIL is a valid function name (regression at 1.0.13.38)
893 ** FILL on lists was missing its return value (regression at 1.0.12.27)
894 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
895 fill pointers properly (regression at 1.0.12.23)
896 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
897 displacement indices properly (regression at 1.0.12.23)
899 changes in sbcl-1.0.17 relative to 1.0.16:
900 * temporary regression: user code can no longer allocate closure
901 variable storage on stack, due to bug 419 without explicitly
902 requesting it. Please consult sbcl-devel for advice if you need to
903 use this feature in the meanwhile.
904 * new feature: runtime argument --control-stack-size can be used to
905 adjust thread default control stack size.
906 * enhancement: improved TIME output
907 ** all times are reported using the measured accuracy (milliseconds
908 for real and GC times, microseconds for everything else.)
909 ** processor cycle counts on x86 and x86-64.
910 ** interpreted forms are counted for both evaluator modes.
911 ** number of lambdas converted by the compiler is reported.
912 ** CPU percentage report (computed from real and total run time.)
913 ** more comprehensive run time reporting, using a condenced format
914 ** interperted form, lambda, and page fault counts are omitted
916 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
918 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
919 * optimization: two argument forms of LAST are upto ~10% faster
920 in normal SPEED policies.
921 * optimization: NCONC no longer needs to heap cons its &REST list
922 in normal SPEED policies.
923 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
925 * bug fix: UNION and NUNION work with :TEST-NOT once more,
926 regression since 1.0.9.1. (thanks to Eric Marsden)
927 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
928 since 1.0.15.36. (thanks to Paul Khuong)
929 * bug fix: LAST when always returned the whole list when given a bignum
930 as the second argument.
931 * bug fix: dynamic extent allocation of nested lists and vectors
932 could leak to otherwise accessible parts.
933 * bug fix: invalid optimization of heap-allocated alien variable
935 * bug fix: fasl header checking is less vulnerable to different
936 platform word lengths.
937 * bug fix: more correct assembler syntax for GNU binutils
938 2.18.50.0.4 support. (thanks to Marijn Schouten)
939 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
941 * bug fix: the fopcompiler can handle LOCALLY forms (with no
942 declarations) successfully. (reported by Attila Lendvai)
944 changes in sbcl-1.0.16 relative to 1.0.15:
945 * minor incompatible change: revert the changes to sb-posix's error
946 signaling added in 1.0.14.
947 * minor incompatible change: change PROBE-FILE back to returning
948 NIL whenever we can't get a truename, as was the case before 1.0.14.
949 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
950 inherits from ERROR instead of just CONDITION.
951 * new feature: SB-INTROSPECT can provide source locations for instances
952 as well. (thanks to Tobias Rittweiler)
953 * optimization: binding special variables now generates smaller code
954 on threaded platforms.
955 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
956 and cases where no :TEST is given but the compiler can infer that
957 the element to search is of type (OR FIXNUM (NOT NUMBER)).
958 * optimization: better LOGNOT on fixnums.
959 * optimization: modular arithmetic for a particular requested width
960 is implemented using a tagged representation unless a better
961 representation is available.
962 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
963 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
964 for wellformedness and returns a specialized vector. (reported by
965 Francois-Rene Rideau)
966 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
967 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
968 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
969 unscheduling timer at the same time as another timer fires.
970 * bug fix: don't reschedule timers for dead threads.
971 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
972 * bug fix: copying output from RUN-PROGRAM to a stream signalled
973 bogus errors if select() was interrupted.
974 * enhancement: add support for fcntl's struct flock to SB-POSIX.
976 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
977 * enhancement: cleaner backtraces for interactive interrupts, as
978 well as other cases where the interesting frames used to be
979 obscured by interrupt handling frames.
980 * enhancement: untracing a whole package using (UNTRACE "FOO") is
981 now supported, and tracing a whole package using (TRACE "FOO") now
982 traces SETF-functions as well.
983 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
984 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
985 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
986 * unreadably printed representation of hash-tables now includes
988 * bug fix: partially fixed #188: type propagation from assignments
989 is now more efficient.
990 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
991 'DOUBLE-FLOAT) are not flushable.
992 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
993 full address of the object, and none of the tag bits.
994 * bug fix: readably printing hash-tables now respects other printer
995 control variables. (reported by Cedric St-Jean)
996 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
997 keyword with MAKE-HASH-TABLE.
998 * bug fix: export SB-POSIX:MKSTEMP.
999 * bug fix: SORT was not interrupt safe.
1000 * bug fix: XREF accounts for the last node of each basic-block as
1002 * bug fix: MAKE-INSTANCE optimizations interacted badly with
1003 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
1004 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
1005 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
1006 create obsolete instances in the case of redefinition or
1007 obsoletion of a superclass. (thanks to Andy Hefner)
1008 * bug fix: Support for the Alpha architecture has been revived; it had
1009 suffered somewhat from lack of maintenance since sbcl-1.0.
1010 * improvements to the Windows port:
1011 ** The system detects the codepage to use at startup, instead of
1012 using the value from the saved core. (thanks to Kei Suzuki)
1014 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
1015 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
1016 (see documentation for details.)
1017 * revived support for OpenBSD (contributed by Josh Elsasser)
1018 * partially fixed bug #108: ROOM no longer suffers from occasional
1019 (AVER (SAP= CURRENT END)) failures.
1020 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
1021 bodies no longer cause a WARNING to be signalled. (reported by
1023 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
1024 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
1025 the bug in TRUNCATE.) (reported by Michael Weber)
1026 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
1027 no samples. (reported by Andy Hefner)
1028 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
1029 no longer appear as (NIL ...) frames in backtraces.
1030 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
1031 is now actually exported from SB-EXT as documented. (reported by
1033 * bug fix: sb-aclrepl now correctly understands how to inspect
1034 single-floats on 64-bit platforms where single-floats are not boxed.
1035 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
1036 yet been finalized. (reported by Levente Meszaros)
1037 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
1038 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
1039 SBCL-specific optimize qualities.
1041 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
1042 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
1043 an executable in the search path, and does so in the child
1044 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
1045 been removed; it can be found in the CVS history, for somebody who
1046 needs that search behavior (see the manual).
1047 * minor incompatible change: compiler policy re. weakening type
1048 checks has changed: now type checks are weakened only if SAFETY < 2
1050 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
1051 unparsing of directory pathnames as files. Analogously,
1052 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
1053 filename to parse into a directory pathname.
1054 * enhancement: implicit generic function creation now signals a
1055 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
1056 users can bind handlers for and muffle around calls to LOAD.
1057 * enhancement: RUN-PROGRAM allows unicode arguments and environments
1058 to be used (using the default stream external format), and allows
1059 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
1060 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
1061 strings and vectors whose element-type or simplicity is not fully
1062 known at compile-time.
1063 * optimization: STRING-TRIM and related functions no longer allocate
1064 a new string when no trimming needs to be performed. These functions
1065 are also faster than before when the input string has been declared
1067 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
1068 * optimization: READ-LINE is significantly faster for files containing
1070 * optimization: non-open coded uses of character comparison operators
1071 (e.g. char=) no longer cons when called with more than one parameter
1072 on platforms supporting dynamic-extent allocation.
1073 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
1074 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
1075 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
1077 * bug fix: some sequence functions elided bounds checking when
1079 * bug fix: too liberal weakening of union-type checks when SPEED >
1081 * bug fix: more bogus fixnum declarations in ROOM implementation
1084 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
1085 * new feature: MAKE-HASH-TABLE now experimentally accepts a
1086 :SYNCHRONIZED argument, which makes the hash-table safe for
1087 concurrent accesses (but not iteration.) See also:
1088 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1089 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1090 * optimization: CONCATENATE on strings is an order of magnitude faster
1091 in code compiled with (> SPEED SPACE).
1092 * optimization: SUBSEQ is ~50% faster on lists.
1093 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1094 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1095 END is smaller then START.
1096 * bug fix: SB-PROFILE will no longer report extra consing for nested
1097 calls to profiled functions.
1098 * bug fix: ROOM implementation had bogus fixnum declarations which
1099 could cause type-errors when calling ROOM on large images.
1100 * bug fix: if file compilation is aborted, the partial fasl is now
1101 deleted, and COMPILE-FILE returns NIL as the primary value.
1102 * bug fix: number of thread safety issues relating to SBCL's internal
1103 hash-table usage have been fixed.
1104 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1105 be returned from its body when the values were being returned
1106 using unknown-values return convection and the W-P-O was wrapped
1107 inside an UNWIND-PROTECT.
1108 * bug fix: sb-posix should now compile again under Windows, enabling
1109 slime to work again.
1111 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1112 * incompatible change: hash-table accessor functions are no longer
1113 automatically protected by locks. Concurrent accesses on the same hash-table
1114 from multiple threads can give inconsistent results or even corrupt the
1115 hash-table completely. Multi-threaded applications should do their own
1116 locking at the correct granularity. In the current implementation it is
1117 still safe to have multiple readers access the same table, but it's not
1118 guaranteed that this property will be maintained in future releases.
1119 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1120 and will signal an error at runtime.
1121 * enhancement: SB-THREAD package now exports a semaphore interface.
1122 * enhancement: CONS can now stack-allocate on x86 and
1123 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1125 * enhancement: nested lists can now be stack allocated on
1126 platforms providing stack allocation support.
1127 * enhancement: dynamic-extent support has been extended to support
1128 cases where there are multiple possible sources for the stack
1130 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1131 if the mutex is uncontested on Linux.
1132 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1133 as specified by the CLHS. (thanks to Tobias Rittweiler)
1134 * bug fix: NaN comparison now works on x86-64.
1135 * bug fix: CLOSE :ABORT T on a stream with pending output now
1137 * bug fix: instances of non-standard metaclasses using standard
1138 instance structure protocol sometimes missed the slot type checks
1140 * bug fix: known functions can be inlined in high-debug code.
1142 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1143 * minor incompatible change: the MSI installer on Windows no longer
1144 associates .lisp and .fasl files with the installed SBCL.
1145 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1146 on Windows. (thanks to Luis Oliviera)
1147 * new feature: SB-CLTL2 contrib module now implements
1148 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1149 * optimization: scavenging weak pointers is now more efficient,
1150 requiring O(1) instead of O(N) per weak pointer to identify
1151 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1152 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1153 method bodies using constant slot-names when the first argument is
1154 a specializer parameter for the method.
1155 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1156 names, when the class of the instance is a direct instance
1157 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1158 as slow as the constant slot-name case.)
1159 * optimization: member type construction is now O(N) instead
1161 * optimization: UNION and NUNION are now O(N+M) for large
1162 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1164 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1165 now more readable in environments like Slime which display it.
1166 (thanks to Tobias C. Rittweiler)
1167 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1168 was able to infer the correct argument type for the object on which
1169 the CAS operation was being performed.
1170 * bug fix: copy propagation interfered with parallel assignment
1171 semantics in local calls. (reported by Paul Khuong)
1172 * bug fix: the signed modular fixnum shift compiled to wrong code on
1173 x86 and x86-64. (spotted by a slight modification to some of
1176 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1177 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1178 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1179 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1181 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1182 and generic function now print the object identity for anonymous
1183 instances. (thanks to Attila Lendvai)
1184 * enhancement: as an extension to MOP, SBCL now supports SETF of
1185 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1186 (thanks to Attila Lendvai)
1187 * enhancement: sb-cover has an alternate annotation mode which puts
1188 the form annotations at the car of the form, instead of annotating
1189 it completely. The benefit of this mode is that it shows better
1190 which forms were instrumented by the compiler.
1191 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1192 by printer control variables. (reported by Dan Corkill)
1193 * bug fix: system leaked memory when delayed output was performed by
1194 the OS in smaller chunks then expected. (thanks to David Smith)
1195 * bug fix: system leaked memory when file streams were not closed
1197 * bug fix: large objects written to slow streams that were modified
1198 after the write could end up with the modified state written to
1199 the underlying file descriptor.
1200 * bug fix: multiple threads operating in parallel on the same stream
1201 could cause buffer-overflows.
1202 * bug fix: source location information is stored correctly
1203 (broken since 1.0.6). This bug would generally show up as the
1204 Slime debugger highlighting the wrong form when the sldb-show-source
1206 * bug fix: Branch forms are again annotated as branches in the sb-cover
1208 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1209 on x86-64. (reported by Gregory Vanuxem)
1210 * bug fix: EQUALP could return wrong results for structures with raw
1211 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1212 word). (reported by Vjacheslav Fyodorov)
1213 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1216 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1217 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1218 atomic compare-and-swap operations on threaded platforms.
1219 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1220 allows assigning a global minimum value to optimization qualities
1221 (overriding proclamations and declarations).
1222 * enhancement: closed over variables can be stack-allocated on x86
1224 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1226 * optimization: slot definition lookup is now O(1). This speeds up
1227 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1228 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1229 * optimization: ASSOC and MEMBER can now be open-coded for all
1230 combinations of keyword arguments when second argument is constant
1231 and SPEED >= SPACE. In other cases a specialized version is
1233 * bug fix: using obsoleted structure instances with TYPEP and
1234 generic functions now signals a sensible error.
1235 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1236 (reported by Kristoffer Kvello)
1237 * bug fix: backtrace construction is now more careful when making
1238 lisp-objects from pointers on the stack, to avoid creating bogus
1239 objects that can be seen by the GC.
1240 * bug fix: defaulting of values in contexts expecting more than 7
1241 variables now works on x86-64. (reported by Christopher Laux)
1242 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1244 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1245 as the property-list of a symbol.
1246 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1247 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1248 situations requested, are once again file-compileable. (reported
1251 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1252 * MOP improvement: support for user-defined subclasses of
1253 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1254 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1255 the expansion of DEFMETHOD, is responsible for generating a form
1256 which creates a list of specializers when evaluated. Additional
1257 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1258 debugging and introspective support.
1259 * minor incompatible change: the (unsupported) spinlock interface
1260 has changed: free spinlock now has the value NIL, and a held spinlock
1261 has the owning thread as its value.
1262 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1263 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1265 * enhancement: name of a socket-stream is now "a socket" instead of
1266 "a constant string".
1267 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1268 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1269 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1270 for generic functions. (Thanks to Tobias C. Rittweiler)
1271 * optimization: bignum printing speed has been improved by 20-40%
1272 (depending on the bignum size.)
1273 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1275 * bug fix: the cache used by the CLOS to store precomputed effective
1276 methods, slot offsets, and constant return values is now thread and
1278 * bug fix: generic function dispatch function updating is now thread
1279 and interrupt safe (in the sense that the known issues have been
1281 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1282 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1283 * bug fix: asdf systems with dependencies to the SB-POSIX or
1284 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1285 * bug fix: interrupt safety of applicable method computation has been
1288 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1289 * new contrib: sb-cover, an experimental code coverage tool, is included
1290 as a contrib module.
1291 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1292 significantly faster.
1293 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1294 produce more efficient code on x86-64 if the compiler can determine
1295 that (+ INDEX <constant>) does not require a bounds check and FOO
1296 has an element type at least 8 bits wide.
1297 * enhancement: a new, experimental synchronous timeout facility is
1298 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1299 * enhancement: when a symbol name conflict error arises, the
1300 conflicting symbols are always printed with a package prefix.
1301 (thanks to Kevin Reid)
1302 * enhancement: stepping is now once again supported on the SPARC and
1303 MIPS platforms. (It is also now more likely to work on CheneyGC
1305 * enhancement: sb-sprof can now also track and report accurate call
1307 * bug fixes: the treatment of non-standard subclasses of
1308 SB-MOP:SPECIALIZER is more correct.
1309 * incompatible change: PURIFY no longer copies the data from the
1310 dynamic space into the static and read-only spaces on platforms
1311 that use the generational garbage collector
1312 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1314 * bug fix: GC race condition occasionally resulting in crashes with
1315 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1317 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1319 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1320 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1321 * bug fix: an asynchronous interrupt could previously leave the
1322 system running with GC inhibited.
1323 * bug fix: a DECLARE form evaluated at top-level now causes an error
1324 rather than silently (or verbosely) returning NIL.
1325 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1326 result in a error being signalled due to source form lookup errors.
1327 (reported by Peter Graves)
1329 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1330 * incompatible change: removed writer methods for host-ent-name,
1331 host-ent-addresses -- changing the values did not update the DNS
1333 * minor incompatible change: changed experimental JOIN-THREAD interface
1334 * documentation: the manual now lists reader and writer methods
1335 in class slot documentation sections. (thanks to Richard M Kreuter)
1336 * documentation: unwinding from asyncronous events has been
1337 documented as unsafe.
1338 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1339 in multithreaded application code.
1340 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1341 platforms. (thanks to James Anderson for the optimization hint)
1342 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1344 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1345 variants no longer cons.
1346 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1347 their NOT- variants no longer cons.
1348 * optimization: EQUAL hash tables no longer use SXHASH for objects
1349 of all data types, but instead use an EQL hash for types for which
1350 EQUAL is the same as EQL.
1351 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1352 are significantly faster.
1353 * optimization: new STRING-OUTPUT-STREAM implementation is much
1354 faster for large outputs and conses 30% less on average.
1355 * enhancement: XREF information is now collected for references made
1356 to global variables using SYMBOL-VALUE and a constant argument.
1357 * enhancement: SIGINT now causes a specific condition
1358 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1359 * bug fix: FILE-POSITION used to signal an error in some cases where
1360 ANSI requires it to return NIL.
1361 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1362 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1363 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1365 * bug fix: bad type declaration in the CLOS implementation has
1366 been fixed. (reported by James Anderson)
1367 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1368 (thanks to Lutz Euler)
1369 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1370 required by the CLHS. (reported by Samium Gromoff)
1371 * bug fix: dead unbound variable references now signal an error.
1372 (reported by Marco Monteiro)
1373 * bug fix: / with an unused value was being deleted in safe code.
1374 (thanks to Marco Monteiro and Kevin Reid)
1375 * bug fix: number of characters that can be written onto a single
1376 line in a file is unlimited.
1377 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1378 been fixed by inhibiting interrupts for when GC is disabled.
1379 * bug fix: some interrupt safety issues with GETHASH, (SETF
1380 GETHASH), CLRHASH and REMHASH have been fixed.
1381 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1382 specifier no longer causes infinite recursion.
1383 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1384 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1385 * bug fix: an &environment argument with the correct variable information
1386 is passed to macros that are expanded during byte compilation.
1387 (reported by Samium Gromoff)
1388 * bug fix: a logic error was causing the hash function for bignums to
1389 have excessive amounts of collisions. (reported by Faré Rideau)
1390 * bug fix: modifying the contents of an array could change the return
1391 value of SXHASH on that array, which is only allowed for strings
1392 and bit vectors. (bug introduced in 0.9.16)
1393 * improvement: the x86-64/darwin port now passes all tests (except
1394 for the debugger tests) but should still be considered
1395 experimental until this is fixed.
1396 * improvement: a style-warning is signaled for CASE (etc) clauses with
1397 duplicate keys. (thanks to Kevin Reid)
1398 * improvement: macos/x86 and macos/x86-64 now use mach exception
1399 handlers for dealing with illegal instructions (for trapping and
1400 error handling) and memory protection violations (for GC).
1402 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1403 * new platform: experimental support for x86-64/darwin (MacOS).
1404 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1405 and getnameinfo sockets functions are used instead of gethostbyaddr
1406 and gethostbyname, on platforms where the newer functions are available.
1407 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1409 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1410 don't print the startup banner, but behave as if --noinform was passed
1411 as a command line argument. (thanks to Kevin Reid)
1412 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1413 * optimization: code using alien values with undeclared types is much faster.
1414 * optimization: the compiler is now able to open code SEARCH in more cases.
1415 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1416 * bug fix: using standardized COMMON-LISP special variables as loop
1417 variables no longer signals bogus package lock violations. (reported
1419 * bug fix: declaring local loop variables to be of a range-limited type
1420 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1421 (reported by Andras Simon)
1422 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1423 bugs remain on x86-64.)
1424 * bug fix: the #= and ## reader macros now interact reasonably with
1425 funcallable instances.
1426 * bug fix: type-checks for function arguments were compiled using the
1427 compiler policy settings of the wrong lexical environment (in 1.0.2
1429 * bug fix: SHADOW accepts characters as string designators, as required
1430 by the spec (thanks to Eric Marsden)
1431 * bug fix: fixed GC safety issues when foreign functions are called with
1432 non-base strings as arguments
1433 * bug fix: more consistent error output for fatal-compiler-warnings, like
1435 * bug fix: functions with proper names were showing up as NIL in
1436 backtraces (reported by Edi Weitz, regression in 1.0.2)
1438 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1439 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1440 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1441 produce more efficient code on the x86 if the compiler can determine
1442 that (+ INDEX <constant>) does not require a bounds check and FOO
1443 has an element type at least 8 bits wide.
1444 * bug fix: references to undefined variables are handled the same way
1445 in toplevel forms as in the normal compiler
1446 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1448 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1449 (thanks to Jon Buller)
1450 * improvement: faster compilation times for complex functions
1451 * improvement: added readlink support to SB-POSIX (thanks to Richard
1454 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1455 * improvement: experimental support for mach exception handling on
1456 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1458 * improvement: support for GBK external format.
1459 (thanks to Chun Tian (binghe))
1460 * improvement: the debugger now displays variables that have been closed
1461 over, in code compiled with (DEBUG 2) or higher
1462 * improvement: support for executable cores on NetBSD (thanks to
1464 * new feature: added a RESTART-FRAME debugger command
1465 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1466 be used to provide an implementation for FILE-POSITION on Gray streams
1467 (thanks to Eric Marsden)
1468 * new feature: add syslog, openlog and closelog support to SB-POSIX
1469 (thanks to Richard Kreuter)
1470 * optimization: the function call overhead in code compiled with
1471 a high DEBUG optimization setting is significantly
1472 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1473 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1474 (thanks to Tony Martinez)
1475 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1476 evaluated once (reported by Kevin Reid)
1477 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1478 works again (thanks to Kilian Sprotte)
1479 * bug fix: an error is signaled for tagbodies with duplicate tags
1480 (thanks to Stephen Wilson)
1481 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1482 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1483 stack frames from alien callbacks.
1484 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1485 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1486 65k pages (thanks to David Woodhouse)
1487 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1489 changes in sbcl-1.0.1 relative to sbcl-1.0:
1490 * new platform: FreeBSD/x86-64, including support for threading.
1491 * new feature: the compiler stores cross-referencing information
1492 about function calls (who-calls), macroexpansion (who-macroexpands)
1493 and special variables (who-binds, who-sets, who-references) for code
1494 compiled with (< SPACE 3). This information is available through the
1495 sb-introspect contrib.
1496 * new feature: users may subclass SEQUENCE, and have instances of
1497 these classes interoperate with standard Common Lisp functions if
1498 a number of methods are defined. (This feature is experimental
1499 and the interface subject to change based on feedback from SBCL
1500 users and the general community)
1501 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1502 x86/x86-64, rather than the previous fixed depth of 8
1503 * improvement: another pthread back-end of mutex "pthread-futex".
