1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.36:
3 * enhancement: Backtrace from THROW to uncaught tag on x86oids now shows
4 stack frame thrown from.
5 * enhancement: WITH-COMPILATION-UNIT :POLICY allows restricting changes to
6 compiler optimization qualities inside dynamic extent of its body.
7 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) take advantage of
8 constraint propgation, allowing better compilation eg. when used to
9 access structures with WITH-SLOTS. (lp#520366)
10 * optimization: the compiler is now more aware of the type of the underlying
11 storage vector for multidimensional simple arrays resulting in better code
12 for accessing such arrays.
13 * optimization: passing NIL as the environment argument to TYPEP no longer
14 inhibits optimizing it. (lp#309788)
15 * optimization: ADJUST-ARRAY and STABLE-SORT on vectors no longer use
16 pre-allocated temporary vectors. (lp#496249)
17 * bug fix: Fix compiler error involving MAKE-ARRAY and IF forms
18 in :INITIAL-CONTENTS. (lp#523612)
19 * bug fix: FUNCTION-LAMBDA-EXPRESSION lost declarations from interpreted
20 functions. (lp#524707)
21 * bug fix: bogus style warnings from certain (SETF SLOT-VALUE) and
22 WITH-SLOTS usages during compilation.
23 * bug fix: SB-C::CLASS-INFO now prints correctly. (lp#514762)
24 * enhancement: Can now build with ud2 instead of int3 as trap instruction on
25 all x86oid platforms with :UD2-BREAKPOINTS target feature.
26 * bug fix: Breakpoints now work when using ud2 instead of int3 as trap
27 instruction (tested on x86oid linux with ud2-breakpoints).
28 * bug fix: slam.sh now works on win32.
29 * bug fix: better differences of numeric types. (lp#309124)
30 * bug fix: arrays declared intersection and union types can have their
31 upgraded element type derived. (lp#316078)
32 * bug fix: SB-SPROF allocation profiling for all threads failed to profile
33 threads started during profiling. (lp#472499)
34 * bug fix: SB-INTROSPECT test failure when building without SB-EVAL feature.
36 * bug fix: SB-CLTL2:DECLARATION-INFORMATION did not take
37 SB-EXT:RESTRICT-COMPILER-POLICY into account. (lp#313337)
39 changes in sbcl-1.0.36 relative to sbcl-1.0.35:
40 * new feature: SB-EXT:TYPEXPAND-1, SB-EXT:TYPEXPAND, and
41 SB-EXT:TYPEXPAND-ALL behave exactly like their MACROEXPAND counterparts
42 but work on type specifiers.
43 * new feature: SB-EXT:DEFINED-TYPE-NAME-P returns whether a symbol is known
44 to name a type specifier.
45 * new feature: SB-EXT:VALID-TYPE-SPECIFIER-P returns whether a given type
46 specifier is valid where "valid" basically means "would be accepted as
47 second argument of TYPEP".
48 * new feature: SB-INTROSPECT:FUNCTION-TYPE takes a function-designator and
49 returns the function's declared, or derived FTYPE.
50 * new feature: SB-POSIX now supports accessing the d_ino member of
51 dirent structures. (Thanks to Philipp Marek and Pierre THEIRRY)
52 * new feature: The function SB-EXT:SEED-RANDOM-STATE has been added to
53 provide for seeding a RANDOM-STATE object with user-provided data or
54 from the operating system's PRNG. Also, (MAKE-RANDOM-STATE T) will
55 attempt to initialize the returned state from the operating system's
56 PRNG where possible. (Thanks to Fare Rideau; launchpad bug lp#310116)
57 * bug fix: Fix SB-SIMPLE-STREAMS:READ-VECTOR to correctly set the
58 FILE-POSITION of the stream being read from. (launchpad bug lp#491087)
59 * bug fix: Fix grammar and style issues for the docstrings of
60 printer-related variables and functions. (Thanks to mon_key; launchpad
62 * bug fix: Fix compilation on chenygc platforms. Thanks to Larry Valkama and
64 * bug fix: SB-THREAD:CONDITION-WAIT sometimes signaled a deadline twice
65 in a row even though a handler defered the deadline long into the
67 * bug fix: A deadline handler was run without interrupts enabled for a
68 deadline signaled within SB-THREAD:CONDITION-WAIT. That could result
69 in infinitely spinning, non-killable threads.
70 * bug fix: Backtrace from internal-errors on x86-64 os x was truncated
71 before reaching the erring stack frame.
72 * bug fix: Fix type derivation for EXPT when raising a fixnum to a
73 real power. (launchpad bug lp#525949)
74 * bug fix: Fix SB-EXT:GENERATION-* accessors for generations > 0 on
75 GENCGC platforms. (launchpad bug lp#529014)
76 * bug fix: More robust checks for invalid DEFMETHOD argument specializers.
77 (launchpad bug lp#525916)
78 * bug fix: Fix building on Darwin when sysctl is not in the user's PATH.
79 (Thanks to Robert Goldman)
81 changes in sbcl-1.0.35 relative to sbcl-1.0.34:
82 * optimization: ROUND with a single single-float or double-float argument
83 is properly inlined when possible.
84 * optimization: Slightly better code is generated for integer<->float
85 conversions and for single-float<->double-float conversions on x86-64.
86 * optimization: SB-ROTATE-BYTE:ROTATE-BYTE now generates more efficient
87 code for 32-bit and 64-bit rotations on x86-64.
88 * bug fix: The install script changes the ownership of directories as well
89 as files for contrib modules using asdf. (thanks to Eugene Ossintsev;
90 launchpad bug lp#508485)
91 * bug fix: TRUNCATE with a single single-float or double-float argument is
92 properly inlined when possible. (launchpad bug lp#489388)
93 * bug fix: Passing a rotation count of zero to SB-ROTATE-BYTE:ROTATE-BYTE
94 no longer causes a compiler error on x86 and ppc.
95 * bug fix: GET-MACRO-CHARACTER bogusly computed its second return value
96 always relative to *READTABLE* rather than the passed argument.
98 changes in sbcl-1.0.34 relative to sbcl-1.0.33:
99 * minor incompatible change: threading support is now enabled by default
101 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
103 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
104 are now caught and reported just like errors during macroexpansion.
105 * enhancement: SB-POSIX now provides access to tcdrain(), tcflow(),
106 tcflush(), tcgetsid(), and tcsendbreak(). (thanks to Jerry James)
107 * enhancement: ASDF systems can now depends on SB-QUEUE.
108 * fixes and improvements related to Unicode and external formats:
109 ** bug fix: error handling and restart usage in the ucs-2 external format
111 ** there is now an implementation of the ucs-4 external format.
112 ** the utf-16 and utf-32 external formats are supported.
113 * bug fix: SB-POSIX wrapper for putenv no longer tries to put lisp strings
114 in the environment. setenv() and unsetenv() are also provided. (reported by
115 Fare Rideau; launchpad bug lp#460455)
116 * bug fix: LOAD of both .fasl and .FASL type files now forces fasl-style
117 loading. This ensures sensible errors for .FASL files from other
118 implementations on case-insensitive filesystems. (reported by Willem
119 Broekema; launchpad bug lp#489417)
120 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
121 Willem Broekema; launchpad bug lp#489698).
122 * bug fix: some minor code rearrangements to reenable warning-free building
123 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
124 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
126 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
129 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
130 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
131 * improvement: support O_LARGEFILE access to files larger than 2GB on
132 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
133 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
134 definitions for methods specializing on the passed class itself.
135 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
136 definitions for methods specializing on the passed class itself, or on
138 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
139 about itself during the build (e.g. for M-? in Slime), if this flag is
140 enabled in customize-target-features.lisp. This will increase the core
141 size by about 5-6mb, though, so it's mostly interesting to SBCL
143 * new feature: various GENCGC tuning parameters have been experimentally
144 documented and exported from SB-EXT. See documentation for details.
145 * fixes and improvements related to Unicode and external formats:
146 ** the Unicode character database has been upgraded to the
147 Unicode 5.2 standard, giving names and properties to a number of new
148 characters, and providing a few extra characters with case
150 ** improvement: restarts for providing replacement input/output on coding
151 errors for fd-stream external formats.
152 ** improvement: where :<encoding> is a keyword corresponding to an
153 external format the system supports, it is now possible to specify
154 (:<encoding> :replacement <character>) as an external format which will
155 automatically substitute <character> on encoding or decoding errors for
156 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
158 ** improvement: the file streams underlying the standard streams (such as
159 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
160 which uses the replacement mechanism to handle encoding errors,
161 preventing various infinite error chains and unrecoverable I/O
163 ** minor incompatible change: the utf-8 external format now correctly
164 refuses to encode Lisp characters in the surrogate range (char-codes
165 between #xd800 and #xdfff).
166 ** fix a typo preventing conversion of strings into octet vectors
167 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
169 ** fix a bug in the octet multibyte handling of decoding errors and the
170 USE-VALUE restart. (launchpad bug lp#314939)
171 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
172 restart on fd-stream decoding errors.
173 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
174 error is near the end of file.
175 ** fix a double-error case in unibyte octet conversions, when the first
176 use of USE-VALUE is ignored.
177 ** fix bugs in handling of undefined code points in unibyte encodings.
178 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
179 streams after an UNREAD-CHAR.
180 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
181 object is allocated in a boxed region of dynamic space.
182 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
183 fork with multiple Lisp threads running is made, instead of going
184 ahead with unpredictable consequences. (reported by Leslie Polzer)
185 * bug fix: uses of slot accessors on specialized method parameters within
186 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
187 error while finalizing the class. This fix may cause classes with slot
188 accessors to be finalized later than previously. (reported by Lars Rune
189 Nøstdal; launchpad bug lp#473699)
190 * bug fix: restore buildability on the MIPS platform. (regression from
191 1.0.30.38, reported by Samium Gromoff)
192 * bug fix: inspecting closures is less likely to fail with a type error.
193 * bug fix: no timer starvation when setting the system clock back.
194 (launchpad bug lp#460283)
195 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
196 standard pprint dispatch table as specified by CLHS.
197 * bug fix: give CLISP a hint about a type declaration to enable it to build
198 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
200 * bug fix: correctly dump literal objects in defaulting forms of arglists.
201 (reported by Attila Lendvai; launchpad bug lp#310132)
202 * bug fix: distinguish in type specifiers between arrays that might be
203 complex and arrays that are definitely complex. (launchpad bug lp#309129)
204 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
205 type. (reported by Levente Mészáros; launchpad bug lp#485972)
206 * bug fix: setting the value of a symbol-macro within a method in the
207 presence of type declarations works properly again. (reported by Iban
208 Hatchondo; launchpad bug lp#485019)
210 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
211 * optimization: faster FIND and POSITION on strings of unknown element type
212 in high SPEED policies. (thanks to Karol Swietlicki)
213 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
215 * improvement: better error signalling for bogus parameter specializer names
216 in DEFMETHOD forms (reported by Pluijzer)
217 * bug fix: DELETE-FILE once again works on logical pathnames (regression
219 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
221 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
222 removes previous default initargs (reported by Lars Rune Nøstdal and
224 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
225 lambda-lists. (reported by Peter Seibel)
226 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
227 correctly when starting from an executable core without saved runtime
228 options (reported by Faré Rideau, thanks to Zach Beane)
229 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
230 an optimization note instead. (reported by Martin Cracauer)
231 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
232 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
233 new process group if it doesn't need to share stdin with the sbcl
234 process. (thanks to Leslie Polzer)
235 * bug fix: SATISFIES could be misoptimized to refer to a local function.
236 (reported by Stanislaw Halik)
238 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
239 * improvement: stack allocation is should now be possible in all nested
240 inlining cases: failure to stack allocate when equivalent code is manually
241 open coded is now considered a bug.
242 * improvements related to Unicode:
243 ** the Unicode character database has been upgraded to the
244 Unicode 5.1 standard, giving names and properties to a number of new
245 characters, and providing a few extra characters with case
247 ** the system now recognizes and produces names for Unicode Hangul
249 ** the EBCDIC-US external-format is now supported for octet operations
250 (as well as for stream operations).
251 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
252 be used to output toplevel compile-time effects into a separate .CFASL
254 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
255 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
257 * optimization: TRUNCATE on known single- and double-floats is upto 25%
259 * optimization: division of floating point numbers by constants uses
260 multiplication by reciprocal when an exact reciprocal exists.
261 * optimization: multiplication of single- and double-floats floats by
262 constant two has been optimized.
263 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
264 sufficient type information is available. (thanks to Leslie Polzer)
265 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
266 known structure objects are as efficient as defstruct generated accessors.
267 * optimization: unused vector creation can now be optimized away.
268 * improvement: ASDF systems can now depends on SB-INTROSPECT.
269 * improvement: a STYLE-WARNING is signalled when a generic function
270 clobbers an earlier FTYPE proclamation.
271 * improvement: the compiler is able to track the effective type of
272 generic function across method addition and removal even in the
273 absence of an explicit DEFGENERIC.
274 * improvement: DESCRIBE now reports on symbols naming undefined
275 but assumed or declared function as well.
276 * improvement: recompilation of systems using SB-GROVEL now works
277 (thanks to Leslie Polzer)
278 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
279 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
281 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
282 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
283 well as user defined declaration names.
284 ** VARIABLE-INFORMATION is now aware of alien variables.
285 * improvement: improved address space layout on OpenBSD (thanks to Josh
287 * improvement: pretty-printing of various Lisp forms has been improved
288 (thanks to Tobias Rittweiler)
289 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
290 unused were deleted in safe code. (reported by John Fremlin)
291 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
293 * bug fix: some out-of-line array predicates were missing (reported by
295 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
297 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
298 (thanks to Larry D'Anna)
299 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
300 as well. (reported by Josh Elsasser)
301 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
302 leading to possible attempts to delete the same file twice. See docstring
303 on DELETE-FILE for details. (reported by John Fremlin)
304 * bug fix: DELETE-FILE once again deletes the file named by the pathname
305 designator argument, rather than its truename. (reported by Luis
307 * bug fix: the low-level debugger had 32-bit assumptions and was missing
308 information about some array types. (thanks to Luis Oliveira)
309 * bug fix: moderately complex combinations of inline expansions could
310 be miscompiled if the result was declared to be dynamic extent.
311 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
312 form (+ <variable> <integer>) were miscompiled under certain
314 * bug fix: in some cases no compiler note about failure to stack allocate
315 was emitted, even if the objects were in fact heap allocated.
316 * bug fix: minor violation of "otherwise inaccessible" rule for stack
317 allocation could cause objects users might reasonably expect to
318 be heap allocated to be stack allocated.
319 * bug fix: DESCRIBE signalled an error for generic functions under
320 certain circumstances. (thanks to Leslie Polzer)
321 * bug fix: Fixed spelling of an error message.
323 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
324 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
325 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
326 of SB-THREAD:THREAD-ERROR-THREAD.
327 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
328 * new feature: docstrings for local and anonymous functions are no longer
329 discarded. (thanks to Leslie Polzer)
330 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
331 values in other threads.
332 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
333 about object allocation.
334 * optimization: division of a real float by a complex float is implemented
335 with a specialised code sequence.
336 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
337 keywords is an order of magnitude faster.
338 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
339 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
340 methods, and similarly for non-standard metaclass classes as long as there
341 are no methods additional on MAKE-INSTANCE.
342 * optimization: more efficient type-checks for FIXNUMs when the value
343 is known to be a signed word on x86 and x86-64.
344 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
345 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
347 * optimization: compiler is smarter about delegating argument type checks to
349 * optimization: several character functions are now compiled somewhat more
350 efficiently. (reported by Lynn Quam)
351 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
352 * improvement: less unsafe constant folding in floating point arithmetic,
353 especially for mixed complex/real -float operations.
354 * optimization: constant double and single floats are stored in native
355 unboxed format on x86[-64].
356 * optimization: smarter code for arithmetic operations with constant floats,
357 complex floats, or integers on x86[-64].
358 * optimization: smarter code for conjugate/multiplication of float complexes
359 and abs/negate of floats on x86-64.
360 * optimization: more efficient complex float and real float operations on
362 * improvement: complex float division is slightly more stable.
363 * improvement: DESCRIBE output has been reworked to be easier to read and
364 contains more pertinent information.
365 * improvement: failure to provide requested stack allocation compiler notes
366 provided in all cases (requested stack allocation not happening without a
367 note being issued is now considered a bug.)
368 * bug fix: SB-POSIX exports the documented types and functions
369 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
370 types. (reported by "abhi")
371 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
372 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
373 (thanks to Paul Khuong)
374 * bug fix: looping around HANDLER-CASE could silently consume stack space
375 on each iteration. (reported by "foobar")
376 * bug fix: better error signalling when calls to functions seeking elements
377 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
378 (reported by Tobias Rittweiler)
379 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
380 directory components now works as it used to. (various prolems reported by
381 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
382 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
383 (reported by Elliot Slaughter, thanks to Stas Boukarev)
384 * bug fix: bogus undefined variable warnings from fopcompiled references to
385 global variables. (thanks to Lars Rune Nøstdal)
386 * bug fix: foreign function names should now appear in backtraces on
387 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
388 * bug fix: SETF compiler macro documentation strings are not discarded
390 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
391 (thanks to Alex Plotnick)
392 * bug fix: SLEEP accepts large integer arguments, truncating them to
393 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
394 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
395 * bug fix: compiler notes for expensive slot type checks could be emitted
396 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
398 changes in sbcl-1.0.29 relative to 1.0.28:
399 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
400 https://bugs.launchpad.net/sbcl
401 Bugs can be reported directly there, or by sending email to
402 sbcl-bugs@lists.sourceforge.net
403 (no subscription required.)
404 * minor incompatible change: under weak type checking policy integer
405 types are weakened less aggressively.
406 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
407 allowed to return, which causes SBCL to quit with exit status 0. Previously
408 if the function returned with a small integer return value, that value
409 was accidentally reused as the exit status.
410 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
411 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
412 :HASH-FUNCTION argument. Refer to user manual for details.
413 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
415 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
416 microseconds since the Unix epoch on all platforms.
417 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
418 allows the compiler to safely elide boundedness checks for special
420 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
421 the symbol, prohibits both lexical and dynamic binding. This is mainly an
422 efficiency measure for threaded platforms, but also valueable in
424 * new feature: UNC pathnames are now understood by the system on Windows.
425 * optimization: the compiler uses a specialized version of FILL when the
426 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
428 * optimization: accesses to potentially non-simple arrays where element type
429 is known are 50% faster.
430 * optimization: compiler now generates faster array typechecking code.
431 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
433 * optimization: multidimensional array accesses in the absence of type
434 information regarding array rank are approximately 10% faster due to
435 open coding of ARRAY-RANK.
436 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
437 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
439 * optimization: result of call to VECTOR can now be stack allocated.
440 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
441 as long as the resulting array is one-dimensional and has a known
442 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
443 length of the list matches the known length of the vector does not
444 allocate the list as an intermediate step. Ditto for VECTOR and simple
446 * optimization: MAKE-ARRAY can now stack allocate in the presence of
447 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
448 known element type, and is known to be simple and one dimensional.
449 * improvement: SBCL now emits a compiler note where stack allocation was
450 requested but could not be provided (not in all cases, unfortunately)
451 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
452 * improvement: pretty-printing loop has been implemented properly. (thanks
453 to Tobias Rittweiler)
454 * documentation: CLOS slot typechecing policy has been documented.
455 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
456 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
457 pointer, leading to memory faults. (thanks to Bart Botta)
458 * bug fix: better floating point exception handling on x86/OpenBSD.
459 (thanks to Josh Elsasser)
460 * bug fix: exit status from QUIT when called under --script was lost
461 (reported by Hubert Kauker)
462 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
463 same implementation of FILL to initialize the array, even if a more
464 efficient one was available (reported by Stas Boukarev, thanks to
466 * bug fix: potential miscompilation of array stack allocation on x86 and
467 x86-64. (reported by Time Tossavainen)
468 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
469 result in their subforms being treated as top level forms. (reported by
471 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
472 than a machine register could leave garbage in the high bits of the
473 result register (bug 316325).
474 * bug fix: disable address space randomization Linux/x86-64 as well,
475 not just x86-64. (reported by Ken Olum)
476 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
477 generate incorrect code.
478 * bug fix: #201; type inference for CONS and ARRAY types could derive
479 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
480 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
481 could not be assigned to or bound with PROGV. (reported by Lorenz
483 * bug fix: the value of CL:- in the inspector was the previous expression
484 evaluated rather than the expression being evaluated.
485 * bug fix: constants can no longer be locally declared special.
486 * bug fix: signals delivered to threads started from foreign land (read:
487 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
488 thread by blocking all signals and resignalling.
489 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
490 from :INITFORM, if any.
492 changes in sbcl-1.0.28 relative to 1.0.27:
493 * a number of bugs in cross-compilation have been fixed, with the ultimate
494 result that building under (at least) clisp should be much more reliable.
495 * minor incompatible changes: echo-streams now propagate unread-char to the
496 underlying input stream, and no longer permit unreading more than one
498 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
499 frames, allowing for instance more reliable backtraces.
500 * improvement: the debugger REPL can now reference lexical variables
501 by name directly for code compiled with (DEBUG 3).
502 * improvement: errors from malformed declarations now have better source
503 paths associated with them. (thanks to Tobias Rittweiler)
504 * optimization: faster local calls on x86/x86-64
505 * bug fix: some error messages for out-of-bound array indexes confused the
506 index and the bound. (thanks to Stas Boukarev)
507 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
509 * bug fix: running regressions tests in shells without OSTYPE set now works.
510 (reported by Harald Hanche-Olsen)
511 * bug fix: more robust static space exhaustion signalling from
512 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
513 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
514 docstring away instead of storing it under names such as (LAMBDA (X)).
515 (reported by Leslie Polzer)
516 * bug fix: timers could go off in the wrong order, be delayed indefinitely
517 (thanks to Ole Arndt for the patch)
518 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
519 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
520 potentially-invalid effective methods in its cache.
521 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
522 instances as well (thanks to Paul Khuong)
523 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
524 SB-POSIX:WAITPID and their C equivalents.
525 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
527 changes in sbcl-1.0.27 relative to 1.0.26:
528 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
529 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
530 * improvement: the system either recovers from stack exhaustion or dies
531 properly as opposed to leaving the user uncertain of whether the handler
532 trampled on some random memory next to the stack or having to rely on
533 --lose-on-corruption (which is still a good idea to use in production
534 because stack exhaustion can happen in signal handlers which will likely
536 * bug fix: fix gc related interrupt handling bug on ppc (regression from
537 1.0.25.37, reported by Harald Hanche-Olsen)
538 * bug fix: work around signal delivery bug in darwin (regression from
539 1.0.25.44, reported by Sidney Markowitz)
540 * bug fix: fix ERROR leaking memory (reported by David Thompson)
542 changes in sbcl-1.0.26 relative to 1.0.25:
543 * incompatible change: an interruption (be it a function passed to
544 INTERRUPT-THREAD or a timer function) runs in an environment where
545 interrupts can be enabled. The interruption can use
546 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
547 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
548 potentially running out of stack. Keep in mind that in the absence
549 of WITHOUT-INTERRUPTS some potentially blocking operation such as
550 acquiring a lock can enable interrupts.