1504 * bug fix: non-ascii command-line arguments are processed correctly
1505 (thanks to Yaroslav Kavenchuk)
1506 * bug fix: non-required arguments were not passed correctly when a method
1507 defined using DEFMETHOD was called from a mop-generated method using
1508 CALL-NEXT-METHOD (reported by Pascal Costanza)
1509 * bug fix: recursion is now permitted in accessors through
1510 SLOT-UNBOUND. (reported by Pascal Costanza)
1511 * bug fix: an error was signaled at startup if the HOME environment
1512 variable was defined, but had an empty value (reported by Peter Van Eynde)
1513 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1514 signal a PROGRAM-ERROR, not a BUG.
1515 * bug fix: SB-POSIX:READDIR works when built with large file support.
1516 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1518 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1519 for better automatic indentation support. (thanks to Matt Pillsbury)
1520 * optimization: loading generic functions no longer takes O(n^2) time,
1521 proportional to the amount of methods in the generic function
1522 (reported by Todd Sabin and Jeremy Brown)
1523 * optimization: the FIND and POSITION family of sequence functions
1524 are significantly faster on arrays whose element types have been
1526 * improvements to the Windows port:
1527 ** Intermittent heap corruption problems have been fixed. (thanks
1528 to Alastair Bridgewater)
1529 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1531 ** Lisp is able to unwind foreign exception frames from alien
1532 callbacks. (thanks to Alastair Bridgewater)
1534 changes in sbcl-1.0 relative to sbcl-0.9.18:
1535 * improvement: experimental support for threading on FreeBSD/x86.
1536 (thanks to NIIMI Satoshi)
1537 * improvement: runtime option --dynamic-space-size can be used
1538 to set the size of the dynamic space reserved on startup.
1539 * improvement: floating point modes in effect are now saved in
1540 core, and restored on startup.
1541 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1542 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1543 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1544 threads are not running after *SAVE-HOOKS* have run.
1545 * improvement: writes to CLOS instance slots are type-checked in code
1546 compiled with (SAFETY 3)
1547 * improvement: floating-point exception handling on FreeBSD (thanks to
1549 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1550 (thanks to Zach Beane)
1551 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1553 * improvement: added support for the Shift-JIS external format.
1554 (contributed by NIIMI Satoshi)
1555 * improvement: callbacks are supported on Linux/PPC. (thanks to
1557 * bug fix: compiler bug triggered by a (non-standard) VALUES
1558 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1559 * bug fix: file compiler no longer confuses validated and already
1560 dumped structurres. (reported by Kaersten Poeck)
1561 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1562 fill-pointer signals a type-error as required. (thanks to
1564 * bug fix: disassemly of funcallable instances works.
1565 * bug fix: single stepping on PPC.
1566 * bug fix: fix thread-safety problems in the type system (generally
1567 manifesting as nonsensical errors like "STRING is a bad type specifier
1568 for sequences" or "The value 1 is not of type FIXNUM")
1569 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1570 (reported by Josip Gracin)
1571 * bug fix: an error is signaled for attempts to displace arrays with
1572 incompatible element types (thanks to Mario Mommer)
1573 * bug fix: more correct handling of wide characters in debug info
1574 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1575 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1576 and don't cause extra consing
1577 * optimization: MAP and MAP-INTO are significantly faster on vectors
1578 whose elements types have been declared.
1579 * Improvements to SB-SPROF:
1580 ** Support for allocation profiling
1581 ** Reduced profiling overhead, especially for long profiling runs
1582 * Improvements to the Windows port:
1583 ** floating point exceptions are now reported correctly.
1584 ** stack exhaustion detection works partially.
1585 ** more accurate GET-INTERNAL-REAL-TIME.
1586 ** better RUN-PROGRAM behaviour in case of errors in the child
1588 ** PROBE-FILE now simplifies pathnames correctly.
1589 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1590 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1591 1st 1970 (although time timezone data will be unreliable).
1592 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1594 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1595 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1596 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1598 * bug fix: two potential GC deadlocks affecting threaded builds.
1599 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1600 returns T (reported by Anton Kazennikov)
1601 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1602 with non-variable places
1603 * bug fix: remove a race condition in the setting of
1604 funcallable-instance functions, this should make threaded CLOS
1605 code more stable against memory faults.
1606 * bug fix: corruption of specials when unbinding is interrupted by an
1607 asynchronous unwind (reported by Hannu Koivisto)
1608 * improvement: the debugger will now also display local variables that
1609 are only used once, for code compiled with a DEBUG optimization quality
1612 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1613 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1614 * incompatible change: External-format support for FFI calls. The
1615 SB-ALIEN:C-STRING no longer implies an ASCII
1616 external-format. Instead the string is subject to external-format
1617 conversion. Additionally return values of type SB-ALIEN:C-STRING
1618 are by default of type (SIMPLE-ARRAY CHARACTER), not
1619 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1620 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1621 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1622 * incompatible change: SB-EXT package no longer contains the
1623 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1624 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1626 * incompatible change: the single-stepper is no longer available
1627 on Alpha, Mips and Sparc platforms.
1628 * minor incompatible change: the direct superclasses of
1629 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1630 not (STANDARD-OBJECT FUNCTION). This makes the
1631 class-precedence-lists of GENERIC-FUNCTION and
1632 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1634 * minor incompatible change: the default stream external format on
1635 non-unicode SBCL is no longer detected from the locale, but is always
1637 * new feature: Add a version of evaluator that uses an interpreter instead
1638 of the compiler. EVAL still uses the compiler by default, to switch it
1639 to use the interpreter, set the value of the variable
1640 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1641 * minor incompatible change: the single-stepper REPL has been merged
1642 with the normal debugger (see the "Stepping" heading of the debugger help
1643 for more details). The debugger command STEP will no longer switch
1644 to the single-stepper REPL.
1645 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1646 for a type now works.
1647 * bug fix: loading of default sysinit file works. (thanks to Leonid
1649 * bug fix: better detection of circularities in the file-compiler.
1650 (reported by Marco Monteiro)
1651 * bug fix: the CL pathname functions now work with files that have
1652 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1653 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1654 systems with Unix98 pty semantics.
1655 * bug fix: ASDF-INSTALL will now work with bsd tar.
1656 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1658 * bug fix: timers expiring in dead threads no longer cause a
1659 type-error (reported by Paul "Nonny Mouse").
1660 * bug fix: thanks to more lightweight single-stepper instrumentation,
1661 code compiled with (DEBUG 3) will compile and execute significantly faster,
1662 and will have more accurate type-inferencing than before
1663 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1664 whose bindings are modified
1665 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1666 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1667 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1668 CL:SOFTWARE-VERSION with proper version
1670 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1671 * feature: implemented the READER-METHOD-CLASS and
1672 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1673 as specified by AMOP.
1674 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1676 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1677 (thanks to Lutz Euler)
1678 * optimization: hashing of general arrays and vectors has been
1679 improved. (reported by Any Fingerhut)
1680 * enhancement: SB-INTROSPECT is now able to find definitions of
1681 profiled functions. (thanks to Troels Henriksen)
1682 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1683 as well, allowing compiler-macros for SETF-functions to expand.
1684 * enhancement: step-instrumentation no longer wraps known
1685 single-value functions with multiple-value context, allowing
1686 better type inference.
1687 * fixed bug #337: use of MAKE-METHOD in method combination now works
1688 even in the presence of user-defined method classes. (reported by
1689 Bruno Haible and Pascal Costanza)
1690 * fixed bug #339(c): if there are applicable methods not part of any
1691 long-form method-combination group, call INVALID-METHOD-ERROR.
1692 (reported by Bruno Haible)
1693 * fixed bug #361: the :FUNCTION initarg in the protocol for
1694 initialization of methods can now be used to override
1695 internally-produced optimized functions. (reported by Bruno
1697 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1698 system-provided lambda expression no longer cause warnings about
1699 unbound #:|pv-table| symbols.
1700 * bug fix: improved the handling of type declarations and the
1701 detection of violations for keyword arguments with non-constant
1703 * bug fix: potentially erronous calls to PATHNAME and
1704 MERGE-PATHNAMES were being flushed in some cases.
1705 (reported by Richard Kreuter)
1706 * bug fix: compiled calls to TYPEP were mishandling obsolete
1707 instances. (reported by James Bielman and Attila Lendvai)
1708 * bug fix: format strings with ~> without matching ~< no longer
1709 trigger an AVER, but signal an understandable error instead.
1710 (reported by Antonio Martinez)
1711 * bug fix: specifying an output-file without a pathname-type for
1712 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1713 of input-file instead of "fasl". (reported by Robert Dodier)
1714 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1715 (reported by James Y Knight).
1716 * bug fix: compiler-macros-function did not consider the environment
1717 argument for shadowing by local functions.
1718 * bug fix: compiler-macros expansion was inhibited by local INLINE
1720 * bug fix: inline expansions of known functions were subject to
1721 step-instrumentation in high debug policies, leading to problems
1722 with type-inference.
1723 * bug fix: compiler failed to differentiate between different CONS
1724 types in some cases.
1725 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1726 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1727 * bug fix: #368: incorrect use of expressed vs. upgraded array
1729 * bug fix: #306a: more precise unions of array types.
1730 * thread-safety improvements:
1731 ** CONDITION-WAIT could return early on Linux, if the thread was
1732 interrupted and subsequently continued with SIGCONT.
1733 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1735 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1736 * added support for the ucs-2 external format. (contributed by Ivan
1738 * minor incompatible change: pretty printing of objects of type
1739 (cons symbol) is, in the default pprint-dispatch-table, now
1740 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1742 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1743 called later in a class's lifetime, possibly as late as when the
1744 first instance of the class is created. Previously,
1745 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1746 class became finalizeable.
1747 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1748 for multibyte external-format streams. (thanks to "vbzoli")
1749 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1750 with :ALLOCATION :CLASS inherited from superclasses of the
1752 * fixed bug: anonymous classes can now be created using the :NAME
1753 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1754 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1755 * fixed bug: core-files saved with :EXECUTABLE T can again be
1756 executed when SBCL_HOME isn't set. (reported by James Knight)
1757 * fixed bug: toplevel LOCALLY forms with declarations could
1758 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1759 * fixed bug: printing from several different threads using different
1760 values of *print-case* could cause invalid output, due to
1761 some internal special variables of the printer not being bound
1762 thread-locally (reported by Max Mikhanosha)
1763 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1764 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1765 (reported by Pascal Costanza)
1766 * fixed bug: FINALIZE-INHERITANCE is called from
1767 REINITIALIZE-INSTANCE on classes when the class has previously
1768 been finalized, as required by AMOP.
1769 * minor code generation optimizations:
1770 ** better register allocation in CLOS dispatching functions
1771 ** overflow detection when coercing signed bytes to fixnums on x86-64
1772 ** is now implemented with one IMUL instruction instead of three shifts
1773 ** more efficient bit-vector access on x86 and x86-64
1774 ** more efficient access to raw structure slots on x86 and x86-64
1775 * fixed some bugs revealed by Paul Dietz' test suite:
1776 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1777 return its argument.
1779 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1780 * feature: thread support on Solaris/x86, and experimental thread support
1782 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1784 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1785 default method for SLOT-UNBOUND.
1786 * minor incompatible change: prevent the user from specializing the
1787 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1788 somewhat counter to the intent of the protocol, I (CSR) think, and
1789 additionally it just doesn't work in SBCL as currently
1790 implemented, thanks to optimizations (that are always valid for
1791 the other three SLOT-VALUEish functions, but not for the setter).
1792 * minor incompatible change: the :SB-LDB feature is now enabled by
1793 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1794 the low-level debugger.
1795 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1796 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1797 down to an SB-SYS:FD-STREAM, allowing direct communication in
1798 more cases. Temporary files and pipes are still used as a fallback
1800 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1801 on the non-reentrant C functions localtime(3) and gmtime(3).
1802 * bug fix: global optimization policy was not visible in LOCALLY and
1804 * bug fix: class objects can be used as specializers in methods.
1805 (reported by Pascal Costanza)
1806 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1807 with a type error. (reported by blitz_ on #lisp)
1808 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1809 failed with a type error. (reported by Pascal Bourguignon)
1810 * bug fix: merging pathnames against defaults with :DIRECTORY
1811 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1812 (reported by James Y Knight)
1813 * bug fix: saving large (>2GB) cores on x86-64 now works
1814 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1815 constant index and a (simple-array (signed-byte 32)) array
1816 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1817 error (patch by Robert J. Macomber)
1818 * feature: TIME now displays the amount of run-time spent in GC
1819 * bug fix: The debugger now does a better job of respecting
1820 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1821 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1822 workaround for bug 403.)
1823 * bug fix: floating point exception handling now works on FreeBSD
1824 (thanks to NIIMI Satoshi)
1825 * fixed some bugs revealed by Paul Dietz' test suite:
1826 ** MISC.641: LET-conversion were not supposed to work in late
1828 * improvements to the Win32/x86 port:
1829 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1830 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1832 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1833 * new feature: source path information is generated for macro-expansion
1834 errors for use in IDE's like Slime (thanks to Helmut Eller)
1835 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1836 * bug fix: compiler does not loop forever on an invalid type in
1838 * improvement: compilation of most CLOS applications is significantly
1840 * optimization: added a limited bytecode compiler for simple toplevel
1841 forms, speeding up compilation and FASL loading
1842 * bug fix: the statistical profiler now properly distinguishes anonymous
1845 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1846 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1847 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1848 into subclasses of the finalized class. Now user calls to
1849 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1850 system return before any subclasses are finalized.
1851 * minor incompatible change: The reader no longer ignores errors
1852 regarding non-existent packages in #+ and #- feature tests.
1853 * new feature: command line options --no-sysinit, --no-userinit to
1854 inhibit loading the corresponding init files
1855 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1856 for sending data through UDP sockets (thanks to François-René Rideau)
1857 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1858 error is signalled instead (thanks to François-René Rideau)
1859 * minor incompatible change: Error signalling behaviour of lexical
1860 operations violating package locks has changed slightly. Refer to
1861 documentation on package locks for details.
1862 * enhancement: EVAL can process IF-expressions without resorting to the
1864 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1865 constant-folded properly due to missing out-of-line functions.
1866 (reported by Mika Pihlajamäki)
1867 * bug fix: LISTEN sometimes returned T even in cases where no data was
1868 immediately available from the stream
1869 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1870 were reversed. (reported by Levente Mészáros)
1871 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1872 when given a vector displaced to an adjustable array. (reported
1874 * bug fix: derivation of float boundaries from numbers outside the
1875 appropriate float range (reported by John Wiseman)
1876 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1878 * bug fix: precomputing cache entries for generic functions with
1879 some subclasses of specializers as yet invalid does not attempt to
1880 fill a cache line with a negative offset. (reported by Levente
1882 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1883 allow condition class objects as arguments to DOCUMENTATION and
1884 (SETF DOCUMENTATION); only find and set documentation for
1885 structure names for the STRUCTURE doc-type. (suggested by Gary
1887 * improvements to the Win32/x86 port:
1888 ** (user-homedir-pathname) and default initialization file
1889 locations now know about the user's "Documents and Settings"
1890 directory (thanks to Yaroslav Kavenchuk)
1891 ** run-program is implemented (thanks to Mike Thomas)
1892 ** sockets support (thanks to Timothy Ritchey)
1893 ** better backtrace support (thanks to Alastair Bridgewater)
1894 ** sb-grovel supported
1895 ** asdf-install and sb-posix work somewhat
1896 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1897 * improvements to the Solaris/x86 port:
1898 ** works on Solaris 11/Solaris Express
1899 ** floating-point exception handling support
1900 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1901 * fixed some bugs revealed by Paul Dietz' test suite:
1902 ** REMOVE-METHOD returns its generic function argument even when
1903 the method is not one of the generic functions' methods.
1904 ** objects declared as MEMBER types can be admissible arguments to
1905 structure accessors.
1906 ** printing characters should simply be printed by the FORMAT ~:C
1908 ** compiler failure when compiling functions with hairy constant
1909 defaults for optional parameters.
1910 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1911 function, which is already optimized.
1913 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1914 * new platform: experimental support for SBCL x86/Darwin, including
1915 MacOS X 10.4.5 on Intel.
1916 * new feature: Unicode character names are now known to the system
1917 (through CHAR-NAME and NAME-CHAR).
1918 * new feature: the filesystem location of SBCL's core file is
1919 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1920 * minor incompatible change: the contrib modules SB-POSIX and
1921 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1922 this change is to make it easier to distribute
1923 location-independent binaries.
1924 * bug fix: as implied by AMOP, standardized classes no longer have
1925 slots named by external symbols of public packages. (reported by
1927 * optimization: calling functions via constant symbols -- as in
1928 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1929 via the function object as in (FUNCALL #'FOO).
1930 * optimization: on x86 and x86-64, the calling convention for the
1931 case of dealing with an unknown number of values has been altered
1932 to be friendlier to the prediction heuristics implemented,
1933 particularly with reference to CALL and RET pairing. (thanks to
1934 Alastair Bridgewater)
1935 * optimization: on x86, the code for access to thread-local symbol
1936 values and binding/unbinding of thread-local symbols is smaller.
1937 (thanks to Alastair Bridgewater)
1938 * enhancement: CONSTANTP is now able to determine constantness of
1939 more complex forms, including calls to constant-foldable standardized
1940 functions and some special forms beyond QUOTE.
1941 * fixed bug: occasional GC crashes on Solaris/x86
1942 * optimization: x86-64 supports stack allocation of results of simple
1943 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1944 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1945 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1946 (thanks to James Knight)
1947 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1948 by the standard (thanks to Stephen Compall)
1950 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1951 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1952 be used for bundling the runtime and the core file into one
1953 executable binary. This feature is not currently supported on all SBCL
1954 platforms. (thanks to James Bielman and NIIMI Satoshi)
1955 * new feature: a generational or ephemeral garbage collector is now
1956 the default on the PowerPC platform (both Linux and Darwin). The
1957 old Cheney (stop and copy) collector is a build-time option.
1958 (thanks to Cyrus Harmon, after Raymond Toy)
1959 * minor incompatible change: the method by which SBCL finds its
1960 contributed modules has changed; it no longer relies on symbolic
1961 links from an $SBCL_HOME/systems directory, but searches directly
1962 in the subdirectories of $SBCL_HOME.
1963 * enhancement: the dynamic heap size on the Linux/PPC platform is
1964 markedly larger, even using the older Cheney garbage collector.
1965 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1966 certain complicated string types. (reported by Gary King)
1967 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1968 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1969 * fixed bug: DOCUMENTATION on structures no longer signals an
1970 error if no structure type of the right name exists. (reported by
1972 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1973 applicable non-standard primary or :AROUND method on
1974 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1975 no longer cause unbound variable errors. (reported by Kevin Reid)
1976 * fixed bug: metacircle resolution in cases where methods have slots
1977 added before the slots from STANDARD-METHOD. (reported by Jean
1979 * fixed bug: the Power PC assembly code for calling into lisp
1980 sometimes computed its return address wrongly (depending on the
1981 15th bit of the address, fixed at link time). This used to
1982 manifest itself in a segmentation violation while building PCL.
1983 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1984 many others over the years)
1985 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1986 the stack pointer (treated as a signed integer) is negative.
1987 (thanks to Peter van Eynde)
1989 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1990 * new platform: experimental support for the Windows operating
1991 system has been added. (thanks to Alastair Bridgewater)
1992 * fixed several bugs in and robustified the PPC FFI (including
1993 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1994 * bug fix: allow non-simple string symbol names (reported by Paul
1996 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1998 * bug fix: saving a core corrupted callbacks on x86/x86-64
1999 * bug fix: closed a loophole in metacircularity detection and
2000 grounding in the PCL implementation of CLOS.
2001 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
2002 (reported by tomppa on #lisp)
2003 * bug fix: clear the direction flag on Lisp -> C transitions, as
2004 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
2005 (reported by Andrej Grozin and Hendrik Maryns)
2006 * optimization: major improvements to GC efficiency on GENCGC platforms
2007 * optimization: faster implementation of EQUAL
2008 * optimization: emit more efficient opcodes for some common
2009 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
2010 * optimization: several other minor code-generation improvements on x86-64
2011 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
2013 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
2014 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2015 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
2016 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
2017 by ANSI as a generic function, and for consistency (SETF
2018 GENERIC-FUNCTION-NAME) is treated likewise.
2019 * fixed bug #233.b: make constraint propagation notice when a variable
2020 value is changed after it is referenced but before it is used
2021 * fixed bug #296: no more arbitrary behaviour from filenames with
2022 odd characters as --load arguments. (reported by Adam Warner)
2023 * fixed bug #390: :CHARACTER-SET pathname components now work as
2024 expected. (reported by Tim Daly Jr)
2025 * fixed bug #391: complicated :TYPE intersections in slot
2026 definitions no longer cause an error in PCL internals.
2027 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
2028 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
2029 returning the number of octets which would be written to the
2030 file-stream. (thanks to Robert J. Macomber)
2031 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
2032 arguments to the internal stream functions. (thanks to David
2034 * bug fix: stack exhaustion is now handled on the x86/SunOS
2035 platform. (thanks to Daisuke Homma)
2036 * bug fix: more accurate ROOM results on GENCGC platforms
2037 * optimization: improved type inference for arithmetic-for
2038 index variables in LOOP
2039 * optimization: faster floating-point SQRT on x86-64
2040 * fixed some bugs revealed by Paul Dietz' test suite:
2041 ** DOCUMENTATION returns NIL instead of "" for method combinations
2042 that don't have a docstring
2044 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
2045 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2046 GENERIC-FUNCTION-NAME) are no longer generic functions, and
2047 therefore are not customizeable by user code (as seems to be at
2048 least permitted and maybe required by AMOP). As a consolation,
2049 however, the SBCL implementation of these functions now calls
2050 REINITIALIZE-INSTANCE as specified by AMOP.
2051 * bug fix: slot-definition documentation is propagated as per ANSI
2052 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
2053 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
2054 argument, as do eslotd initialization methods. (from Pascal
2055 Costanza's "Closer" project)
2056 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
2057 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
2059 * bug fix: it is now possible to have more than one subclass of
2060 STANDARD-GENERIC-FUNCTION without causing stack overflow.
2061 (reported by Bruno Haible, Pascal Costanza and others)
2062 * bug fix: the dependent update protocol now works for generic
2063 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
2064 and Pascal Costanza)
2065 * bug fix: condition-class instances corresponding to
2066 DEFINE-CONDITION forms are now created eagerly. (reported by
2067 Kalle Olavi Niemitalo on comp.lang.lisp)
2068 * bug fix: floating point printing is more accurate in some
2069 circumstances. (thanks to Simon Alexander)
2070 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
2071 merged with *DEFAULT-PATHNAME-DEFAULTS*.
2072 * bug fix: callbacks on OS X now preserve stack-alignment.