551 * incompatible change: GC-OFF and GC-ON are removed, as they were
552 always unsafe. Use WITHOUT-GCING instead.
553 * new feature: runtime option --disable-ldb
554 * new feature: runtime option --lose-on-corruption to die at the
555 slightest hint of possibly non-recoverable errors: running out of
556 memory, stack, alien stack, binding stack, encountering a memory
557 fault, etc. In the absence of --lose-on-corruption a warning is
559 * enhancement: detect binding stack exhaustion
560 * enhancement: detect alien stack exhaustion on x86/x86-64
561 * improvement: generally more stable and reliable interrupt handling
562 * improvement: there is a per thread interruption queue,
563 interruptions are executed in order of arrival
564 * improvement: a repeating timer reschedules itself when the it has
565 finished, but expiration times are spaced equally. If an
566 expiration time is in the past it will trigger after a short grace
567 period that may give a chance to other things to run.
568 * optimization: slightly faster gc on multithreaded builds
569 * optimization: faster WITHOUT-GCING
570 * bug fix: when JOIN-THREAD signals an error, do it when not holding
571 important locks so that the debugger/handler doesn't produce
572 recursive errors or deadlock.
573 * bug fix: real-time signals are not used anymore, so no more
574 hanging when the system wide real-time signal queue gets full.
575 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
577 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
578 * bug fix: fix random memory faults related to interrupts on alpha
579 * bug fix: fix random memory faults related to interrupts on sparc
580 * bug fix: fix deadlocks related to starting threads
581 * bug fix: fix deadlines on locks on futex platforms
582 * bug fix: restore errno in signal handlers
583 * bug fix: fix deadlocks related to hash tables
584 * bug fix: fix deadlocks in pcl
586 changes in sbcl-1.0.25 relative to 1.0.24:
587 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
588 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
589 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
590 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
591 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
592 better name and does not return values so stale on multiprocessor systems.
593 Also, HOLDING-MUTEX-P was added for about the only sane usage of
595 * improvement: unithread builds keep track of MUTEX-VALUE.
596 * improvement: reading from a TWO-WAY-STREAM does not touch the output
597 stream anymore making it thread safe to have a concurrent reader and
598 a writer, for instance, in a pipe.
599 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
600 arguments, making compiler-macros for SETF-functions able to inspect
601 their constant arguments.
602 * improvement: COMPILE-FILE reports times with millisecond accuracy
603 (thanks to Luis Oliveira)
604 * optimization: CHAR-CODE type derivation has been improved, making
605 TYPEP elimination on subtypes of CHARACTER work better. (reported
606 by Tobias Rittweiler, patch by Paul Khuong)
607 * bug fix: setting alien structure fields of type struct by value now
608 computes the right offset for the memory copy.
609 * bug fix: compilation problem involving inlined calls to aliens with
610 result type VOID. (reported by Ken Olum)
611 * bug fix: #235a; sequential inline expasion in different policies no
612 longer reuses the functional from the previous expansion site.
613 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
614 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
615 * bug fix: named ENUMs in multiply-referenced alien record types no longer
616 cause an implied type redefinition. Regression from 1.0.21.29.
617 * bug fix: sign-extension of small signed return values now works with C
618 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
619 * improvements to the Windows port:
620 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
621 Westfold, thanks to Rudi Schlatte)
623 changes in sbcl-1.0.24 relative to 1.0.23:
624 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
625 vector of a multidimensional SIMPLE-ARRAY.
626 * new feature: the system now signals a continuable error if standard
627 readtable modification is attempted. (thanks to Tobias Rittweiler)
628 * new feature: DIRECTORY has been extended with a non-standard keyword
629 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
630 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
631 has been robustified and documented better. (thanks to Robert Goldman)
632 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
633 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
634 * tradeoff: constant FORMAT control strings are now compiled unless
635 SPACE > SPEED (previously only when SPEED > SPACE.)
636 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
638 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
640 * bug fix: lexical type declarations are now correctly reported by
641 SB-CLTL2. (reported by Larry D'Anna)
642 * bug fix: STRING-TO-OCTETS did not handle :START properly when
643 using UTF-8 as external format. (thanks to Luis Oliveira)
644 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
645 are signalled correctly. (thanks to Stas Boukarev)
646 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
647 designator. (thanks to Tobias Rittweiler)
648 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
649 designator, and returns T instead of the function. (thanks to
651 * bug fix: direct superclasses of STANDARD-CLASS and
652 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
653 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
654 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
655 when no class with the named slot yet exists no longer causes a
656 compile-time style-warning.
657 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
658 in safe code. (reported by Didier Verna)
659 * bug fix: #430; nested structure constructors can stack allocate.
660 * bug fix: on some 64-bit platforms dynamic space size was truncated
661 to #xffff0000 bytes. (reported by Benjamin Lambert)
662 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
663 REPL unusable. (reported by Daniel Herring)
664 * bug fix: return values of READ-SEQUENCE did not take :START into
665 account on file streams, regressions since 1.0.12.22. (reported by
666 Thomas Russ, patch by Paul Khuong)
667 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
668 method specializer used to confuse permuation vector optimization.
669 * bug fix: system inserted bogus implicit type declarations for local
670 special variables in DEFMETHOD bodies.
671 * bug fix: #354; duplicated frames in backtraces due to
672 non-tail-call-optimized XEPs to functions with return type NIL
674 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
675 initializes structure object slots according to DEFSTRUCT initforms,
676 and DEFSTRUCT forms :INCLUDEind structure classes defined using
677 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
678 (reported by Bruno Haible and Stephen Wilson)
679 * bug fix: #395; fill-pointer output streams used now support
680 element-type BASE-CHAR as well.
681 * bug fix: compiler error when attempting to derive return value of
682 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
684 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
686 * bug fix: Mac OS X binaries should now be portable between Leopard
689 changes in sbcl-1.0.23 relative to 1.0.22:
690 * enhancement: when disassembling method functions, disassembly
691 for the associated fast function is also produced.
692 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
694 * optimization: printing with *PRINT-PRETTY* true is now more
695 efficient as long as the object being printed doesn't require
696 special handling by the pretty printer.
697 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
698 now interact correctly with type declarations.
699 * partial bug fix: PCL detects infinite recursion during wrapper
700 validation. (thanks to Attila Lendvai)
701 * bug fix: #426; nested function calls are inlined properly.
702 Previously if FOO was an inline function, in calls of the form
703 (FOO (FOO ...)) the outer call was not inlined.
704 * bug fix: long long arguments passed in registers to alien
705 functions on PowerPC Linux and NetBSD are now handled correctly
706 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
708 changes in sbcl-1.0.22 relative to 1.0.21:
709 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
710 for the shared object in the current directory, but passes the native
711 namestring of the designated pathname to the operation system's shared
712 object loading function as-it.
713 * minor incompatible change: --disable-debugger toplevel option now takes
714 effect before processing of initialization files and --eval or --load
716 * new feature: new commandline argument: --script, which supports
717 shebang lines. See documentation for details. (based on work by
719 * new feature: SAVE-LISP-AND-DIE can save current values of
720 --dynamic-space-size and --control-stack-size in the executable core,
721 causing it to skip normal runtime option processing. See documentation
722 for details. (thanks to Zach Beane)
723 * enhancement: inoccous calls to EVAL or generic functions dispatching
724 on subclasses of eg. STREAM no longer cause compiler notes to appear.
725 * enhancement: the system no longer resignals errors from --load and
726 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
727 associated with the original error to be lost. (thanks to Ariel
729 * enhancement: :DONT-SAVE keyword argument has been added to
730 LOAD-SHARED-OBJECT for controlling interaction with
731 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
732 the effects of an earlier LOAD-SHARED-OBJECT call.
733 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
734 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
735 provided. (reported by Cedric St-Jean)
736 * bug fix: circularity handling in the reader did not treat raw
737 structure slots correctly. (reported by Cedric St-Jean)
738 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
739 file descriptors when there were none.
740 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
741 was broken. (reporteed by Willem Broekema)
742 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
743 pathnames without a directory.
744 * bug fix: COMPILE-FILE did not warn about undefined variable
745 references at toplevel, and LOAD of the resulting fasl did
747 * bug fix: functions with non-required arguments used to end up with
748 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
749 * bug fix: redefining a function with non-required arguments didn't
750 update the system's knowledge about its call signature properly.
751 * bug fix: fixed #431; incompatible alien record type redefinitions
752 are detected and handled. (reported by Neil Haven)
753 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
754 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
756 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
757 after alien stack frames.
758 * bug fix: strings passed to mk{s,d}temp lacked null termination
760 changes in sbcl-1.0.21 relative to 1.0.20:
761 * new feature: the compiler is able to track the effective type of a
762 generic function across method addition and removal.
763 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
764 appropriately typed structure slots without locking.
765 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
766 information like those gathered by TIME using a programming-friendly
768 * new feature: TIME reports time taken even if the form performs a
769 non-local transfer of control.
770 * enhancement: reduced conservativism on GENCGC platforms: on
771 average 45% less pages pinned (measured from SBCL self build).
772 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
773 mutate constant symbols or violate declaimed type of the symbol.
774 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
775 sleeping on the semaphore in heavily contested semaphores.
776 * bug fix: semaphores and condition variables were not interrupt
778 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
779 owned by other threads anymore.
780 * bug fix: FIND on lists called KEY outside the specified
781 subsequence. (reported by budden)
782 * bug fix: LOG doesn't use single-float intermediate results when
783 given mixed integer and double-float arguments, leading to better
784 precision. (reported by Bob Felts)
785 * bug fix: LOG with base zero returned values of inconsistent type.
787 changes in sbcl-1.0.20 relative to 1.0.19:
788 * minor incompatible change: OPTIMIZE qualities
789 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
790 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
791 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
792 * documentation: some slot access efficiency guidelines have been
793 added to the user manual.
794 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
795 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
796 as ASSOC and MEMEBER.
797 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
798 to more efficient EQ-comparison versions more often.
799 * optimization: enhanced derivation of DOLIST iteration variable type
801 * optimization: constant folding of simple (LIST ...) forms as DOLIST
803 * optimization: runtime lookup of function definitions can be
804 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
805 * optimization: compiler is able to derive the return type of
806 (AREF (THE STRING X) Y) as being CHARACTER.
807 * optimization: CLRHASH on empty hash-tables no longer does pointless
808 work. (thanks to Alec Berryman)
809 * optimization: REPLACE deftransforms don't punt when :START1 or
811 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
812 no longer emits a code deletion note. (reported by Andrew
814 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
815 functions with an odd number of &OPTIONAL arguments, a &REST
816 argument, and one or more &KEY arguments at the call site.
817 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
818 arguments appeared at call sites as well.
819 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
820 READ-LINE &co can no longer read from them afterwards. (reported
822 * bug fix: fixed #427: unused local aliens no longer cause compiler
823 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
825 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
826 alien-stack leakage. (reported by Andy Hefner)
827 * bug fix: PROGV signals an error when an attempt to violate declared
828 type of a variable or bind a constant is made.
829 * bug fix: SET signals an error when an attempt to violate declared
830 type of a variable is made.
831 * bug fix: restart computation during the execution of a restart
832 test function no longer causes infinite recursion. (reported by
834 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
835 signals an error. (thanks to Pierre Mai)
836 * bug fix: EXPORT left symbol unexported in conflict situations.
837 (thanks to Michael Weber)
838 * bug fix: correctly handle name conflicts involving CL:NIL.
839 (thanks to Michael Weber)
840 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
841 conflicts arising from USEing package with conflicting symbols
842 correctly. (thanks to Michael Weber)
844 changes in sbcl-1.0.19 relative to 1.0.18:
845 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
846 warnings that go otherwise unhandled will be muffled if they are
847 of the type that's the value of this variable.
848 * optimization: stack allocation is slightly more efficient on x86
850 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
851 where a raw slot always is initialized using the initform whose
852 type is not know sufficiently well a compile-time are now compiled
853 correctly. (reported by John Morrison)
854 * bug fix: compiler no longer makes erronous assumptions in the
855 presence of non-foldable SATISFIES types.
856 * bug fix: stack analysis missed cleanups of dynamic-extent
857 arguments in non-let-converted calls to local functions.
858 * improvements to the Windows port:
859 ** adjusted address spaces for building on both Win32 and
860 Win64. (thanks for John Connors)
861 * fixed some bugs revealed by Paul Dietz' test suite:
862 ** interval arithmetic during type derivation used inexact integer
863 to single-float coercions.
864 ** arithmetic operations involving large integers and single
865 floats give the same results in compiled and interpreted code.
866 ** deriving the result type of COERCE no longer signals an error
867 if the derived type of the second argument is a MEMBER type
868 containing invalid type specifiers.
869 ** ADJOIN with constant NIL as second argument works correctly.
870 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
872 changes in sbcl-1.0.18 relative to 1.0.17:
873 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
874 profiles only the current thread.
875 * minor incompatible change: changes to SYMBOL-VALUE of constants
876 defined with DEFCONSTANT now signal an error.
877 * enhancement: SB-SPROF now has support for wallclock profiling,
878 and is also able to profile specific threads. REPORT output
879 has also additional sorting options.
880 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
882 * optimization: structure allocation has been improved
883 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
884 ** out of line constructors are ~10% faster.
885 ** inline constructors are ~15% faster.
886 ** inline constructors are capable of dynamic extent allocation
887 (generally on x86 and x86-64, in some cases on other platforms
889 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
891 * optimization: file compiler is now able to coalesce non-circular
892 lists, non-base strings, and bit-vectors. Additionally, constants
893 are never referenced through SYMBOL-VALUE at runtime.
894 * optimization: code defining methods on PRINT-OBJECT (and other
895 generic functions in the COMMON-LISP package) now loads faster.
896 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
898 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
899 fasl is now deleted. (reported by Attila Lendvai)
900 * bug fix: READ-LINE always returned NIL for the last line in files.
901 (reported by Yoshinori Tahara)
902 * bug fix: more accurate disassembly annotations of foreign function
903 calls. (thanks to Andy Hefner)
904 * bug fix: trimming non-simple strings and non-string string
905 designators when the there is nothing to trim works properly.
906 (thanks to James Knight)
907 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
908 (thanks to Travis Cross)
909 * fixed some bugs revealed by Paul Dietz' test suite:
910 ** NIL is a valid function name (regression at 1.0.13.38)
911 ** FILL on lists was missing its return value (regression at 1.0.12.27)
912 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
913 fill pointers properly (regression at 1.0.12.23)
914 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
915 displacement indices properly (regression at 1.0.12.23)
917 changes in sbcl-1.0.17 relative to 1.0.16:
918 * temporary regression: user code can no longer allocate closure
919 variable storage on stack, due to bug 419 without explicitly
920 requesting it. Please consult sbcl-devel for advice if you need to
921 use this feature in the meanwhile.
922 * new feature: runtime argument --control-stack-size can be used to
923 adjust thread default control stack size.
924 * enhancement: improved TIME output
925 ** all times are reported using the measured accuracy (milliseconds
926 for real and GC times, microseconds for everything else.)
927 ** processor cycle counts on x86 and x86-64.
928 ** interpreted forms are counted for both evaluator modes.
929 ** number of lambdas converted by the compiler is reported.
930 ** CPU percentage report (computed from real and total run time.)
931 ** more comprehensive run time reporting, using a condenced format
932 ** interperted form, lambda, and page fault counts are omitted
934 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
936 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
937 * optimization: two argument forms of LAST are upto ~10% faster
938 in normal SPEED policies.
939 * optimization: NCONC no longer needs to heap cons its &REST list
940 in normal SPEED policies.
941 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
943 * bug fix: UNION and NUNION work with :TEST-NOT once more,
944 regression since 1.0.9.1. (thanks to Eric Marsden)
945 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
946 since 1.0.15.36. (thanks to Paul Khuong)
947 * bug fix: LAST when always returned the whole list when given a bignum
948 as the second argument.
949 * bug fix: dynamic extent allocation of nested lists and vectors
950 could leak to otherwise accessible parts.
951 * bug fix: invalid optimization of heap-allocated alien variable
953 * bug fix: fasl header checking is less vulnerable to different
954 platform word lengths.
955 * bug fix: more correct assembler syntax for GNU binutils
956 2.18.50.0.4 support. (thanks to Marijn Schouten)
957 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
959 * bug fix: the fopcompiler can handle LOCALLY forms (with no
960 declarations) successfully. (reported by Attila Lendvai)
962 changes in sbcl-1.0.16 relative to 1.0.15:
963 * minor incompatible change: revert the changes to sb-posix's error
964 signaling added in 1.0.14.
965 * minor incompatible change: change PROBE-FILE back to returning
966 NIL whenever we can't get a truename, as was the case before 1.0.14.
967 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
968 inherits from ERROR instead of just CONDITION.
969 * new feature: SB-INTROSPECT can provide source locations for instances
970 as well. (thanks to Tobias Rittweiler)
971 * optimization: binding special variables now generates smaller code
972 on threaded platforms.
973 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
974 and cases where no :TEST is given but the compiler can infer that
975 the element to search is of type (OR FIXNUM (NOT NUMBER)).
976 * optimization: better LOGNOT on fixnums.
977 * optimization: modular arithmetic for a particular requested width
978 is implemented using a tagged representation unless a better
979 representation is available.
980 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
981 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
982 for wellformedness and returns a specialized vector. (reported by
983 Francois-Rene Rideau)
984 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
985 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
986 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
987 unscheduling timer at the same time as another timer fires.
988 * bug fix: don't reschedule timers for dead threads.
989 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
990 * bug fix: copying output from RUN-PROGRAM to a stream signalled
991 bogus errors if select() was interrupted.
992 * enhancement: add support for fcntl's struct flock to SB-POSIX.
994 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
995 * enhancement: cleaner backtraces for interactive interrupts, as
996 well as other cases where the interesting frames used to be
997 obscured by interrupt handling frames.
998 * enhancement: untracing a whole package using (UNTRACE "FOO") is
999 now supported, and tracing a whole package using (TRACE "FOO") now
1000 traces SETF-functions as well.
1001 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
1002 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
1003 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
1004 * unreadably printed representation of hash-tables now includes
1006 * bug fix: partially fixed #188: type propagation from assignments
1007 is now more efficient.
1008 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
1009 'DOUBLE-FLOAT) are not flushable.
1010 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
1011 full address of the object, and none of the tag bits.
1012 * bug fix: readably printing hash-tables now respects other printer
1013 control variables. (reported by Cedric St-Jean)
1014 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
1015 keyword with MAKE-HASH-TABLE.
1016 * bug fix: export SB-POSIX:MKSTEMP.
1017 * bug fix: SORT was not interrupt safe.
1018 * bug fix: XREF accounts for the last node of each basic-block as
1020 * bug fix: MAKE-INSTANCE optimizations interacted badly with
1021 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
1022 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
1023 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
1024 create obsolete instances in the case of redefinition or
1025 obsoletion of a superclass. (thanks to Andy Hefner)
1026 * bug fix: Support for the Alpha architecture has been revived; it had
1027 suffered somewhat from lack of maintenance since sbcl-1.0.
1028 * improvements to the Windows port:
1029 ** The system detects the codepage to use at startup, instead of
1030 using the value from the saved core. (thanks to Kei Suzuki)
1032 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
1033 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
1034 (see documentation for details.)
1035 * revived support for OpenBSD (contributed by Josh Elsasser)
1036 * partially fixed bug #108: ROOM no longer suffers from occasional
1037 (AVER (SAP= CURRENT END)) failures.
1038 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
1039 bodies no longer cause a WARNING to be signalled. (reported by
1041 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
1042 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
1043 the bug in TRUNCATE.) (reported by Michael Weber)
1044 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
1045 no samples. (reported by Andy Hefner)
1046 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
1047 no longer appear as (NIL ...) frames in backtraces.
1048 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
1049 is now actually exported from SB-EXT as documented. (reported by
1051 * bug fix: sb-aclrepl now correctly understands how to inspect
1052 single-floats on 64-bit platforms where single-floats are not boxed.
1053 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
1054 yet been finalized. (reported by Levente Meszaros)
1055 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
1056 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
1057 SBCL-specific optimize qualities.
1059 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
1060 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
1061 an executable in the search path, and does so in the child
1062 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
1063 been removed; it can be found in the CVS history, for somebody who
1064 needs that search behavior (see the manual).
1065 * minor incompatible change: compiler policy re. weakening type
1066 checks has changed: now type checks are weakened only if SAFETY < 2
1068 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
1069 unparsing of directory pathnames as files. Analogously,
1070 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
1071 filename to parse into a directory pathname.
1072 * enhancement: implicit generic function creation now signals a
1073 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
1074 users can bind handlers for and muffle around calls to LOAD.
1075 * enhancement: RUN-PROGRAM allows unicode arguments and environments
1076 to be used (using the default stream external format), and allows
1077 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
1078 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
1079 strings and vectors whose element-type or simplicity is not fully
1080 known at compile-time.
1081 * optimization: STRING-TRIM and related functions no longer allocate
1082 a new string when no trimming needs to be performed. These functions
1083 are also faster than before when the input string has been declared
1085 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
1086 * optimization: READ-LINE is significantly faster for files containing
1088 * optimization: non-open coded uses of character comparison operators
1089 (e.g. char=) no longer cons when called with more than one parameter
1090 on platforms supporting dynamic-extent allocation.
1091 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
1092 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
1093 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
1095 * bug fix: some sequence functions elided bounds checking when
1097 * bug fix: too liberal weakening of union-type checks when SPEED >
1099 * bug fix: more bogus fixnum declarations in ROOM implementation
1102 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
1103 * new feature: MAKE-HASH-TABLE now experimentally accepts a
1104 :SYNCHRONIZED argument, which makes the hash-table safe for
1105 concurrent accesses (but not iteration.) See also:
1106 SB-EXT:WITH-LOCKED-HASH-TABLE, and
1107 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1108 * optimization: CONCATENATE on strings is an order of magnitude faster
1109 in code compiled with (> SPEED SPACE).
1110 * optimization: SUBSEQ is ~50% faster on lists.
1111 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1112 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1113 END is smaller then START.
1114 * bug fix: SB-PROFILE will no longer report extra consing for nested
1115 calls to profiled functions.
1116 * bug fix: ROOM implementation had bogus fixnum declarations which
1117 could cause type-errors when calling ROOM on large images.
1118 * bug fix: if file compilation is aborted, the partial fasl is now
1119 deleted, and COMPILE-FILE returns NIL as the primary value.
1120 * bug fix: number of thread safety issues relating to SBCL's internal
1121 hash-table usage have been fixed.
1122 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1123 be returned from its body when the values were being returned
1124 using unknown-values return convection and the W-P-O was wrapped
1125 inside an UNWIND-PROTECT.
1126 * bug fix: sb-posix should now compile again under Windows, enabling
1127 slime to work again.
1129 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1130 * incompatible change: hash-table accessor functions are no longer
1131 automatically protected by locks. Concurrent accesses on the same hash-table
1132 from multiple threads can give inconsistent results or even corrupt the
1133 hash-table completely. Multi-threaded applications should do their own
1134 locking at the correct granularity. In the current implementation it is
1135 still safe to have multiple readers access the same table, but it's not
1136 guaranteed that this property will be maintained in future releases.
1137 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1138 and will signal an error at runtime.
1139 * enhancement: SB-THREAD package now exports a semaphore interface.
1140 * enhancement: CONS can now stack-allocate on x86 and
1141 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1143 * enhancement: nested lists can now be stack allocated on
1144 platforms providing stack allocation support.
1145 * enhancement: dynamic-extent support has been extended to support
1146 cases where there are multiple possible sources for the stack
1148 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1149 if the mutex is uncontested on Linux.
1150 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1151 as specified by the CLHS. (thanks to Tobias Rittweiler)
1152 * bug fix: NaN comparison now works on x86-64.
1153 * bug fix: CLOSE :ABORT T on a stream with pending output now
1155 * bug fix: instances of non-standard metaclasses using standard
1156 instance structure protocol sometimes missed the slot type checks
1158 * bug fix: known functions can be inlined in high-debug code.
1160 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1161 * minor incompatible change: the MSI installer on Windows no longer
1162 associates .lisp and .fasl files with the installed SBCL.
1163 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1164 on Windows. (thanks to Luis Oliviera)
1165 * new feature: SB-CLTL2 contrib module now implements
1166 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1167 * optimization: scavenging weak pointers is now more efficient,
1168 requiring O(1) instead of O(N) per weak pointer to identify
1169 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1170 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1171 method bodies using constant slot-names when the first argument is
1172 a specializer parameter for the method.
1173 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1174 names, when the class of the instance is a direct instance
1175 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1176 as slow as the constant slot-name case.)
1177 * optimization: member type construction is now O(N) instead
1179 * optimization: UNION and NUNION are now O(N+M) for large
1180 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1182 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1183 now more readable in environments like Slime which display it.
1184 (thanks to Tobias C. Rittweiler)
1185 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1186 was able to infer the correct argument type for the object on which
1187 the CAS operation was being performed.
1188 * bug fix: copy propagation interfered with parallel assignment
1189 semantics in local calls. (reported by Paul Khuong)
1190 * bug fix: the signed modular fixnum shift compiled to wrong code on
1191 x86 and x86-64. (spotted by a slight modification to some of
1194 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1195 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1196 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1197 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1199 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1200 and generic function now print the object identity for anonymous
1201 instances. (thanks to Attila Lendvai)
1202 * enhancement: as an extension to MOP, SBCL now supports SETF of
1203 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1204 (thanks to Attila Lendvai)
1205 * enhancement: sb-cover has an alternate annotation mode which puts
1206 the form annotations at the car of the form, instead of annotating
1207 it completely. The benefit of this mode is that it shows better
1208 which forms were instrumented by the compiler.
1209 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1210 by printer control variables. (reported by Dan Corkill)
1211 * bug fix: system leaked memory when delayed output was performed by
1212 the OS in smaller chunks then expected. (thanks to David Smith)
1213 * bug fix: system leaked memory when file streams were not closed
1215 * bug fix: large objects written to slow streams that were modified
1216 after the write could end up with the modified state written to
1217 the underlying file descriptor.
1218 * bug fix: multiple threads operating in parallel on the same stream
1219 could cause buffer-overflows.
1220 * bug fix: source location information is stored correctly
1221 (broken since 1.0.6). This bug would generally show up as the
1222 Slime debugger highlighting the wrong form when the sldb-show-source
1224 * bug fix: Branch forms are again annotated as branches in the sb-cover
1226 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1227 on x86-64. (reported by Gregory Vanuxem)
1228 * bug fix: EQUALP could return wrong results for structures with raw
1229 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1230 word). (reported by Vjacheslav Fyodorov)
1231 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1234 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1235 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1236 atomic compare-and-swap operations on threaded platforms.
1237 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1238 allows assigning a global minimum value to optimization qualities
1239 (overriding proclamations and declarations).
1240 * enhancement: closed over variables can be stack-allocated on x86
1242 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1244 * optimization: slot definition lookup is now O(1). This speeds up
1245 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1246 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1247 * optimization: ASSOC and MEMBER can now be open-coded for all
1248 combinations of keyword arguments when second argument is constant
1249 and SPEED >= SPACE. In other cases a specialized version is
1251 * bug fix: using obsoleted structure instances with TYPEP and
1252 generic functions now signals a sensible error.
1253 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1254 (reported by Kristoffer Kvello)
1255 * bug fix: backtrace construction is now more careful when making
1256 lisp-objects from pointers on the stack, to avoid creating bogus
1257 objects that can be seen by the GC.
1258 * bug fix: defaulting of values in contexts expecting more than 7
1259 variables now works on x86-64. (reported by Christopher Laux)
1260 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1262 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1263 as the property-list of a symbol.
1264 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1265 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1266 situations requested, are once again file-compileable. (reported
1269 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1270 * MOP improvement: support for user-defined subclasses of
1271 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1272 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1273 the expansion of DEFMETHOD, is responsible for generating a form
1274 which creates a list of specializers when evaluated. Additional
1275 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1276 debugging and introspective support.
1277 * minor incompatible change: the (unsupported) spinlock interface
1278 has changed: free spinlock now has the value NIL, and a held spinlock
1279 has the owning thread as its value.
1280 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1281 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1283 * enhancement: name of a socket-stream is now "a socket" instead of
1284 "a constant string".
1285 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1286 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1287 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1288 for generic functions. (Thanks to Tobias C. Rittweiler)
1289 * optimization: bignum printing speed has been improved by 20-40%
1290 (depending on the bignum size.)
1291 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1293 * bug fix: the cache used by the CLOS to store precomputed effective
1294 methods, slot offsets, and constant return values is now thread and
1296 * bug fix: generic function dispatch function updating is now thread
1297 and interrupt safe (in the sense that the known issues have been
1299 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1300 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1301 * bug fix: asdf systems with dependencies to the SB-POSIX or
1302 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1303 * bug fix: interrupt safety of applicable method computation has been
1306 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1307 * new contrib: sb-cover, an experimental code coverage tool, is included
1308 as a contrib module.
1309 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1310 significantly faster.
1311 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1312 produce more efficient code on x86-64 if the compiler can determine
1313 that (+ INDEX <constant>) does not require a bounds check and FOO
1314 has an element type at least 8 bits wide.
1315 * enhancement: a new, experimental synchronous timeout facility is
1316 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1317 * enhancement: when a symbol name conflict error arises, the
1318 conflicting symbols are always printed with a package prefix.
1319 (thanks to Kevin Reid)
1320 * enhancement: stepping is now once again supported on the SPARC and
1321 MIPS platforms. (It is also now more likely to work on CheneyGC
1323 * enhancement: sb-sprof can now also track and report accurate call
1325 * bug fixes: the treatment of non-standard subclasses of
1326 SB-MOP:SPECIALIZER is more correct.
1327 * incompatible change: PURIFY no longer copies the data from the
1328 dynamic space into the static and read-only spaces on platforms
1329 that use the generational garbage collector
1330 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1332 * bug fix: GC race condition occasionally resulting in crashes with
1333 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1335 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1337 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1338 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1339 * bug fix: an asynchronous interrupt could previously leave the
1340 system running with GC inhibited.
1341 * bug fix: a DECLARE form evaluated at top-level now causes an error
1342 rather than silently (or verbosely) returning NIL.
1343 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1344 result in a error being signalled due to source form lookup errors.
1345 (reported by Peter Graves)
1347 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1348 * incompatible change: removed writer methods for host-ent-name,
1349 host-ent-addresses -- changing the values did not update the DNS
1351 * minor incompatible change: changed experimental JOIN-THREAD interface
1352 * documentation: the manual now lists reader and writer methods
1353 in class slot documentation sections. (thanks to Richard M Kreuter)
1354 * documentation: unwinding from asyncronous events has been
1355 documented as unsafe.
1356 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1357 in multithreaded application code.
1358 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1359 platforms. (thanks to James Anderson for the optimization hint)
1360 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1362 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1363 variants no longer cons.
1364 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1365 their NOT- variants no longer cons.
1366 * optimization: EQUAL hash tables no longer use SXHASH for objects
1367 of all data types, but instead use an EQL hash for types for which
1368 EQUAL is the same as EQL.
1369 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1370 are significantly faster.
1371 * optimization: new STRING-OUTPUT-STREAM implementation is much
1372 faster for large outputs and conses 30% less on average.
1373 * enhancement: XREF information is now collected for references made
1374 to global variables using SYMBOL-VALUE and a constant argument.
1375 * enhancement: SIGINT now causes a specific condition
1376 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1377 * bug fix: FILE-POSITION used to signal an error in some cases where
1378 ANSI requires it to return NIL.
1379 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1380 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1381 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1383 * bug fix: bad type declaration in the CLOS implementation has
1384 been fixed. (reported by James Anderson)
1385 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1386 (thanks to Lutz Euler)
1387 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1388 required by the CLHS. (reported by Samium Gromoff)
1389 * bug fix: dead unbound variable references now signal an error.
1390 (reported by Marco Monteiro)
1391 * bug fix: / with an unused value was being deleted in safe code.
1392 (thanks to Marco Monteiro and Kevin Reid)
1393 * bug fix: number of characters that can be written onto a single
1394 line in a file is unlimited.
1395 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1396 been fixed by inhibiting interrupts for when GC is disabled.
1397 * bug fix: some interrupt safety issues with GETHASH, (SETF
1398 GETHASH), CLRHASH and REMHASH have been fixed.
1399 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1400 specifier no longer causes infinite recursion.
1401 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1402 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1403 * bug fix: an &environment argument with the correct variable information
1404 is passed to macros that are expanded during byte compilation.
1405 (reported by Samium Gromoff)
1406 * bug fix: a logic error was causing the hash function for bignums to
1407 have excessive amounts of collisions. (reported by Faré Rideau)
1408 * bug fix: modifying the contents of an array could change the return
1409 value of SXHASH on that array, which is only allowed for strings
1410 and bit vectors. (bug introduced in 0.9.16)
1411 * improvement: the x86-64/darwin port now passes all tests (except
1412 for the debugger tests) but should still be considered
1413 experimental until this is fixed.
1414 * improvement: a style-warning is signaled for CASE (etc) clauses with
1415 duplicate keys. (thanks to Kevin Reid)
1416 * improvement: macos/x86 and macos/x86-64 now use mach exception
1417 handlers for dealing with illegal instructions (for trapping and
1418 error handling) and memory protection violations (for GC).
1420 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1421 * new platform: experimental support for x86-64/darwin (MacOS).
1422 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1423 and getnameinfo sockets functions are used instead of gethostbyaddr
1424 and gethostbyname, on platforms where the newer functions are available.
1425 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1427 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1428 don't print the startup banner, but behave as if --noinform was passed
1429 as a command line argument. (thanks to Kevin Reid)
1430 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1431 * optimization: code using alien values with undeclared types is much faster.
1432 * optimization: the compiler is now able to open code SEARCH in more cases.
1433 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1434 * bug fix: using standardized COMMON-LISP special variables as loop
1435 variables no longer signals bogus package lock violations. (reported
1437 * bug fix: declaring local loop variables to be of a range-limited type
1438 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1439 (reported by Andras Simon)
1440 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1441 bugs remain on x86-64.)
1442 * bug fix: the #= and ## reader macros now interact reasonably with
1443 funcallable instances.
1444 * bug fix: type-checks for function arguments were compiled using the
1445 compiler policy settings of the wrong lexical environment (in 1.0.2
1447 * bug fix: SHADOW accepts characters as string designators, as required
1448 by the spec (thanks to Eric Marsden)
1449 * bug fix: fixed GC safety issues when foreign functions are called with
1450 non-base strings as arguments
1451 * bug fix: more consistent error output for fatal-compiler-warnings, like
1453 * bug fix: functions with proper names were showing up as NIL in
1454 backtraces (reported by Edi Weitz, regression in 1.0.2)
1456 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1457 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1458 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1459 produce more efficient code on the x86 if the compiler can determine
1460 that (+ INDEX <constant>) does not require a bounds check and FOO
1461 has an element type at least 8 bits wide.
1462 * bug fix: references to undefined variables are handled the same way
1463 in toplevel forms as in the normal compiler
1464 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1466 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1467 (thanks to Jon Buller)
1468 * improvement: faster compilation times for complex functions
1469 * improvement: added readlink support to SB-POSIX (thanks to Richard
1472 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1473 * improvement: experimental support for mach exception handling on
1474 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1476 * improvement: support for GBK external format.
1477 (thanks to Chun Tian (binghe))
1478 * improvement: the debugger now displays variables that have been closed
1479 over, in code compiled with (DEBUG 2) or higher
1480 * improvement: support for executable cores on NetBSD (thanks to
1482 * new feature: added a RESTART-FRAME debugger command
1483 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1484 be used to provide an implementation for FILE-POSITION on Gray streams
1485 (thanks to Eric Marsden)
1486 * new feature: add syslog, openlog and closelog support to SB-POSIX
1487 (thanks to Richard Kreuter)
1488 * optimization: the function call overhead in code compiled with
1489 a high DEBUG optimization setting is significantly
1490 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1491 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1492 (thanks to Tony Martinez)
1493 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1494 evaluated once (reported by Kevin Reid)
1495 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1496 works again (thanks to Kilian Sprotte)
1497 * bug fix: an error is signaled for tagbodies with duplicate tags
1498 (thanks to Stephen Wilson)
1499 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1500 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1501 stack frames from alien callbacks.
1502 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1503 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1504 65k pages (thanks to David Woodhouse)
1505 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1507 changes in sbcl-1.0.1 relative to sbcl-1.0:
1508 * new platform: FreeBSD/x86-64, including support for threading.
1509 * new feature: the compiler stores cross-referencing information
1510 about function calls (who-calls), macroexpansion (who-macroexpands)
1511 and special variables (who-binds, who-sets, who-references) for code
1512 compiled with (< SPACE 3). This information is available through the
1513 sb-introspect contrib.
1514 * new feature: users may subclass SEQUENCE, and have instances of
1515 these classes interoperate with standard Common Lisp functions if
1516 a number of methods are defined. (This feature is experimental
1517 and the interface subject to change based on feedback from SBCL
1518 users and the general community)
1519 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1520 x86/x86-64, rather than the previous fixed depth of 8
1521 * improvement: another pthread back-end of mutex "pthread-futex".
1522 * bug fix: non-ascii command-line arguments are processed correctly
1523 (thanks to Yaroslav Kavenchuk)
1524 * bug fix: non-required arguments were not passed correctly when a method
1525 defined using DEFMETHOD was called from a mop-generated method using
1526 CALL-NEXT-METHOD (reported by Pascal Costanza)
1527 * bug fix: recursion is now permitted in accessors through
1528 SLOT-UNBOUND. (reported by Pascal Costanza)
1529 * bug fix: an error was signaled at startup if the HOME environment
1530 variable was defined, but had an empty value (reported by Peter Van Eynde)
1531 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1532 signal a PROGRAM-ERROR, not a BUG.
1533 * bug fix: SB-POSIX:READDIR works when built with large file support.
1534 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1536 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1537 for better automatic indentation support. (thanks to Matt Pillsbury)
1538 * optimization: loading generic functions no longer takes O(n^2) time,
1539 proportional to the amount of methods in the generic function
1540 (reported by Todd Sabin and Jeremy Brown)
1541 * optimization: the FIND and POSITION family of sequence functions
1542 are significantly faster on arrays whose element types have been
1544 * improvements to the Windows port:
1545 ** Intermittent heap corruption problems have been fixed. (thanks
1546 to Alastair Bridgewater)
1547 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1549 ** Lisp is able to unwind foreign exception frames from alien
1550 callbacks. (thanks to Alastair Bridgewater)
1552 changes in sbcl-1.0 relative to sbcl-0.9.18:
1553 * improvement: experimental support for threading on FreeBSD/x86.
1554 (thanks to NIIMI Satoshi)
1555 * improvement: runtime option --dynamic-space-size can be used
1556 to set the size of the dynamic space reserved on startup.
1557 * improvement: floating point modes in effect are now saved in
1558 core, and restored on startup.
1559 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1560 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1561 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1562 threads are not running after *SAVE-HOOKS* have run.
1563 * improvement: writes to CLOS instance slots are type-checked in code
1564 compiled with (SAFETY 3)
1565 * improvement: floating-point exception handling on FreeBSD (thanks to
1567 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1568 (thanks to Zach Beane)
1569 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1571 * improvement: added support for the Shift-JIS external format.
1572 (contributed by NIIMI Satoshi)
1573 * improvement: callbacks are supported on Linux/PPC. (thanks to
1575 * bug fix: compiler bug triggered by a (non-standard) VALUES
1576 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1577 * bug fix: file compiler no longer confuses validated and already
1578 dumped structurres. (reported by Kaersten Poeck)
1579 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1580 fill-pointer signals a type-error as required. (thanks to
1582 * bug fix: disassemly of funcallable instances works.
1583 * bug fix: single stepping on PPC.
1584 * bug fix: fix thread-safety problems in the type system (generally
1585 manifesting as nonsensical errors like "STRING is a bad type specifier
1586 for sequences" or "The value 1 is not of type FIXNUM")
1587 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1588 (reported by Josip Gracin)
1589 * bug fix: an error is signaled for attempts to displace arrays with
1590 incompatible element types (thanks to Mario Mommer)
1591 * bug fix: more correct handling of wide characters in debug info
1592 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1593 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1594 and don't cause extra consing
1595 * optimization: MAP and MAP-INTO are significantly faster on vectors
1596 whose elements types have been declared.
1597 * Improvements to SB-SPROF:
1598 ** Support for allocation profiling
1599 ** Reduced profiling overhead, especially for long profiling runs
1600 * Improvements to the Windows port:
1601 ** floating point exceptions are now reported correctly.
1602 ** stack exhaustion detection works partially.
1603 ** more accurate GET-INTERNAL-REAL-TIME.
1604 ** better RUN-PROGRAM behaviour in case of errors in the child
1606 ** PROBE-FILE now simplifies pathnames correctly.
1607 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1608 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1609 1st 1970 (although time timezone data will be unreliable).
1610 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1612 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1613 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1614 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1616 * bug fix: two potential GC deadlocks affecting threaded builds.
1617 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1618 returns T (reported by Anton Kazennikov)
1619 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1620 with non-variable places
1621 * bug fix: remove a race condition in the setting of
1622 funcallable-instance functions, this should make threaded CLOS
1623 code more stable against memory faults.
1624 * bug fix: corruption of specials when unbinding is interrupted by an
1625 asynchronous unwind (reported by Hannu Koivisto)
1626 * improvement: the debugger will now also display local variables that
1627 are only used once, for code compiled with a DEBUG optimization quality
1630 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1631 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1632 * incompatible change: External-format support for FFI calls. The
1633 SB-ALIEN:C-STRING no longer implies an ASCII
1634 external-format. Instead the string is subject to external-format
1635 conversion. Additionally return values of type SB-ALIEN:C-STRING
1636 are by default of type (SIMPLE-ARRAY CHARACTER), not
1637 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1638 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1639 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1640 * incompatible change: SB-EXT package no longer contains the
1641 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1642 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1644 * incompatible change: the single-stepper is no longer available
1645 on Alpha, Mips and Sparc platforms.
1646 * minor incompatible change: the direct superclasses of
1647 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1648 not (STANDARD-OBJECT FUNCTION). This makes the
1649 class-precedence-lists of GENERIC-FUNCTION and
1650 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1652 * minor incompatible change: the default stream external format on
1653 non-unicode SBCL is no longer detected from the locale, but is always
1655 * new feature: Add a version of evaluator that uses an interpreter instead
1656 of the compiler. EVAL still uses the compiler by default, to switch it
1657 to use the interpreter, set the value of the variable
1658 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1659 * minor incompatible change: the single-stepper REPL has been merged
1660 with the normal debugger (see the "Stepping" heading of the debugger help
1661 for more details). The debugger command STEP will no longer switch
1662 to the single-stepper REPL.
1663 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1664 for a type now works.
1665 * bug fix: loading of default sysinit file works. (thanks to Leonid
1667 * bug fix: better detection of circularities in the file-compiler.
1668 (reported by Marco Monteiro)
1669 * bug fix: the CL pathname functions now work with files that have
1670 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1671 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1672 systems with Unix98 pty semantics.
1673 * bug fix: ASDF-INSTALL will now work with bsd tar.
1674 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1676 * bug fix: timers expiring in dead threads no longer cause a
1677 type-error (reported by Paul "Nonny Mouse").
1678 * bug fix: thanks to more lightweight single-stepper instrumentation,
1679 code compiled with (DEBUG 3) will compile and execute significantly faster,
1680 and will have more accurate type-inferencing than before
1681 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1682 whose bindings are modified
1683 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1684 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1685 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1686 CL:SOFTWARE-VERSION with proper version
1688 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1689 * feature: implemented the READER-METHOD-CLASS and
1690 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1691 as specified by AMOP.
1692 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1694 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1695 (thanks to Lutz Euler)
1696 * optimization: hashing of general arrays and vectors has been
1697 improved. (reported by Any Fingerhut)
1698 * enhancement: SB-INTROSPECT is now able to find definitions of
1699 profiled functions. (thanks to Troels Henriksen)
1700 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1701 as well, allowing compiler-macros for SETF-functions to expand.
1702 * enhancement: step-instrumentation no longer wraps known
1703 single-value functions with multiple-value context, allowing
1704 better type inference.
1705 * fixed bug #337: use of MAKE-METHOD in method combination now works
1706 even in the presence of user-defined method classes. (reported by
1707 Bruno Haible and Pascal Costanza)
1708 * fixed bug #339(c): if there are applicable methods not part of any
1709 long-form method-combination group, call INVALID-METHOD-ERROR.