2073 (thanks to Cyrus Harmon)
2074 * enhancement: the x86-64 disassembler is much better at
2075 disassembling SSE instructions. (thanks to Lutz Euler)
2076 * enhancement: CHECK-TYPE on a variable now allows type-inference
2077 (thanks to Vincent Arkesteijn)
2078 * optimization: improved performance of EUC-JP external format.
2079 (thanks to NIIMI Satoshi)
2080 * optimization: performance improvements to IO on file streams of
2081 :ELEMENT-TYPE CHARACTER
2082 * optimization: much faster memory allocation on x86-64
2083 * optimization: faster garbage collections (latency and throughput)
2085 * optimization: faster bignum subtraction and division on x86 and x86-64
2086 (thanks to Lutz Euler)
2087 * optimization: more accurate type-inference for loops that use a
2088 floating point index variable or a negative step.
2090 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2091 * bug fix: add a workaround to SBCL looping infinitely at startup on
2092 Linux kernels with apparently buggy implementations of personality().
2093 (thanks to Svein Ove Aas)
2094 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2095 (thanks to David Lichteblau)
2096 * bug fix: local bindings shadow global symbol macros for
2097 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2098 * optimization: non-open coded uses of numeric comparison operators
2099 (e.g. >) no longer cons when called with more than one parameter
2100 on platforms supporting dynamic-extent allocation.
2101 * enhancement: saving cores with foreign code loaded is now supported
2102 on MIPS/Linux in addition to the previously supported platforms.
2103 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2104 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2105 * bug fix: division by zero in sb-sprof when no samples were collected
2106 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2107 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2108 (thanks to Alexander Kjeldaas)
2109 * new feature: ignoring the return values of destructive functions
2110 like NREVERSE signals a compile-time style-warning.
2111 (thanks to Kevin Reid)
2112 * new feature: the alignment of alien structure fields can be
2113 explicitly specified. (thanks to Cyrus Harmon)
2114 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2115 and dump core on SIGQUIT
2117 ** incompatible change: threads do not inherit values of specials
2118 from their parents (see manual)
2119 ** bug fix: threads stacks belonging to dead threads are freed by the
2120 next exiting thread, no need to gc to collect thread stacks anymore
2121 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2122 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2123 ** bug fix: dynamic variable and thread start related gc lossage
2124 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2126 * fixed some bugs revealed by Paul Dietz' test suite:
2127 ** SUBTYPEP is slightly more accurate on heinously complicated
2128 CONS types where some of the members have uncertain (in the
2129 NIL, NIL sense) type relationships to each other.
2131 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2132 no longer purified unless :PURIFY T is explicitly specified.
2133 ** Non-purified cores are significantly smaller than before
2135 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2136 * new feature: timers based on Zach Beane's excellent timer package
2137 * added support for the following external formats: koi8-u,
2138 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2139 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2140 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2141 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2142 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2143 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2144 * incompatible change: a threaded SBCL will no longer revert to
2145 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2146 * bug fix: interrupts are disabled until startup is complete; no
2147 more sigsegvs when receiving a signal to soon
2148 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2150 * bug fix: add a workaround for the memory randomization features in
2151 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2152 workaround will only be in effect on systems with the proc filesystem
2154 * bug fix: printing objects of type HASH-TABLE signals a
2155 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2157 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2158 have been running for over 50 days. (reported by Gilbert Baumann)
2159 * bug fix: the logic for getting names of functions gets less
2160 confused when confronded with alternate-metaclass
2161 funcallable-instances. (reported by Cyrus Harmon)
2162 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2163 consistent, even on internal alternate-metaclass objects.
2164 * bug fix: flush closure information collected by physical
2165 environment analysis prepass before the main pass. (bug reported
2167 * bug fix: compiler pack phase does not modify a hashtable, which is
2168 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2169 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2170 may be expanded. (reported by Alexander <malishev> on c.l.l)
2171 * bug fix: no more style-warnings when compiling code using LAST
2172 under high-SPEED low-SPACE settings. (reported by David Wallin
2174 * fixed some incompatibilities between SBCL's MOP and the MOP
2176 ** the METAOBJECT class is now implemented;
2177 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2178 STANDARD-OBJECT, as required;
2179 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2180 now compatible (as required by VALIDATE-SUPERCLASS); there
2181 remains a consistency requirement over the presence or absence
2182 of the FUNCTION class in the superclass list of finalized
2183 classes; see the manual for more details;
2184 ** the :AROUND method for COMPUTE-SLOTS on
2185 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2186 requested slot ordering.
2188 ** bug fix: parent thread now can be gc'ed even with a live
2190 ** bug fix: binding a special with PROGV to no value is not
2191 the same as that symbol not having been bound (thanks to
2193 ** bug fix: binding specials is thread safe (thanks to
2195 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2196 and SB-SPROF do not die with 'no handler for signal XX in
2197 interrupt_handle_now(..)' anymore
2198 ** bug fix: WITH-TIMEOUT works with multiple running threads
2199 * fixed some bugs revealed by Paul Dietz' test suite:
2200 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2201 the :method-class keyword argument.
2203 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2204 * new port: the Solaris operating system on x86 processors is now
2205 mostly supported, though some rough edges in the environment
2206 remain. (thanks to Daisuke Homma)
2207 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2208 space for its heap. (thanks to Thiemo Seufer)
2209 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2210 closures. (thanks to Thiemo Seufer)
2211 * minor incompatible change: eof selects abort in the debugger.
2212 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2213 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2215 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2216 deprecated, and will go away in a future revision of SBCL.
2217 * minor incompatible change: GC-ON and GC-OFF are no longer
2218 implemented with a counter, it does not matter how many times gc
2219 is switched on or off
2220 * bug fix: discriminating functions for generic function classes
2221 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2222 make invalid assumptions about method precedence order. (reported
2224 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2225 platforms. (thanks to Thiemo Seufer)
2226 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2227 * bug fix: correct pathname printing with printer escaping is on.
2228 (thanks to Kevin Reid)
2229 * bug fix: complex VOP definitions in "user-space" no longer trigger
2230 package locks. (reported by Zach Beane)
2231 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2232 for funcallable-instance functions, and is no different from
2234 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2235 characters in the current readtable. (reported by Nicholas Neuss)
2236 * bug fix: SUBTYPEP on various CONS types returns more a more
2237 accurate acknowledgment of its certainty.
2238 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2239 lists in some cases. This partially fixes bug 384.
2240 * flush all standard streams before prompting in the REPL and the
2242 * bug fix: signal handling and triggering gc do not conflict
2243 directly anymore, in particular a high frequency sb-sprof does
2244 not prevent gc from running
2245 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2246 approximation for timezone and DST information between the
2247 universal time epoch and the smallest negative 32-bit time_t.
2248 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2249 year 1899 so long as the encoded time is non-negative.
2250 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2251 the correct answer for non-constant rotate amounts when performing
2252 an inline 32-bit rotation.
2254 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2255 there is only one thread in the session
2256 ** bug fix: memory leak for streams created in one thread and
2257 written to in another
2258 ** bug fix: lockup when compiled with gcc4
2259 ** bug fix: race that allows the gc to be triggered when gc is
2261 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2262 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2264 ** bug fix: finished threads are gc'ed properly
2265 * fixed some bugs revealed by Paul Dietz' test suite:
2266 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2267 has arguments with a different set of applicable methods from
2268 the orignal arguments.
2269 ** The type error thrown by MAP now has a correct expected-type
2271 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2272 name a compiled function.
2273 ** Three MISC tests where a large bignum was improperly coerced to
2274 a float that couldn't represent that bignum during type
2275 derivation were fixed.
2276 ** SUBTYPEP can now handle types involving the negation of a
2277 list-form FUNCTION type.
2278 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2279 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2280 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2282 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2283 * New feature: Experimental support for bivalent streams: streams
2284 opened with :element-type :default now allow character and binary
2285 (unsigned-byte 8) I/O
2286 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2287 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2289 * bug fix: no more highly sporadic "couldn't check whether ~S is
2290 readable" when reading a stream and an interrupt hits in the middle
2291 of a select system call
2292 * compiler better recognizes complex arrays (reported by Eduardo
2294 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2295 for variables on SBCL built with linkage-tables. (reported by Luis
2297 * various error reporting improvements.
2298 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2299 (Thanks to Thiemo Seufer)
2300 * bug fix: nested reader invokations work correctly
2301 * bug fix: it is possible to have simultaneous references to foreign
2302 code and foreign data with the same name.
2304 ** added x86-64 support
2305 ** incompatible change: the threading api now works with thread
2306 objects instead of thread ids
2307 ** bug fix: threads are protected from signals and interruption when
2308 starting up or going down
2309 ** bug fix: a race where an exiting thread could lose its stack to gc
2310 ** bug fix: don't halt on infinite error in threads if possible
2311 ** fixed numerous gc deadlocks introduced in the pthread merge
2312 ** bug fix: fixed thread safety issues in read and print
2313 ** bug fix: debugger doesn't hang on session lock if interrupted at
2314 an inappropriate moment
2315 ** bug fix: run-program is now thread safe(r)
2316 ** bug fix: inner with-recursive-lock no longer releases the mutex
2317 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2318 * fixed some bugs revealed by Paul Dietz' test suite:
2319 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2320 EXPECTED-TYPE slots filled.
2321 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2323 ** MAKE-CONDITION accepts classes as type-designators.
2324 ** COMPILE may never return NIL.
2325 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2326 range before calling Unix time functions
2328 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2329 * numerous signal handling fixes to increase stability
2330 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2331 * minor incompatible change: we now correctly canonize default
2332 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2333 as per the MOP, rather than the historical (INITARG INITFUNCTION
2334 INITFORM). (reported by Bruno Haible)
2335 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2336 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2337 for more information.
2338 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2339 pathname is a directory pathname.
2340 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2341 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2343 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2344 :METACLASS argument in addition to a class name. (reported by
2345 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2346 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2347 (reported by Bruno Haible; patch from Gerd Moellmann)
2348 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2350 * bug fix: more cleanups to the floating point exception handling on
2351 x86-64 (thanks to James Knight)
2352 * bug fix: the compiler does not try to constant fold calls of
2353 COERCE and BIT-* functions when they return freshly constructed
2354 objects. (reported by Nikodemus Siivola and Paul Dietz)
2355 * optimization: improved the MIPS versions of generic (in the
2356 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2357 * optimization: direct conversion of (unsigned-byte 32) to floats on
2358 the PowerPC platform.
2359 * optimization: structure instances with raw slots now use less
2360 memory, and probably show better memory locality. (thanks to
2362 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2363 are treated as requests for stack allocation on the x86-64,
2364 Alpha-32, PPC and SPARC platforms.
2365 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2366 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2367 (thanks to Tony Martinez)
2369 ** gcing a dead thread can no longer lead to lockups
2370 ** threads block signals until they are set up properly
2371 ** errno is no longer shared by threads
2372 ** interrupt-thread restores the eflags register on x86
2373 ** fixed some lockups due to gc/thread interaction
2374 * fixed some bugs revealed by Paul Dietz' test suite:
2375 ** invalid dotted lists no longer raise a read error when
2376 *READ-SUPPRESS* is T
2377 ** signal an error if a symbol that names a declaration is used
2378 as the name of a type, or vice versa
2379 ** allow using the (declare (typespec var*)) abbreviation for
2380 (declare (type typespec var*)) with all type specifiers
2381 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2382 ** always return NIL from PPRINT-POP when OBJECT is NIL
2383 ** don't signal errors when pretty-printing malformed LABELS,
2384 FLET or MACROLET forms
2385 ** declarations in a DOLIST body are also in scope for the
2387 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2389 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2392 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2393 * fixed cross-compiler leakages that prevented building a 32-bit
2394 target with a 64-bit host compiler.
2395 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2396 opened with :IF-EXISTS :OVERWRITE.
2397 * fixed bug 281: error for an invalid qualifier in a short-form method
2398 combination method is not signalled until the faulty method is called.
2399 * bug fix: iteration variable type inferrer failed to deal with open
2400 intervals. (reported by Alan Shields)
2401 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2402 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2403 or more runtime options were provided to the sbcl binary.
2404 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2405 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2406 * optimization: there's now a fast-path for fixnum arguments in the
2407 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2408 * optimization: the code generated on x86-64 is more compact thanks
2409 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2410 * bug fix: floating-point exception handling now partly works on
2411 x86-64. (Thanks to James Knight)
2412 * improvement to the MIPS backend from Thiemo Seufer: C-style
2413 64-bit long long arguments and return values to alien functions
2415 * fixed some bugs revealed by Paul Dietz' test suite:
2416 ** the type-error signalled from WARN has a filled-in DATUM slot.
2417 ** the type-error required when a stream is not associated with
2418 a file has the stream as its datum.
2419 ** type-errors on single-floats on x86-64 no longer have
2420 :INVALID-OBJECT as the datum
2421 ** the type-errors signalled for invalid function names now have
2422 a correct expected type
2423 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2424 for structures defined with a :TYPE. Documentation strings for
2425 typed structures are no longer immediately discarded
2426 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2427 broadcast streams and synonym streams. FILE-LENGTH now also works
2428 on broadcast streams.
2430 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2431 * incompatible change: the --noprogrammer option, deprecated since
2432 version 0.7.5, has been removed. Please use the equivalent
2433 --disable-debugger option instead.
2434 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2436 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2437 inoperational for a while now) has been completely removed.
2438 * null lexical environments are now printed as #<NULL-LEXENV>,
2439 significantly reducing the amount of clutter in typical backtraces.
2440 * documentation on weak pointers, finalization, and after GC hooks
2441 has been added to the manual.
2442 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2443 as other specialized array types, is much faster. SUBSEQ and
2444 COPY-SEQ on such arrays have also been sped up.
2445 * optimization: EQL is now more efficient when at least other argument
2446 is known to be of type (OR FIXNUM (NOT NUMBER)).
2447 * fixed bug: the runtime is now less vulnerable to changes in the
2448 size of the SBCL object on OS X, and virtual memory is reserved for
2449 all spaces that need to be at a fixed address.
2450 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2451 * fixed bug: finalizers and after GC hooks that cause consing are now
2453 * fixed bug: compiler error messages and summaries are now printed to
2454 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2455 * fixed inference of the upper bound of an iteration variable.
2456 (reported by Rajat Datta).
2457 * fixed bug 211e: calling local functions with duplicated constant
2458 keyword argument no longer causes a bogus style warning about an
2460 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2461 ftype declaration to be disregarded. (reported by Dave Roberts)
2462 * fixed bug 373: caused by erronous compilation of references to alien
2463 variables in the runtime on ppc/darwin.
2464 * fixed bug 376: CONJUGATE type deriver.
2465 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2466 (reported by Baughn on #lisp)
2467 * a cross-compiler bug on non-x86 platforms has been identified and
2468 fixed. (thanks to Bruno Haible)
2469 * improvements to the MIPS runtime code for increased stability.
2470 (thanks to Thiemo Seufer)
2471 * increased the maximimum compact environment size to allow
2472 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2473 * partial workaround for bug 135: don't name the function we're
2474 calling for hairy cases of EVAL, so as not to accumulate one environment
2475 entry per call to EVAL. (reported by Kevin Reid)
2476 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2477 * optimization: type testing for non-vector arrays should be faster.
2478 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2480 * bug fix: setting 31st element of a bit vector to zero did not work
2482 * bug fix: redefining a class definition which failed due to a
2483 previous accessor / function clash now works (but see BUGS entry
2484 #380 for more problems in this area). (thanks to Zach Beane)
2485 * the long form of DEFINE-METHOD-COMBINATION disables method group
2486 checking when given a single method group with pattern *. (thanks
2488 * on x86 compiler supports stack allocation of results of simple
2489 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2490 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2491 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2492 * fixed some bugs related to Unicode integration:
2493 ** the restarts for recovering from input and output encoding
2494 errors only appear when there is in fact such an error to
2496 * fixed some bugs revealed by Paul Dietz' test suite:
2497 ** MISC.549 and similar: late transformation of unsafe type
2498 assertions into derived types caused unexpected code
2500 ** SCALE-FLOAT type deriver is less wrong.
2501 ** type derivers for EXP, LOG and similar functions compute result
2502 types for complex arguments better.
2503 ** (MISC.563) CONJUGATE type deriver works for very restricted
2505 ** out-of-line type testers for character strings are available.
2506 ** EQUAL compiler transform understands specialness of objects
2508 ** accessing double-floats stored on the stack now works on x86-64.
2509 ** debugger internals could sometimes create invalid lispobjs,
2510 resulting in GC crashes.
2511 ** MISC.548: type check weakening can convert required type into
2513 ** initialization forms for bindings are not in scope of free special
2516 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2517 * incompatible change: thread support for non-NPTL systems has
2518 been removed - locking is buggy and unreliable. A threaded
2519 SBCL build will now warn at startup and refuse to create
2520 new threads, unless futex support is detected in the kernel
2521 * incompatible change: the top level REPL now has only an ABORT
2522 restart associated with it, not TOPLEVEL and ABORT as it used to.
2523 TOP and TOPLEVEL are now available as debugger commands for
2524 returning to the top level.
2525 * incompatible change: forms evaluated in the REPL now use the
2526 global optimization policy.
2527 * incompatible change: user- and system-initialization files are
2528 no longer processed with LOAD, but by READ and EVAL; hence the
2529 global optimization policy, startup package, readtable, etc,
2531 * The .fasl file format number has been incremented because of
2532 various incompatible changes.
2533 * internal entry point details and argument counts no longer appear
2534 in backtraces unless explicitly requested by setting
2535 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2536 * built-in and standard functions no longer have names like "top
2537 level local call to FOO".
2538 * fixed bug 32: functions defined in non-null lexical environments
2539 now have more legible printed representation
2540 * fixed bug 33: functions defined in non-null lexical environemnts
2541 are now more amenable to inspection by INSPECT.
2542 * workaround for bug 354: XEPs no longer appear in backtraces unless
2543 explicitly requested.
2544 * fixed bug: receiving the signal which results from attempting to
2545 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2546 SIGBUS on other *BSDs) on architectures where the C stack is also
2547 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2548 to debugger confusion or wild instability. (reported by Cheuksan
2550 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2551 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2552 (reported by Lutz Euler)
2553 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2554 compiled code. (reported by Paul Dietz)
2555 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2556 specialized parameter to a method to something that is not TYPEP
2557 the specializer is now possible.
2558 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2559 face of package deletion.
2560 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2561 pathname. (reported by David Tolpin on c.l.l)
2562 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2563 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2564 STRING-STREAM and FILE-STREAM, respectively.
2565 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2566 than previously; in addition, its test suite is now run on build.
2567 (thanks to Luke Gorrie)
2568 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2569 * optimization: added a immediate representation for single-floats
2571 * fixed some bugs revealed by Paul Dietz' test suite:
2572 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2573 ** attempting to create a package with a colliding nickname causes
2574 correctable errors to be signalled.
2575 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2576 ** several x86-64 backend bugs related to sign-extension of immediate
2579 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2580 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2581 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2583 * fixed loading of multiply forward-referenced layouts.
2584 (thanks to Cheuksan Wang)
2585 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2586 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2587 (see COMPILE-FILE documentation for details of :PRINT option)
2588 * fixed bugs 19 and 317: fixed-format floating point printing is
2589 more accurate. This also fixes a bug reported by Adam Warner
2590 related to the ~@F format directive.
2591 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2593 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2594 dispatch table if the from-char is a dispatch macro character.
2595 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2596 by the wordsize no longer ignore the last word. (reported by Lutz
2598 * fixed bug in type checking of dynamic-extent variables. (reported
2600 * optimization: sequence traversal functions use their freedom to
2601 coerce function designators to functions.
2602 * optimization: code with many calls to CLOS methods specialized on
2603 CLOS classes has had redundant type checks removed.
2604 * optimization: added declarations to speed up operations that access
2605 the internal character database (for example STRING-UPCASE)
2606 * optimization: comparison operations between floats and sufficiently small
2607 fixnums no longer create extra rationals
2608 * fixed some bugs related to Unicode integration:
2609 ** portions of multibyte characters at the end of buffers for
2610 character-based file input are correctly transferred to the
2611 start of the buffer at the next read.
2612 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2613 passing it through to OPEN.
2614 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2615 argument given to internal calls to OPEN.
2616 ** the built-in comment readers (introduced by character sequences
2617 ";" and "#|") are more forgiving to encoding errors; they will
2618 STYLE-WARN and then attempt to resync the stream at a character
2619 boundary. (thanks to Teemu Kalvas)
2620 * fixed some bugs in the x86-64 port:
2621 ** Negative short int return values from c-calls are sign-extended
2623 ** The stack is aligned to 16-bytes for c-calls, as required by
2624 the ABI. (thanks to Cheuksan Wang)
2625 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2626 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2627 for enums. (thanks to Vincent Arkesteijn)
2628 ** Multiple small optimizations and bugfixes for floating point
2630 * fixed some bugs revealed by Paul Dietz' test suite:
2631 ** Space, Tab, Linefeed, Return and Page have the invalid
2632 secondary constituent character trait.
2633 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2635 ** WITH-INPUT-FROM-STRING should only update the index place on
2637 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2638 works more reliably.
2639 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2640 with one-dimensional arrays with fill pointers.
2641 ** TYPE-OF failed on a complex with an integer realpart and a
2643 ** compiler failure during type inference for the code of form
2644 (IF (EQL X (THE ...)) ...) (MISC.535).
2646 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2647 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2648 platform. The port supports 61-bit fixnums, large memory spaces
2649 and reloading shared object files.
2650 * enhancement: saving cores with foreign code loaded is now
2651 supported on ppc/Darwin in addition to the previously supported
2653 * enhancement: the statistical profiler now walks deeper into the
2654 call stack for more meaningful call-graphs and accrued time
2655 reports (x86/x86-64 only). It also now reports time spent in
2657 * enhancement: it is now possible to trace most individual methods
2658 of a generic function in addition to tracing the generic function
2660 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2662 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2663 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2664 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2665 produces an error. (thanks to Vincent Arkesteijn)
2666 * bug fix: NAMESTRING on pathnames with :WILD components in their
2667 directories works correctly. (thanks to Artem V. Andreev)
2668 * fixed bug 125: compiler preserves identity of closures. (reported
2670 * bug fixed: functions with &REST arguments sometimes failed with
2671 "Undefined function" when compiled with (DEBUG 3). (reported by
2673 * bug fix: overflow during compiling of setting element of a bit
2674 vector with constant index and value. (reported by Timmy Douglas)
2675 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2676 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2677 * on x86 compiler partially supports stack allocation of dynamic-extent
2679 * GO and RETURN-FROM do not check the extent of their exit points
2680 when compiled with SAFETY 0.
2681 * fixed some bugs related to Unicode integration:
2682 ** encoding and decoding errors are now much more robustly
2683 handled; it should now be possible to recover even from invalid
2684 input or output to the terminal. (thanks to Teemu Kalvas)
2685 ** provided a first cut at implementing STRING-TO-OCTETS and
2686 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2687 ** altered the SB-MD5 contributed module slightly, changing the
2688 interface just enough for it to be supportable for builds where
2689 lisp characters are not eight bits.