1710 (reported by Bruno Haible)
1711 * fixed bug #361: the :FUNCTION initarg in the protocol for
1712 initialization of methods can now be used to override
1713 internally-produced optimized functions. (reported by Bruno
1715 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1716 system-provided lambda expression no longer cause warnings about
1717 unbound #:|pv-table| symbols.
1718 * bug fix: improved the handling of type declarations and the
1719 detection of violations for keyword arguments with non-constant
1721 * bug fix: potentially erronous calls to PATHNAME and
1722 MERGE-PATHNAMES were being flushed in some cases.
1723 (reported by Richard Kreuter)
1724 * bug fix: compiled calls to TYPEP were mishandling obsolete
1725 instances. (reported by James Bielman and Attila Lendvai)
1726 * bug fix: format strings with ~> without matching ~< no longer
1727 trigger an AVER, but signal an understandable error instead.
1728 (reported by Antonio Martinez)
1729 * bug fix: specifying an output-file without a pathname-type for
1730 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1731 of input-file instead of "fasl". (reported by Robert Dodier)
1732 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1733 (reported by James Y Knight).
1734 * bug fix: compiler-macros-function did not consider the environment
1735 argument for shadowing by local functions.
1736 * bug fix: compiler-macros expansion was inhibited by local INLINE
1738 * bug fix: inline expansions of known functions were subject to
1739 step-instrumentation in high debug policies, leading to problems
1740 with type-inference.
1741 * bug fix: compiler failed to differentiate between different CONS
1742 types in some cases.
1743 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1744 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1745 * bug fix: #368: incorrect use of expressed vs. upgraded array
1747 * bug fix: #306a: more precise unions of array types.
1748 * thread-safety improvements:
1749 ** CONDITION-WAIT could return early on Linux, if the thread was
1750 interrupted and subsequently continued with SIGCONT.
1751 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1753 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1754 * added support for the ucs-2 external format. (contributed by Ivan
1756 * minor incompatible change: pretty printing of objects of type
1757 (cons symbol) is, in the default pprint-dispatch-table, now
1758 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1760 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1761 called later in a class's lifetime, possibly as late as when the
1762 first instance of the class is created. Previously,
1763 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1764 class became finalizeable.
1765 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1766 for multibyte external-format streams. (thanks to "vbzoli")
1767 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1768 with :ALLOCATION :CLASS inherited from superclasses of the
1770 * fixed bug: anonymous classes can now be created using the :NAME
1771 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1772 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1773 * fixed bug: core-files saved with :EXECUTABLE T can again be
1774 executed when SBCL_HOME isn't set. (reported by James Knight)
1775 * fixed bug: toplevel LOCALLY forms with declarations could
1776 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1777 * fixed bug: printing from several different threads using different
1778 values of *print-case* could cause invalid output, due to
1779 some internal special variables of the printer not being bound
1780 thread-locally (reported by Max Mikhanosha)
1781 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1782 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1783 (reported by Pascal Costanza)
1784 * fixed bug: FINALIZE-INHERITANCE is called from
1785 REINITIALIZE-INSTANCE on classes when the class has previously
1786 been finalized, as required by AMOP.
1787 * minor code generation optimizations:
1788 ** better register allocation in CLOS dispatching functions
1789 ** overflow detection when coercing signed bytes to fixnums on x86-64
1790 ** is now implemented with one IMUL instruction instead of three shifts
1791 ** more efficient bit-vector access on x86 and x86-64
1792 ** more efficient access to raw structure slots on x86 and x86-64
1793 * fixed some bugs revealed by Paul Dietz' test suite:
1794 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1795 return its argument.
1797 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1798 * feature: thread support on Solaris/x86, and experimental thread support
1800 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1802 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1803 default method for SLOT-UNBOUND.
1804 * minor incompatible change: prevent the user from specializing the
1805 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1806 somewhat counter to the intent of the protocol, I (CSR) think, and
1807 additionally it just doesn't work in SBCL as currently
1808 implemented, thanks to optimizations (that are always valid for
1809 the other three SLOT-VALUEish functions, but not for the setter).
1810 * minor incompatible change: the :SB-LDB feature is now enabled by
1811 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1812 the low-level debugger.
1813 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1814 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1815 down to an SB-SYS:FD-STREAM, allowing direct communication in
1816 more cases. Temporary files and pipes are still used as a fallback
1818 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1819 on the non-reentrant C functions localtime(3) and gmtime(3).
1820 * bug fix: global optimization policy was not visible in LOCALLY and
1822 * bug fix: class objects can be used as specializers in methods.
1823 (reported by Pascal Costanza)
1824 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1825 with a type error. (reported by blitz_ on #lisp)
1826 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1827 failed with a type error. (reported by Pascal Bourguignon)
1828 * bug fix: merging pathnames against defaults with :DIRECTORY
1829 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1830 (reported by James Y Knight)
1831 * bug fix: saving large (>2GB) cores on x86-64 now works
1832 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1833 constant index and a (simple-array (signed-byte 32)) array
1834 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1835 error (patch by Robert J. Macomber)
1836 * feature: TIME now displays the amount of run-time spent in GC
1837 * bug fix: The debugger now does a better job of respecting
1838 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1839 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1840 workaround for bug 403.)
1841 * bug fix: floating point exception handling now works on FreeBSD
1842 (thanks to NIIMI Satoshi)
1843 * fixed some bugs revealed by Paul Dietz' test suite:
1844 ** MISC.641: LET-conversion were not supposed to work in late
1846 * improvements to the Win32/x86 port:
1847 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1848 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1850 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1851 * new feature: source path information is generated for macro-expansion
1852 errors for use in IDE's like Slime (thanks to Helmut Eller)
1853 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1854 * bug fix: compiler does not loop forever on an invalid type in
1856 * improvement: compilation of most CLOS applications is significantly
1858 * optimization: added a limited bytecode compiler for simple toplevel
1859 forms, speeding up compilation and FASL loading
1860 * bug fix: the statistical profiler now properly distinguishes anonymous
1863 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1864 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1865 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1866 into subclasses of the finalized class. Now user calls to
1867 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1868 system return before any subclasses are finalized.
1869 * minor incompatible change: The reader no longer ignores errors
1870 regarding non-existent packages in #+ and #- feature tests.
1871 * new feature: command line options --no-sysinit, --no-userinit to
1872 inhibit loading the corresponding init files
1873 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1874 for sending data through UDP sockets (thanks to François-René Rideau)
1875 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1876 error is signalled instead (thanks to François-René Rideau)
1877 * minor incompatible change: Error signalling behaviour of lexical
1878 operations violating package locks has changed slightly. Refer to
1879 documentation on package locks for details.
1880 * enhancement: EVAL can process IF-expressions without resorting to the
1882 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1883 constant-folded properly due to missing out-of-line functions.
1884 (reported by Mika Pihlajamäki)
1885 * bug fix: LISTEN sometimes returned T even in cases where no data was
1886 immediately available from the stream
1887 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1888 were reversed. (reported by Levente Mészáros)
1889 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1890 when given a vector displaced to an adjustable array. (reported
1892 * bug fix: derivation of float boundaries from numbers outside the
1893 appropriate float range (reported by John Wiseman)
1894 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1896 * bug fix: precomputing cache entries for generic functions with
1897 some subclasses of specializers as yet invalid does not attempt to
1898 fill a cache line with a negative offset. (reported by Levente
1900 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1901 allow condition class objects as arguments to DOCUMENTATION and
1902 (SETF DOCUMENTATION); only find and set documentation for
1903 structure names for the STRUCTURE doc-type. (suggested by Gary
1905 * improvements to the Win32/x86 port:
1906 ** (user-homedir-pathname) and default initialization file
1907 locations now know about the user's "Documents and Settings"
1908 directory (thanks to Yaroslav Kavenchuk)
1909 ** run-program is implemented (thanks to Mike Thomas)
1910 ** sockets support (thanks to Timothy Ritchey)
1911 ** better backtrace support (thanks to Alastair Bridgewater)
1912 ** sb-grovel supported
1913 ** asdf-install and sb-posix work somewhat
1914 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1915 * improvements to the Solaris/x86 port:
1916 ** works on Solaris 11/Solaris Express
1917 ** floating-point exception handling support
1918 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1919 * fixed some bugs revealed by Paul Dietz' test suite:
1920 ** REMOVE-METHOD returns its generic function argument even when
1921 the method is not one of the generic functions' methods.
1922 ** objects declared as MEMBER types can be admissible arguments to
1923 structure accessors.
1924 ** printing characters should simply be printed by the FORMAT ~:C
1926 ** compiler failure when compiling functions with hairy constant
1927 defaults for optional parameters.
1928 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1929 function, which is already optimized.
1931 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1932 * new platform: experimental support for SBCL x86/Darwin, including
1933 MacOS X 10.4.5 on Intel.
1934 * new feature: Unicode character names are now known to the system
1935 (through CHAR-NAME and NAME-CHAR).
1936 * new feature: the filesystem location of SBCL's core file is
1937 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1938 * minor incompatible change: the contrib modules SB-POSIX and
1939 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1940 this change is to make it easier to distribute
1941 location-independent binaries.
1942 * bug fix: as implied by AMOP, standardized classes no longer have
1943 slots named by external symbols of public packages. (reported by
1945 * optimization: calling functions via constant symbols -- as in
1946 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1947 via the function object as in (FUNCALL #'FOO).
1948 * optimization: on x86 and x86-64, the calling convention for the
1949 case of dealing with an unknown number of values has been altered
1950 to be friendlier to the prediction heuristics implemented,
1951 particularly with reference to CALL and RET pairing. (thanks to
1952 Alastair Bridgewater)
1953 * optimization: on x86, the code for access to thread-local symbol
1954 values and binding/unbinding of thread-local symbols is smaller.
1955 (thanks to Alastair Bridgewater)
1956 * enhancement: CONSTANTP is now able to determine constantness of
1957 more complex forms, including calls to constant-foldable standardized
1958 functions and some special forms beyond QUOTE.
1959 * fixed bug: occasional GC crashes on Solaris/x86
1960 * optimization: x86-64 supports stack allocation of results of simple
1961 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1962 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1963 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1964 (thanks to James Knight)
1965 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1966 by the standard (thanks to Stephen Compall)
1968 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1969 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1970 be used for bundling the runtime and the core file into one
1971 executable binary. This feature is not currently supported on all SBCL
1972 platforms. (thanks to James Bielman and NIIMI Satoshi)
1973 * new feature: a generational or ephemeral garbage collector is now
1974 the default on the PowerPC platform (both Linux and Darwin). The
1975 old Cheney (stop and copy) collector is a build-time option.
1976 (thanks to Cyrus Harmon, after Raymond Toy)
1977 * minor incompatible change: the method by which SBCL finds its
1978 contributed modules has changed; it no longer relies on symbolic
1979 links from an $SBCL_HOME/systems directory, but searches directly
1980 in the subdirectories of $SBCL_HOME.
1981 * enhancement: the dynamic heap size on the Linux/PPC platform is
1982 markedly larger, even using the older Cheney garbage collector.
1983 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1984 certain complicated string types. (reported by Gary King)
1985 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1986 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1987 * fixed bug: DOCUMENTATION on structures no longer signals an
1988 error if no structure type of the right name exists. (reported by
1990 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1991 applicable non-standard primary or :AROUND method on
1992 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1993 no longer cause unbound variable errors. (reported by Kevin Reid)
1994 * fixed bug: metacircle resolution in cases where methods have slots
1995 added before the slots from STANDARD-METHOD. (reported by Jean
1997 * fixed bug: the Power PC assembly code for calling into lisp
1998 sometimes computed its return address wrongly (depending on the
1999 15th bit of the address, fixed at link time). This used to
2000 manifest itself in a segmentation violation while building PCL.
2001 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
2002 many others over the years)
2003 * fixed bug: ROOM no longer reports silly numbers for stack usage if
2004 the stack pointer (treated as a signed integer) is negative.
2005 (thanks to Peter van Eynde)
2007 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
2008 * new platform: experimental support for the Windows operating
2009 system has been added. (thanks to Alastair Bridgewater)
2010 * fixed several bugs in and robustified the PPC FFI (including
2011 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
2012 * bug fix: allow non-simple string symbol names (reported by Paul
2014 * bug fix: interrupt handling on NetBSD (thanks to Richard M
2016 * bug fix: saving a core corrupted callbacks on x86/x86-64
2017 * bug fix: closed a loophole in metacircularity detection and
2018 grounding in the PCL implementation of CLOS.
2019 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
2020 (reported by tomppa on #lisp)
2021 * bug fix: clear the direction flag on Lisp -> C transitions, as
2022 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
2023 (reported by Andrej Grozin and Hendrik Maryns)
2024 * optimization: major improvements to GC efficiency on GENCGC platforms
2025 * optimization: faster implementation of EQUAL
2026 * optimization: emit more efficient opcodes for some common
2027 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
2028 * optimization: several other minor code-generation improvements on x86-64
2029 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
2031 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
2032 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2033 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
2034 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
2035 by ANSI as a generic function, and for consistency (SETF
2036 GENERIC-FUNCTION-NAME) is treated likewise.
2037 * fixed bug #233.b: make constraint propagation notice when a variable
2038 value is changed after it is referenced but before it is used
2039 * fixed bug #296: no more arbitrary behaviour from filenames with
2040 odd characters as --load arguments. (reported by Adam Warner)
2041 * fixed bug #390: :CHARACTER-SET pathname components now work as
2042 expected. (reported by Tim Daly Jr)
2043 * fixed bug #391: complicated :TYPE intersections in slot
2044 definitions no longer cause an error in PCL internals.
2045 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
2046 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
2047 returning the number of octets which would be written to the
2048 file-stream. (thanks to Robert J. Macomber)
2049 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
2050 arguments to the internal stream functions. (thanks to David
2052 * bug fix: stack exhaustion is now handled on the x86/SunOS
2053 platform. (thanks to Daisuke Homma)
2054 * bug fix: more accurate ROOM results on GENCGC platforms
2055 * optimization: improved type inference for arithmetic-for
2056 index variables in LOOP
2057 * optimization: faster floating-point SQRT on x86-64
2058 * fixed some bugs revealed by Paul Dietz' test suite:
2059 ** DOCUMENTATION returns NIL instead of "" for method combinations
2060 that don't have a docstring
2062 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
2063 * minor incompatible change: (SETF CLASS-NAME) and (SETF
2064 GENERIC-FUNCTION-NAME) are no longer generic functions, and
2065 therefore are not customizeable by user code (as seems to be at
2066 least permitted and maybe required by AMOP). As a consolation,
2067 however, the SBCL implementation of these functions now calls
2068 REINITIALIZE-INSTANCE as specified by AMOP.
2069 * bug fix: slot-definition documentation is propagated as per ANSI
2070 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
2071 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
2072 argument, as do eslotd initialization methods. (from Pascal
2073 Costanza's "Closer" project)
2074 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
2075 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
2077 * bug fix: it is now possible to have more than one subclass of
2078 STANDARD-GENERIC-FUNCTION without causing stack overflow.
2079 (reported by Bruno Haible, Pascal Costanza and others)
2080 * bug fix: the dependent update protocol now works for generic
2081 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
2082 and Pascal Costanza)
2083 * bug fix: condition-class instances corresponding to
2084 DEFINE-CONDITION forms are now created eagerly. (reported by
2085 Kalle Olavi Niemitalo on comp.lang.lisp)
2086 * bug fix: floating point printing is more accurate in some
2087 circumstances. (thanks to Simon Alexander)
2088 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
2089 merged with *DEFAULT-PATHNAME-DEFAULTS*.
2090 * bug fix: callbacks on OS X now preserve stack-alignment.
2091 (thanks to Cyrus Harmon)
2092 * enhancement: the x86-64 disassembler is much better at
2093 disassembling SSE instructions. (thanks to Lutz Euler)
2094 * enhancement: CHECK-TYPE on a variable now allows type-inference
2095 (thanks to Vincent Arkesteijn)
2096 * optimization: improved performance of EUC-JP external format.
2097 (thanks to NIIMI Satoshi)
2098 * optimization: performance improvements to IO on file streams of
2099 :ELEMENT-TYPE CHARACTER
2100 * optimization: much faster memory allocation on x86-64
2101 * optimization: faster garbage collections (latency and throughput)
2103 * optimization: faster bignum subtraction and division on x86 and x86-64
2104 (thanks to Lutz Euler)
2105 * optimization: more accurate type-inference for loops that use a
2106 floating point index variable or a negative step.
2108 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2109 * bug fix: add a workaround to SBCL looping infinitely at startup on
2110 Linux kernels with apparently buggy implementations of personality().
2111 (thanks to Svein Ove Aas)
2112 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2113 (thanks to David Lichteblau)
2114 * bug fix: local bindings shadow global symbol macros for
2115 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2116 * optimization: non-open coded uses of numeric comparison operators
2117 (e.g. >) no longer cons when called with more than one parameter
2118 on platforms supporting dynamic-extent allocation.
2119 * enhancement: saving cores with foreign code loaded is now supported
2120 on MIPS/Linux in addition to the previously supported platforms.
2121 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2122 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2123 * bug fix: division by zero in sb-sprof when no samples were collected
2124 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2125 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2126 (thanks to Alexander Kjeldaas)
2127 * new feature: ignoring the return values of destructive functions
2128 like NREVERSE signals a compile-time style-warning.
2129 (thanks to Kevin Reid)
2130 * new feature: the alignment of alien structure fields can be
2131 explicitly specified. (thanks to Cyrus Harmon)
2132 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2133 and dump core on SIGQUIT
2135 ** incompatible change: threads do not inherit values of specials
2136 from their parents (see manual)
2137 ** bug fix: threads stacks belonging to dead threads are freed by the
2138 next exiting thread, no need to gc to collect thread stacks anymore
2139 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2140 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2141 ** bug fix: dynamic variable and thread start related gc lossage
2142 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2144 * fixed some bugs revealed by Paul Dietz' test suite:
2145 ** SUBTYPEP is slightly more accurate on heinously complicated
2146 CONS types where some of the members have uncertain (in the
2147 NIL, NIL sense) type relationships to each other.
2149 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2150 no longer purified unless :PURIFY T is explicitly specified.
2151 ** Non-purified cores are significantly smaller than before
2153 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2154 * new feature: timers based on Zach Beane's excellent timer package
2155 * added support for the following external formats: koi8-u,
2156 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2157 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2158 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2159 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2160 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2161 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2162 * incompatible change: a threaded SBCL will no longer revert to
2163 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2164 * bug fix: interrupts are disabled until startup is complete; no
2165 more sigsegvs when receiving a signal to soon
2166 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2168 * bug fix: add a workaround for the memory randomization features in
2169 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2170 workaround will only be in effect on systems with the proc filesystem
2172 * bug fix: printing objects of type HASH-TABLE signals a
2173 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2175 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2176 have been running for over 50 days. (reported by Gilbert Baumann)
2177 * bug fix: the logic for getting names of functions gets less
2178 confused when confronded with alternate-metaclass
2179 funcallable-instances. (reported by Cyrus Harmon)
2180 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2181 consistent, even on internal alternate-metaclass objects.
2182 * bug fix: flush closure information collected by physical
2183 environment analysis prepass before the main pass. (bug reported
2185 * bug fix: compiler pack phase does not modify a hashtable, which is
2186 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2187 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2188 may be expanded. (reported by Alexander <malishev> on c.l.l)
2189 * bug fix: no more style-warnings when compiling code using LAST
2190 under high-SPEED low-SPACE settings. (reported by David Wallin
2192 * fixed some incompatibilities between SBCL's MOP and the MOP
2194 ** the METAOBJECT class is now implemented;
2195 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2196 STANDARD-OBJECT, as required;
2197 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2198 now compatible (as required by VALIDATE-SUPERCLASS); there
2199 remains a consistency requirement over the presence or absence
2200 of the FUNCTION class in the superclass list of finalized
2201 classes; see the manual for more details;
2202 ** the :AROUND method for COMPUTE-SLOTS on
2203 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2204 requested slot ordering.
2206 ** bug fix: parent thread now can be gc'ed even with a live
2208 ** bug fix: binding a special with PROGV to no value is not
2209 the same as that symbol not having been bound (thanks to
2211 ** bug fix: binding specials is thread safe (thanks to
2213 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2214 and SB-SPROF do not die with 'no handler for signal XX in
2215 interrupt_handle_now(..)' anymore
2216 ** bug fix: WITH-TIMEOUT works with multiple running threads
2217 * fixed some bugs revealed by Paul Dietz' test suite:
2218 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2219 the :method-class keyword argument.
2221 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2222 * new port: the Solaris operating system on x86 processors is now
2223 mostly supported, though some rough edges in the environment
2224 remain. (thanks to Daisuke Homma)
2225 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2226 space for its heap. (thanks to Thiemo Seufer)
2227 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2228 closures. (thanks to Thiemo Seufer)
2229 * minor incompatible change: eof selects abort in the debugger.
2230 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2231 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2233 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2234 deprecated, and will go away in a future revision of SBCL.
2235 * minor incompatible change: GC-ON and GC-OFF are no longer
2236 implemented with a counter, it does not matter how many times gc
2237 is switched on or off
2238 * bug fix: discriminating functions for generic function classes
2239 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2240 make invalid assumptions about method precedence order. (reported
2242 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2243 platforms. (thanks to Thiemo Seufer)
2244 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2245 * bug fix: correct pathname printing with printer escaping is on.
2246 (thanks to Kevin Reid)
2247 * bug fix: complex VOP definitions in "user-space" no longer trigger
2248 package locks. (reported by Zach Beane)
2249 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2250 for funcallable-instance functions, and is no different from
2252 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2253 characters in the current readtable. (reported by Nicholas Neuss)
2254 * bug fix: SUBTYPEP on various CONS types returns more a more
2255 accurate acknowledgment of its certainty.
2256 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2257 lists in some cases. This partially fixes bug 384.
2258 * flush all standard streams before prompting in the REPL and the
2260 * bug fix: signal handling and triggering gc do not conflict
2261 directly anymore, in particular a high frequency sb-sprof does
2262 not prevent gc from running
2263 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2264 approximation for timezone and DST information between the
2265 universal time epoch and the smallest negative 32-bit time_t.
2266 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2267 year 1899 so long as the encoded time is non-negative.
2268 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2269 the correct answer for non-constant rotate amounts when performing
2270 an inline 32-bit rotation.
2272 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2273 there is only one thread in the session
2274 ** bug fix: memory leak for streams created in one thread and
2275 written to in another
2276 ** bug fix: lockup when compiled with gcc4
2277 ** bug fix: race that allows the gc to be triggered when gc is
2279 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2280 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2282 ** bug fix: finished threads are gc'ed properly
2283 * fixed some bugs revealed by Paul Dietz' test suite:
2284 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2285 has arguments with a different set of applicable methods from
2286 the orignal arguments.
2287 ** The type error thrown by MAP now has a correct expected-type
2289 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2290 name a compiled function.
2291 ** Three MISC tests where a large bignum was improperly coerced to
2292 a float that couldn't represent that bignum during type
2293 derivation were fixed.
2294 ** SUBTYPEP can now handle types involving the negation of a
2295 list-form FUNCTION type.
2296 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2297 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2298 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2300 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2301 * New feature: Experimental support for bivalent streams: streams
2302 opened with :element-type :default now allow character and binary
2303 (unsigned-byte 8) I/O
2304 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2305 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2307 * bug fix: no more highly sporadic "couldn't check whether ~S is
2308 readable" when reading a stream and an interrupt hits in the middle
2309 of a select system call
2310 * compiler better recognizes complex arrays (reported by Eduardo
2312 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2313 for variables on SBCL built with linkage-tables. (reported by Luis
2315 * various error reporting improvements.
2316 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2317 (Thanks to Thiemo Seufer)
2318 * bug fix: nested reader invokations work correctly
2319 * bug fix: it is possible to have simultaneous references to foreign
2320 code and foreign data with the same name.
2322 ** added x86-64 support
2323 ** incompatible change: the threading api now works with thread
2324 objects instead of thread ids
2325 ** bug fix: threads are protected from signals and interruption when
2326 starting up or going down
2327 ** bug fix: a race where an exiting thread could lose its stack to gc
2328 ** bug fix: don't halt on infinite error in threads if possible
2329 ** fixed numerous gc deadlocks introduced in the pthread merge
2330 ** bug fix: fixed thread safety issues in read and print
2331 ** bug fix: debugger doesn't hang on session lock if interrupted at
2332 an inappropriate moment
2333 ** bug fix: run-program is now thread safe(r)
2334 ** bug fix: inner with-recursive-lock no longer releases the mutex
2335 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2336 * fixed some bugs revealed by Paul Dietz' test suite:
2337 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2338 EXPECTED-TYPE slots filled.
2339 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2341 ** MAKE-CONDITION accepts classes as type-designators.
2342 ** COMPILE may never return NIL.
2343 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2344 range before calling Unix time functions
2346 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2347 * numerous signal handling fixes to increase stability
2348 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2349 * minor incompatible change: we now correctly canonize default
2350 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2351 as per the MOP, rather than the historical (INITARG INITFUNCTION
2352 INITFORM). (reported by Bruno Haible)
2353 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2354 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2355 for more information.
2356 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2357 pathname is a directory pathname.
2358 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2359 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2361 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2362 :METACLASS argument in addition to a class name. (reported by
2363 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2364 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2365 (reported by Bruno Haible; patch from Gerd Moellmann)
2366 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2368 * bug fix: more cleanups to the floating point exception handling on
2369 x86-64 (thanks to James Knight)
2370 * bug fix: the compiler does not try to constant fold calls of
2371 COERCE and BIT-* functions when they return freshly constructed
2372 objects. (reported by Nikodemus Siivola and Paul Dietz)
2373 * optimization: improved the MIPS versions of generic (in the
2374 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2375 * optimization: direct conversion of (unsigned-byte 32) to floats on
2376 the PowerPC platform.
2377 * optimization: structure instances with raw slots now use less
2378 memory, and probably show better memory locality. (thanks to
2380 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2381 are treated as requests for stack allocation on the x86-64,
2382 Alpha-32, PPC and SPARC platforms.
2383 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2384 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2385 (thanks to Tony Martinez)
2387 ** gcing a dead thread can no longer lead to lockups
2388 ** threads block signals until they are set up properly
2389 ** errno is no longer shared by threads
2390 ** interrupt-thread restores the eflags register on x86
2391 ** fixed some lockups due to gc/thread interaction
2392 * fixed some bugs revealed by Paul Dietz' test suite:
2393 ** invalid dotted lists no longer raise a read error when
2394 *READ-SUPPRESS* is T
2395 ** signal an error if a symbol that names a declaration is used
2396 as the name of a type, or vice versa
2397 ** allow using the (declare (typespec var*)) abbreviation for
2398 (declare (type typespec var*)) with all type specifiers
2399 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2400 ** always return NIL from PPRINT-POP when OBJECT is NIL
2401 ** don't signal errors when pretty-printing malformed LABELS,
2402 FLET or MACROLET forms
2403 ** declarations in a DOLIST body are also in scope for the
2405 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2407 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2410 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2411 * fixed cross-compiler leakages that prevented building a 32-bit
2412 target with a 64-bit host compiler.
2413 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2414 opened with :IF-EXISTS :OVERWRITE.
2415 * fixed bug 281: error for an invalid qualifier in a short-form method
2416 combination method is not signalled until the faulty method is called.
2417 * bug fix: iteration variable type inferrer failed to deal with open
2418 intervals. (reported by Alan Shields)
2419 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2420 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2421 or more runtime options were provided to the sbcl binary.
2422 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2423 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2424 * optimization: there's now a fast-path for fixnum arguments in the
2425 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2426 * optimization: the code generated on x86-64 is more compact thanks
2427 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2428 * bug fix: floating-point exception handling now partly works on
2429 x86-64. (Thanks to James Knight)
2430 * improvement to the MIPS backend from Thiemo Seufer: C-style
2431 64-bit long long arguments and return values to alien functions
2433 * fixed some bugs revealed by Paul Dietz' test suite:
2434 ** the type-error signalled from WARN has a filled-in DATUM slot.
2435 ** the type-error required when a stream is not associated with
2436 a file has the stream as its datum.
2437 ** type-errors on single-floats on x86-64 no longer have
2438 :INVALID-OBJECT as the datum
2439 ** the type-errors signalled for invalid function names now have
2440 a correct expected type
2441 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2442 for structures defined with a :TYPE. Documentation strings for
2443 typed structures are no longer immediately discarded
2444 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2445 broadcast streams and synonym streams. FILE-LENGTH now also works
2446 on broadcast streams.
2448 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2449 * incompatible change: the --noprogrammer option, deprecated since
2450 version 0.7.5, has been removed. Please use the equivalent
2451 --disable-debugger option instead.
2452 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2454 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2455 inoperational for a while now) has been completely removed.
2456 * null lexical environments are now printed as #<NULL-LEXENV>,
2457 significantly reducing the amount of clutter in typical backtraces.
2458 * documentation on weak pointers, finalization, and after GC hooks
2459 has been added to the manual.
2460 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2461 as other specialized array types, is much faster. SUBSEQ and
2462 COPY-SEQ on such arrays have also been sped up.
2463 * optimization: EQL is now more efficient when at least other argument
2464 is known to be of type (OR FIXNUM (NOT NUMBER)).
2465 * fixed bug: the runtime is now less vulnerable to changes in the
2466 size of the SBCL object on OS X, and virtual memory is reserved for
2467 all spaces that need to be at a fixed address.
2468 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2469 * fixed bug: finalizers and after GC hooks that cause consing are now
2471 * fixed bug: compiler error messages and summaries are now printed to
2472 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2473 * fixed inference of the upper bound of an iteration variable.
2474 (reported by Rajat Datta).
2475 * fixed bug 211e: calling local functions with duplicated constant
2476 keyword argument no longer causes a bogus style warning about an
2478 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2479 ftype declaration to be disregarded. (reported by Dave Roberts)
2480 * fixed bug 373: caused by erronous compilation of references to alien
2481 variables in the runtime on ppc/darwin.
2482 * fixed bug 376: CONJUGATE type deriver.
2483 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2484 (reported by Baughn on #lisp)
2485 * a cross-compiler bug on non-x86 platforms has been identified and
2486 fixed. (thanks to Bruno Haible)
2487 * improvements to the MIPS runtime code for increased stability.
2488 (thanks to Thiemo Seufer)
2489 * increased the maximimum compact environment size to allow
2490 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2491 * partial workaround for bug 135: don't name the function we're
2492 calling for hairy cases of EVAL, so as not to accumulate one environment
2493 entry per call to EVAL. (reported by Kevin Reid)
2494 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2495 * optimization: type testing for non-vector arrays should be faster.
2496 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2498 * bug fix: setting 31st element of a bit vector to zero did not work
2500 * bug fix: redefining a class definition which failed due to a
2501 previous accessor / function clash now works (but see BUGS entry
2502 #380 for more problems in this area). (thanks to Zach Beane)
2503 * the long form of DEFINE-METHOD-COMBINATION disables method group
2504 checking when given a single method group with pattern *. (thanks
2506 * on x86 compiler supports stack allocation of results of simple
2507 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2508 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2509 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2510 * fixed some bugs related to Unicode integration:
2511 ** the restarts for recovering from input and output encoding
2512 errors only appear when there is in fact such an error to
2514 * fixed some bugs revealed by Paul Dietz' test suite:
2515 ** MISC.549 and similar: late transformation of unsafe type
2516 assertions into derived types caused unexpected code
2518 ** SCALE-FLOAT type deriver is less wrong.
2519 ** type derivers for EXP, LOG and similar functions compute result
2520 types for complex arguments better.
2521 ** (MISC.563) CONJUGATE type deriver works for very restricted
2523 ** out-of-line type testers for character strings are available.
2524 ** EQUAL compiler transform understands specialness of objects
2526 ** accessing double-floats stored on the stack now works on x86-64.
2527 ** debugger internals could sometimes create invalid lispobjs,
2528 resulting in GC crashes.
2529 ** MISC.548: type check weakening can convert required type into
2531 ** initialization forms for bindings are not in scope of free special
2534 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2535 * incompatible change: thread support for non-NPTL systems has
2536 been removed - locking is buggy and unreliable. A threaded
2537 SBCL build will now warn at startup and refuse to create
2538 new threads, unless futex support is detected in the kernel
2539 * incompatible change: the top level REPL now has only an ABORT
2540 restart associated with it, not TOPLEVEL and ABORT as it used to.
2541 TOP and TOPLEVEL are now available as debugger commands for
2542 returning to the top level.
2543 * incompatible change: forms evaluated in the REPL now use the
2544 global optimization policy.
2545 * incompatible change: user- and system-initialization files are
2546 no longer processed with LOAD, but by READ and EVAL; hence the
2547 global optimization policy, startup package, readtable, etc,
2549 * The .fasl file format number has been incremented because of
2550 various incompatible changes.
2551 * internal entry point details and argument counts no longer appear
2552 in backtraces unless explicitly requested by setting
2553 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2554 * built-in and standard functions no longer have names like "top
2555 level local call to FOO".
2556 * fixed bug 32: functions defined in non-null lexical environments
2557 now have more legible printed representation
2558 * fixed bug 33: functions defined in non-null lexical environemnts
2559 are now more amenable to inspection by INSPECT.
2560 * workaround for bug 354: XEPs no longer appear in backtraces unless
2561 explicitly requested.
2562 * fixed bug: receiving the signal which results from attempting to
2563 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2564 SIGBUS on other *BSDs) on architectures where the C stack is also
2565 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2566 to debugger confusion or wild instability. (reported by Cheuksan
2568 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2569 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2570 (reported by Lutz Euler)
2571 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2572 compiled code. (reported by Paul Dietz)
2573 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2574 specialized parameter to a method to something that is not TYPEP
2575 the specializer is now possible.
2576 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2577 face of package deletion.
2578 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2579 pathname. (reported by David Tolpin on c.l.l)
2580 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2581 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2582 STRING-STREAM and FILE-STREAM, respectively.
2583 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2584 than previously; in addition, its test suite is now run on build.
2585 (thanks to Luke Gorrie)
2586 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2587 * optimization: added a immediate representation for single-floats
2589 * fixed some bugs revealed by Paul Dietz' test suite:
2590 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2591 ** attempting to create a package with a colliding nickname causes
2592 correctable errors to be signalled.
2593 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2594 ** several x86-64 backend bugs related to sign-extension of immediate
2597 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2598 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2599 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2601 * fixed loading of multiply forward-referenced layouts.
2602 (thanks to Cheuksan Wang)
2603 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2604 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2605 (see COMPILE-FILE documentation for details of :PRINT option)
2606 * fixed bugs 19 and 317: fixed-format floating point printing is
2607 more accurate. This also fixes a bug reported by Adam Warner
2608 related to the ~@F format directive.
2609 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2611 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2612 dispatch table if the from-char is a dispatch macro character.
2613 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2614 by the wordsize no longer ignore the last word. (reported by Lutz
2616 * fixed bug in type checking of dynamic-extent variables. (reported
2618 * optimization: sequence traversal functions use their freedom to
2619 coerce function designators to functions.
2620 * optimization: code with many calls to CLOS methods specialized on
2621 CLOS classes has had redundant type checks removed.
2622 * optimization: added declarations to speed up operations that access
2623 the internal character database (for example STRING-UPCASE)
2624 * optimization: comparison operations between floats and sufficiently small
2625 fixnums no longer create extra rationals
2626 * fixed some bugs related to Unicode integration:
2627 ** portions of multibyte characters at the end of buffers for
2628 character-based file input are correctly transferred to the
2629 start of the buffer at the next read.
2630 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2631 passing it through to OPEN.
2632 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2633 argument given to internal calls to OPEN.
2634 ** the built-in comment readers (introduced by character sequences
2635 ";" and "#|") are more forgiving to encoding errors; they will
2636 STYLE-WARN and then attempt to resync the stream at a character
2637 boundary. (thanks to Teemu Kalvas)
2638 * fixed some bugs in the x86-64 port:
2639 ** Negative short int return values from c-calls are sign-extended
2641 ** The stack is aligned to 16-bytes for c-calls, as required by
2642 the ABI. (thanks to Cheuksan Wang)
2643 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2644 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2645 for enums. (thanks to Vincent Arkesteijn)
2646 ** Multiple small optimizations and bugfixes for floating point
2648 * fixed some bugs revealed by Paul Dietz' test suite:
2649 ** Space, Tab, Linefeed, Return and Page have the invalid
2650 secondary constituent character trait.
2651 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2653 ** WITH-INPUT-FROM-STRING should only update the index place on
2655 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2656 works more reliably.
2657 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2658 with one-dimensional arrays with fill pointers.
2659 ** TYPE-OF failed on a complex with an integer realpart and a
2661 ** compiler failure during type inference for the code of form
2662 (IF (EQL X (THE ...)) ...) (MISC.535).
2664 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2665 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2666 platform. The port supports 61-bit fixnums, large memory spaces
2667 and reloading shared object files.
2668 * enhancement: saving cores with foreign code loaded is now
2669 supported on ppc/Darwin in addition to the previously supported
2671 * enhancement: the statistical profiler now walks deeper into the
2672 call stack for more meaningful call-graphs and accrued time
2673 reports (x86/x86-64 only). It also now reports time spent in
2675 * enhancement: it is now possible to trace most individual methods
2676 of a generic function in addition to tracing the generic function
2678 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2680 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2681 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2682 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2683 produces an error. (thanks to Vincent Arkesteijn)
2684 * bug fix: NAMESTRING on pathnames with :WILD components in their
2685 directories works correctly. (thanks to Artem V. Andreev)
2686 * fixed bug 125: compiler preserves identity of closures. (reported
2688 * bug fixed: functions with &REST arguments sometimes failed with
2689 "Undefined function" when compiled with (DEBUG 3). (reported by
2691 * bug fix: overflow during compiling of setting element of a bit
2692 vector with constant index and value. (reported by Timmy Douglas)
2693 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2694 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2695 * on x86 compiler partially supports stack allocation of dynamic-extent
2697 * GO and RETURN-FROM do not check the extent of their exit points
2698 when compiled with SAFETY 0.
2699 * fixed some bugs related to Unicode integration:
2700 ** encoding and decoding errors are now much more robustly
2701 handled; it should now be possible to recover even from invalid
2702 input or output to the terminal. (thanks to Teemu Kalvas)
2703 ** provided a first cut at implementing STRING-TO-OCTETS and
2704 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2705 ** altered the SB-MD5 contributed module slightly, changing the
2706 interface just enough for it to be supportable for builds where
2707 lisp characters are not eight bits.
2708 * fixed some bugs revealed by Paul Dietz' test suite:
2709 ** the FORMATTER-generated functions for ~V[ conditionals require
2710 the correct number of arguments.
2711 ** READ-FROM-STRING returns the mandated second value when applied
2712 to displaced strings.
2713 ** the #\Rubout and #\Backspace characters are treated as invalid
2714 constituent characters by the tokenizer.
2716 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2717 * new feature: reloading changed shared object files with
2718 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2719 * new feature: references to foreign variables and functions
2720 can now be compiled and loaded before the corresponding shared
2721 object file is loaded, as long as the foreign definitions are
2722 available at runtime.
2723 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2724 supported. (thanks to Dan Debertin)
2725 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2726 just keywords. (thanks to Vincent Arkesteijn)
2727 * fixed bug #331: structure-class instances corresponding to
2728 DEFSTRUCT forms are now created eagerly.
2729 * fixed bug #345: backtraces from calls to undefined functions work
2730 on x86 as well. Related bug #61 is now also partially fixed on x86
2731 (backtraces from throws to unknown catch tags.)
2732 * bug fix: lambda-list parsing is now stricter vrt. order and number
2733 of lambda-list keywords.
2734 * bug fix: as specified by AMOP, an error is signalled if a
2735 class-option appears multiple times in a DEFCLASS form. (reported
2737 * bug fix: RANDOM can be compiled when the compiler derives the type
2738 of its numeric argument as a disjoint set of small integers.
2739 (reported by Paul Dietz)
2740 * bug fix: starting a core saved with shared objects loaded when
2741 those objects are not available no longer causes threaded SBCL to
2742 hang. (reported by Sean Ross)
2743 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2745 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2746 argument, which must always be NIL. (reported by Kalle Niemitalo)
2747 * bug fix: printing 1.0d+23 no longer results in an error.
2748 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2749 * bug fix: structure slot setters preserve evaluation order. (thanks
2751 * fixed some bugs related to Unicode integration:
2752 ** RUN-PROGRAM can allow its child to take input from a Lisp
2753 stream. (reported by Stefan Scholl)
2754 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2755 locales. (reported by Ken Causey)
2756 * fixed some bugs revealed by Paul Dietz' test suite:
2757 ** INCF, DECF and REMF evaluate their place form as specified in
2759 ** FORMATTER expands ~{ ~} iteration directives with V or #
2760 parameters correctly.
2761 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2762 consequent uses no arguments correctly.
2763 ** the system has a partial understanding of the (COMPLEX RATIO)
2765 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2768 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2769 * new feature: a build-time option (controlled by the :SB-UNICODE
2770 keyword feature, enabled by default) for building the system with
2771 support for the entire 21-bit character space defined by the
2773 * new feature: the system now has rudimentary external-format
2774 support; the primary user-visible change at this time is that
2775 characters with the high bit set (such as lower-case-e-acute) will
2776 print correctly to a terminal in a UTF-8 environment.
2777 * minor incompatible change: BASE-CHAR no longer names a class;
2778 however, CHARACTER continues to do so, as required by ANSI.
2779 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2780 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2781 has been moved to the SB-EXT package (temporarily re-exported from
2783 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2784 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2786 * fixed bug #351: better error-handlind and reporting for malformed
2788 * fixed bug #350: bignum-printing is now more memory-efficient,
2789 allowing printing of very large bignums, eg. (expt 2 10000000).
2790 (reported by Bruno Haible)
2791 * fixed bug #302: better primitive-type selection for intersection
2793 * fixed bug #308: non-graphic characters now all have names, as
2794 required. (reported by Bruno Haible)
2795 * bug fix: redefining a class with different superclasses now correctly
2796 removes it from the direct-subclasses of its previous superclasses.
2797 (reported by David Morse)
2798 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2799 new value now works. (reported by Bruno Haible)
2800 * bug fix: correct canonicalization of multiple non-standard slot
2801 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2802 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2803 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2804 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2805 now exists, an signals an error.
2806 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2807 during macroexpansion for non-iteration variables as well. (reported
2808 by Bruno Haible for CMUCL)
2809 * bug fix: Cyclic structures and unprintable objects in compiler
2810 messages no longer cause errors. (reported by Bruno Haible)
2811 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2812 and READ-FROM-STRING all now return a primary value of NIL if
2813 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2814 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2815 (reported by Bruno Haible for CMUCL)
2816 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2817 do not designate a condition. (reported by Bruno Haible for
2819 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2820 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2821 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2822 permitting resolution in favour of any of the conflicting symbols.
2823 (reported by Bruno Haible for CMUCL)
2824 * bug fix: EQUAL compiler optimizations is less aggressive on
2825 strings which can potentially compare true despite having distinct
2826 specialized array element types.
2827 * bug fix: unit enumerations can be defined without dividing by
2828 zero. (thanks to Vincent Arkesteijn)
2829 * FORMAT compile-time argument count checking has been enhanced.
2830 (report from Bruno Haible for CMUCL)
2831 * a partial workaround for the bug 262: the compiler does not try to
2832 inline-expand a local function doing RETURN-FROM from a deleted
2833 BLOCK. (thanks to Peter Denno for the bug report and to David
2834 Wragg for the simple test case)
2835 * fixed some bugs revealed by Paul Dietz' test suite:
2836 ** RENAME-PACKAGE allows all package designators as new package
2838 ** constraint propagation and lambda variable substitution are
2839 more cautious in dealing with partially deleted code.
2840 ** compiler sometimes lost reoptimization passes.
2841 ** CERROR, given a condition as condition designator, passes its
2842 remaining arguments to the continue format control without
2844 ** Case-altering FORMAT directives work correctly on non-ASCII
2846 ** The REAL type specifier handles bounds outside the floating
2847 point ranges without signalling FLOATING-POINT-OVERFLOW.
2848 ** Functions with IR1-transformations can create intercomponent
2849 references to global functions.
2850 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2852 ** FORMAT ~R treats a nil value for its first parameter correctly.
2854 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2855 * enhancement: saving cores with foreign code loaded is now
2856 supported on x86/NetBSD and sparc/Linux in addition to the previously
2857 supported platforms.