2690 * fixed some bugs revealed by Paul Dietz' test suite:
2691 ** the FORMATTER-generated functions for ~V[ conditionals require
2692 the correct number of arguments.
2693 ** READ-FROM-STRING returns the mandated second value when applied
2694 to displaced strings.
2695 ** the #\Rubout and #\Backspace characters are treated as invalid
2696 constituent characters by the tokenizer.
2698 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2699 * new feature: reloading changed shared object files with
2700 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2701 * new feature: references to foreign variables and functions
2702 can now be compiled and loaded before the corresponding shared
2703 object file is loaded, as long as the foreign definitions are
2704 available at runtime.
2705 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2706 supported. (thanks to Dan Debertin)
2707 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2708 just keywords. (thanks to Vincent Arkesteijn)
2709 * fixed bug #331: structure-class instances corresponding to
2710 DEFSTRUCT forms are now created eagerly.
2711 * fixed bug #345: backtraces from calls to undefined functions work
2712 on x86 as well. Related bug #61 is now also partially fixed on x86
2713 (backtraces from throws to unknown catch tags.)
2714 * bug fix: lambda-list parsing is now stricter vrt. order and number
2715 of lambda-list keywords.
2716 * bug fix: as specified by AMOP, an error is signalled if a
2717 class-option appears multiple times in a DEFCLASS form. (reported
2719 * bug fix: RANDOM can be compiled when the compiler derives the type
2720 of its numeric argument as a disjoint set of small integers.
2721 (reported by Paul Dietz)
2722 * bug fix: starting a core saved with shared objects loaded when
2723 those objects are not available no longer causes threaded SBCL to
2724 hang. (reported by Sean Ross)
2725 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2727 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2728 argument, which must always be NIL. (reported by Kalle Niemitalo)
2729 * bug fix: printing 1.0d+23 no longer results in an error.
2730 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2731 * bug fix: structure slot setters preserve evaluation order. (thanks
2733 * fixed some bugs related to Unicode integration:
2734 ** RUN-PROGRAM can allow its child to take input from a Lisp
2735 stream. (reported by Stefan Scholl)
2736 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2737 locales. (reported by Ken Causey)
2738 * fixed some bugs revealed by Paul Dietz' test suite:
2739 ** INCF, DECF and REMF evaluate their place form as specified in
2741 ** FORMATTER expands ~{ ~} iteration directives with V or #
2742 parameters correctly.
2743 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2744 consequent uses no arguments correctly.
2745 ** the system has a partial understanding of the (COMPLEX RATIO)
2747 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2750 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2751 * new feature: a build-time option (controlled by the :SB-UNICODE
2752 keyword feature, enabled by default) for building the system with
2753 support for the entire 21-bit character space defined by the
2755 * new feature: the system now has rudimentary external-format
2756 support; the primary user-visible change at this time is that
2757 characters with the high bit set (such as lower-case-e-acute) will
2758 print correctly to a terminal in a UTF-8 environment.
2759 * minor incompatible change: BASE-CHAR no longer names a class;
2760 however, CHARACTER continues to do so, as required by ANSI.
2761 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2762 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2763 has been moved to the SB-EXT package (temporarily re-exported from
2765 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2766 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2768 * fixed bug #351: better error-handlind and reporting for malformed
2770 * fixed bug #350: bignum-printing is now more memory-efficient,
2771 allowing printing of very large bignums, eg. (expt 2 10000000).
2772 (reported by Bruno Haible)
2773 * fixed bug #302: better primitive-type selection for intersection
2775 * fixed bug #308: non-graphic characters now all have names, as
2776 required. (reported by Bruno Haible)
2777 * bug fix: redefining a class with different superclasses now correctly
2778 removes it from the direct-subclasses of its previous superclasses.
2779 (reported by David Morse)
2780 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2781 new value now works. (reported by Bruno Haible)
2782 * bug fix: correct canonicalization of multiple non-standard slot
2783 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2784 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2785 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2786 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2787 now exists, an signals an error.
2788 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2789 during macroexpansion for non-iteration variables as well. (reported
2790 by Bruno Haible for CMUCL)
2791 * bug fix: Cyclic structures and unprintable objects in compiler
2792 messages no longer cause errors. (reported by Bruno Haible)
2793 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2794 and READ-FROM-STRING all now return a primary value of NIL if
2795 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2796 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2797 (reported by Bruno Haible for CMUCL)
2798 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2799 do not designate a condition. (reported by Bruno Haible for
2801 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2802 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2803 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2804 permitting resolution in favour of any of the conflicting symbols.
2805 (reported by Bruno Haible for CMUCL)
2806 * bug fix: EQUAL compiler optimizations is less aggressive on
2807 strings which can potentially compare true despite having distinct
2808 specialized array element types.
2809 * bug fix: unit enumerations can be defined without dividing by
2810 zero. (thanks to Vincent Arkesteijn)
2811 * FORMAT compile-time argument count checking has been enhanced.
2812 (report from Bruno Haible for CMUCL)
2813 * a partial workaround for the bug 262: the compiler does not try to
2814 inline-expand a local function doing RETURN-FROM from a deleted
2815 BLOCK. (thanks to Peter Denno for the bug report and to David
2816 Wragg for the simple test case)
2817 * fixed some bugs revealed by Paul Dietz' test suite:
2818 ** RENAME-PACKAGE allows all package designators as new package
2820 ** constraint propagation and lambda variable substitution are
2821 more cautious in dealing with partially deleted code.
2822 ** compiler sometimes lost reoptimization passes.
2823 ** CERROR, given a condition as condition designator, passes its
2824 remaining arguments to the continue format control without
2826 ** Case-altering FORMAT directives work correctly on non-ASCII
2828 ** The REAL type specifier handles bounds outside the floating
2829 point ranges without signalling FLOATING-POINT-OVERFLOW.
2830 ** Functions with IR1-transformations can create intercomponent
2831 references to global functions.
2832 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2834 ** FORMAT ~R treats a nil value for its first parameter correctly.
2836 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2837 * enhancement: saving cores with foreign code loaded is now
2838 supported on x86/NetBSD and sparc/Linux in addition to the previously
2839 supported platforms.
2840 * bug fix: on some platforms repeated installations caused multiple
2841 copies of HTML documentation to be installed -- should not happen
2842 any more. (reported by Stefan Scholl)
2843 * bug fix: parsing self-recursive alien record types multiple times
2844 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2845 original patch by Helmut Eller for CMUCL)
2846 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2847 (thanks to Richard Kreuter)
2848 * bug fix: defining classes whose accessors are methods on existing
2849 generic functions in other (locked) packages no longer signals
2850 bogus package lock violations. (reported by François-René Rideau)
2851 * bug fix: special variables as DEFMETHOD parameters no longer have
2852 associated bogus type declarations. (reported by David Wragg and
2854 * bug fix: read-write consistency on streams of element-type
2855 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2856 * bug fix: redefiniton of the only method of a generic function with
2857 no DEFGENERIC no longer emits a full WARNING. In addition,
2858 redefinition of generic functions with no DEFGENERIC to an
2859 incompatible lambda list now signals an error. (thanks to Zach
2861 * bug fix: DEFGENERIC now works even when there's a function of the
2862 same name in an enclosing lexical environment. (thanks to Zach
2864 * fixed compiler failure, caused by instrumenting code during
2865 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2866 * optimization: added loop analysis and improved register allocation
2867 * fixed some bugs revealed by Paul Dietz' test suite:
2868 ** POSITION on displaced vectors with non-zero displacement
2869 returns the right answer.
2870 ** (SIMPLE-STRING) is a valid type specifier for sequence
2872 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2874 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2875 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2877 ** The justification version of the FORMAT ~< directive treats
2878 non-zero minpad parameter correctly.
2880 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2881 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2882 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2883 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2884 the supported interface.
2885 * new feature: Single-stepping of code compiled with DEBUG 2 or
2886 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2887 * new feature: saving cores with foreign code loaded is now
2888 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2889 Timothy Moore's work for CMUCL)
2890 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2891 parameters to * instead of NIL if no initform is supplied.
2892 (reported by Johan Bockgaard)
2893 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2894 correctly when FUNCALL appears as the car of the form. Note:
2895 despite this FUNCALL forms are not currently subject to
2896 compiler-macro expansion. (port of Raymond Toy's fix for the
2897 same from CMUCL, reported by Johan Bockgaard)
2898 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2899 (thanks for Teemu Kalvas)
2900 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2901 case instead of terminating the iteration (thanks for Julian
2902 Squires, Sean Champ and Raymond Toy)
2903 * bug fix: incorrect expansion of defgeneric that caused a style
2904 warning. (thanks for Zach Beane)
2905 * bug fix: slot accessor effective method computation works properly
2906 for classes with multiple non-standard applicable methods on
2907 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2908 * on x86 compiler supports stack allocation of results of LIST and
2909 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2910 CMUCL implementation by Gerd Moellmann)
2911 * fixed some bugs revealed by Paul Dietz' test suite:
2912 ** FORMAT strings with both the ~<~:;~> form of the justification
2913 directive and pretty-printing directives cause an error.
2915 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2916 * incompatible change: the internal functions
2917 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2918 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2919 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2920 instead of the old functions.
2921 * new feature: on platforms where "dladdr" is available foreign
2922 function names now appear in backtraces. (based on Helmut Eller's
2924 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2925 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2927 * documentation: ASDF manual is now distributed with SBCL.
2928 * bug fix: SBCL can now load its contributed modules with REQUIRE
2929 even if the system-provided entries have been removed from
2930 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2932 * bug fix: backtraces involving undefined functions or assembly
2933 routines are more informative. (thanks to Brian Downing)
2934 * bug fix: mutually referent alien structures now work correctly.
2935 (reported by Rick Taube)
2936 * bug fix: structures defined by WITH-ALIEN can be referred to
2937 within other definitions in the same WITH-ALIEN.
2938 * bug fix: division operators (MOD, TRUNCATE and the like) with
2939 constant zero divisors and integer dividends no longer generate
2941 * bug fix: provide default methods for INPUT-STREAM-P and
2942 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2943 * bug fix: improve the detection and resolution of MOP metacycles
2944 (where effective method computation depends on the generic function
2945 having its effective method computed). (reported by Bruno Haible)
2946 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2947 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2948 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2949 * bug fix: pathnames with relative directory components can be
2950 represented relative to default pathnames.
2951 * optimization: in taking the GCD of bignums, reduce the two bignums
2952 to approximately the same size (using Euclid's algorithm) before
2953 applying the more sophisticated binary GCD. (thanks to Juho
2955 * optimization: COUNT on bitvectors now operates word-at-a-time.
2956 * optimization: ASH with a positive, but not necessarily constant,
2957 (leftwards) shift, when in a modular context, is compiled to a
2959 * fixed some bugs revealed by Paul Dietz' test suite:
2960 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2961 if the corresponding argument is NIL.
2962 ** FORMAT directives accepting a minpad parameter treat negative
2964 ** literal commas as character parameters to FORMAT directives are
2966 ** literal spaces directly after ~<Newline> directives within a
2967 format-logical-block (~:< ~@:>) do not induce :FILL-style
2968 conditional newlines.
2969 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2970 as specified: it no longer includes conditional newlines).
2971 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2973 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2974 :PER-LINE-PREFIX argument does not evaluate to a string.
2976 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2977 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2978 the manual for details; for now, package locks can be disabled by
2979 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2980 if no major problems are found then it is likely that they will be
2981 compiled in unconditionally.
2982 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2983 unsupported operators on all platforms. To load a shared library
2984 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2985 object file, link it into a shared library outside of SBCL and
2986 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2988 * bug workaround: ROOM T and the GC were not working together
2989 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2990 reporting facility aren't preserved. That reporting has been
2991 disabled, so now until and unless someone figures out how to make it
2992 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2993 * minor incompatible change: as threatened around sbcl-0.8.0, the
2994 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2995 The symbols are also exported from SB-PCL for backwards
2996 compatibility, but more so than before SB-PCL should be treated as
2997 an implementation-internal package.
2998 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
3000 * the SB-SPROF contrib now works on (most) non-x86 architectures.
3001 It is known as of this release not to work on the Alpha, however.
3002 * fixed bug #167: errors signalled due to illegal syntax in method
3003 bodies are now more legible.
3004 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
3005 designators and can hence be used with TYPEP.
3006 * fixed bug #333: CHECK-TYPE now ensures that the type error
3007 signalled, if any, has the right object to be accessed by
3008 TYPE-ERROR-DATUM. (reported by Tony Martinez)
3009 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
3010 5.1.2.3 for multiple-value place subforms. (reported by Kalle
3012 * fixed bug #334: programmatic addition of slots using specialized
3013 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
3014 and :ALLOCATION :CLASS slots.
3015 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
3016 float exponents. (rereported by Peter Seibel)
3017 * fixed a bug: #\Space (and other whitespace characters) are no
3018 longer considered to be macro characters in standard syntax by
3019 GET-MACRO-CHARACTER.
3020 * fixed bug: initialization of condition class metaobjects no longer
3021 causes an instance of the condition to be created. (reported by
3023 * fixed bug: it is now possible to have slots such that
3024 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
3025 description is neither :INSTANCE nor :CLASS.
3026 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
3027 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
3028 SLOT-VALUE-USING-CLASS), and no longer causes errors with
3029 non-standard SLOT-DEFINITION-ALLOCATION values.
3030 * fixed bugs: various race conditions handling exiting threads.
3031 CL-PPCRE's thread torture test now passes.
3032 * fixed arguably-a-bug: GC time proportional to number of threads in
3033 system even when most of them are idle
3034 * optimization: improved performance of BIT and SBIT on bit-vectors.
3035 * .fasl file incompatibility: The fasl file version number has
3036 been incremented because of changes associated with package locks.
3038 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
3039 * minor incompatible change: the system no longer provides
3040 optimization hints (conditions which are TYPEP
3041 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
3042 mode; these hints are still emitted when higher SPEED optimization
3044 * new contrib module: a sampling profiler (profiling by statistical
3045 sampling, rather than by instrumenting functions) is available as
3046 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
3047 * the behaviour of the standard function ED is now customizeable by
3048 third parties through a hook variable: see ED's documentation
3049 string for information on the protocol.
3050 * the compiler no longer emits efficiency notes for (FUNCALL X)
3051 when the type of X is uncertain under default optimization
3053 * fixed bug 276: mutating a binding of a specialized parameter to a
3054 method to something that is not TYPEP the specializer is now
3056 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
3057 exhibit the required behaviour on the x86 platform. (thanks to
3058 Peter van Eynde, Eric Marsden and Bruno Haible)
3059 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
3060 even for difficult arguments. (reported by Peter Graves)
3061 * fixed bug 141a: the backquote printer now descends quoted
3063 * fixed another bug in backquote printing: no more destructive
3064 modification of the form's list structure. (reported by Brian
3066 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
3067 move between its address being taken and the call to
3068 interrupt_thread, fixing a crashing race condition.
3069 * the SB-POSIX contrib implementation has been adjusted so that it
3070 no longer exhibits ridiculously poor performance when constructing
3071 instances corresponding to C structs.
3073 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
3074 * minor incompatible change: the sb-grovel contrib now treats C
3075 structures as alien (in the sense of SB-ALIEN) objects rather than
3076 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
3077 has implications for memory management of client code
3078 (sb-grovel-returned objects must now be manually managed) and for
3079 type safety (alien objects now have full types).
3080 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
3081 used to control emission of compiler diagnostics, rather than the
3082 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
3083 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
3084 quality should be considered deprecated.
3085 * install.sh now installs the user manual as well
3086 * (not quite a new documentable feature, but worth considering in
3087 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3088 beginnings of a semantically meaningful condition hierarchy is
3089 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3091 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3092 designator as the defaults argument.
3093 * fixed bug: Displaced arrays whose displaced-to array has become
3094 too small now cause ARRAY-DIMENSION to signal an error, providing
3095 for safer bounds-checking. (reported by Bruno Haible)
3096 * fixed bug: DEFCLASS slot definitions with identical :READER and
3097 :WRITER names now signal a reasonable error. (reported by Thomas
3099 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3101 * fixed bug: Invalid slot specification errors now print correctly.
3102 (thanks to Zach Beane)
3103 * fixed bug 320: Shared to local slot value transfers in class
3104 redefinitions now happen corrently from superclasses as
3105 well. (reported by Bruno Haible)
3106 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3108 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3109 improper lists correctly. (reported by Bruno Haible)
3110 * fixed bug 313: source-transform for <fun-name> was erroneously
3111 applied to a call of a value of a variable with name <fun-name>.
3112 (reported by Antonio Menezes Leitao)
3113 * fixed bug 307: The obsolete instance protocol ensures that
3114 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3115 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3116 remove dead unknown-values globs from the middle of the stack.
3117 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3118 easier to resume long computations after using *BREAK-ON-SIGNALS*
3119 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3120 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3121 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3122 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3123 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3124 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3126 * fixed a (fairly theoretical) bug in string printing: if
3127 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3128 does not have array-element-type equal to the most general string
3130 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3131 function name. (thanks to Nikodemus Siivola)
3132 * fixed bug: compile-time format string checker failed on
3133 non-closed ~{. (reported by Thomas F Burdick)
3134 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3135 don't warn on legal loop code involving more than one aggregate
3136 boolean. (thanks to Nikodemus Siivola)
3137 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3138 in keyword position to MAKE-PACKAGE was not being reported as
3140 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3141 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3143 * fixed bugs 280 and 312: the checking for multiple definitions in a
3144 file is less likely to become confused by uses of inline
3146 * fixed bug: the #S reader macro performs the keyword coercion
3147 specified for slot names. (reported by Kalle Niemitalo)
3148 * fixed bug: lambda lists may contain symbols whose names start with
3149 & but are not lambda-list-keywords; their occurrence triggers a
3151 * fixed bug 321: define-method-combination argument lambda lists do
3152 not cause invalid code to be generated when &optional or &aux
3153 variables have default values. (reported by Bruno Haible)
3154 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3155 have CLOS classes; this appears also to have fixed reported
3156 instability in stack exhaustion detection.
3157 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3158 MUFFLE-WARNING when no associated restart is present is now
3160 * optimization: rearranged the expansion of various defining macros
3161 so that each expands into only one top-level form in a
3162 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3164 * optimization: used a previously little-used slot in symbols to
3165 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3167 * fixed some bugs revealed by Paul Dietz' test suite:
3168 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3170 ** Fixed an optimization bug related to inheritance of initforms
3171 from local to shared slots.
3172 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3173 ** CAST optimizer forgot to flush argument derived type.
3174 ** print/read consistency on floats is now orders of magnitude
3175 more likely. (thanks also to Bruno Haible for a similar report
3177 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3178 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3179 for objects of type REAL. Make it so.
3180 ** SXHASH is specified (infelicitously) to respect similarity,
3181 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3182 it so. (thanks to Markus Ziegler)
3183 ** on the Alpha, the compiler succeeds in compiling functions
3184 returning a known number of arguments greater than 63.
3185 ** fixed handling of invalid NIL arguments in keyword position
3187 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3189 ** provided a readably-printable representation for RANDOM-STATE
3191 ** ensured that pathnames, where they have a namestring, always
3192 print using #P"..." syntax.
3194 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3195 * Support for the forthcoming 2.0 version of the NetBSD kernel
3196 running on x86 hardware has been added. (thanks to Perry
3197 E. Metzger most immediately, and others for their past work)
3198 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3199 current release version; SBCL's OpenBSD support had been broken
3200 since about the time of OpenBSD's switch to ELF binary format.)
3201 * [placeholder for DX summary]
3202 ** user code with &REST lists declared dynamic-extent, under high
3203 speed or space and low safety and debug optimization policy.
3204 * The manual has been converted to Texinfo format and the debugger
3205 chapter from the cmucl manual has been added.
3206 * A facility has been added to extract documentation strings from
3207 sbcl and store them as Texinfo-formatted snippets for inclusion in
3208 the manual (via Texinfo's @include directive)
3209 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3210 the test case to Dave Roberts)
3211 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3212 pointers. (reported by Sean Ross)
3213 * bug fix: PROFILE output is printed nicely even for large numerical
3214 values. (thanks to Zach Beane)
3215 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3216 greater than 32 handle EOF correctly.
3217 * bug fix: on X86 an immediate argument of the IMUL instruction is
3218 correctly printed in disassembly. (thanks to Lutz Euler)
3219 * bug fix: class slots in redefined classes preserve their old
3220 values. (thanks to Bruno Haible and Nikodemus Siivola)
3221 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3222 succeeds. (reported by Marco Baringer)
3223 * fixed some bugs revealed by Paul Dietz' test suite:
3224 ** READ-SEQUENCE now works on ECHO-STREAMs.
3225 ** RATIONALIZE works more according to its specification. (thanks
3228 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3229 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3230 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3231 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3232 should matter to you only if you rebind the printer control
3233 variables and then find you want different bindings in the
3234 debugger than in the ordinary execution of your program.)
3235 * The runtime build system has been tweaked to support building
3236 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3237 assembler and linker. (thanks to Nikodemus Siivola)
3238 * Unbound, undefined, undeclared variables now trigger full
3239 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3240 more often programmer error than deliberate exploitation of undefined
3242 * optimization: the hash algorithm for strings has changed to one
3243 that is less vulnerable to spurious collisions. (thanks to Juho
3245 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3246 needless bounds checking. (thanks to Juho Snellman)
3247 * optimization: implemented multiplication as a modular
3248 (UNSIGNED-BYTE 32) operation on the PPC backend.
3249 * fixed some bugs revealed by Paul Dietz' test suite:
3250 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3252 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3253 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3254 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3256 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3258 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3260 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3261 their output stream on EOF from read.
3262 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3263 have been read to end-of-file.
3264 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3266 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3267 description of determination of which consecutive characters
3269 ** Printing the "Space" character with escaping on now yields "#\\ ",
3270 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3271 ** Reading floating-point numbers with *READ-BASE* set to a number
3272 less than 10 works correctly.
3273 ** Reading floating-point numbers with *READ-BASE* set to a number
3274 more than 10 works correctly.
3275 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3276 the readtable currently in effect.
3278 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3279 * minor incompatible change: parsing of namestrings on a physical
3280 (Unix) host has changed; numbers after the final #\. in a
3281 namestring are no longer interpreted as a version field. This is
3282 intented to be largely invisible to the user, except that the
3283 meaning of the namestring "*.*.*" has changed: it now refers to a
3284 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3285 should usually be replaced by
3286 (make-pathname :name :wild :type :wild :version :wild)
3287 with the added benefit that this is more likely to be portable.
3288 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3289 option to OPEN now signals an error if the file being opened
3290 exists; this may have an impact on existing code.
3291 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3292 signals received, on the PowerPC platforms (both Linux and
3293 Darwin). (thanks to Pierre Mai for pointing out the location of
3295 * several fixes on OS X: The system now builds and runs cleanly on
3296 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3297 (thanks to Brian Mastenbrook)
3298 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3299 second-resolution: integer multiples of 1/3600 between -24 and 24.