2858 * bug fix: on some platforms repeated installations caused multiple
2859 copies of HTML documentation to be installed -- should not happen
2860 any more. (reported by Stefan Scholl)
2861 * bug fix: parsing self-recursive alien record types multiple times
2862 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2863 original patch by Helmut Eller for CMUCL)
2864 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2865 (thanks to Richard Kreuter)
2866 * bug fix: defining classes whose accessors are methods on existing
2867 generic functions in other (locked) packages no longer signals
2868 bogus package lock violations. (reported by François-René Rideau)
2869 * bug fix: special variables as DEFMETHOD parameters no longer have
2870 associated bogus type declarations. (reported by David Wragg and
2872 * bug fix: read-write consistency on streams of element-type
2873 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2874 * bug fix: redefiniton of the only method of a generic function with
2875 no DEFGENERIC no longer emits a full WARNING. In addition,
2876 redefinition of generic functions with no DEFGENERIC to an
2877 incompatible lambda list now signals an error. (thanks to Zach
2879 * bug fix: DEFGENERIC now works even when there's a function of the
2880 same name in an enclosing lexical environment. (thanks to Zach
2882 * fixed compiler failure, caused by instrumenting code during
2883 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2884 * optimization: added loop analysis and improved register allocation
2885 * fixed some bugs revealed by Paul Dietz' test suite:
2886 ** POSITION on displaced vectors with non-zero displacement
2887 returns the right answer.
2888 ** (SIMPLE-STRING) is a valid type specifier for sequence
2890 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2892 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2893 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2895 ** The justification version of the FORMAT ~< directive treats
2896 non-zero minpad parameter correctly.
2898 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2899 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2900 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2901 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2902 the supported interface.
2903 * new feature: Single-stepping of code compiled with DEBUG 2 or
2904 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2905 * new feature: saving cores with foreign code loaded is now
2906 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2907 Timothy Moore's work for CMUCL)
2908 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2909 parameters to * instead of NIL if no initform is supplied.
2910 (reported by Johan Bockgaard)
2911 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2912 correctly when FUNCALL appears as the car of the form. Note:
2913 despite this FUNCALL forms are not currently subject to
2914 compiler-macro expansion. (port of Raymond Toy's fix for the
2915 same from CMUCL, reported by Johan Bockgaard)
2916 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2917 (thanks for Teemu Kalvas)
2918 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2919 case instead of terminating the iteration (thanks for Julian
2920 Squires, Sean Champ and Raymond Toy)
2921 * bug fix: incorrect expansion of defgeneric that caused a style
2922 warning. (thanks for Zach Beane)
2923 * bug fix: slot accessor effective method computation works properly
2924 for classes with multiple non-standard applicable methods on
2925 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2926 * on x86 compiler supports stack allocation of results of LIST and
2927 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2928 CMUCL implementation by Gerd Moellmann)
2929 * fixed some bugs revealed by Paul Dietz' test suite:
2930 ** FORMAT strings with both the ~<~:;~> form of the justification
2931 directive and pretty-printing directives cause an error.
2933 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2934 * incompatible change: the internal functions
2935 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2936 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2937 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2938 instead of the old functions.
2939 * new feature: on platforms where "dladdr" is available foreign
2940 function names now appear in backtraces. (based on Helmut Eller's
2942 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2943 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2945 * documentation: ASDF manual is now distributed with SBCL.
2946 * bug fix: SBCL can now load its contributed modules with REQUIRE
2947 even if the system-provided entries have been removed from
2948 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2950 * bug fix: backtraces involving undefined functions or assembly
2951 routines are more informative. (thanks to Brian Downing)
2952 * bug fix: mutually referent alien structures now work correctly.
2953 (reported by Rick Taube)
2954 * bug fix: structures defined by WITH-ALIEN can be referred to
2955 within other definitions in the same WITH-ALIEN.
2956 * bug fix: division operators (MOD, TRUNCATE and the like) with
2957 constant zero divisors and integer dividends no longer generate
2959 * bug fix: provide default methods for INPUT-STREAM-P and
2960 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2961 * bug fix: improve the detection and resolution of MOP metacycles
2962 (where effective method computation depends on the generic function
2963 having its effective method computed). (reported by Bruno Haible)
2964 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2965 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2966 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2967 * bug fix: pathnames with relative directory components can be
2968 represented relative to default pathnames.
2969 * optimization: in taking the GCD of bignums, reduce the two bignums
2970 to approximately the same size (using Euclid's algorithm) before
2971 applying the more sophisticated binary GCD. (thanks to Juho
2973 * optimization: COUNT on bitvectors now operates word-at-a-time.
2974 * optimization: ASH with a positive, but not necessarily constant,
2975 (leftwards) shift, when in a modular context, is compiled to a
2977 * fixed some bugs revealed by Paul Dietz' test suite:
2978 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2979 if the corresponding argument is NIL.
2980 ** FORMAT directives accepting a minpad parameter treat negative
2982 ** literal commas as character parameters to FORMAT directives are
2984 ** literal spaces directly after ~<Newline> directives within a
2985 format-logical-block (~:< ~@:>) do not induce :FILL-style
2986 conditional newlines.
2987 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2988 as specified: it no longer includes conditional newlines).
2989 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2991 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2992 :PER-LINE-PREFIX argument does not evaluate to a string.
2994 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2995 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2996 the manual for details; for now, package locks can be disabled by
2997 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2998 if no major problems are found then it is likely that they will be
2999 compiled in unconditionally.
3000 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
3001 unsupported operators on all platforms. To load a shared library
3002 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
3003 object file, link it into a shared library outside of SBCL and
3004 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
3006 * bug workaround: ROOM T and the GC were not working together
3007 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
3008 reporting facility aren't preserved. That reporting has been
3009 disabled, so now until and unless someone figures out how to make it
3010 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
3011 * minor incompatible change: as threatened around sbcl-0.8.0, the
3012 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
3013 The symbols are also exported from SB-PCL for backwards
3014 compatibility, but more so than before SB-PCL should be treated as
3015 an implementation-internal package.
3016 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
3018 * the SB-SPROF contrib now works on (most) non-x86 architectures.
3019 It is known as of this release not to work on the Alpha, however.
3020 * fixed bug #167: errors signalled due to illegal syntax in method
3021 bodies are now more legible.
3022 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
3023 designators and can hence be used with TYPEP.
3024 * fixed bug #333: CHECK-TYPE now ensures that the type error
3025 signalled, if any, has the right object to be accessed by
3026 TYPE-ERROR-DATUM. (reported by Tony Martinez)
3027 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
3028 5.1.2.3 for multiple-value place subforms. (reported by Kalle
3030 * fixed bug #334: programmatic addition of slots using specialized
3031 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
3032 and :ALLOCATION :CLASS slots.
3033 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
3034 float exponents. (rereported by Peter Seibel)
3035 * fixed a bug: #\Space (and other whitespace characters) are no
3036 longer considered to be macro characters in standard syntax by
3037 GET-MACRO-CHARACTER.
3038 * fixed bug: initialization of condition class metaobjects no longer
3039 causes an instance of the condition to be created. (reported by
3041 * fixed bug: it is now possible to have slots such that
3042 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
3043 description is neither :INSTANCE nor :CLASS.
3044 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
3045 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
3046 SLOT-VALUE-USING-CLASS), and no longer causes errors with
3047 non-standard SLOT-DEFINITION-ALLOCATION values.
3048 * fixed bugs: various race conditions handling exiting threads.
3049 CL-PPCRE's thread torture test now passes.
3050 * fixed arguably-a-bug: GC time proportional to number of threads in
3051 system even when most of them are idle
3052 * optimization: improved performance of BIT and SBIT on bit-vectors.
3053 * .fasl file incompatibility: The fasl file version number has
3054 been incremented because of changes associated with package locks.
3056 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
3057 * minor incompatible change: the system no longer provides
3058 optimization hints (conditions which are TYPEP
3059 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
3060 mode; these hints are still emitted when higher SPEED optimization
3062 * new contrib module: a sampling profiler (profiling by statistical
3063 sampling, rather than by instrumenting functions) is available as
3064 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
3065 * the behaviour of the standard function ED is now customizeable by
3066 third parties through a hook variable: see ED's documentation
3067 string for information on the protocol.
3068 * the compiler no longer emits efficiency notes for (FUNCALL X)
3069 when the type of X is uncertain under default optimization
3071 * fixed bug 276: mutating a binding of a specialized parameter to a
3072 method to something that is not TYPEP the specializer is now
3074 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
3075 exhibit the required behaviour on the x86 platform. (thanks to
3076 Peter van Eynde, Eric Marsden and Bruno Haible)
3077 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
3078 even for difficult arguments. (reported by Peter Graves)
3079 * fixed bug 141a: the backquote printer now descends quoted
3081 * fixed another bug in backquote printing: no more destructive
3082 modification of the form's list structure. (reported by Brian
3084 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
3085 move between its address being taken and the call to
3086 interrupt_thread, fixing a crashing race condition.
3087 * the SB-POSIX contrib implementation has been adjusted so that it
3088 no longer exhibits ridiculously poor performance when constructing
3089 instances corresponding to C structs.
3091 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
3092 * minor incompatible change: the sb-grovel contrib now treats C
3093 structures as alien (in the sense of SB-ALIEN) objects rather than
3094 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
3095 has implications for memory management of client code
3096 (sb-grovel-returned objects must now be manually managed) and for
3097 type safety (alien objects now have full types).
3098 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
3099 used to control emission of compiler diagnostics, rather than the
3100 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
3101 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
3102 quality should be considered deprecated.
3103 * install.sh now installs the user manual as well
3104 * (not quite a new documentable feature, but worth considering in
3105 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
3106 beginnings of a semantically meaningful condition hierarchy is
3107 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3109 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3110 designator as the defaults argument.
3111 * fixed bug: Displaced arrays whose displaced-to array has become
3112 too small now cause ARRAY-DIMENSION to signal an error, providing
3113 for safer bounds-checking. (reported by Bruno Haible)
3114 * fixed bug: DEFCLASS slot definitions with identical :READER and
3115 :WRITER names now signal a reasonable error. (reported by Thomas
3117 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3119 * fixed bug: Invalid slot specification errors now print correctly.
3120 (thanks to Zach Beane)
3121 * fixed bug 320: Shared to local slot value transfers in class
3122 redefinitions now happen corrently from superclasses as
3123 well. (reported by Bruno Haible)
3124 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3126 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3127 improper lists correctly. (reported by Bruno Haible)
3128 * fixed bug 313: source-transform for <fun-name> was erroneously
3129 applied to a call of a value of a variable with name <fun-name>.
3130 (reported by Antonio Menezes Leitao)
3131 * fixed bug 307: The obsolete instance protocol ensures that
3132 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3133 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3134 remove dead unknown-values globs from the middle of the stack.
3135 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3136 easier to resume long computations after using *BREAK-ON-SIGNALS*
3137 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3138 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3139 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3140 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3141 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3142 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3144 * fixed a (fairly theoretical) bug in string printing: if
3145 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3146 does not have array-element-type equal to the most general string
3148 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3149 function name. (thanks to Nikodemus Siivola)
3150 * fixed bug: compile-time format string checker failed on
3151 non-closed ~{. (reported by Thomas F Burdick)
3152 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3153 don't warn on legal loop code involving more than one aggregate
3154 boolean. (thanks to Nikodemus Siivola)
3155 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3156 in keyword position to MAKE-PACKAGE was not being reported as
3158 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3159 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3161 * fixed bugs 280 and 312: the checking for multiple definitions in a
3162 file is less likely to become confused by uses of inline
3164 * fixed bug: the #S reader macro performs the keyword coercion
3165 specified for slot names. (reported by Kalle Niemitalo)
3166 * fixed bug: lambda lists may contain symbols whose names start with
3167 & but are not lambda-list-keywords; their occurrence triggers a
3169 * fixed bug 321: define-method-combination argument lambda lists do
3170 not cause invalid code to be generated when &optional or &aux
3171 variables have default values. (reported by Bruno Haible)
3172 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3173 have CLOS classes; this appears also to have fixed reported
3174 instability in stack exhaustion detection.
3175 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3176 MUFFLE-WARNING when no associated restart is present is now
3178 * optimization: rearranged the expansion of various defining macros
3179 so that each expands into only one top-level form in a
3180 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3182 * optimization: used a previously little-used slot in symbols to
3183 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3185 * fixed some bugs revealed by Paul Dietz' test suite:
3186 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3188 ** Fixed an optimization bug related to inheritance of initforms
3189 from local to shared slots.
3190 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3191 ** CAST optimizer forgot to flush argument derived type.
3192 ** print/read consistency on floats is now orders of magnitude
3193 more likely. (thanks also to Bruno Haible for a similar report
3195 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3196 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3197 for objects of type REAL. Make it so.
3198 ** SXHASH is specified (infelicitously) to respect similarity,
3199 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3200 it so. (thanks to Markus Ziegler)
3201 ** on the Alpha, the compiler succeeds in compiling functions
3202 returning a known number of arguments greater than 63.
3203 ** fixed handling of invalid NIL arguments in keyword position
3205 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3207 ** provided a readably-printable representation for RANDOM-STATE
3209 ** ensured that pathnames, where they have a namestring, always
3210 print using #P"..." syntax.
3212 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3213 * Support for the forthcoming 2.0 version of the NetBSD kernel
3214 running on x86 hardware has been added. (thanks to Perry
3215 E. Metzger most immediately, and others for their past work)
3216 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3217 current release version; SBCL's OpenBSD support had been broken
3218 since about the time of OpenBSD's switch to ELF binary format.)
3219 * [placeholder for DX summary]
3220 ** user code with &REST lists declared dynamic-extent, under high
3221 speed or space and low safety and debug optimization policy.
3222 * The manual has been converted to Texinfo format and the debugger
3223 chapter from the cmucl manual has been added.
3224 * A facility has been added to extract documentation strings from
3225 sbcl and store them as Texinfo-formatted snippets for inclusion in
3226 the manual (via Texinfo's @include directive)
3227 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3228 the test case to Dave Roberts)
3229 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3230 pointers. (reported by Sean Ross)
3231 * bug fix: PROFILE output is printed nicely even for large numerical
3232 values. (thanks to Zach Beane)
3233 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3234 greater than 32 handle EOF correctly.
3235 * bug fix: on X86 an immediate argument of the IMUL instruction is
3236 correctly printed in disassembly. (thanks to Lutz Euler)
3237 * bug fix: class slots in redefined classes preserve their old
3238 values. (thanks to Bruno Haible and Nikodemus Siivola)
3239 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3240 succeeds. (reported by Marco Baringer)
3241 * fixed some bugs revealed by Paul Dietz' test suite:
3242 ** READ-SEQUENCE now works on ECHO-STREAMs.
3243 ** RATIONALIZE works more according to its specification. (thanks
3246 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3247 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3248 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3249 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3250 should matter to you only if you rebind the printer control
3251 variables and then find you want different bindings in the
3252 debugger than in the ordinary execution of your program.)
3253 * The runtime build system has been tweaked to support building
3254 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3255 assembler and linker. (thanks to Nikodemus Siivola)
3256 * Unbound, undefined, undeclared variables now trigger full
3257 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3258 more often programmer error than deliberate exploitation of undefined
3260 * optimization: the hash algorithm for strings has changed to one
3261 that is less vulnerable to spurious collisions. (thanks to Juho
3263 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3264 needless bounds checking. (thanks to Juho Snellman)
3265 * optimization: implemented multiplication as a modular
3266 (UNSIGNED-BYTE 32) operation on the PPC backend.
3267 * fixed some bugs revealed by Paul Dietz' test suite:
3268 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3270 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3271 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3272 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3274 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3276 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3278 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3279 their output stream on EOF from read.
3280 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3281 have been read to end-of-file.
3282 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3284 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3285 description of determination of which consecutive characters
3287 ** Printing the "Space" character with escaping on now yields "#\\ ",
3288 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3289 ** Reading floating-point numbers with *READ-BASE* set to a number
3290 less than 10 works correctly.
3291 ** Reading floating-point numbers with *READ-BASE* set to a number
3292 more than 10 works correctly.
3293 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3294 the readtable currently in effect.
3296 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3297 * minor incompatible change: parsing of namestrings on a physical
3298 (Unix) host has changed; numbers after the final #\. in a
3299 namestring are no longer interpreted as a version field. This is
3300 intented to be largely invisible to the user, except that the
3301 meaning of the namestring "*.*.*" has changed: it now refers to a
3302 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3303 should usually be replaced by
3304 (make-pathname :name :wild :type :wild :version :wild)
3305 with the added benefit that this is more likely to be portable.
3306 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3307 option to OPEN now signals an error if the file being opened
3308 exists; this may have an impact on existing code.
3309 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3310 signals received, on the PowerPC platforms (both Linux and
3311 Darwin). (thanks to Pierre Mai for pointing out the location of
3313 * several fixes on OS X: The system now builds and runs cleanly on
3314 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3315 (thanks to Brian Mastenbrook)
3316 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3317 second-resolution: integer multiples of 1/3600 between -24 and 24.
3318 (thanks to Vincent Arkesteijn)
3319 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3320 type when called with 1 argument; PEEK-CHAR checked type of
3321 PEEK-TYPE only after having read first character from a
3322 stream. (reported by Peter Graves)
3323 * bug fix: the garbage collector now has much better locality
3324 behaviour, and in particular no longer treats all memory as being
3325 exhausted when many small objects point to each other in a deeply
3327 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3328 recognized as being TYPEP their class.
3329 * bug fix: the PUSHNEW documentation string has been corrected.
3330 (thanks to Vincent Arkesteijn)
3331 * bug fix: defaulting of the value for the last of an atypically
3332 large number of multiple values being bound was not being
3333 performed correctly on the Alpha or PPC platforms
3334 * optimization: implemented multiplication as a modular
3335 (UNSIGNED-BYTE 32) operation on the x86 backend.
3336 * optimization: SEARCH on simple-base-strings can now be open-coded.
3337 (see also contrib/compiler-extras.lisp for inspiration for
3338 teaching the compiler about the Boyer-Moore algorithm).
3339 * value, returned by MAX (and MIN) called with several EQUALP, but
3340 not EQL, arguments now does not depend on compiler settings.
3341 * fixed some bugs revealed by Paul Dietz' test suite:
3342 ** in stack analysis liveness information is propagated from
3343 non-local entry points.
3344 ** pathwise CAST removing failed when the CAST node did not start
3346 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3347 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3349 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3350 host is already defined.
3351 ** RENAME-FILE works on streams instead of signalling an internal
3353 ** PEEK-CHAR uses the current readtable when determining whether
3354 or not a character is whitespace.
3355 ** MERGE-PATHNAMES handles the case when the pathname does not
3356 specify a name while the default-pathname specifies a version.
3357 ** Pathnames now stand a better chance of respecting print/read
3359 ** Attempting to use standardized file system operators with a
3360 pathname with invalid :DIRECTORY components signals a
3362 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3364 ** DIRECTORY on logical pathnames is more correct.
3365 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3366 signal a TYPE-ERROR if their argument is not a stream.
3367 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3368 designator argument does not designate a stream.
3369 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3370 examining the synonym.
3371 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3373 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3374 element-type larger than ([UN]SIGNED-BYTE 32).
3376 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3377 * When built with the :SB-FUTEX feature, threaded builds now take
3378 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3379 for faster/more reliable mutex and condition variable support.
3380 * Incompatible change (but one you probably shouldn't have been using
3381 anyway): the interface and code for arbitrating between multiple
3382 threads in the same user session has been redesigned.
3383 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3384 error when called without an explicit environment argument.
3385 (thanks to Nikodemus Siivola)
3386 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3387 confused about their position. (thanks to Adam Warner and Gerd
3389 * bug fix: Pretty printing backquoted forms with unquotations in the
3390 argument list position of various code constructs such as LAMBDA
3391 now works correctly. (reported by Paul Dietz)
3392 * bug fix: Pretty printing unquotations no longer loses all
3393 stream position information.
3394 * optimization: performance of string output streams is now less
3395 poor for multiple small sequence writes.
3396 * optimization: performance of CSUBTYPEP in the presence of complex
3397 expressions involving CONS and NOT many times has been improved.
3398 (reported by Paul Dietz)
3399 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3401 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3402 streams for the REPL, for the debugger, and so on. (thanks to
3404 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3405 * fixed some bugs revealed by Paul Dietz' test suite:
3406 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3407 is ignored if it is too small, rather than propagating through
3408 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3409 ** extremely complex negations of CONS types were not being
3410 sufficiently canonicalized, leading to inconsistencies in
3412 ** VALUES tranformer lost derived type.
3414 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3415 * fixed a bootstrapping bug: the build process no longer assumes
3416 that the various BOOLE-related constants have the same value in
3417 host and target lisps. (noted by Paul Dietz' test suite on an
3418 SBCL binary built from CLISP)
3419 * The system can now be dynamically linked on the MIPS platform,
3420 which enables dynamic loading of foreign code from Lisp. (thanks
3421 to Ralf Baechle for discussions on the MIPS ABI)
3422 * The system now records debugging information for its own source
3423 files in a filesystem-position-independent manner, relative to
3424 the "SYS" logical host.
3425 * fixed a compiler bug: MV-LET convertion did not check references
3426 to the "max args" entry point. (reported by Brian Downing)
3427 * tweaked disassembly notes to be less confident about proclaiming
3428 some instruction as an LRA. (thanks to Brian Downing)
3429 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3430 now each have their own history, command character, and other
3431 characteristics. (thanks to David Lichteblau)
3432 * fixed some bugs revealed by Paul Dietz' test suite:
3433 ** compiler failure in compiling LOGAND expressions including a
3435 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3436 shift greater than 32.
3437 ** FLUSH-DEST did not mark blocks for type check regeneration.
3438 ** HANDLER-CASE failed to accept declarations in handler clauses
3439 in some circumstances.
3441 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3442 * New code in contrib/sb-introspect (still probably not entirely
3443 stable yet) provides some support for smart Lisp development
3444 environments like SLIME.
3445 * The conditions signalled for errors occurring when loading .fasl
3446 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3447 in a way which should help ASDF recover gracefully.
3448 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3449 stuff has been cleaned up. If you code contrib/ stuff, this might
3450 affect you, and you can look at contrib/README, contrib/STANDARDS,
3451 and/or the 0.8.4.27 diff to check.
3452 * In full calls the compiler now does not generate checks for declared
3453 argument types for all arguments.
3454 * various threading fixes
3455 ** and some experimental patches which didn't make it into
3456 the main tree for this release, but which are shipped in
3457 contrib/experimental-thread.patch as a possible fix for some
3458 failures (deadlock, spinning...) in GC-intensive multithreaded
3460 * fixed PPC build problem (source code incompatibility of different
3461 library versions): added offsetof() hackery which attempts to divine
3462 where glibc maintainers put uc_mcontext today
3463 * fixed bug 282: compiler does not trust type assertions while passing
3464 arguments to a full call.
3465 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3466 &OPTIONAL VALUES type parameter.
3467 * fix bug 214: algorithm for noting rejected templates is now more
3468 similar to that of template seletion. (also reported by rydis on
3470 * fixed bug 141b: printing backquoted information readably and prettily
3471 inserts a space where necessary.
3472 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3473 (CHARACTER 10) are now reported as errors, rather than propagated
3474 as unknown types. (reported by piso on #lisp)
3475 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3476 with values NIL and :ERROR. (thanks to Milan Zamazal)
3477 * fixed bug 191c: CLOS now does proper keyword argument checking as
3478 described in CLHS 7.6.5 and 7.6.5.1.
3479 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3480 longer raise an error; further, using a list as a for-as-arithmetic
3481 counter now raises a meaningful error.
3482 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3483 now understood by sequence creation functions such as MAKE-SEQUENCE
3485 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3486 a STRING-INPUT-STREAM.
3487 * compiler enhancement: SIGNUM is now better able to derive the type
3489 * type declarations inside WITH-SLOTS are checked. (reported by
3491 * fixed some bugs revealed by Paul Dietz' test suite:
3492 ** incorrect optimization of TRUNCATE for a positive first
3493 argument and negative second.
3494 ** compiler failure in let-convertion during flushing dead code.
3495 ** compiler failure while deriving type of TRUNCATE on an
3496 interval, containing 0.
3497 ** ASH of a negative bignum by a negative bignum count now returns
3499 ** intersection of CONS types now canonicalizes properly, fixing
3500 inconsistencies in SUBTYPEP.
3502 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3503 * incompatible change: The --disable-debugger command line
3504 option now clobbers the debugger at a more fundamental
3505 level, by redefining #'INVOKE-DEBUGGER instead of by
3506 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3507 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3508 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3509 enter the debugger REPL and then suffer recursive errors
3510 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3511 while under the new system, BREAK in --disable-debugger mode
3512 terminates the system just as an unhandled error would.
3513 * fixed compiler performance when processing loops with a step >1;
3514 * bug fix: DOCUMENTATION now retrieves generic function
3515 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3516 support has been systematized, and now supports the methods
3517 specified by ANSI, along with a default method and a method for
3518 slot documentation. (reported by Nathan Froyd)
3519 * bug fix: effective methods associated with a generic function are
3520 no longer cached over a change of that generic function's method
3521 combination. (reported by Andreas Fuchs)
3522 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3523 names a non-existent directory. (thanks to Andreas Fuchs)
3524 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3525 the PPC platform, lead to stack corruption; this has been fixed.
3526 (reported by Rainer Joswig)
3527 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3528 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3529 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3530 arithmetic in general, is now much more reliable on the PPC
3532 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3533 platform now returns the right answer.
3534 * optimization: restored some effective method precomputation in
3535 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3536 precomputation is now tunable.
3537 * optimization: compiler-internal data structure use has been
3538 reviewed, and changes have been made that should improve the
3539 performance of the compiler by about 20%.
3540 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3541 simple-base-strings and simple-bit-vectors is improved.
3542 * optimization: the optimization of 32-bit logical and arithmetic
3543 functions introduced in version 0.8.3 on the x86 has been
3544 implemented on the mips, ppc and sparc platforms; an
3545 implementation of the same facility, but for 64-bit arithmetic,
3546 has been added for the alpha.
3547 * microoptimization: the compiler is better able to make use of the
3548 x86 LEA instruction for multiplication by constants.
3549 * bug fix: in some situations compiler did not report usage of
3550 generic arithmetic in (SPEED 3) policy.
3551 * bug 145b fix: compiler used wrong type specifier while converting
3552 MEMBER-types to numeric.
3553 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3555 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3556 index argument now works properly on the Alpha platform.
3557 * bug fix: floating point exception treatment on the Alpha platform
3559 * bug fix: FILE-POSITION works much better on string input and
3560 output streams. (thanks to Nikodemus Siivola)
3561 * bug fix: many threading/garbage collection symptoms sorted.
3562 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3563 might be pseudo-atomic.
3564 * internal change: Stopping for GC is now done with signals not
3565 ptrace. GC is now done in whichever thread wanted it, instead of
3567 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3569 * bug fix: result form in DO is not contained in the implicit
3571 * incompatible change: ICR structure is changed; the value part of
3572 CONTINUATION is now called LVAR; corresponding functions are
3573 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3574 * added type deriver for ISQRT (thanks to Robert E. Brown).
3575 * bug fix: better support for loading from the command line when an
3576 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3578 * fixed some bugs revealed by Paul Dietz' test suite:
3579 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3580 ** ROUND and FROUND now give the right answer when given very
3581 small float arguments.
3582 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3584 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3585 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3586 ** LCM should return a non-negative integer.
3587 ** PARSE-INTEGER returned the index of a terminator instead of the
3588 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3589 ** PARSE-INTEGER returned an incorrect index being applied to a
3591 ** LCM with two arguments of 0 returns 0 rather than signalling
3593 ** unsigned addition of a 32-bit constant with the high bit set no
3594 longer causes an internal compiler error on the x86.
3595 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3596 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3597 with negative last argument.
3598 ** byte specifiers with zero size and position no longer cause
3599 an error during type derivation.
3600 ** bignum multiplication on the Alpha platform now returns the
3602 * porting: The system now builds on SuSE AMD64, although it still
3603 generates a 32-bit binary.
3604 * .fasl file incompatibility: The fasl file version number has
3605 been incremented (because of the changes to internal compiler
3606 data structures referred to above).
3608 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3609 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3610 more accurately, on the Darwin kernel running on PowerPC hardware.
3611 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3612 * Compiler code deletion notes now signal a condition of type
3613 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3614 an associated MUFFLE-WARNING restart.
3615 * The compiler now performs limited argument count validation of
3616 constant format strings in FORMAT, and where appropriate in ERROR,
3617 CERROR and WARN. (thanks to Gerd Moellmann)
3618 * New ASDF-INSTALL contrib can be used for automatic download and
3619 installation of third-party Lisp code from CCLAN or other sites
3621 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3622 which forces another thread to execute a function supplied by the
3624 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3625 now accept and act upon their :ELEMENT-TYPE keyword argument.
3626 (reported by Martin Atzmueller, Edi Weitz)
3627 * bug fix: FILE-POSITION now accepts position designators up to
3628 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3629 is the greater. (thanks to Patrik Nordebo)
3630 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3631 Kalle Olavi Niemitalo)
3632 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3633 printable. (reported by Eric Marsden)
3634 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3635 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3636 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3637 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3638 (thanks to Stig Sandoe)
3639 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3640 UNREAD-CHAR now correctly includes the unread character in the
3641 target sequence. (thanks to Gerd Moellmann)
3642 * bug fix in threaded builds: the system can now be suspended and
3643 resumed by shell job control with minimal disruption.
3644 * bug fixes in times and timezones >2038AD
3645 * better handling of "where is GNU make?" problem in build scripts
3646 (thanks to Nikodemus Siivola)
3647 * new optimization: inside a named function any reference to a
3648 function with the same name is considered to be a self-reference;
3649 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3650 optimization quality.
3651 * new optimization on x86: logical functions and + now have
3652 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3653 used when the result is truncated to 32 bits.
3654 * VALUES declaration is partially enabled.
3655 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3656 * fixed some bugs revealed by Paul Dietz' test suite:
3657 ** The system now obeys the constraint imposed by
3658 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3659 types form a lattice under type intersection.
3660 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3661 ** ASSOC now ignores NIL elements in an alist.
3662 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3663 and (1+ MOST-POSITIVE-FIXNUM) answers.
3664 ** The addition of a method with invalid qualifiers to a generic
3665 function does not cause an error to be signalled immediately;
3666 a warning is signalled, and the error is generated only on
3667 calling the generic function.
3668 * changed .fasl file version number, in part to add type codes for
3669 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3670 obscure ANSI requirements
3672 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3673 * fixed bug 148: failure to inline-expand a local function left
3674 garbage, confusing the compiler.
3675 * fixed bugs 3cd: structure slot readers perform type check if the
3676 slot can have an invalid value (i.e. it is either not initialized
3677 or can be written with a less specific slot writer).
3678 * bug fix: the compiler now traps array references to elements off
3679 the end of an array; previously, the bounds checking in some
3680 circumstances could go off-by-one.
3681 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3683 * type declarations for array element types now obey the description
3684 on the CLHS page "Declaration TYPE", as per discussions on
3685 sbcl-help around 2003-05-08. This means that a declaration
3686 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3687 declaration, all references to BAR will be asserted or assumed
3688 (with THE, so dependent on compiler policy) to involve objects of
3689 type FOO. Note that no such declaration is implied in
3690 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3691 * declared types of functions from the "Conditions"
3692 chapter. (reported by Paul Dietz)
3693 * bug fix: CERROR accepts a function as its first argument.
3694 * bug fix: NTH an NTHCDR accept a bignum as index
3695 arguments. (reported by Adam Warner)
3696 * optimization: character compare routines now optimize comparing
3697 against a constant character. (reported by Gilbert Baumann)
3698 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3699 argument now works properly on the MIPS platform.
3700 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3701 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3702 declared to have a functional type, does not cause an error
3704 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3705 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3706 * fixed bug: if last continuation of a deleted block has a
3707 destination, this destination should be deleted too. (reported by
3709 * fixed a bug in the bootstrap process: the host compiler's values
3710 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3711 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3712 test case from Patrik Nordebo)
3713 * improved the ability of the disassembler on the PPC platform to
3714 provide helpful disassembly notes.
3715 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3716 the class in more cases than previously.
3717 * bug fix: FILE-POSITION now understands :START and :END for
3718 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3719 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3720 8) data. (thanks to David Lichteblau)
3721 * bug fix: it is possible to add a method to a generic function
3722 without lambda list.
3723 * bug fix: reader failed to signal END-OF-FILE inside an
3724 object representation. (reported by Nikodemus Siivola)
3725 * fixed some bugs revealed by Paul Dietz' test suite:
3726 ** LAST and [N]BUTLAST should accept a bignum.
3727 ** condition slot accessors are methods.
3728 ** (VECTOR NIL) is a subtype of STRING.
3730 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3731 * minor incompatible change: some nonsensical specialized lambda
3732 lists (used in DEFMETHOD) which were previously ignored now signal
3734 * minor incompatible change: the system is now aware of the types of
3735 variables in the COMMON-LISP package, and will signal errors for
3736 most violations of these type constraints (where previously they
3737 were silently accepted).
3738 * minor incompatible change: COMPILE-FILE now uses the freedom
3739 afforded (ANSI 3.2.2.3) to use derived function types for
3740 functions defined in the same file. This also permits the system
3741 to warn on static type mismatches and function
3742 redefinition. (Currently it does not work with high DEBUG level.)
3743 * minor incompatible change: VALUES declaration is disabled.
3744 * When issuing notes, the compiler now signals a condition of type
3745 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3746 restart for use in user handlers. It is expected that the
3747 COMPILER-NOTE condition will eventually become a condition
3748 supertype to a hierarchy of note types, which will then be
3749 handleable in a similar fashion. However, at the moment, no such
3750 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3751 but it's an implementation detail, not a classification for the
3753 * Changes in type checking closed the following bugs:
3754 ** type checking of unused values (192b, 194d, 203);
3755 ** template selection based on unsafe type assertions (192c, 236);
3756 ** type checking in branches (194bc).
3757 * A short form of VALUES type specifier has ANSI meaning (it has
3758 increased the number of situations when SBCL cannot perform type
3760 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3761 keywords or constants is permissible.
3762 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3763 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3764 classes. (thanks to Antonio Martinez)
3765 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3766 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3767 argument to a pretty-print stream. Instead, it leaves any such
3768 operations to DESCRIBE-OBJECT methods.
3769 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3771 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3772 lambda lists. (thanks to David Lichteblau)
3773 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3774 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3775 (thanks to Gerd Moellmann)
3776 * fixed bug 246: increased compilation speed of long
3777 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3779 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3781 * DEFCONSTANT now throws a condition of type
3782 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3783 constant to a non-EQL value; CONTINUE and ABORT restarts
3784 respectively change and preserve the value.
3785 * fixed bug 63: The code walker, part of the implementation of CLOS,
3786 is now better at handling symbol macros.
3787 * bug fix: There is no longer an internal implementation type named
3788 CL:LENGTH. (reported by Raymond Toy)
3789 * bug fix: In macro-like defining macros/special operators the
3790 implicit block does not enclose the lambda list.
3791 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3792 enclosed in parentheses) are not suitable as type specifiers, and
3793 their use properly signals an error now.
3794 * bug fix: An argument count mismatch for a type specifier in code
3795 being compiled no longer causes an unhandled error at compile
3796 time, but signals a compile-time warning.
3797 * fixed simple vector readable printing
3798 * bug fix: DESCRIBE takes more care over whether the class
3799 precedence list slot of a class is bound before accessing it.
3800 (reported by Markus Krummenacker)
3801 * bug fix: FORMATTER can successfully compile pretty-printer format
3802 strings which use variants of the ~* directive inside.
3803 * bug fix: SEARCH now applies its TEST predicate to the elements of
3804 the arguments in the correct order. (thanks to Wolfhard Buss)
3805 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3806 in type inference. (thanks to Robert E. Brown)
3807 * bug fix: Reading in symbols with an explicit package name of ""
3808 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3809 * fixed some bugs revealed by Paul Dietz' test suite:
3810 ** NIL is now allowed as a structure slot name.
3811 ** Arbitrary numbers, not just REALs, are allowed in certain
3812 circumstances in LOOP for-as-arithmetic clauses.
3813 ** Multiple class redefinitions before slot access no longer
3814 causes a type error.
3815 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3816 association between the name and a class.
3817 ** Generic functions with non-standard method-combination and over
3818 five methods all of which return constants no longer return NIL
3819 after the first few invocations. (thanks to Gerd Moellmann)
3820 ** CALL-NEXT-METHOD with no arguments now passes the original
3821 values of the arguments, even in the presence of assignment.
3822 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3824 ** DISASSEMBLE works with closures and funcallable instances.
3825 ** ADD-METHOD now returns the generic function, not the new method.
3826 ** FIND-METHOD signals an error if the lengths of the specializers
3827 is incompatible with the generic function, even if the ERRORP
3829 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3830 which its argument is a member.
3831 ** DEFCLASS only redefines the class named by its class-name
3832 argument if that name is the proper name of the class;
3833 otherwise, it creates a new class.
3834 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3835 of the UNBOUND-SLOT condition to the name of the slot.
3836 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3837 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3838 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3839 SLOT-MAKUNBOUND in the specified fashion.
3841 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3842 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3843 cross-compilation host. As a consequence, we can now bootstrap our
3844 way up to SBCL starting with a bare gcc toolchain and human-readable
3845 source code (first the source to CLISP, then the source to SBCL).
3846 * A contributed module containing a partial implementation of the
3847 simple-streams interface has been included. (thanks to Rudi
3849 * A contributed module implementing the RFC1321 Message Digest
3850 Algorithm, known as MD5, has been included.
3851 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3852 no longer has any effect, as the code controlled by this feature
3853 has been deleted. (As far as we know, no-one has ever built using
3854 this feature, and its semantics were confused in any case).
3855 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3856 work on conditions (as required by the ANSI specification),
3857 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3858 expected behaviour on conditions. Users should note, however,
3859 that such behaviour is not required by the ANSI specification,
3860 and so use of this behaviour may render their code unportable.
3861 * fixed some bugs revealed by Paul Dietz' test suite:
3862 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3864 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3865 ** SXHASH on condition objects no longer returns NIL.
3866 ** :ALLOCATION :CLASS slots are better treated; their values are
3867 updated on class redefinition, and initforms inherited from
3868 superclasses are applied.
3869 ** REMOVE-METHOD returns its generic function argument even when
3870 no method was removed.
3871 ** SHARED-INITIALIZE now initializes the values of the requested
3872 slots, including those with :ALLOCATION :CLASS.
3873 ** ALLOCATE-INSTANCE now works on structure classes defined via
3874 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3876 ** SLOT-EXISTS-P now works on conditions, as well as structures
3878 ** MAKE-LOAD-FORM now has the required methods on
3879 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3880 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3881 keyword, and now implements the SLOT-NAMES argument.
3882 ** methods with &OPTIONAL arguments no longer allow too many
3883 arguments to be passed in the call without error.
3884 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3885 option is consistent with the required arguments of the generic
3886 function lambda list.
3887 * bug fix: REQUIRE accepts a string designator. (Thanks to
3889 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3890 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3891 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3892 * bug fix: The compiler now checks for duplicated variables in macro
3894 * bug fix: SETQ on globals returns the correct value.
3895 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3896 not a COMPILER-ERROR (followed by some other strange error on
3897 choosing the CONTINUE restart).
3898 * bug fix: make.sh and friends are now more consistent in the way that
3899 they look for GNU "make".
3901 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3902 * experimental native threads support (on x86 Linux >=2.4 only).
3903 This is not compiled in by default: you need to add :SB-THREAD to
3904 the target features. See the "Beyond ANSI" chapter of the manual
3906 * fix for longstanding nonANSIism: The old distinction between
3907 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3908 The return value from CL:FIND-CLASS is now a CLOS class, and
3909 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3910 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3911 * An interface to the MetaObject Protocol, as described in Kiczales,
3912 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3913 MIT Press, 1991, is available from the SB-MOP package.
3914 * incompatible change: the SB-PCL package should now be considered
3915 a private implementation detail, and no longer a semi-private MOP
3917 * minor incompatible change: due to rearrangement for threads, the
3918 control stack and binding stack are now allocated at arbitrary
3919 addresses instead of being hardcoded per-port. Users affected by
3920 this probably have to be doing advanced things with shared
3921 libraries, and will know who they are.
3922 * minor incompatible change: Previously, all --eval forms used were
3923 processed with READ before any of them were processed with EVAL.
3924 Now each --eval form is processed with both READ and EVAL before
3925 the next --eval form is processed. (Thus package operations like
3926 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3927 work as the user might reasonably expect.)
3928 * minor incompatible change: *STANDARD-INPUT* is now only an
3929 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3931 * minor incompatible change: Y-OR-N-P is now character-oriented,
3932 not line oriented. Also, YES-OR-NO-P now works without errors.
3933 (thanks to Antonio Martinez)
3934 * sb-aclrepl module improvements: an integrated inspector, added
3935 repl features, and a bug fix to :trace command.
3936 * Known functions, which cannot be open coded by the backend, are
3937 considered to be able to check types of their arguments. (fixing
3938 a bug report by Nathan J. Froyd)
3939 * fixed a bug in computing method discriminating functions: It is
3940 now possible to define methods specialized on classes which have
3941 forward-referenced superclasses. (thanks to Gerd Moellmann)
3942 * fixed evaluation order in optional entries (reported by Gilbert
3944 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3945 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3946 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3947 required-by-AMOP NAME argument, as well as CLASS and
3948 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3949 * fixed bug 20: DEFMETHOD can define methods using names that are
3950 not the proper names of classes to designate class specializers.
3951 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3952 Unix file descriptors, instead of blowing up. (thanks to Antonio
3954 * Garbage collection refactoring: user-visible change is that a
3955 call to the GC function during WITHOUT-GCING will not do garbage
3956 collection until the end of the WITHOUT-GCING. If you were doing
3957 this you were probably losing anyway.
3958 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3959 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3960 TYPEP the latter but not the former.
3961 * The compiler issues a full WARNING for calls to undefined functions
3962 with names from the CL package.
3963 * MAP-INTO for a vector destination is open coded. (reported by
3964 Brian Downing on c.l.l)
3965 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3966 documentation string.
3967 * fixed some bugs revealed by Paul Dietz' test suite:
3968 ** COPY-ALIST now signals an error if its argument is a dotted
3970 ** Condition slots are now accessed more correctly in the presence
3971 of multiple initargs for a given slot.
3972 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3973 exclude restarts of the same name associated with a different
3975 ** DEFCLASS of forward-referenced classes with another
3976 forward-referenced class in the superclasses list no longer
3978 ** Condition slots are now initialized once each, not multiple
3979 times. (thanks to Gerd Moellmann)
3980 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3982 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3983 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3984 arguments contain duplicated elements.
3985 ** RESTART-CASE understands local macros.
3986 ** RESTART-CASE associates exactly its own restarts with a condition.
3987 ** ENDP in safe mode checks its argument to be of type LIST.
3988 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3989 in question is unbound.
3990 ** Optimized MAKE-INSTANCE functions no longer cause internal
3991 assertion failures in the presence of duplicate initargs.
3992 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3993 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3994 propagated even in the ctor optimized implementation of
3996 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3998 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3999 * a better implementation of SXHASH on (simple) bit vectors,
4000 measured both in execution speed and in distribution of results
4001 over the positive fixnums, has been installed. Likewise, a better
4002 implementation of EQUAL for simple bit vectors is now available.
4003 * fixed CEILING optimization for a divisor of form 2^k.
4004 * fixed bug 240 (emitting extra style warnings "using the lexical
4005 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
4006 by Antonio Martinez)
4007 * fixed SXHASH, giving different results for NIL depending on type
4008 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
4009 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
4010 variable. (found by Rolf Wester)
4011 * SBCL does not ignore type declarations for special
4012 variables. (reported by rif on c.l.l 2003-03-05)
4013 * some bug fixes in contrib/sb-aclrepl/
4014 * fixed some bugs revealed by Paul Dietz' test suite:
4015 ** a bug in the CONS type specifier, whereby the CAR and CDR
4016 types got intertwined, has been fixed;
4017 ** the type system is now able to reason about the interaction
4018 between INTEGER and RATIO types more completely;
4019 ** APPEND, [N]REVERSE and NRECONC check that those their
4020 arguments, which must be proper lists, are really so;
4021 ** An array specialized to be unable to hold elements has been
4022 implemented, as required -- yes, really -- by ANSI;
4023 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
4024 on malformed property lists;
4026 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
4027 * incompatible packaging change: in line with Unix convention,
4028 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
4029 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
4030 * REQUIRE and PROVIDE are now optionally capable of doing something
4031 useful. See the documentation string for REQUIRE.
4032 * infrastructure for a managed SBCL contrib system: contributed
4033 modules in this release include:
4034 ** the ASDF system definition facility;
4035 ** an interface to the BSD Sockets API;
4036 ** an ACL-like convenience interface to the repl;
4037 (thanks to Kevin Rosenberg)
4038 ** an implementation of ROTATE-BYTE, with efficient implementation
4040 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
4041 gives a closer approximation to the right answer than previously.
4042 (thanks to Raymond Toy)
4043 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
4044 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
4045 optional environment arguments, as required by ANSI.
4046 * fixed bugs in other functions taking environment objects, allowing
4047 calls with an explicit NIL environment argument to be compiled
4049 * fixed bug 228: primary return values from
4050 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
4051 COMPILE or FUNCTION.
4052 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
4053 :TYPE will no longer signal errors on innocuous objects.