3300 (thanks to Vincent Arkesteijn)
3301 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3302 type when called with 1 argument; PEEK-CHAR checked type of
3303 PEEK-TYPE only after having read first character from a
3304 stream. (reported by Peter Graves)
3305 * bug fix: the garbage collector now has much better locality
3306 behaviour, and in particular no longer treats all memory as being
3307 exhausted when many small objects point to each other in a deeply
3309 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3310 recognized as being TYPEP their class.
3311 * bug fix: the PUSHNEW documentation string has been corrected.
3312 (thanks to Vincent Arkesteijn)
3313 * bug fix: defaulting of the value for the last of an atypically
3314 large number of multiple values being bound was not being
3315 performed correctly on the Alpha or PPC platforms
3316 * optimization: implemented multiplication as a modular
3317 (UNSIGNED-BYTE 32) operation on the x86 backend.
3318 * optimization: SEARCH on simple-base-strings can now be open-coded.
3319 (see also contrib/compiler-extras.lisp for inspiration for
3320 teaching the compiler about the Boyer-Moore algorithm).
3321 * value, returned by MAX (and MIN) called with several EQUALP, but
3322 not EQL, arguments now does not depend on compiler settings.
3323 * fixed some bugs revealed by Paul Dietz' test suite:
3324 ** in stack analysis liveness information is propagated from
3325 non-local entry points.
3326 ** pathwise CAST removing failed when the CAST node did not start
3328 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3329 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3331 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3332 host is already defined.
3333 ** RENAME-FILE works on streams instead of signalling an internal
3335 ** PEEK-CHAR uses the current readtable when determining whether
3336 or not a character is whitespace.
3337 ** MERGE-PATHNAMES handles the case when the pathname does not
3338 specify a name while the default-pathname specifies a version.
3339 ** Pathnames now stand a better chance of respecting print/read
3341 ** Attempting to use standardized file system operators with a
3342 pathname with invalid :DIRECTORY components signals a
3344 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3346 ** DIRECTORY on logical pathnames is more correct.
3347 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3348 signal a TYPE-ERROR if their argument is not a stream.
3349 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3350 designator argument does not designate a stream.
3351 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3352 examining the synonym.
3353 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3355 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3356 element-type larger than ([UN]SIGNED-BYTE 32).
3358 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3359 * When built with the :SB-FUTEX feature, threaded builds now take
3360 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3361 for faster/more reliable mutex and condition variable support.
3362 * Incompatible change (but one you probably shouldn't have been using
3363 anyway): the interface and code for arbitrating between multiple
3364 threads in the same user session has been redesigned.
3365 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3366 error when called without an explicit environment argument.
3367 (thanks to Nikodemus Siivola)
3368 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3369 confused about their position. (thanks to Adam Warner and Gerd
3371 * bug fix: Pretty printing backquoted forms with unquotations in the
3372 argument list position of various code constructs such as LAMBDA
3373 now works correctly. (reported by Paul Dietz)
3374 * bug fix: Pretty printing unquotations no longer loses all
3375 stream position information.
3376 * optimization: performance of string output streams is now less
3377 poor for multiple small sequence writes.
3378 * optimization: performance of CSUBTYPEP in the presence of complex
3379 expressions involving CONS and NOT many times has been improved.
3380 (reported by Paul Dietz)
3381 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3383 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3384 streams for the REPL, for the debugger, and so on. (thanks to
3386 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3387 * fixed some bugs revealed by Paul Dietz' test suite:
3388 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3389 is ignored if it is too small, rather than propagating through
3390 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3391 ** extremely complex negations of CONS types were not being
3392 sufficiently canonicalized, leading to inconsistencies in
3394 ** VALUES tranformer lost derived type.
3396 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3397 * fixed a bootstrapping bug: the build process no longer assumes
3398 that the various BOOLE-related constants have the same value in
3399 host and target lisps. (noted by Paul Dietz' test suite on an
3400 SBCL binary built from CLISP)
3401 * The system can now be dynamically linked on the MIPS platform,
3402 which enables dynamic loading of foreign code from Lisp. (thanks
3403 to Ralf Baechle for discussions on the MIPS ABI)
3404 * The system now records debugging information for its own source
3405 files in a filesystem-position-independent manner, relative to
3406 the "SYS" logical host.
3407 * fixed a compiler bug: MV-LET convertion did not check references
3408 to the "max args" entry point. (reported by Brian Downing)
3409 * tweaked disassembly notes to be less confident about proclaiming
3410 some instruction as an LRA. (thanks to Brian Downing)
3411 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3412 now each have their own history, command character, and other
3413 characteristics. (thanks to David Lichteblau)
3414 * fixed some bugs revealed by Paul Dietz' test suite:
3415 ** compiler failure in compiling LOGAND expressions including a
3417 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3418 shift greater than 32.
3419 ** FLUSH-DEST did not mark blocks for type check regeneration.
3420 ** HANDLER-CASE failed to accept declarations in handler clauses
3421 in some circumstances.
3423 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3424 * New code in contrib/sb-introspect (still probably not entirely
3425 stable yet) provides some support for smart Lisp development
3426 environments like SLIME.
3427 * The conditions signalled for errors occurring when loading .fasl
3428 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3429 in a way which should help ASDF recover gracefully.
3430 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3431 stuff has been cleaned up. If you code contrib/ stuff, this might
3432 affect you, and you can look at contrib/README, contrib/STANDARDS,
3433 and/or the 0.8.4.27 diff to check.
3434 * In full calls the compiler now does not generate checks for declared
3435 argument types for all arguments.
3436 * various threading fixes
3437 ** and some experimental patches which didn't make it into
3438 the main tree for this release, but which are shipped in
3439 contrib/experimental-thread.patch as a possible fix for some
3440 failures (deadlock, spinning...) in GC-intensive multithreaded
3442 * fixed PPC build problem (source code incompatibility of different
3443 library versions): added offsetof() hackery which attempts to divine
3444 where glibc maintainers put uc_mcontext today
3445 * fixed bug 282: compiler does not trust type assertions while passing
3446 arguments to a full call.
3447 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3448 &OPTIONAL VALUES type parameter.
3449 * fix bug 214: algorithm for noting rejected templates is now more
3450 similar to that of template seletion. (also reported by rydis on
3452 * fixed bug 141b: printing backquoted information readably and prettily
3453 inserts a space where necessary.
3454 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3455 (CHARACTER 10) are now reported as errors, rather than propagated
3456 as unknown types. (reported by piso on #lisp)
3457 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3458 with values NIL and :ERROR. (thanks to Milan Zamazal)
3459 * fixed bug 191c: CLOS now does proper keyword argument checking as
3460 described in CLHS 7.6.5 and 7.6.5.1.
3461 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3462 longer raise an error; further, using a list as a for-as-arithmetic
3463 counter now raises a meaningful error.
3464 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3465 now understood by sequence creation functions such as MAKE-SEQUENCE
3467 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3468 a STRING-INPUT-STREAM.
3469 * compiler enhancement: SIGNUM is now better able to derive the type
3471 * type declarations inside WITH-SLOTS are checked. (reported by
3473 * fixed some bugs revealed by Paul Dietz' test suite:
3474 ** incorrect optimization of TRUNCATE for a positive first
3475 argument and negative second.
3476 ** compiler failure in let-convertion during flushing dead code.
3477 ** compiler failure while deriving type of TRUNCATE on an
3478 interval, containing 0.
3479 ** ASH of a negative bignum by a negative bignum count now returns
3481 ** intersection of CONS types now canonicalizes properly, fixing
3482 inconsistencies in SUBTYPEP.
3484 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3485 * incompatible change: The --disable-debugger command line
3486 option now clobbers the debugger at a more fundamental
3487 level, by redefining #'INVOKE-DEBUGGER instead of by
3488 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3489 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3490 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3491 enter the debugger REPL and then suffer recursive errors
3492 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3493 while under the new system, BREAK in --disable-debugger mode
3494 terminates the system just as an unhandled error would.
3495 * fixed compiler performance when processing loops with a step >1;
3496 * bug fix: DOCUMENTATION now retrieves generic function
3497 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3498 support has been systematized, and now supports the methods
3499 specified by ANSI, along with a default method and a method for
3500 slot documentation. (reported by Nathan Froyd)
3501 * bug fix: effective methods associated with a generic function are
3502 no longer cached over a change of that generic function's method
3503 combination. (reported by Andreas Fuchs)
3504 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3505 names a non-existent directory. (thanks to Andreas Fuchs)
3506 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3507 the PPC platform, lead to stack corruption; this has been fixed.
3508 (reported by Rainer Joswig)
3509 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3510 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3511 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3512 arithmetic in general, is now much more reliable on the PPC
3514 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3515 platform now returns the right answer.
3516 * optimization: restored some effective method precomputation in
3517 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3518 precomputation is now tunable.
3519 * optimization: compiler-internal data structure use has been
3520 reviewed, and changes have been made that should improve the
3521 performance of the compiler by about 20%.
3522 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3523 simple-base-strings and simple-bit-vectors is improved.
3524 * optimization: the optimization of 32-bit logical and arithmetic
3525 functions introduced in version 0.8.3 on the x86 has been
3526 implemented on the mips, ppc and sparc platforms; an
3527 implementation of the same facility, but for 64-bit arithmetic,
3528 has been added for the alpha.
3529 * microoptimization: the compiler is better able to make use of the
3530 x86 LEA instruction for multiplication by constants.
3531 * bug fix: in some situations compiler did not report usage of
3532 generic arithmetic in (SPEED 3) policy.
3533 * bug 145b fix: compiler used wrong type specifier while converting
3534 MEMBER-types to numeric.
3535 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3537 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3538 index argument now works properly on the Alpha platform.
3539 * bug fix: floating point exception treatment on the Alpha platform
3541 * bug fix: FILE-POSITION works much better on string input and
3542 output streams. (thanks to Nikodemus Siivola)
3543 * bug fix: many threading/garbage collection symptoms sorted.
3544 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3545 might be pseudo-atomic.
3546 * internal change: Stopping for GC is now done with signals not
3547 ptrace. GC is now done in whichever thread wanted it, instead of
3549 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3551 * bug fix: result form in DO is not contained in the implicit
3553 * incompatible change: ICR structure is changed; the value part of
3554 CONTINUATION is now called LVAR; corresponding functions are
3555 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3556 * added type deriver for ISQRT (thanks to Robert E. Brown).
3557 * bug fix: better support for loading from the command line when an
3558 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3560 * fixed some bugs revealed by Paul Dietz' test suite:
3561 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3562 ** ROUND and FROUND now give the right answer when given very
3563 small float arguments.
3564 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3566 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3567 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3568 ** LCM should return a non-negative integer.
3569 ** PARSE-INTEGER returned the index of a terminator instead of the
3570 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3571 ** PARSE-INTEGER returned an incorrect index being applied to a
3573 ** LCM with two arguments of 0 returns 0 rather than signalling
3575 ** unsigned addition of a 32-bit constant with the high bit set no
3576 longer causes an internal compiler error on the x86.
3577 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3578 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3579 with negative last argument.
3580 ** byte specifiers with zero size and position no longer cause
3581 an error during type derivation.
3582 ** bignum multiplication on the Alpha platform now returns the
3584 * porting: The system now builds on SuSE AMD64, although it still
3585 generates a 32-bit binary.
3586 * .fasl file incompatibility: The fasl file version number has
3587 been incremented (because of the changes to internal compiler
3588 data structures referred to above).
3590 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3591 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3592 more accurately, on the Darwin kernel running on PowerPC hardware.
3593 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3594 * Compiler code deletion notes now signal a condition of type
3595 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3596 an associated MUFFLE-WARNING restart.
3597 * The compiler now performs limited argument count validation of
3598 constant format strings in FORMAT, and where appropriate in ERROR,
3599 CERROR and WARN. (thanks to Gerd Moellmann)
3600 * New ASDF-INSTALL contrib can be used for automatic download and
3601 installation of third-party Lisp code from CCLAN or other sites
3603 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3604 which forces another thread to execute a function supplied by the
3606 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3607 now accept and act upon their :ELEMENT-TYPE keyword argument.
3608 (reported by Martin Atzmueller, Edi Weitz)
3609 * bug fix: FILE-POSITION now accepts position designators up to
3610 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3611 is the greater. (thanks to Patrik Nordebo)
3612 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3613 Kalle Olavi Niemitalo)
3614 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3615 printable. (reported by Eric Marsden)
3616 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3617 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3618 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3619 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3620 (thanks to Stig Sandoe)
3621 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3622 UNREAD-CHAR now correctly includes the unread character in the
3623 target sequence. (thanks to Gerd Moellmann)
3624 * bug fix in threaded builds: the system can now be suspended and
3625 resumed by shell job control with minimal disruption.
3626 * bug fixes in times and timezones >2038AD
3627 * better handling of "where is GNU make?" problem in build scripts
3628 (thanks to Nikodemus Siivola)
3629 * new optimization: inside a named function any reference to a
3630 function with the same name is considered to be a self-reference;
3631 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3632 optimization quality.
3633 * new optimization on x86: logical functions and + now have
3634 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3635 used when the result is truncated to 32 bits.
3636 * VALUES declaration is partially enabled.
3637 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3638 * fixed some bugs revealed by Paul Dietz' test suite:
3639 ** The system now obeys the constraint imposed by
3640 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3641 types form a lattice under type intersection.
3642 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3643 ** ASSOC now ignores NIL elements in an alist.
3644 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3645 and (1+ MOST-POSITIVE-FIXNUM) answers.
3646 ** The addition of a method with invalid qualifiers to a generic
3647 function does not cause an error to be signalled immediately;
3648 a warning is signalled, and the error is generated only on
3649 calling the generic function.
3650 * changed .fasl file version number, in part to add type codes for
3651 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3652 obscure ANSI requirements
3654 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3655 * fixed bug 148: failure to inline-expand a local function left
3656 garbage, confusing the compiler.
3657 * fixed bugs 3cd: structure slot readers perform type check if the
3658 slot can have an invalid value (i.e. it is either not initialized
3659 or can be written with a less specific slot writer).
3660 * bug fix: the compiler now traps array references to elements off
3661 the end of an array; previously, the bounds checking in some
3662 circumstances could go off-by-one.
3663 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3665 * type declarations for array element types now obey the description
3666 on the CLHS page "Declaration TYPE", as per discussions on
3667 sbcl-help around 2003-05-08. This means that a declaration
3668 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3669 declaration, all references to BAR will be asserted or assumed
3670 (with THE, so dependent on compiler policy) to involve objects of
3671 type FOO. Note that no such declaration is implied in
3672 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3673 * declared types of functions from the "Conditions"
3674 chapter. (reported by Paul Dietz)
3675 * bug fix: CERROR accepts a function as its first argument.
3676 * bug fix: NTH an NTHCDR accept a bignum as index
3677 arguments. (reported by Adam Warner)
3678 * optimization: character compare routines now optimize comparing
3679 against a constant character. (reported by Gilbert Baumann)
3680 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3681 argument now works properly on the MIPS platform.
3682 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3683 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3684 declared to have a functional type, does not cause an error
3686 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3687 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3688 * fixed bug: if last continuation of a deleted block has a
3689 destination, this destination should be deleted too. (reported by
3691 * fixed a bug in the bootstrap process: the host compiler's values
3692 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3693 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3694 test case from Patrik Nordebo)
3695 * improved the ability of the disassembler on the PPC platform to
3696 provide helpful disassembly notes.
3697 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3698 the class in more cases than previously.
3699 * bug fix: FILE-POSITION now understands :START and :END for
3700 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3701 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3702 8) data. (thanks to David Lichteblau)
3703 * bug fix: it is possible to add a method to a generic function
3704 without lambda list.
3705 * bug fix: reader failed to signal END-OF-FILE inside an
3706 object representation. (reported by Nikodemus Siivola)
3707 * fixed some bugs revealed by Paul Dietz' test suite:
3708 ** LAST and [N]BUTLAST should accept a bignum.
3709 ** condition slot accessors are methods.
3710 ** (VECTOR NIL) is a subtype of STRING.
3712 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3713 * minor incompatible change: some nonsensical specialized lambda
3714 lists (used in DEFMETHOD) which were previously ignored now signal
3716 * minor incompatible change: the system is now aware of the types of
3717 variables in the COMMON-LISP package, and will signal errors for
3718 most violations of these type constraints (where previously they
3719 were silently accepted).
3720 * minor incompatible change: COMPILE-FILE now uses the freedom
3721 afforded (ANSI 3.2.2.3) to use derived function types for
3722 functions defined in the same file. This also permits the system
3723 to warn on static type mismatches and function
3724 redefinition. (Currently it does not work with high DEBUG level.)
3725 * minor incompatible change: VALUES declaration is disabled.
3726 * When issuing notes, the compiler now signals a condition of type
3727 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3728 restart for use in user handlers. It is expected that the
3729 COMPILER-NOTE condition will eventually become a condition
3730 supertype to a hierarchy of note types, which will then be
3731 handleable in a similar fashion. However, at the moment, no such
3732 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3733 but it's an implementation detail, not a classification for the
3735 * Changes in type checking closed the following bugs:
3736 ** type checking of unused values (192b, 194d, 203);
3737 ** template selection based on unsafe type assertions (192c, 236);
3738 ** type checking in branches (194bc).
3739 * A short form of VALUES type specifier has ANSI meaning (it has
3740 increased the number of situations when SBCL cannot perform type
3742 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3743 keywords or constants is permissible.
3744 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3745 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3746 classes. (thanks to Antonio Martinez)
3747 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3748 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3749 argument to a pretty-print stream. Instead, it leaves any such
3750 operations to DESCRIBE-OBJECT methods.
3751 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3753 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3754 lambda lists. (thanks to David Lichteblau)
3755 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3756 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3757 (thanks to Gerd Moellmann)
3758 * fixed bug 246: increased compilation speed of long
3759 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3761 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3763 * DEFCONSTANT now throws a condition of type
3764 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3765 constant to a non-EQL value; CONTINUE and ABORT restarts
3766 respectively change and preserve the value.
3767 * fixed bug 63: The code walker, part of the implementation of CLOS,
3768 is now better at handling symbol macros.
3769 * bug fix: There is no longer an internal implementation type named
3770 CL:LENGTH. (reported by Raymond Toy)
3771 * bug fix: In macro-like defining macros/special operators the
3772 implicit block does not enclose the lambda list.
3773 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3774 enclosed in parentheses) are not suitable as type specifiers, and
3775 their use properly signals an error now.
3776 * bug fix: An argument count mismatch for a type specifier in code
3777 being compiled no longer causes an unhandled error at compile
3778 time, but signals a compile-time warning.
3779 * fixed simple vector readable printing
3780 * bug fix: DESCRIBE takes more care over whether the class
3781 precedence list slot of a class is bound before accessing it.
3782 (reported by Markus Krummenacker)
3783 * bug fix: FORMATTER can successfully compile pretty-printer format
3784 strings which use variants of the ~* directive inside.
3785 * bug fix: SEARCH now applies its TEST predicate to the elements of
3786 the arguments in the correct order. (thanks to Wolfhard Buss)
3787 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3788 in type inference. (thanks to Robert E. Brown)
3789 * bug fix: Reading in symbols with an explicit package name of ""
3790 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3791 * fixed some bugs revealed by Paul Dietz' test suite:
3792 ** NIL is now allowed as a structure slot name.
3793 ** Arbitrary numbers, not just REALs, are allowed in certain
3794 circumstances in LOOP for-as-arithmetic clauses.
3795 ** Multiple class redefinitions before slot access no longer
3796 causes a type error.
3797 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3798 association between the name and a class.
3799 ** Generic functions with non-standard method-combination and over
3800 five methods all of which return constants no longer return NIL
3801 after the first few invocations. (thanks to Gerd Moellmann)
3802 ** CALL-NEXT-METHOD with no arguments now passes the original
3803 values of the arguments, even in the presence of assignment.
3804 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3806 ** DISASSEMBLE works with closures and funcallable instances.
3807 ** ADD-METHOD now returns the generic function, not the new method.
3808 ** FIND-METHOD signals an error if the lengths of the specializers
3809 is incompatible with the generic function, even if the ERRORP
3811 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3812 which its argument is a member.
3813 ** DEFCLASS only redefines the class named by its class-name
3814 argument if that name is the proper name of the class;
3815 otherwise, it creates a new class.
3816 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3817 of the UNBOUND-SLOT condition to the name of the slot.
3818 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3819 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3820 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3821 SLOT-MAKUNBOUND in the specified fashion.
3823 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3824 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3825 cross-compilation host. As a consequence, we can now bootstrap our
3826 way up to SBCL starting with a bare gcc toolchain and human-readable
3827 source code (first the source to CLISP, then the source to SBCL).
3828 * A contributed module containing a partial implementation of the
3829 simple-streams interface has been included. (thanks to Rudi
3831 * A contributed module implementing the RFC1321 Message Digest
3832 Algorithm, known as MD5, has been included.
3833 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3834 no longer has any effect, as the code controlled by this feature
3835 has been deleted. (As far as we know, no-one has ever built using
3836 this feature, and its semantics were confused in any case).
3837 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3838 work on conditions (as required by the ANSI specification),
3839 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3840 expected behaviour on conditions. Users should note, however,
3841 that such behaviour is not required by the ANSI specification,
3842 and so use of this behaviour may render their code unportable.
3843 * fixed some bugs revealed by Paul Dietz' test suite:
3844 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3846 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3847 ** SXHASH on condition objects no longer returns NIL.
3848 ** :ALLOCATION :CLASS slots are better treated; their values are
3849 updated on class redefinition, and initforms inherited from
3850 superclasses are applied.
3851 ** REMOVE-METHOD returns its generic function argument even when
3852 no method was removed.
3853 ** SHARED-INITIALIZE now initializes the values of the requested
3854 slots, including those with :ALLOCATION :CLASS.
3855 ** ALLOCATE-INSTANCE now works on structure classes defined via
3856 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3858 ** SLOT-EXISTS-P now works on conditions, as well as structures
3860 ** MAKE-LOAD-FORM now has the required methods on
3861 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3862 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3863 keyword, and now implements the SLOT-NAMES argument.
3864 ** methods with &OPTIONAL arguments no longer allow too many
3865 arguments to be passed in the call without error.
3866 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3867 option is consistent with the required arguments of the generic
3868 function lambda list.
3869 * bug fix: REQUIRE accepts a string designator. (Thanks to
3871 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3872 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3873 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3874 * bug fix: The compiler now checks for duplicated variables in macro
3876 * bug fix: SETQ on globals returns the correct value.
3877 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3878 not a COMPILER-ERROR (followed by some other strange error on
3879 choosing the CONTINUE restart).
3880 * bug fix: make.sh and friends are now more consistent in the way that
3881 they look for GNU "make".
3883 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3884 * experimental native threads support (on x86 Linux >=2.4 only).
3885 This is not compiled in by default: you need to add :SB-THREAD to
3886 the target features. See the "Beyond ANSI" chapter of the manual
3888 * fix for longstanding nonANSIism: The old distinction between
3889 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3890 The return value from CL:FIND-CLASS is now a CLOS class, and
3891 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3892 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3893 * An interface to the MetaObject Protocol, as described in Kiczales,
3894 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3895 MIT Press, 1991, is available from the SB-MOP package.
3896 * incompatible change: the SB-PCL package should now be considered
3897 a private implementation detail, and no longer a semi-private MOP
3899 * minor incompatible change: due to rearrangement for threads, the
3900 control stack and binding stack are now allocated at arbitrary
3901 addresses instead of being hardcoded per-port. Users affected by
3902 this probably have to be doing advanced things with shared
3903 libraries, and will know who they are.
3904 * minor incompatible change: Previously, all --eval forms used were
3905 processed with READ before any of them were processed with EVAL.
3906 Now each --eval form is processed with both READ and EVAL before
3907 the next --eval form is processed. (Thus package operations like
3908 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3909 work as the user might reasonably expect.)
3910 * minor incompatible change: *STANDARD-INPUT* is now only an
3911 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3913 * minor incompatible change: Y-OR-N-P is now character-oriented,
3914 not line oriented. Also, YES-OR-NO-P now works without errors.
3915 (thanks to Antonio Martinez)
3916 * sb-aclrepl module improvements: an integrated inspector, added
3917 repl features, and a bug fix to :trace command.
3918 * Known functions, which cannot be open coded by the backend, are
3919 considered to be able to check types of their arguments. (fixing
3920 a bug report by Nathan J. Froyd)
3921 * fixed a bug in computing method discriminating functions: It is
3922 now possible to define methods specialized on classes which have
3923 forward-referenced superclasses. (thanks to Gerd Moellmann)
3924 * fixed evaluation order in optional entries (reported by Gilbert
3926 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3927 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3928 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3929 required-by-AMOP NAME argument, as well as CLASS and
3930 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3931 * fixed bug 20: DEFMETHOD can define methods using names that are
3932 not the proper names of classes to designate class specializers.
3933 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3934 Unix file descriptors, instead of blowing up. (thanks to Antonio
3936 * Garbage collection refactoring: user-visible change is that a
3937 call to the GC function during WITHOUT-GCING will not do garbage
3938 collection until the end of the WITHOUT-GCING. If you were doing
3939 this you were probably losing anyway.
3940 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3941 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3942 TYPEP the latter but not the former.
3943 * The compiler issues a full WARNING for calls to undefined functions
3944 with names from the CL package.
3945 * MAP-INTO for a vector destination is open coded. (reported by
3946 Brian Downing on c.l.l)
3947 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3948 documentation string.
3949 * fixed some bugs revealed by Paul Dietz' test suite:
3950 ** COPY-ALIST now signals an error if its argument is a dotted
3952 ** Condition slots are now accessed more correctly in the presence
3953 of multiple initargs for a given slot.
3954 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3955 exclude restarts of the same name associated with a different
3957 ** DEFCLASS of forward-referenced classes with another
3958 forward-referenced class in the superclasses list no longer
3960 ** Condition slots are now initialized once each, not multiple
3961 times. (thanks to Gerd Moellmann)
3962 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3964 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3965 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3966 arguments contain duplicated elements.
3967 ** RESTART-CASE understands local macros.
3968 ** RESTART-CASE associates exactly its own restarts with a condition.
3969 ** ENDP in safe mode checks its argument to be of type LIST.
3970 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3971 in question is unbound.
3972 ** Optimized MAKE-INSTANCE functions no longer cause internal
3973 assertion failures in the presence of duplicate initargs.
3974 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3975 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3976 propagated even in the ctor optimized implementation of
3978 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3980 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3981 * a better implementation of SXHASH on (simple) bit vectors,
3982 measured both in execution speed and in distribution of results
3983 over the positive fixnums, has been installed. Likewise, a better
3984 implementation of EQUAL for simple bit vectors is now available.
3985 * fixed CEILING optimization for a divisor of form 2^k.
3986 * fixed bug 240 (emitting extra style warnings "using the lexical
3987 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3988 by Antonio Martinez)
3989 * fixed SXHASH, giving different results for NIL depending on type
3990 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3991 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3992 variable. (found by Rolf Wester)
3993 * SBCL does not ignore type declarations for special
3994 variables. (reported by rif on c.l.l 2003-03-05)
3995 * some bug fixes in contrib/sb-aclrepl/
3996 * fixed some bugs revealed by Paul Dietz' test suite:
3997 ** a bug in the CONS type specifier, whereby the CAR and CDR
3998 types got intertwined, has been fixed;
3999 ** the type system is now able to reason about the interaction
4000 between INTEGER and RATIO types more completely;
4001 ** APPEND, [N]REVERSE and NRECONC check that those their
4002 arguments, which must be proper lists, are really so;
4003 ** An array specialized to be unable to hold elements has been
4004 implemented, as required -- yes, really -- by ANSI;
4005 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
4006 on malformed property lists;
4008 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
4009 * incompatible packaging change: in line with Unix convention,
4010 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
4011 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
4012 * REQUIRE and PROVIDE are now optionally capable of doing something
4013 useful. See the documentation string for REQUIRE.
4014 * infrastructure for a managed SBCL contrib system: contributed
4015 modules in this release include:
4016 ** the ASDF system definition facility;
4017 ** an interface to the BSD Sockets API;
4018 ** an ACL-like convenience interface to the repl;
4019 (thanks to Kevin Rosenberg)
4020 ** an implementation of ROTATE-BYTE, with efficient implementation
4022 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
4023 gives a closer approximation to the right answer than previously.
4024 (thanks to Raymond Toy)
4025 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
4026 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
4027 optional environment arguments, as required by ANSI.
4028 * fixed bugs in other functions taking environment objects, allowing
4029 calls with an explicit NIL environment argument to be compiled
4031 * fixed bug 228: primary return values from
4032 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
4033 COMPILE or FUNCTION.
4034 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
4035 :TYPE will no longer signal errors on innocuous objects.
4036 * fixed bug 231b: SETQ is better at respecting type declarations in
4037 the lexical environment.
4038 * fixed a bug in DEFCLASS: classes named by symbols with no or
4039 unprintable packages can now be defined.
4040 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
4041 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
4042 * fixed compiler failure related to checking types of functions.
4043 (reported by Robert E. Brown)
4044 * the compiler is now much more consistent in its error-checking
4045 treatment of bounding index arguments to sequence functions: in
4046 (SAFETY 3) code, errors will be signalled in almost all cases if
4047 invalid sequence bounding indices are passed to functions defined
4048 by ANSI to operate on sequences.
4049 * fixed a bug in the build procedure: documentation of SBCL-specific
4050 packages is now preserved and available in the final Lisp image.
4051 * lifted FDEFINITION lookup out of loops in the implementation of
4052 many list operations. (thanks to Robert E. Brown)
4053 * fixed a bug in the reader: the #n# reader macro now works for
4054 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
4055 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
4056 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
4057 E. Brown and Tony Martinez respectively)
4058 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
4059 SLOT-BOUNDP: in particular, it is now able to optimize them much
4060 better, and is now not vulnerable to having packages renamed.
4061 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
4062 always called when a slot is not present in an instance. (thanks
4064 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
4065 longer signal an error on unreading a character following EOF on
4066 the previous constituent stream. (thanks to Tony Martinez)
4067 * fixed some bugs revealed by Paul Dietz' test suite:
4068 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
4069 not just nonnegative fixnums;
4070 ** the logical bit-array operators such as BIT-AND now accept an
4071 explicit NIL for their "opt-arg" argument (to indicate a
4072 freshly-consed result bit-array);
4073 ** ELT now signals an error on an invalid sequence index in safe
4075 ** the type system is now cleverer about negations of numeric
4076 types, and consequently understands the BIGNUM and RATIO types
4078 ** the type system is now cleverer about the interaction between
4079 INTEGER and RATIO types: while bugs still remain, many more
4080 cases are accurately computed;
4081 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
4082 if it is in the last clause;
4083 ** CONSTANTLY now correctly returns a side-effect-free function in
4085 ** DECLARE is no longer treated as a special-operator; in
4086 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
4087 * incremented fasl file version number due to the change in the
4088 DEFSTRUCT-SLOT-DESCRIPTION structure.
4090 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4091 * minor incompatible change: code processed by the "interpreter" or
4092 EVAL now has a compilation optimization policy of (DEBUG 2)
4093 (changed from (DEBUG 1)) to improve debuggability of interactive
4094 development, and to allow the use of the debug RETURN command in
4096 * an experimental implementation of the RETURN command for the
4097 debugger has been included. (thanks to Frederik Kuivinen)
4098 * fixed bug 62: constraints were not propagated into a loop.
4099 * fixed bug in embedded calls of SORT (reported and investigated by
4101 * fixed some bugs revealed by Paul Dietz' test suite:
4102 ** printing and reading of arrays with some dimensions having
4103 length 0 (thanks to Gerd Moellmann);
4104 ** BOA constructor with &AUX argument without a default value does
4105 not cause a type error;
4106 ** CONSTANTP now returns true for all self-evaluating objects.
4108 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4109 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4110 accessors that are related by inheritance, as specified in the
4111 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4112 Valtteri Vuorikoski)
4113 * The compiler is now able to inline functions that were defined in
4114 a complex lexical environment (e.g. inside a MACROLET).
4115 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4117 * Support for the upcoming FreeBSD-5.0 release has been included.
4118 (thanks to Dag-Erling Smorgrav)
4119 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4120 effect when it is not in a toplevel context.
4121 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4122 stand a better chance of being correct. (thanks to Gerd
4124 * fixed bug in COERCE, which now signals an error on coercing a
4125 rational to a bounded real type which excludes the expected
4127 * The compiler is now able to derive types more accurately from the
4128 COERCE and COMPILE functions.
4129 * fixed bug 223: functional binding is considered to be constant
4130 only for symbols in the CL package.
4131 * fixed bug 231: SETQ did not check the type of a variable being set
4132 (reported by Robert E. Brown)
4133 * A new optimization for MAKE-INSTANCE has been included, fixing
4134 various bugs (including relating to :ALLOCATION :CLASS slots and
4135 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4137 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4138 ** As required by ANSI, LOOP now disallows anonymous collection
4139 clauses such as COLLECT I in conjunction with aggregate boolean
4140 clauses such as THEREIS (= I 1);
4141 ** LOOP now signals an error when any variable is reused in the
4142 same loop (including the potentially useful construct analogous
4143 to WITH A = 1 WITH A = (1+ A);
4144 ** IT is only a special loop symbol within the first clause of a
4145 conditional loop clause;
4146 ** LOOP with a typed iteration variable over a hashtable now
4147 signals a type error iff it should.
4148 * fixed some other bugs revealed by Paul Dietz' test suite:
4149 ** FILE-STREAM now names the class previously known as FD-STREAM;
4150 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4151 argument) no longer signals an error;
4152 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4153 of the slot symbol, rather than using the current package
4154 ((:CONC-NAME "") continues to intern the slot's name in the
4156 * incremented fasl file version number, because of the incompatible
4157 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4158 of the new implementation of DEFINE-COMPILER-MACRO.
4160 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4161 * Support for building SBCL for MIPS platforms running in
4162 little-endian mode has now been checked in, and basic
4163 functionality on said platforms verified.
4164 * minor incompatible change: PCL now records the pathname of a file
4165 in which methods and the like are defined, rather than its
4167 * minor incompatible change: TRUENAME now considers the truename of
4168 a file naming a directory to be the pathname with :DIRECTORY
4169 component indicating that directory.
4170 * minor incompatible change: a NAMED clause in the extended form of
4171 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4172 reason for the previous behaviour is unclear.
4173 * more systematization and improvement of CLOS and MOP conformance
4174 in PCL (thanks to Gerd Moellman and Pierre Mai):
4175 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4177 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4178 primary methods with no specializers;
4179 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4181 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4182 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4183 FORWARD-REFERENCED-CLASSes; error reporting on
4184 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4186 ** SXHASH on CLOS instances now uses a slot internal to the
4187 instance to return different numbers on distinct instances,
4188 while preserving the same return value through invocations of
4190 ** DEFMETHOD signals errors when methods with longer incongruent
4191 lambda lists are added to generic functions;
4192 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4193 CLASS, as specified in AMOP;
4194 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4195 on the order returned by the primary method for classes of
4196 class STANDARD-CLASS;
4197 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4198 * fixed some bugs shown by Paul Dietz' test suite:
4199 ** DOLIST puts its body in TAGBODY;
4200 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4202 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4203 value producing form;
4204 ** if more variables are given to PROGV than values, extra
4205 variables are bound and made to have no value;
4206 ** NSUBSTITUTE on list arguments gets the right answer with
4208 ** ELT signals an error of type TYPE-ERROR when the index argument
4209 is not a valid sequence index;
4210 ** LOOP signals (at macroexpansion time) an error of type
4211 PROGRAM-ERROR when duplicate variable names are found;
4212 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4213 ** FUNCALL of special-operators now cause an error of type
4215 ** PSETQ now works as required in the presence of side-effecting
4216 symbol-macro places;
4217 ** NCONC accepts any object as its last argument;
4218 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4220 ** loop-for-as-package does not require a package to be explicitely
4222 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4224 * fixed bug 166: compiler preserves "there is a way to go"
4225 invariant when deleting code.
4226 * fixed bug 172: macro lambda lists with required arguments after
4227 &REST arguments now cause an error to be signalled. (thanks to
4229 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4230 bug. (thanks to Matthew Danish)
4231 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4233 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4234 arguments are lexically rebound. (thanks to Gerd Moellmann and
4236 * fixed bug 194: error messages are now more informative when there
4237 is no primary method applicable in a call to a generic
4238 function. (thanks to Gerd Moellmann)
4239 * fixed bug in command line argument checking (thanks to Julian
4241 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4242 (thanks to Matthew Danish)
4243 * incremented fasl file version number, because of the
4244 SXHASH-related changes in the layout of CLOS data structures
4246 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4247 * minor incompatible change: The runtime (the Unix executable named
4248 "sbcl") is now much pickier about the .core files it will load.
4249 Essentially it now requires .core files to descend from the same
4250 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4251 as the runtime does. (The intent is to prevent the crashes which
4252 can occur, and which can even be reported as mysterious failures,
4253 when people patch the sources or change the build parameters
4254 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4255 sbcl and .core files.)
4256 * fixed bug: VALUES-LIST is no longer optimized away.
4257 * fixed bug 142: The FFI conversion of C string values to Lisp
4258 string values no longer conses excessively. (thanks to Nathan
4259 Froyd porting Raymond Toy's fix to CMU CL)
4260 * began to systematize and improve MOP conformance in PCL (thanks to
4261 Nathan Froyd, Gerd Moellman and Pierre Mai):
4262 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4264 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4265 ** FINALIZE-INHERITANCE is now called on class finalization;
4266 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4267 argument precedence order.
4268 * fixed bug 202: The compiler no longer fails on functions whose
4269 derived types contradict their declared type.
4270 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4271 so it can be non-toplevel.
4272 * The fasl file version number has changed (because of the new
4273 implementation of DEFMACRO).
4274 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4275 safe code, that any length requirement by their type-specifier
4276 argument is valid. The exceptions to this are described in bug
4278 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4279 functions now check the types of their inputs as required by ANSI.
4280 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4281 introduced symbol is DECLAREd to be SPECIAL.
4282 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4283 * fixed bug: :COUNT argument to sequence functions may be negative
4284 * fixed bug: body of DO-SYMBOLS may contain declarations
4285 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4286 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4287 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4288 symbol macro only once
4289 * fixed printing of call frame when argument list is unavailable
4290 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4291 * compiler no longer signals WARNING on unknown keyword
4294 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4295 * A beta-quality port to the mips architecture running Linux,
4296 based on the old CMUCL backend, has been made. It has been tested
4297 on a big-endian kernel, and works sufficiently well to be able to
4298 rebuild itself; it has not been tested in little-endian mode.
4299 * fixed an inconsistency between gencgc.c and purify.c which made
4300 dumping/loading .core files unreliable
4301 * fixed bug 120a: The compiler now deals correctly with IFs where
4302 the consequent is the same as the alternative, instead of
4303 misderiving the return type. (thanks to Alexey Dejneka)
4304 * fixed bug 113: Logical pathnames are now dumpable (the logical
4305 host is resolved at load-time, throwing an error if it is not
4307 * fixed bug 174: FORMAT's error message is slightly clearer when a
4308 non-printing character is used in a format directive.
4309 * fixed several bugs in compiler checking of type declarations, i.e.
4310 violations of the Python "declarations are assertions" principle
4311 (thanks to Alexey Dejneka)
4312 * fixed several bugs in PCL's error checking (thanks to Gerd
4314 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4316 * fixed bug in compilation of functions as first class values
4317 (thanks to Antonio Martinez-Shotton)
4318 * The compiler's handling TYPE-ERRORs which it can prove will
4319 inevitably happen at runtime has been cleaned up and corrected
4320 in several ways. (thanks to Alexey Dejneka)
4321 * improved argument type checking for various basic arithmetic
4322 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4323 lavished on them in the past that they can be compiled in many
4324 ways in different special cases
4325 * fixed bug 181: compiler checks validity of user supplied type
4327 * cleaned up code flushing in optimization: Function calls which
4328 should signal errors for safety purposes (e.g. which ANSI says
4329 should signal errors when their arguments are of incorrect type)
4330 are no longer optimized away.
4331 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4332 * incremented fasl file version number, because changes in the
4333 implementation of sequence functions like COERCE caused
4334 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4335 in old inline expansions) to become undefined. (Actually these
4336 changes were later undone, so we might very well be binary
4337 compatible with 0.7.7 after all, but leaving the version number
4338 incremented seemed like the simplest and most conservative
4341 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4342 * An alpha-quality port to the parisc architecture running Linux,
4343 based on the old CMUCL backend, has been made. This, even more so
4344 than the other backends, should be considered still a work in
4345 progress; known problems include that the Linux kernel in 64-bit
4346 mode does not propagate the correct sigcontext structure to
4347 userspace, and consequently SBCL on a parisc64 kernel will not
4349 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4350 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4351 suppressed.) Also now that the compiler is looking at declarations
4352 in the environment, it checks optimization declarations as well,
4353 and suppresses inlining when (> DEBUG SPEED).
4354 * More fixes have been made to treatment of floating point exception
4355 treatment and other Unix signals. In particular, floating point
4356 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4357 * The detection and handling of control stack exhaustion (infinite
4358 or very deeply nested recursion) has changed. Stack exhaustion
4359 detection is now done by write-protecting pages at the OS level
4360 and applies at all optimization settings; when found, a
4361 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4362 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4363 be caught using IGNORE-ERRORS.
4364 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4365 correctly when transforming recursive LABELS functions to LETs.
4366 (thanks to Alexey Dejneka)
4367 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4368 that are names of constants or global variables.
4369 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4370 alien routines with docstrings.
4371 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4372 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4374 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4375 (thanks to Eric Marsden)
4376 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4377 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4378 * Bug fix: The SPARC backend can now compile functions involving
4379 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4380 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4381 to a functions expecting a PATHNAME-DESIGNATOR.
4382 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4383 lambda lists. (thanks to Alexey Dejneka)
4384 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4385 functions so that EXPORTing the name of the class doesn't cause
4386 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4387 to Antonio Martinez for reporting this)
4388 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4389 constant-folded without failing an assertion. (thanks to Einar
4390 Floystad Dorum for reporting this)
4391 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4392 types are now dealt with more correctly by the compiler.
4393 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4394 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4395 behaviour analogous to RENAME-FILE. This puts its behaviour more
4396 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4398 * The fasl file version number has changed again. (because of the
4399 bug fix involving the names of PCL MAKE-INSTANCE functions)
4401 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4402 * bug fix: Floating point exceptions are treated much more
4403 consistently on the x86/Linux and PPC/Linux platforms.
4404 * Array initialization with :INITIAL-ELEMENT is now much faster for
4405 cases when the compiler cannot open code the array creation, but
4406 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4407 array accesses have also seen a speed increase.
4408 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4409 specified. (This was at the root of some bad interactions between
4410 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4412 * bug fix: Internal error arguments for undefined functions are now
4413 computed correctly on the PPC/Linux platform.
4414 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4415 Raymond Toy's patch for CMU CL)
4416 * Support for the Solaris 9 operating environment has been included
4417 (thanks to Daniel Merritt)
4418 * A very ugly but hopefully complete draft of the missing FFI chapter
4419 of the manual has been created by reformatting the corresponding
4420 CMU CL manual chapter into (currently very ugly and incoherent)
4421 DocBook and bringing it up to date for SBCL behavior. Thus, the
4422 manual is now essentially complete, at least by my extreme
4423 once-and-only-once standards, whereby it's acceptable to refer to
4424 the doc strings of SB-EXT functions as the primary documentation.
4425 * The fasl file version number has changed again, due to cleanup of
4426 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4427 is no longer a static symbol.)
4429 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4430 * SBCL now builds with OpenMCL (version 0.12) as the
4431 cross-compilation host; also, more progress has been made toward
4432 bootstrapping under CLISP.
4433 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4435 * bug 158 fixed: The compiler can now deal with integer loop
4436 increments different from 1; fixing this turned out also to fix
4438 * bug 169 fixed: no more bogus warnings about using lexical bindings
4439 despite the presence of perfectly good SPECIAL declarations (thanks
4440 to David Lichteblau)
4441 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4442 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4443 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4445 * bug 180 fixed: Method combination specifications no longer ignore
4446 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4447 * bug fix: Structure type predicate functions now check their argument
4448 count as they should.
4449 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4450 correctly. (thanks to Pierre Mai)
4451 * minor incompatible change: The --noprogrammer option is deprecated
4452 in favor of the new --disable-debugger option, which is very similar.
4453 (The major difference is that it takes effect at a slightly different
4454 time at startup, causing handling of errors in --sysinit and
4455 --userinit files will be affected differently.) The
4456 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4457 been added to allow this functionality to be controlled from ordinary
4458 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4459 maintainers, who might want to run non-interactive scripts to
4460 build SBCL cores which will later be used interactively.)
4461 * minor incompatible change: The LOAD function no longer, when given
4462 a wild pathname to load, loads all files matching that pathname.
4463 Instead, an error of type FILE-ERROR is signalled.
4465 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4466 * bug 147 fixed: The compiler preserves its block link/count
4467 invariants more correctly now so that it doesn't crash. (thanks
4469 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4471 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4472 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4473 no-value-for-this-character as NIL (as specified by ANSI).
4474 * HOST-NAMESTRING on physical pathnames now returns a string that is
4475 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4476 (thanks to Christophe Rhodes)
4477 * The Alpha port handles icache flushing more correctly. (thanks to
4479 * More progress has been made toward bootstrapping under CLISP. (thanks
4480 to Christophe Rhodes)
4481 * The fasl file format has changed again, because dynamic loading
4482 on OpenBSD (which has non-ELF object files) motivated some cleanups
4483 in the way that foreign symbols are transformed and passed around.
4484 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4485 is no longer treated as whitespace by the reader, but instead as
4486 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4487 |A
\7fB|, instead of A as it used to.
4489 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4490 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4491 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4492 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4493 this as of 0.7.1.45, but was left out of the previous news section
4494 (thanks to Dan Barlow)
4495 * SBCL now runs on the Solaris operating system on SPARC architectures
4496 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4497 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4498 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4499 * SPARC backend cleanups, allowing builds of cores optimized for V8
4500 and V9 SPARCS, and also emission of code targeted to a particular
4501 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4503 * SBCL is closer to bootstrapping under CLISP, thanks to various
4504 fixes by Christophe Rhodes.
4505 * The fasl file format has changed again, to allow the compiler's
4506 INFO database to support symbol macros.
4507 * The user manual (in doc/) is formatted into HTML more nicely.
4508 (thanks to coreythomas)
4509 * The system is smarter about SUBTYPEP relationships, especially
4510 those involving NOT types (including types such as ATOM which are
4511 represented internally using NOT types). Thus SUBTYPEP is less
4512 likely to return (VALUES NIL NIL) in general, and in particular
4513 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4515 * The fasl file format has changed again, because the internal
4516 representation of types now includes a new slot to support the new
4517 SUBTYPEP-of-NOT-types logic.
4518 * (not a change in the main branch of SBCL, but a related prototype
4519 which can hopefully be merged into the main branch of SBCL in the
4520 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4521 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4522 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4523 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4524 dispatch macro character combinations. (thanks to Alexey Dejneka)
4525 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4526 that superficially look like logical namestrings correctly.
4527 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4529 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4530 default supplied-p) for &optional and &key arguments. (thanks to
4533 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4534 * incompatible change: The compiler is now less aggressive about
4535 tail call optimization, doing it only when (> SPACE DEBUG) or
4536 (> SPEED DEBUG). (This is an incompatible change because there are
4537 programs which relied on the old CMU-CL-style behavior to optimize
4538 away their unbounded recursion which will now die of stack overflow.)
4539 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4540 for non-GENCGC systems has been increased to 20M (since that
4541 seems much closer to the likely performance optimum for modern
4542 systems than the old 4M value was)
4543 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4544 and *DEBUG-PRINT-LEVEL*
4545 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4546 of CMU CL's support for SPARC, and various endianness and other
4547 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4548 * new syntactic sugar for the Unix command line: --load foo.bar is now
4549 an alternate notation for --eval '(load "foo.bar")'.
4551 ** The system now detects stack overflow and handles it gracefully,
4552 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4553 optimization settings. (This is a good thing in general, and
4554 its introduction in this version should be particularly timely
4555 for anyone whose code fails because of suppression of tail
4557 ** The system now hunts for the C variable "environ" in a more
4558 devious way, to avoid segfaults when the C library version differs
4559 between compile time and run time. (thanks to Christophe Rhodes)
4560 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4561 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4562 ** The compiler no longer issues bogus style warnings for undefined
4563 classes in the same source file as the DEFCLASSes which defined
4564 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4566 ** fixes in CONDITION class precedence list for undefined function
4567 errors (thanks to Alexei Dejneka)
4568 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4569 correctly. (thanks to Dan Barlow)
4570 ** portability fixes aiming at bootstrapping under CLISP (thanks
4571 to Dave McDonald and Christophe Rhodes)
4572 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4573 ** fixes in type translation and and type inference (thanks to
4575 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4576 ** various fixes in the new ports (thanks to Dan Barlow)
4577 * several changes related to debugging:
4578 ** suppression of tail recursion, as noted above
4579 ** stack overflow detection, as noted above
4580 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4581 is now the default. (For some time encapsulation has been more
4582 reliable than the breakpoint-based :ENCAPSULATE NIL
4583 implementation, at least on X86 systems; and I just noticed that
4584 encapsulation also seems closer to the spirit of the ANSI
4587 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4589 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4590 up properly again. (There was a packaging bug in 0.7.0 which
4591 left their definitions in SB-SYS::LOAD-FOREIGN and
4592 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4593 vital for most things which interface to C-level interfaces,
4594 like extensions working with sockets or databases or
4595 Perl-compatible regexes or whatever, and the need to fix
4596 this bug is the main reason that 0.7.1 was released so
4598 ** DEFGENERIC is now choosier about the methods it redefines, so that
4599 reLOADing a previously-LOADed file containing DEFGENERICs does
4600 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4601 cycle now works as it should. (thanks to Alexey Dejneka)
4602 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4603 fixed by Christophe Rhodes. (He actually submitted this patch
4604 months ago, and I delayed until after 0.7.0.)
4605 ** Bug 111 (internal compiler confusion about runtime checks on
4606 FUNCTION types) was fixed by Alexey Dejneka.
4607 * Some internal cleanups (getting rid of variables which aren't
4608 needed now that the byte interpreter is gone) caused the fasl
4609 file format number to change again.
4611 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4612 * major incompatible change: The default fasl file extension, i.e. the
4613 default extension for files produced by COMPILE-FILE, has changed
4614 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4616 ** There are many changes in the implementation of the compiler.
4617 SBCL is now essentially a compiler-only implementation of ANSI
4618 Common Lisp. EVAL still "interprets" a few special cases, but
4619 almost all the interesting cases are handled by creating
4620 a LAMBDA expression, calling COMPILE on it, then calling
4621 FUNCALL on the result.
4622 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4623 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4624 Since the code is newer, there might still be some new bugs
4625 (though not as many as before Martin Atzmueller's fixes:-). But
4626 the new code is substantially simpler and clearer, and hopefully
4627 any remaining bugs will be simpler, less fundamental, and more
4628 fixable then the bugs in the old code.
4629 ** The revised compiler is still a little unsteady on its feet.
4631 *** The debugging information it produces (particularly the names
4632 of FUNCTION objects) is sometimes much less useful than what
4633 the old compiler produced.
4634 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4635 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4636 within a MACROLET) has been temporarily weakened.
4637 ** There are new compiler optimizations for various functions:
4638 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4639 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4640 *** the math functions TRUNCATE, FLOOR, and CEILING
4641 *** the function-of-all-trades COERCE
4642 Mostly these should be transparent, but there's one
4643 potentially-annoying problem (bug 117): when the compiler
4644 inline-expands a function and does type analysis on the result,
4645 it can create control paths which have type mismatches, and
4646 when it can't prove that those control paths aren't taken,
4647 it will issue WARNINGs about the type mismatches. This is
4648 a particular problem in practice for the new sequence functions.
4649 It's not clear how this should be fixed, and for now, a
4650 workaround is given in the entry for 117 in the BUGS file.
4651 ** (Because of the interaction between the two previous items --
4652 occasional inlining problems and new inline expansions -- some
4653 of the new sequence function optimizations won't really kick in
4654 completely until debugging information, and then inlining, are
4655 straightened out in some future version.)
4656 * minor incompatible changes:
4657 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4658 DIRECTORY no longer implicitly promotes NIL slots of its
4659 pathname argument to :WILD. In particular, when you ask for the
4660 contents of a directory (which you used to be able to do without
4661 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4662 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4663 ** changes in behavior that ANSI explicitly defines to be
4664 implementation dependent:
4665 *** The new compiler-only implementation still conforms with ANSI,
4666 but acts a little different than before. Besides the obvious
4667 changes in performance tradeoffs (that the cost per form passed
4668 to EVAL has gone up, and the cost per form executed by EVAL
4669 has gone down), the behavior of the system changes a little
4670 because there are no longer any interpreted function objects.
4671 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4672 e.g. doing COMPILE on the output of interactive DEFUN is
4674 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4676 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4677 has changed from (:CL) to NIL.
4678 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4679 e.g. #\Newline and #\Tab, don't have names specified in the
4680 ANSI Common Lisp standard, is now based on their ASCII symbolic
4681 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4682 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4683 are no longer used for output.
4684 ** changes in internal implementation constants:
4685 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4686 4 million. (If your application spends a lot of time GCing and
4687 you have a lot of RAM, you might want to experiment with
4688 increasing it even more.)
4689 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4690 However, almost all old code should still continue to work without
4691 immediate update, as SB-C-CALL is now a (deprecated) nickname
4693 ** Old operator names in the style DEF-FOO are now deprecated in
4694 favor of new corresponding names DEFINE-FOO, for consistency with
4695 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4696 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4697 internal symbols, but a few supported extensions like
4698 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4699 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4700 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4701 etc. as you get deeper into recursive calls to the debugger
4702 command loop, instead of the old "5]", "5]]", "5]]]"
4703 sequence. (I was motivated to do this when squabbles between
4704 ILISP and SBCL left me very deeply nested in the debugger. In the
4705 short term, this change will probably provoke more ILISP/SBCL
4706 squabbles, but hopefully it will be an improvement in the long run.)
4707 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4708 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4709 function, since declaiming return types involving aliens is
4710 (1) annoyingly messy to do by hand and (2) vital to efficient
4711 compilation of code which calls such functions.
4712 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4713 longer reexported by the SB-EXT package. They're solely useful
4714 for alien code, so it seems more logical that you should get
4715 them from the SB-ALIEN package, not in SB-EXT.
4716 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4717 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4718 features. Instead, the code that they used to control is always
4719 built into the system.
4720 * many other bug fixes
4721 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4722 advantage of the new EVAL-WHEN stuff and to clean them up in
4723 general, and they are now more ANSI-compliant in a number of
4724 ways. Martin Atzmueller is responsible for a lot of this.
4725 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4727 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4728 *** correct ERROR type for various file operations
4729 *** some fixes for Lisp streams
4730 *** DEFMETHOD syntax checking
4731 *** changing old weird representation of debug information as
4732 strings (which, among their other deficiencies, don't transform
4733 correctly when you rename packages, and don't change their
4734 print representation when you change things like *PACKAGE*
4735 and *PRINT-LENGTH*) to symbols and lists of symbols
4736 He also made several improvements and fixed several bugs in DESCRIBE.
4737 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4739 *** misbehavior of WRITE-STRING/WRITE-LINE
4740 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4741 and several other LOOP problems as well
4742 *** DIRECTORY when similar filenames are present
4743 *** DEFGENERIC with :METHOD options
4744 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4745 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4746 *** argument ordering in FIND with :TEST option
4747 *** mishandled package designator argument in APROPOS-LIST
4748 *** various problems in the backquote readmacro
4749 *** a bug in APROPOS
4750 *** probably some others that I'm not describing very well here,
4751 since the CVS log documents them by reference to sbcl-devel
4752 messages, and the SourceForge archives aren't working well.:-(
4753 ** Dan Barlow improved the Alpha port (and is making progress on the
4754 PPC port, for those of you who think different).
4755 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4756 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4757 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4758 continues to work on the SPARC port (for those of you in a position
4759 to look down upon our little PC-compatible boxes from a great height).
4760 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4761 and uses that copy, rather than the current dynamic value, when
4762 it's trying to decide whether to truncate output. Thus e.g.
4763 (let ((*print-lines* 50))
4764 (pprint-logical-block (stream nil)
4766 (let ((*print-lines* 8))
4767 (print (aref possiblybigthings i) stream)))))
4768 should now truncate the logical block only at 50 lines, instead of
4769 often truncating it at 8 lines, as it did before.
4770 * The doc/cmucl/ directory, containing old CMU CL documentation
4771 from the time of the fork, is no longer part of the base system.
4772 SourceForge has shut down its anonymous FTP service, and with it
4773 my original plan for distributing the old CMU CL documentation
4774 there. For now, if you need these files you can download an old
4775 SBCL source release and extract them from it.
4776 * The fasl file version number changed again, for dozens of reasons,
4777 some of which are apparent above.
4779 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4780 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4781 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4782 patch, so that DISASSEMBLE output is much nicer.
4783 * The code in the SB-PROFILE package now seems reasonably stable.
4784 I still haven't decided what the final interface should look like
4785 (I'd like PROFILE to interact cleanly with TRACE, since both
4786 facilities use function encapsulation) but if you have a need
4787 for profiling now, you can probably use it successfully with
4788 the current CMU-CL-style interface.
4789 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4790 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4791 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4792 link now returns the dangling link itself, and for similar
4793 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4794 link itself. (In these cases the old code signalled an error and
4795 looped endlessly, respectively.) Thus, DIRECTORY now works even
4796 in the presence of dangling and cyclic symbolic links.
4797 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4798 is now a supported extension again, since the consensus on
4799 sbcl-devel was that it can be useful for ordinary development
4800 work, not just for debugging SBCL itself.
4801 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4802 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4803 that currently-defined functions might be redefined later with
4804 different return types.
4805 * Hash tables can be printed readably, as inspired by CMU CL code
4806 of Eric Marsden and SBCL code of Martin Atzmueller.
4807 * better error handling in CLOS method combination, thanks to
4808 Martin Atzmueller porting Pierre Mai's CMU CL patches
4809 * more overflow fixes for >16Mbyte I/O buffers
4810 * A bug in READ has been fixed, so that now a single Ctrl-D
4811 character suffices to cause end-of-file on character streams.
4812 In particular, now you only need one Ctrl-D at the command
4813 line (not two) to exit SBCL.
4814 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4816 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4817 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4818 does the right thing.
4819 * The implementation of some type tests, especially for CONDITION
4820 types, is now tidier and maybe faster, due to CMU CL code
4821 originally by Douglas Crosher, ported by Martin Atzmueller.
4822 * Some math functions have been fixed, and there are new
4823 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4824 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4825 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4826 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4827 likely to be built into sbcl-0.7.0.)
4828 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4829 CL, ported to SBCL by Martin Atzmueller)
4830 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4831 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4832 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4833 * There's a new slam.sh hack to shorten the edit/compile/debug
4834 cycle for low-level changes to SBCL itself, and a new
4835 :SB-AFTER-XC-CORE target feature to control the generation of
4836 the after-xc.core file needed by slam.sh.
4837 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4838 COMPILE-FILE is no longer supported, so that now every function
4839 gets an entry point, so that block compilation looks a little
4840 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4841 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4842 returns the number of bytes consed since the system started,
4843 rather than the number consed since the first time the function
4844 was called. (The new definition parallels ANSI functions like
4845 CL:GET-INTERNAL-RUN-TIME.)
4846 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4847 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4848 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4849 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4850 * Fasl file version numbers are now independent of the target CPU,
4851 since historically most system changes which required version
4852 number changes have affected all CPUs equally. Similarly,
4853 the byte fasl file version is now equal to the ordinary
4856 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4857 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4858 is no longer recognized. I apologize for this, because it was
4859 listed in SB-EXT as a supported extension, but I found that
4860 its existing behavior was poorly specified, as well as incorrectly
4861 specified, and it looked like too much of a mess to straighten it
4862 out. I have enough on my hands trying to get ANSI stuff to work..
4863 * many patches ported from CMU CL by Martin Atzmueller, with
4864 half a dozen bug fixes in pretty-printing and the debugger, and
4865 half a dozen others elsewhere
4866 * fixed bug 13: Floating point infinities are now supported again.
4867 They might still be a little bit flaky, but thanks to bug reports
4868 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4869 as flaky as they were.
4870 * The --noprogrammer command line option is now supported. (Its
4871 behavior is slightly different in detail from what the old man
4872 page claimed it would do, but it's still appropriate under the
4873 same circumstances that the man page talks about.)
4874 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4875 are now supported, and enabled by default. Thus, the compiler can
4876 handle many floating point and complex operations much less
4877 inefficiently. (Thus e.g. you can implement a complex FFT
4879 * The compiler now detects type mismatches between DECLAIM FTYPE
4880 and DEFUN better, and implements CHECK-TYPE more correctly, and
4881 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4882 thanks to patches from Martin Atzmueller.
4883 * various fixes to make the cross-compiler more portable to
4884 ANSI-conforming-but-different cross-compilation hosts (notably
4885 Lispworks for Windows, following bug reports from Arthur Lemmens)
4886 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4887 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4888 fixed thanks to Pierre Mai's CMU CL patches.
4889 * fixes to keep the system from overflowing internal counters when
4890 it tries to use i/o buffers larger than 16M bytes
4891 * fixed bug 45a: Various internal functions required to support
4892 complex special functions have been merged from CMU CL sources.
4893 (When I was first setting up SBCL, I misunderstood a compile-time
4894 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4895 * improved support for type intersection and union, fixing bug 12
4896 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4897 more obscure bugs as well
4898 * some steps toward byte-compiling non-performance-critical
4899 parts of the system, courtesy of patches from Martin Atzmueller
4900 * Christophe Rhodes has made some debian packages of sbcl at
4901 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4902 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4903 stable, but are nonetheless usable. When he's ready, I'd be happy
4904 to add them to the SourceForge "File Releases" section. (And if
4905 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4906 * new fasl file format version number (because of changes in
4907 internal representation of (OR ..) types to accommodate the new
4908 support for (AND ..) types, among other things)
4910 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4911 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4913 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4915 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4916 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4917 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4918 * DESCRIBE now works on structure objects again.
4919 * Most function call argument type mismatches are now handled as
4920 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4921 know whether the function will be redefined before the call is
4922 executed. (The compiler could flag local calls with full WARNINGs,
4923 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4924 it doesn't keep track of enough information to know whether calls
4925 are local in this sense.)
4926 * Compiler output is now more verbose, with messages truncated
4927 later than before. (There should be some supported way for users
4928 to override the default verbosity, but I haven't decided how to
4929 provide it yet, so this behavior is still controlled by the internal
4930 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4931 src/compiler/ir1util.lisp.)
4932 * Fasl file format version numbers have increased again, because
4933 support for the Gray streams extension changes the layout of the
4934 system's STREAM objects.
4935 * The Gray subclassable streams extension now works, thanks to a
4936 patch from Martin Atzmueller.
4937 * The full LOAD-FOREIGN extension (not just the primitive
4938 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4939 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4940 but like most other programs, it defaults to copying the Unix
4941 environment from the original process instead of starting the
4942 new process in an empty environment.
4943 * Extensions which manipulate the Unix environment now support
4944 an :ENVIRONMENT keyword option which doesn't smash case or
4945 do other bad things. The CMU-CL-style :ENV option is retained
4946 for porting convenience.
4947 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4948 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4950 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4952 * A patch from Martin Atzmueller seems to have solved the SIGINT
4953 problem, and as far as we know, signal-handling now works cleanly.
4954 (If you find any new bugs, please report them!)
4955 * The system no longer defaults Lisp source file names to types
4956 ".l", ".cl", or ".lsp", but only to ".lisp".
4957 * The compiler no longer uses special default file extensions for
4958 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4959 seems to expect a single default extension for all compiled code,
4960 and there's no compelling reason to try to stretch the standard
4961 to allow two different extensions.) Instead, byte-compiled files
4962 default to the same extension as native-compiled files.
4963 * Fasl file format version numbers have increased again, because
4964 a rearrangement of internal implementation packages made some
4965 dumped symbols in old fasl files unreadable in new cores.
4966 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4967 many fewer weird special cases.
4968 * Bug #17 (differing COMPILE-FILE behavior between logical and
4969 physical pathnames) has been fixed, and some related misbehavior too,
4970 thanks to a patch from Martin Atzmueller.
4971 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4972 by Tim Moore, ported to SBCL by Martin Atzmueller.
4973 * Martin Atzmueller fixed several filesystem-related problems,
4974 including bug #36, in part by porting CMU CL patches, which were
4975 written in part by Paul Werkowski.
4976 * More compiler warnings in src/runtime/ are gone, thanks to
4977 more patches from Martin Atzmueller.
4978 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4981 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4983 * DESCRIBE now works on CONDITION objects.
4984 * The debugger now handles errors which arise when trying to print
4985 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4987 * The build system now uses an additional file, customize-target-features.lisp,
4988 to allow local modifications to the target *FEATURES* list. (The point of
4989 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4990 debugging features enabled, without having to worry about propagating it
4991 into everyone's system when I do a "cvs update".) When no
4992 customize-target-features.lisp file exists, the target *FEATURES* list
4993 should be constructed the same way as before.
4994 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4995 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4996 warn about behavior which is undefined under the ANSI spec. Note: This
4997 is specified by ANSI, but it's not very popular with programmers.
4998 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4999 macro in the SBCL sources for an example of a workaround which you
5000 might use to make portable ANSI-standard code which does what you want.
5001 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
5002 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
5003 when it's used as a non-toplevel form. (This is required in order
5004 to implement the DEFCONSTANT-EQX macro.)
5005 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
5006 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
5007 funny things, due to bugs in the implementation of EVAL-WHEN
5008 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
5009 non-toplevel DEFCONSTANTs are uncommon.)
5010 * The core file version number and fasl file version number have been
5011 incremented, because the old noncompliant DEFCONSTANT behavior involved
5012 calling functions which no longer exist, and because I also took the
5013 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
5014 * fixed bug 1 (error handling before read-eval-print loop starts), and
5015 redid debugger restarts and related debugger commands somewhat while
5017 ** The QUIT debugger command is gone, since it did something
5018 rather different than the SB-EXT:QUIT command, and since it never
5019 worked properly outside the main toplevel read/eval/print loop.
5020 Invoking the new TOPLEVEL restart provides the same functionality.
5021 ** The GO debugger command is also gone, since you can just invoke
5022 the CONTINUE restart directly instead.
5023 ** The TOP debugger command is also gone, since it's redundant with the
5024 FRAME 0 command, and since it interfered with abbreviations for the
5026 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
5028 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
5029 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
5030 some porting work by Martin Atzmueller)
5031 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
5032 with a boatload of patches which he ported to SBCL. Now that those
5034 ** The system tries to make sure that its low-priority messages
5035 are prefixed by semicolons, to help people who like to use
5036 syntax highlighting in their ILISP buffer. (This patch
5037 was originally due to Raymond Toy.)
5038 ** The system now optimizes INTEGER-LENGTH better, thanks to more
5039 patches originally written by Raymond Toy.
5040 ** The compiler understands coercion between single-value and
5041 multiple-VALUES type expressions better, getting rid of some very
5042 weird behavior, thanks to patches originally by Robert MacLachlan
5043 and Douglas Crosher.
5044 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
5045 lambda lists, thanks to a patch originally by Pierre Mai.
5046 ** The system no longer bogusly warns about "abbreviated type
5048 ** The compiler gets less confused by inlining and RETURN-FROM,
5049 thanks to some patches originally by Tim Moore.
5050 ** The system no longer hangs when dumping circular lists to fasl
5051 files, thanks to a patch originally from Douglas Crosher.
5052 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
5053 undefined function error.
5054 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
5055 in structure slot initforms) for now, documented workaround instead:-|
5056 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
5057 * fixed bug 5: added stubs for various Gray stream functions called
5058 in the not-a-CL:STREAM case, so that even when Gray streams aren't
5059 installed, at least appropriate type errors are generated
5060 * fixed bug 8: better reporting of various PROGRAM-ERRORs
5061 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
5062 consistently in DEFMETHOD forms.
5063 * removed bug 21 from BUGS, since Martin Atzmueller points out that
5064 it doesn't seem to affect SBCL after all
5065 * The C runtime system now builds with better optimization and many
5066 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
5068 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
5070 * The system is now under CVS at SourceForge (instead of the
5071 CVS repository on my home machine).
5072 * The new signal handling code has been tweaked to treat register
5073 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
5074 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
5075 downstream, e.g. in debug-int.lisp, has implicit dependencies
5076 on the unsignedness of integer representation of machine words,
5077 and that caused the system to bomb out with infinite regress
5078 when trying to recover from type errors involving signed values,
5079 e.g. (BUTLAST '(1 2 3) -1).)
5080 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
5081 (This was one of the bugs Peter Van Eynde reported back in July.)
5082 * The system now uses code inspired by Colin Walters' O(N)
5083 implementation of MAP (from the cmucl-imp@cons.org mailing
5084 list, 2 September 2000) when it can't use a DEFTRANSFORM to
5085 inline the MAP operation, and there is more than one
5086 sequence argument to the MAP call (so that it can't just
5087 do ETYPECASE once and for all based on the type of the
5088 single sequence argument). (The old non-inline implementation
5089 of the general M-argument sequence-of-length-N case required
5090 O(M*N*N) time when any of the sequence arguments were LISTs.)
5091 * The QUIT :UNIX-CODE keyword argument has been renamed to
5092 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5093 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5095 * Raymond Wiker's patches to port dynamic loading from Linux to
5096 FreeBSD have been added.
5097 * The BUGS file is now more nearly up to date, thanks in large part
5098 to Martin Atzmueller's review of it.
5099 * The debugger now flushes standard output streams before it begins
5100 its output ("debugger invoked" and so forth).
5101 * The core version number and fasl file version number have both
5102 been incremented, because of incompatible changes in the layout
5104 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5105 used to not be called for a saved Lisp image.)
5106 * Martin Atzmueller's version of a patch to fix a compiler crash,
5107 as posted on sbcl-devel 13 September 2000, has been installed.
5108 * Instead of installing Martin Atzmueller's patch for the
5109 compiler transform for SUBSEQ, I deleted the compiler transform,
5110 and transforms for some similar consing operations.
5111 * A bug in signal handling which kept TRACE from working on OpenBSD
5113 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5114 away by constant folding
5115 * The system now defines its address space constants in one place
5116 (in the Lisp sources), and propagates them automatically elsewhere
5117 (through GENESIS and the sbcl.h file). Therefore, patching the
5118 address map is less unnecessarily tedious and error-prone. The
5119 Lisp names of address space constants have also been systematized.
5120 * CVS tags like dollar-Header-dollar have been removed from
5121 the sources, because they have never saved me trouble and
5122 they've been source of trouble working with patches and other
5123 diff-related operations.
5124 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5125 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5127 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5129 * The system has been ported to OpenBSD.
5130 * The system now compiles with a simple "sh make.sh" on the systems
5131 that it's supported on. I.e., now you no longer need to tweak
5132 text in the target-features.lisp-expr and symlinks in src/runtime/
5133 by hand, the make.sh takes care of it for you.
5134 * The system is no longer so grossly inefficient when compiling code
5135 involving vectors implemented as general (not simple) vectors (VECTOR T),
5136 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5137 which dares to use the various sequence functions on non-simple
5138 vectors, takes less of a performance hit.
5139 * There is now a primitive type predicate VECTOR-T-P
5140 to test for the (VECTOR T) type, so that e.g.
5141 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5142 can now be compiled with some semblance of efficiency. (The old code
5143 turned the type declaration into a full call to %TYPEP at runtime!)
5144 * AREF on (VECTOR T) is still not fast, since it's still compiled
5145 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5146 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5147 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5148 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5149 without making full calls to SUBTYPEP at runtime.
5150 (Some analogous efficiency issues for non-simple vectors specialized to
5151 element types other than T, or for non-simple multidimensional arrays,
5152 have not been addressed. They could almost certainly be handled the
5153 same way if anyone is motivated to do so.)
5154 * The changes in array handling break binary compatibility, so
5155 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5156 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5157 instead of (VALUES T T).
5158 * By following the instructions that Dan Barlow posted to sbcl-devel
5159 on 2 July 2000, I was able to enable primitive dynamic object
5160 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5161 functionality is not implemented (since it calls ld to resolve
5162 library references automatically, requiring RUN-PROGRAM for its
5163 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5164 try to resolve library references) is now supported.
5165 * The system now flushes the standard output streams when it terminates,
5166 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5167 them at several other probably-convenient times, e.g. in each pass of
5168 the toplevel read-eval-print loop, and after evaluating a form given
5169 as an "--eval" command-line option. (These changes were motivated by a
5170 discussion of stream flushing issues on cmucl-imp in August 2000.)
5171 * The source transform for TYPEP of array types no longer assumes
5172 that an array whose element type is a not-yet-defined type
5173 is implemented as an array of T, but instead punts, so that the
5174 type will be interpreted at runtime.
5175 * There is now some support for cross-compiling in make.sh: each of
5176 the phases of make.sh has its own script. (This should be transparent
5177 to people doing ordinary, non-cross-compile builds.)
5178 * Since my laptop doesn't have hundreds of megabytes of memory like
5179 my desktop machine, I became more motivated to do some items on
5180 my to-do list in order to reduce the size of the system a little:
5181 ** Arrange for various needed-only-at-cold-init things to be
5182 uninterned after cold init. To support this, those things have
5183 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5184 symbols with such names are now uninterned after cold init).
5185 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5186 fasl files for cold load.
5187 ** Remove the old compiler structure pooling code (which used to
5188 be conditional on the target feature :SB-ALLOC) completely.
5189 ** Redo the representation of some data in cold init to be more compact.
5190 (I also looked into supporting byte compiled code at bootstrap time,
5191 which would probably reduce the size of the system a lot, but that
5192 looked too complicated, so I punted for now.)
5193 * The maximum signal nesting depth in the src/runtime/ support code has
5194 been reduced from 4096 to 256. (I don't know any reason for the very
5195 large old value. If the new smaller value turns out to break something,
5196 I'll probably just bump it back up.)
5197 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5199 * Many, many bugs reported by Peter Van Eynde have been added to
5200 the BUGS list; some have even been fixed.
5201 * While enabling dynamic object file loading, I tried to make the
5202 code easier to understand, renaming various functions and variables
5203 with less ambiguous names, and changing some function calling
5204 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5205 * While trying to figure out how to do the OpenBSD port, I tried to
5206 clean up some of the code in src/runtime/. In particular, I dropped
5207 support for non-POSIX signal handling, added various comments,
5208 tweaked the code to reduce the number of compilation warnings, and
5209 renamed some files to increase consistency.
5210 * To support the new automatic configuration functionality in make.sh,
5211 the source file target-features.lisp-expr has been replaced with the
5212 source file base-target-features.lisp-expr and the machine-generated
5213 file local-target-features.lisp-expr.
5214 * fixed a stupid quoting error in make.sh so that using CMU CL
5215 "lisp -batch" as cross-compilation host works again
5217 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5219 * DESCRIBE no longer tries to call itself recursively to describe
5220 bound/fbound values, so that it no longer fails on symbols which are
5221 bound to themselves (like keywords, T, and NIL).
5222 * DESCRIBE now works on generic functions.
5223 * The printer now prints less-screwed-up representations of closures
5224 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5225 * A private symbol is used instead of the :EMPTY keyword previously
5226 used to mark empty slots in hash tables. Thus
5227 (DEFVAR *HT* (MAKE-HASH-TABLE))
5228 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5229 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5230 now does what ANSI says that it should. (You can still get
5231 similar noncompliant behavior if bang on the hash table
5232 implementation with all the symbols you get back from
5233 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5234 This breaks binary compatibility, since tests for equality to
5235 :EMPTY are wired into things like the macroexpansion of
5236 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5238 * There's now a minimal placeholder implementation for CL:STEP,
5239 as required by ANSI.
5240 * An obscure bug in the interaction of the normal compiler, the byte
5241 compiler, inlining, and structure predicates has been patched
5242 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5243 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5245 * Missing ordinary arguments in a macro call are now detected even
5246 when the macro lambda list contains &KEY or &REST.
5247 * The debugger no longer complains about encountering the top of the
5248 stack when you type "FRAME 0" to explicitly instruct it to go to
5249 the top of the stack. And it now prints the frame you request even
5250 if it's the current frame (instead of saying "You are here.").
5251 * As specified by ANSI, the system now always prints keywords
5252 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5254 * The default initial SIZE of HASH-TABLEs is now smaller.
5255 * Type information from CLOS class dispatch is now propagated
5256 into DEFMETHOD bodies, so that e.g.
5257 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5259 is now basically equivalent to
5260 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5261 (DECLARE (TYPE SINGLE-FLOAT X))
5263 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5264 operation, without having to do run-time type dispatch.
5265 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5266 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5267 * Since I don't seem to be making big file reorganizations very often
5268 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5269 is overflowing my ability to conveniently back them up), I've finally
5270 checked the system into CVS. (The CVS repository is on my home system,
5271 not at SourceForge -- putting it on SourceForge might come later.)
5272 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5273 high-level GC-NOTIFY-FOO functions send their output. (There's
5274 still very little control of where low-level verbose GC functions
5275 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5276 less than it used to -- the GC-NOTIFY-FOO functions are now under
5277 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5278 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5279 in only one place in the source code, and propagates it automatically
5280 everywhere that it's needed. Thus e.g. when I bump the version from
5281 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5282 * The C source files now include boilerplate legalese and documentation
5283 at the head of each file (just as the Lisp source files already did).
5284 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5285 to his page will be replaced with a link to his new CLiki service.
5287 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5289 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5290 * The build process now looks for GNU make under the default name "gmake",
5291 instead of "make" as it used to. If GNU make is not available as "gmake"
5292 on your system, you can change this default behavior by setting the
5293 GNUMAKE environment variable.
5294 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5295 can build without error under CMU CL.
5297 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5299 * There is now a partial SBCL user manual (with some new text and some
5300 text cribbed from the CMU CL manual).
5301 * The beginnings of a profiler have been added (starting with the
5302 CMU CL profiler and simplifying and cleaning up). Eventually the
5303 main interface should be through the TRACE macro, but for now,
5304 it's still accessed through vaguely CMU-CL-style functions and macros
5305 exported from the package SB-PROFILE.
5306 * Some problems left over from porting CMU CL to the new
5307 cross-compilation bootstrap process have been cleaned up:
5308 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5309 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5310 part when getting rid of the compiler *BACKEND* stuff.)
5311 ** The value of *NULL-TYPE* was screwed up, because it was
5312 being initialized before the type system knew the final
5313 definition of the 'NULL type. This screwed up several key
5314 optimizations in the compiler, causing inefficiency in all sorts
5315 of places. (I found it because I wanted to understand why
5316 GET-INTERNAL-RUN-TIME was consing.)
5317 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5318 PROCLAIM FTYPE information. Unfortunately this broke binary
5319 compatibility again, since now the forms output by DEFGENERIC
5320 to refer to functions which didn't exist in 0.6.3.
5321 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5322 can use the new (as of 0.6.3) transform for SOME into MAP into
5324 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5325 values to (INTEGER 0 1000000), so that the time code will no longer
5326 occasionally get blown up by Linux returning 1000000 microseconds
5327 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5328 its output conform to the ANSI spec. (Alas, this makes its output
5329 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5330 * A full call to MAP NIL with a single sequence argument no longer conses.
5331 * fixes to problems pointed out by Martin Atzmueller:
5332 * The manual page no longer talks about multiprocessing as though
5333 it were currently supported.
5334 * The ILISP support patches have been removed from the distribution,
5335 because as of version 5.10.1, ILISP now supports SBCL without us
5336 having to maintain patches.
5337 * added a modified version of Raymond Toy's recent CMU CL patch for
5338 EQUALP comparison of HASH-TABLE
5340 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5342 * The system still can't cross-compile itself with
5343 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5344 entails), but at least it can compile more of itself that way
5345 than it used to be able to, and various buglets which were uncovered
5346 by trying to cross-compile itself that way have now been fixed.
5347 * This release breaks binary compatibility again. This time
5348 at least I've incremented the FASL file format version to 2, so that the
5349 problem can be detected reliably instead of just causing weird errors.
5350 * various new style warnings:
5351 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5352 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5353 ** using the lexical binding of a variable named in the *FOO* style
5354 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5355 as specified by ANSI.
5356 * *RANDOM-STATE* is no longer automatically initialized from
5357 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5358 default behavior of the system is to repeat its behavior every time
5359 it's run. If you'd like to change this behavior, you can always
5360 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5361 old convention there was no comparably easy way to get the system to
5362 repeat its behavior every time it was run.
5363 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5364 FTYPE declarations has been removed, in favor of their ANSI
5365 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5366 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5367 the types of their sequence arguments can be determined at compile time.
5368 This is done through a new open code expansion for MAP which eliminates
5369 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5370 argument types can be determined at compile time.
5371 * The optimizer now transforms COERCE into an identity operation when it
5372 can prove that the coerced object is already of the correct type. (This
5373 can be a win for machine generated code, including the output of other
5374 optimization transforms, such as the MAP transform above.)
5375 * Credit information has been moved from source file headers into CREDITS.
5376 * Source file headers have been made more standard.
5377 * The CASE macro now compiles without complaining even when it has
5380 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5382 * (Note that the way that the PCL macroexpansions were rewritten
5383 to accommodate the change in DEFGENERIC below breaks binary
5384 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5385 not run under sbcl-0.6.2. Once we get out of alpha releases,
5386 i.e. hit release 1.0.0, we'll probably try to maintain binary
5387 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5388 and sbcl-1.4.4. Until then, however, it might be fairly common
5389 for maintenance releases to break binary compatibility.)
5390 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5392 * The reporting of unhandled signals has been changed to print some
5393 explanatory text as well as the report form. (Previously only
5394 the report form was printed.)
5395 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5396 it defines, so that the compiler no longer issues undefined function
5397 warnings for compiled-but-not-yet-loaded generic functions.
5398 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5399 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5400 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5401 section of the ANSI spec are supported.
5402 * The "" nickname for the "KEYWORD" package has been removed.
5403 The reader still handles symbol tokens which begin with a package marker
5404 as keywords, but it doesn't expose its mechanism for doing so in the
5405 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5406 * The system now issues STYLE-WARNINGs for contradictory TYPE
5407 proclamations. (Warnings for contradictory FTYPE proclamations would
5408 be nice too, but those can't be done usefully unless the type system
5409 is made smarter about FUNCTION types.)
5410 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5411 systematized, so that "*target-*.lisp is supposed to exist only on the
5412 target and imply that there's a related file which exists on the
5413 host, and *host-*.lisp is supposed to exist only on the host and imply
5414 that there's a related file which exists on the target. This involves a
5415 lot of renaming. Hopefully the acute confusion caused by the renaming
5416 will be justified by the reduction in chronic confusion..
5417 ** runtime-type.lisp -> early-target-type.lisp
5418 ** target-type.lisp -> late-target-type.lisp
5419 ** early-host-format.lisp -> early-format.lisp
5420 ** late-host-format.lisp -> late-format.lisp
5421 ** host-error.lisp -> misc-error.lisp
5422 ** early-error.lisp -> early-target-error.lisp
5423 ** late-error.lisp -> late-target-error.lisp
5424 ** host-defboot.lisp -> early-defboot.lisp
5425 ** code/misc.lisp -> code/target-misc.lisp
5426 ** code/host-misc.lisp -> code/misc.lisp
5427 ** code/numbers.lisp -> code/target-numbers.lisp
5428 ** code/early-numbers.lisp -> numbers.lisp
5429 ** early-host-type.lisp -> early-type.lisp
5430 ** late-host-type.lisp -> late-type.lisp
5431 ** host-typep.lisp -> typep.lisp
5432 ** load.lisp -> target-load.lisp
5433 ** host-load.lisp -> load.lisp
5434 ** host-disassem.lisp -> disassem.lisp
5435 ** host-insts.lisp -> insts.lisp
5436 ** byte-comp.lisp -> target-byte-comp.lisp
5437 ** host-byte-comp.lisp -> byte-comp.lisp
5438 ** host-signal.lisp -> signal.lisp
5439 ** host-defstruct.lisp -> defstruct.lisp
5440 ** late-target-type.lisp -> deftypes-for-target.lisp
5441 Furthermore, several other previously target-only files foo.lisp (e.g.
5442 hash-table.lisp and random.lisp) have been split into a target-and-host
5443 foo.lisp file and a target-only target-foo.lisp file, with their key type
5444 definitions in the target-and-host part, so that the cross-compiler will
5445 know more about target types.
5446 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5447 gone away. In their place are various *BACKEND-FOO* variables
5448 corresponding to the slots of the old structure.
5449 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5450 into the target SBCL has been fixed.
5451 * The chill.lisp system for loading cold code into a running SBCL
5453 * Support for the CMU CL "scavenger hook" extension has been removed.
5454 (It was undocumented and unused in the CMU CL sources that SBCL was
5455 derived from, and stale in sbcl-0.6.1.)
5456 * Various errors in the cross-compiler type system were detected
5457 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5458 (enabling various consistency checks). Many of them were fixed,
5459 but some hard problems remain, so the compiler is back to
5460 running without *TYPE-SYSTEM-INITIALIZED* for now.
5461 * As part of the cross-compiler type system cleanup, I implemented
5462 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5463 * I have started adding UNCROSS calls throughout the type system
5464 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5465 on cross-compiler input files will be able to go away, and various
5467 * CONSTANTP now returns true for quoted forms (as explicitly required
5470 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5472 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5473 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5474 (MAKE-LIST -1). In the longer term, it ought to become true
5475 that declarations are assertions even at SAFETY 1. For now, it's not
5476 quite true even at SAFETY 3, but it's at least more nearly true..
5477 (Note that this change seems to increases the size of the system by
5478 O(5%) and to decrease the speed of the compiler by 20% or more.)
5479 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5480 for the problem of printing dozens of lines of distracting information
5481 about low-level system machinery as part of the top stack frame
5482 on entry to the debugger when an undefined function was called.
5483 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5484 is not reset to COMMON-LISP-USER.
5485 * Compilation of stuff related to dyncount.lisp has been made conditional
5486 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5487 smaller. The various dyncount-related symbols have been moved into
5488 a new "SB-DYNCOUNT" package.
5489 * tty-inspect.lisp has been renamed to inspect.lisp.
5490 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5491 feature has gone away. (When we eventually port to other flavors of
5492 libc and/or Unix, we'll try to make the differences between flavors
5493 invisible at the user level.)
5494 * Various other *FEATURES* tags, and/or their associated conditionals,
5495 have been removed if obsolescent, or given better documentation, or
5496 sometimes given more-mnemonic names.
5498 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5500 * tidied up "make.sh" script
5501 * tidied up system directory structure
5502 * better "clean.sh" behavior
5503 * added doc/FOR-CMUCL-DEVELOPERS
5504 * many many small tweaks to output format, e.g. removing possibly-confusing
5505 trailing #\. character in DESCRIBE-INSTANCE
5506 * (EQUALP #\A 'A) no longer signals an error.
5507 * new hashing code, including EQUALP hashing
5508 * tidied up Lisp initialization and toplevel
5509 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5510 * command line argument processing
5511 * added POSIX-GETENV function to deal with Unix-ish environment variables
5512 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5513 terminating SBCL on EOF
5514 * non-verbose GC by default
5515 * There is no more "sbcl" shell script; the sbcl file is now the C
5516 runtime executable (just like CMU CL).
5517 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5519 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5520 symbol and package information as Lisp native symbol and package objects
5521 instead of strings naming symbols and strings naming packages. This way,
5522 whenever packages are renamed (as in warm init), debug information is
5523 transformed along with everything else.
5524 * tweaked the optimization policy declarations which control the building
5525 of SBCL itself. Now, among other things, the system no longer saves
5526 source location debugging information. (This helps two problems at once
5527 by reducing SBCL size and by keeping SBCL from trying to look for its
5528 sources -- which may not exist -- when reporting errors.)
5529 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5530 debugging and testing purposes
5531 * cleaned up printing, making the printer call PRINT-OBJECT for
5532 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5533 methods, giving nearly-ANSI behavior
5534 * converted almost all special variables to use *FOO* naming convention
5535 * deleted PARSE-TIME functionality, since it can be done portably
5536 * moved some files out of cold init into warm init
5537 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5539 * regularized formatting of source files
5540 * added an install.sh script
5541 * fixed ridiculous memory usage of cross-compiler by making
5542 compiler/alloc.lisp not try to do pooling unless it can hook
5543 itself into the GC of the cross-compilation host. Now the system
5544 builds nicely on my old laptop.
5545 * added :SB-ALLOC in target-features.lisp-expr
5546 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5547 was not implemented)
5548 * re-did condition handling and note reporting in the compiler. Notes
5549 are no longer handled by signalling conditions. Style warnings
5550 and warnings are handled more correctly and reported in such a way
5551 that it's easy to find one or the other in your output (so that you
5552 can e.g. figure out which of many problems caused COMPILE-FILE to
5554 * changed the severity of several compiler warnings from full WARNING
5555 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5556 compiler note reporting so that it doesn't use the condition system
5557 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5558 COMPILE-FILE command)
5559 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5560 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5561 no longer supported, since their implementation was deeply intertwingled
5562 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5563 * removed bogus "support" for compiler macros named (SETF FOO), and
5564 removed the compiler macro for SETF INFO (but only after making a fool
5565 of myself on the cmucl-imp mailing list by posting a bogus patch for
5566 DEFINE-COMPILER-MACRO..)
5567 * Compiled files containing forms which have side effects on the Lisp
5568 reader (such as DEFPACKAGE forms) are now handled more correctly.
5569 (Compiler queuing of top level lambdas has been suppressed by setting
5570 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5571 * deleted various currently-unused source files, e.g. gengc.lisp. They
5572 may be added back at some point e.g. when porting to other architectures,
5573 but until they are it's distracting to distribute them and to try to
5575 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5576 there were so many of them they're just distractions, and UNCROSS is
5577 known to be able to handle the current sources
5578 * moved PROFILE functionality into TRACE, so that it will be clear
5579 how the wrapping and unwrapping of functions when you profile them
5580 interacts with the wrapping and unwrapping of functions when you
5581 trace them. (Actually, the functionality isn't there yet, but at least
5582 the interface specification is there. Hopefully, the functionality will
5583 arrive with some future maintenance release.)
5584 * removed host-oops.lisp
5585 * changed signature of QUIT function to allow UNIX-CODE argument
5586 * fixed READ-SEQUENCE bug
5587 * tweaked verbose GC output so that it looks more like the progress
5588 output that ANSI specifies for functions like LOAD
5589 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5590 * added <http://sbcl.sourceforge.com> to the banner information printed by