4054 * fixed bug 231b: SETQ is better at respecting type declarations in
4055 the lexical environment.
4056 * fixed a bug in DEFCLASS: classes named by symbols with no or
4057 unprintable packages can now be defined.
4058 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
4059 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
4060 * fixed compiler failure related to checking types of functions.
4061 (reported by Robert E. Brown)
4062 * the compiler is now much more consistent in its error-checking
4063 treatment of bounding index arguments to sequence functions: in
4064 (SAFETY 3) code, errors will be signalled in almost all cases if
4065 invalid sequence bounding indices are passed to functions defined
4066 by ANSI to operate on sequences.
4067 * fixed a bug in the build procedure: documentation of SBCL-specific
4068 packages is now preserved and available in the final Lisp image.
4069 * lifted FDEFINITION lookup out of loops in the implementation of
4070 many list operations. (thanks to Robert E. Brown)
4071 * fixed a bug in the reader: the #n# reader macro now works for
4072 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
4073 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
4074 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
4075 E. Brown and Tony Martinez respectively)
4076 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
4077 SLOT-BOUNDP: in particular, it is now able to optimize them much
4078 better, and is now not vulnerable to having packages renamed.
4079 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
4080 always called when a slot is not present in an instance. (thanks
4082 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
4083 longer signal an error on unreading a character following EOF on
4084 the previous constituent stream. (thanks to Tony Martinez)
4085 * fixed some bugs revealed by Paul Dietz' test suite:
4086 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
4087 not just nonnegative fixnums;
4088 ** the logical bit-array operators such as BIT-AND now accept an
4089 explicit NIL for their "opt-arg" argument (to indicate a
4090 freshly-consed result bit-array);
4091 ** ELT now signals an error on an invalid sequence index in safe
4093 ** the type system is now cleverer about negations of numeric
4094 types, and consequently understands the BIGNUM and RATIO types
4096 ** the type system is now cleverer about the interaction between
4097 INTEGER and RATIO types: while bugs still remain, many more
4098 cases are accurately computed;
4099 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
4100 if it is in the last clause;
4101 ** CONSTANTLY now correctly returns a side-effect-free function in
4103 ** DECLARE is no longer treated as a special-operator; in
4104 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
4105 * incremented fasl file version number due to the change in the
4106 DEFSTRUCT-SLOT-DESCRIPTION structure.
4108 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4109 * minor incompatible change: code processed by the "interpreter" or
4110 EVAL now has a compilation optimization policy of (DEBUG 2)
4111 (changed from (DEBUG 1)) to improve debuggability of interactive
4112 development, and to allow the use of the debug RETURN command in
4114 * an experimental implementation of the RETURN command for the
4115 debugger has been included. (thanks to Frederik Kuivinen)
4116 * fixed bug 62: constraints were not propagated into a loop.
4117 * fixed bug in embedded calls of SORT (reported and investigated by
4119 * fixed some bugs revealed by Paul Dietz' test suite:
4120 ** printing and reading of arrays with some dimensions having
4121 length 0 (thanks to Gerd Moellmann);
4122 ** BOA constructor with &AUX argument without a default value does
4123 not cause a type error;
4124 ** CONSTANTP now returns true for all self-evaluating objects.
4126 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4127 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4128 accessors that are related by inheritance, as specified in the
4129 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4130 Valtteri Vuorikoski)
4131 * The compiler is now able to inline functions that were defined in
4132 a complex lexical environment (e.g. inside a MACROLET).
4133 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4135 * Support for the upcoming FreeBSD-5.0 release has been included.
4136 (thanks to Dag-Erling Smorgrav)
4137 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4138 effect when it is not in a toplevel context.
4139 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4140 stand a better chance of being correct. (thanks to Gerd
4142 * fixed bug in COERCE, which now signals an error on coercing a
4143 rational to a bounded real type which excludes the expected
4145 * The compiler is now able to derive types more accurately from the
4146 COERCE and COMPILE functions.
4147 * fixed bug 223: functional binding is considered to be constant
4148 only for symbols in the CL package.
4149 * fixed bug 231: SETQ did not check the type of a variable being set
4150 (reported by Robert E. Brown)
4151 * A new optimization for MAKE-INSTANCE has been included, fixing
4152 various bugs (including relating to :ALLOCATION :CLASS slots and
4153 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4155 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4156 ** As required by ANSI, LOOP now disallows anonymous collection
4157 clauses such as COLLECT I in conjunction with aggregate boolean
4158 clauses such as THEREIS (= I 1);
4159 ** LOOP now signals an error when any variable is reused in the
4160 same loop (including the potentially useful construct analogous
4161 to WITH A = 1 WITH A = (1+ A);
4162 ** IT is only a special loop symbol within the first clause of a
4163 conditional loop clause;
4164 ** LOOP with a typed iteration variable over a hashtable now
4165 signals a type error iff it should.
4166 * fixed some other bugs revealed by Paul Dietz' test suite:
4167 ** FILE-STREAM now names the class previously known as FD-STREAM;
4168 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4169 argument) no longer signals an error;
4170 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4171 of the slot symbol, rather than using the current package
4172 ((:CONC-NAME "") continues to intern the slot's name in the
4174 * incremented fasl file version number, because of the incompatible
4175 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4176 of the new implementation of DEFINE-COMPILER-MACRO.
4178 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4179 * Support for building SBCL for MIPS platforms running in
4180 little-endian mode has now been checked in, and basic
4181 functionality on said platforms verified.
4182 * minor incompatible change: PCL now records the pathname of a file
4183 in which methods and the like are defined, rather than its
4185 * minor incompatible change: TRUENAME now considers the truename of
4186 a file naming a directory to be the pathname with :DIRECTORY
4187 component indicating that directory.
4188 * minor incompatible change: a NAMED clause in the extended form of
4189 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4190 reason for the previous behaviour is unclear.
4191 * more systematization and improvement of CLOS and MOP conformance
4192 in PCL (thanks to Gerd Moellman and Pierre Mai):
4193 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4195 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4196 primary methods with no specializers;
4197 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4199 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4200 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4201 FORWARD-REFERENCED-CLASSes; error reporting on
4202 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4204 ** SXHASH on CLOS instances now uses a slot internal to the
4205 instance to return different numbers on distinct instances,
4206 while preserving the same return value through invocations of
4208 ** DEFMETHOD signals errors when methods with longer incongruent
4209 lambda lists are added to generic functions;
4210 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4211 CLASS, as specified in AMOP;
4212 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4213 on the order returned by the primary method for classes of
4214 class STANDARD-CLASS;
4215 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4216 * fixed some bugs shown by Paul Dietz' test suite:
4217 ** DOLIST puts its body in TAGBODY;
4218 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4220 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4221 value producing form;
4222 ** if more variables are given to PROGV than values, extra
4223 variables are bound and made to have no value;
4224 ** NSUBSTITUTE on list arguments gets the right answer with
4226 ** ELT signals an error of type TYPE-ERROR when the index argument
4227 is not a valid sequence index;
4228 ** LOOP signals (at macroexpansion time) an error of type
4229 PROGRAM-ERROR when duplicate variable names are found;
4230 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4231 ** FUNCALL of special-operators now cause an error of type
4233 ** PSETQ now works as required in the presence of side-effecting
4234 symbol-macro places;
4235 ** NCONC accepts any object as its last argument;
4236 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4238 ** loop-for-as-package does not require a package to be explicitely
4240 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4242 * fixed bug 166: compiler preserves "there is a way to go"
4243 invariant when deleting code.
4244 * fixed bug 172: macro lambda lists with required arguments after
4245 &REST arguments now cause an error to be signalled. (thanks to
4247 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4248 bug. (thanks to Matthew Danish)
4249 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4251 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4252 arguments are lexically rebound. (thanks to Gerd Moellmann and
4254 * fixed bug 194: error messages are now more informative when there
4255 is no primary method applicable in a call to a generic
4256 function. (thanks to Gerd Moellmann)
4257 * fixed bug in command line argument checking (thanks to Julian
4259 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4260 (thanks to Matthew Danish)
4261 * incremented fasl file version number, because of the
4262 SXHASH-related changes in the layout of CLOS data structures
4264 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4265 * minor incompatible change: The runtime (the Unix executable named
4266 "sbcl") is now much pickier about the .core files it will load.
4267 Essentially it now requires .core files to descend from the same
4268 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4269 as the runtime does. (The intent is to prevent the crashes which
4270 can occur, and which can even be reported as mysterious failures,
4271 when people patch the sources or change the build parameters
4272 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4273 sbcl and .core files.)
4274 * fixed bug: VALUES-LIST is no longer optimized away.
4275 * fixed bug 142: The FFI conversion of C string values to Lisp
4276 string values no longer conses excessively. (thanks to Nathan
4277 Froyd porting Raymond Toy's fix to CMU CL)
4278 * began to systematize and improve MOP conformance in PCL (thanks to
4279 Nathan Froyd, Gerd Moellman and Pierre Mai):
4280 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4282 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4283 ** FINALIZE-INHERITANCE is now called on class finalization;
4284 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4285 argument precedence order.
4286 * fixed bug 202: The compiler no longer fails on functions whose
4287 derived types contradict their declared type.
4288 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4289 so it can be non-toplevel.
4290 * The fasl file version number has changed (because of the new
4291 implementation of DEFMACRO).
4292 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4293 safe code, that any length requirement by their type-specifier
4294 argument is valid. The exceptions to this are described in bug
4296 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4297 functions now check the types of their inputs as required by ANSI.
4298 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4299 introduced symbol is DECLAREd to be SPECIAL.
4300 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4301 * fixed bug: :COUNT argument to sequence functions may be negative
4302 * fixed bug: body of DO-SYMBOLS may contain declarations
4303 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4304 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4305 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4306 symbol macro only once
4307 * fixed printing of call frame when argument list is unavailable
4308 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4309 * compiler no longer signals WARNING on unknown keyword
4312 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4313 * A beta-quality port to the mips architecture running Linux,
4314 based on the old CMUCL backend, has been made. It has been tested
4315 on a big-endian kernel, and works sufficiently well to be able to
4316 rebuild itself; it has not been tested in little-endian mode.
4317 * fixed an inconsistency between gencgc.c and purify.c which made
4318 dumping/loading .core files unreliable
4319 * fixed bug 120a: The compiler now deals correctly with IFs where
4320 the consequent is the same as the alternative, instead of
4321 misderiving the return type. (thanks to Alexey Dejneka)
4322 * fixed bug 113: Logical pathnames are now dumpable (the logical
4323 host is resolved at load-time, throwing an error if it is not
4325 * fixed bug 174: FORMAT's error message is slightly clearer when a
4326 non-printing character is used in a format directive.
4327 * fixed several bugs in compiler checking of type declarations, i.e.
4328 violations of the Python "declarations are assertions" principle
4329 (thanks to Alexey Dejneka)
4330 * fixed several bugs in PCL's error checking (thanks to Gerd
4332 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4334 * fixed bug in compilation of functions as first class values
4335 (thanks to Antonio Martinez-Shotton)
4336 * The compiler's handling TYPE-ERRORs which it can prove will
4337 inevitably happen at runtime has been cleaned up and corrected
4338 in several ways. (thanks to Alexey Dejneka)
4339 * improved argument type checking for various basic arithmetic
4340 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4341 lavished on them in the past that they can be compiled in many
4342 ways in different special cases
4343 * fixed bug 181: compiler checks validity of user supplied type
4345 * cleaned up code flushing in optimization: Function calls which
4346 should signal errors for safety purposes (e.g. which ANSI says
4347 should signal errors when their arguments are of incorrect type)
4348 are no longer optimized away.
4349 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4350 * incremented fasl file version number, because changes in the
4351 implementation of sequence functions like COERCE caused
4352 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4353 in old inline expansions) to become undefined. (Actually these
4354 changes were later undone, so we might very well be binary
4355 compatible with 0.7.7 after all, but leaving the version number
4356 incremented seemed like the simplest and most conservative
4359 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4360 * An alpha-quality port to the parisc architecture running Linux,
4361 based on the old CMUCL backend, has been made. This, even more so
4362 than the other backends, should be considered still a work in
4363 progress; known problems include that the Linux kernel in 64-bit
4364 mode does not propagate the correct sigcontext structure to
4365 userspace, and consequently SBCL on a parisc64 kernel will not
4367 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4368 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4369 suppressed.) Also now that the compiler is looking at declarations
4370 in the environment, it checks optimization declarations as well,
4371 and suppresses inlining when (> DEBUG SPEED).
4372 * More fixes have been made to treatment of floating point exception
4373 treatment and other Unix signals. In particular, floating point
4374 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4375 * The detection and handling of control stack exhaustion (infinite
4376 or very deeply nested recursion) has changed. Stack exhaustion
4377 detection is now done by write-protecting pages at the OS level
4378 and applies at all optimization settings; when found, a
4379 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4380 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4381 be caught using IGNORE-ERRORS.
4382 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4383 correctly when transforming recursive LABELS functions to LETs.
4384 (thanks to Alexey Dejneka)
4385 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4386 that are names of constants or global variables.
4387 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4388 alien routines with docstrings.
4389 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4390 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4392 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4393 (thanks to Eric Marsden)
4394 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4395 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4396 * Bug fix: The SPARC backend can now compile functions involving
4397 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4398 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4399 to a functions expecting a PATHNAME-DESIGNATOR.
4400 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4401 lambda lists. (thanks to Alexey Dejneka)
4402 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4403 functions so that EXPORTing the name of the class doesn't cause
4404 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4405 to Antonio Martinez for reporting this)
4406 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4407 constant-folded without failing an assertion. (thanks to Einar
4408 Floystad Dorum for reporting this)
4409 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4410 types are now dealt with more correctly by the compiler.
4411 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4412 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4413 behaviour analogous to RENAME-FILE. This puts its behaviour more
4414 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4416 * The fasl file version number has changed again. (because of the
4417 bug fix involving the names of PCL MAKE-INSTANCE functions)
4419 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4420 * bug fix: Floating point exceptions are treated much more
4421 consistently on the x86/Linux and PPC/Linux platforms.
4422 * Array initialization with :INITIAL-ELEMENT is now much faster for
4423 cases when the compiler cannot open code the array creation, but
4424 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4425 array accesses have also seen a speed increase.
4426 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4427 specified. (This was at the root of some bad interactions between
4428 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4430 * bug fix: Internal error arguments for undefined functions are now
4431 computed correctly on the PPC/Linux platform.
4432 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4433 Raymond Toy's patch for CMU CL)
4434 * Support for the Solaris 9 operating environment has been included
4435 (thanks to Daniel Merritt)
4436 * A very ugly but hopefully complete draft of the missing FFI chapter
4437 of the manual has been created by reformatting the corresponding
4438 CMU CL manual chapter into (currently very ugly and incoherent)
4439 DocBook and bringing it up to date for SBCL behavior. Thus, the
4440 manual is now essentially complete, at least by my extreme
4441 once-and-only-once standards, whereby it's acceptable to refer to
4442 the doc strings of SB-EXT functions as the primary documentation.
4443 * The fasl file version number has changed again, due to cleanup of
4444 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4445 is no longer a static symbol.)
4447 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4448 * SBCL now builds with OpenMCL (version 0.12) as the
4449 cross-compilation host; also, more progress has been made toward
4450 bootstrapping under CLISP.
4451 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4453 * bug 158 fixed: The compiler can now deal with integer loop
4454 increments different from 1; fixing this turned out also to fix
4456 * bug 169 fixed: no more bogus warnings about using lexical bindings
4457 despite the presence of perfectly good SPECIAL declarations (thanks
4458 to David Lichteblau)
4459 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4460 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4461 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4463 * bug 180 fixed: Method combination specifications no longer ignore
4464 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4465 * bug fix: Structure type predicate functions now check their argument
4466 count as they should.
4467 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4468 correctly. (thanks to Pierre Mai)
4469 * minor incompatible change: The --noprogrammer option is deprecated
4470 in favor of the new --disable-debugger option, which is very similar.
4471 (The major difference is that it takes effect at a slightly different
4472 time at startup, causing handling of errors in --sysinit and
4473 --userinit files will be affected differently.) The
4474 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4475 been added to allow this functionality to be controlled from ordinary
4476 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4477 maintainers, who might want to run non-interactive scripts to
4478 build SBCL cores which will later be used interactively.)
4479 * minor incompatible change: The LOAD function no longer, when given
4480 a wild pathname to load, loads all files matching that pathname.
4481 Instead, an error of type FILE-ERROR is signalled.
4483 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4484 * bug 147 fixed: The compiler preserves its block link/count
4485 invariants more correctly now so that it doesn't crash. (thanks
4487 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4489 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4490 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4491 no-value-for-this-character as NIL (as specified by ANSI).
4492 * HOST-NAMESTRING on physical pathnames now returns a string that is
4493 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4494 (thanks to Christophe Rhodes)
4495 * The Alpha port handles icache flushing more correctly. (thanks to
4497 * More progress has been made toward bootstrapping under CLISP. (thanks
4498 to Christophe Rhodes)
4499 * The fasl file format has changed again, because dynamic loading
4500 on OpenBSD (which has non-ELF object files) motivated some cleanups
4501 in the way that foreign symbols are transformed and passed around.
4502 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4503 is no longer treated as whitespace by the reader, but instead as
4504 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4505 |A
\7fB|, instead of A as it used to.
4507 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4508 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4509 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4510 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4511 this as of 0.7.1.45, but was left out of the previous news section
4512 (thanks to Dan Barlow)
4513 * SBCL now runs on the Solaris operating system on SPARC architectures
4514 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4515 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4516 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4517 * SPARC backend cleanups, allowing builds of cores optimized for V8
4518 and V9 SPARCS, and also emission of code targeted to a particular
4519 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4521 * SBCL is closer to bootstrapping under CLISP, thanks to various
4522 fixes by Christophe Rhodes.
4523 * The fasl file format has changed again, to allow the compiler's
4524 INFO database to support symbol macros.
4525 * The user manual (in doc/) is formatted into HTML more nicely.
4526 (thanks to coreythomas)
4527 * The system is smarter about SUBTYPEP relationships, especially
4528 those involving NOT types (including types such as ATOM which are
4529 represented internally using NOT types). Thus SUBTYPEP is less
4530 likely to return (VALUES NIL NIL) in general, and in particular
4531 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4533 * The fasl file format has changed again, because the internal
4534 representation of types now includes a new slot to support the new
4535 SUBTYPEP-of-NOT-types logic.
4536 * (not a change in the main branch of SBCL, but a related prototype
4537 which can hopefully be merged into the main branch of SBCL in the
4538 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4539 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4540 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4541 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4542 dispatch macro character combinations. (thanks to Alexey Dejneka)
4543 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4544 that superficially look like logical namestrings correctly.
4545 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4547 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4548 default supplied-p) for &optional and &key arguments. (thanks to
4551 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4552 * incompatible change: The compiler is now less aggressive about
4553 tail call optimization, doing it only when (> SPACE DEBUG) or
4554 (> SPEED DEBUG). (This is an incompatible change because there are
4555 programs which relied on the old CMU-CL-style behavior to optimize
4556 away their unbounded recursion which will now die of stack overflow.)
4557 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4558 for non-GENCGC systems has been increased to 20M (since that
4559 seems much closer to the likely performance optimum for modern
4560 systems than the old 4M value was)
4561 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4562 and *DEBUG-PRINT-LEVEL*
4563 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4564 of CMU CL's support for SPARC, and various endianness and other
4565 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4566 * new syntactic sugar for the Unix command line: --load foo.bar is now
4567 an alternate notation for --eval '(load "foo.bar")'.
4569 ** The system now detects stack overflow and handles it gracefully,
4570 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4571 optimization settings. (This is a good thing in general, and
4572 its introduction in this version should be particularly timely
4573 for anyone whose code fails because of suppression of tail
4575 ** The system now hunts for the C variable "environ" in a more
4576 devious way, to avoid segfaults when the C library version differs
4577 between compile time and run time. (thanks to Christophe Rhodes)
4578 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4579 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4580 ** The compiler no longer issues bogus style warnings for undefined
4581 classes in the same source file as the DEFCLASSes which defined
4582 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4584 ** fixes in CONDITION class precedence list for undefined function
4585 errors (thanks to Alexei Dejneka)
4586 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4587 correctly. (thanks to Dan Barlow)
4588 ** portability fixes aiming at bootstrapping under CLISP (thanks
4589 to Dave McDonald and Christophe Rhodes)
4590 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4591 ** fixes in type translation and and type inference (thanks to
4593 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4594 ** various fixes in the new ports (thanks to Dan Barlow)
4595 * several changes related to debugging:
4596 ** suppression of tail recursion, as noted above
4597 ** stack overflow detection, as noted above
4598 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4599 is now the default. (For some time encapsulation has been more
4600 reliable than the breakpoint-based :ENCAPSULATE NIL
4601 implementation, at least on X86 systems; and I just noticed that
4602 encapsulation also seems closer to the spirit of the ANSI
4605 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4607 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4608 up properly again. (There was a packaging bug in 0.7.0 which
4609 left their definitions in SB-SYS::LOAD-FOREIGN and
4610 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4611 vital for most things which interface to C-level interfaces,
4612 like extensions working with sockets or databases or
4613 Perl-compatible regexes or whatever, and the need to fix
4614 this bug is the main reason that 0.7.1 was released so
4616 ** DEFGENERIC is now choosier about the methods it redefines, so that
4617 reLOADing a previously-LOADed file containing DEFGENERICs does
4618 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4619 cycle now works as it should. (thanks to Alexey Dejneka)
4620 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4621 fixed by Christophe Rhodes. (He actually submitted this patch
4622 months ago, and I delayed until after 0.7.0.)
4623 ** Bug 111 (internal compiler confusion about runtime checks on
4624 FUNCTION types) was fixed by Alexey Dejneka.
4625 * Some internal cleanups (getting rid of variables which aren't
4626 needed now that the byte interpreter is gone) caused the fasl
4627 file format number to change again.
4629 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4630 * major incompatible change: The default fasl file extension, i.e. the
4631 default extension for files produced by COMPILE-FILE, has changed
4632 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4634 ** There are many changes in the implementation of the compiler.
4635 SBCL is now essentially a compiler-only implementation of ANSI
4636 Common Lisp. EVAL still "interprets" a few special cases, but
4637 almost all the interesting cases are handled by creating
4638 a LAMBDA expression, calling COMPILE on it, then calling
4639 FUNCALL on the result.
4640 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4641 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4642 Since the code is newer, there might still be some new bugs
4643 (though not as many as before Martin Atzmueller's fixes:-). But
4644 the new code is substantially simpler and clearer, and hopefully
4645 any remaining bugs will be simpler, less fundamental, and more
4646 fixable then the bugs in the old code.
4647 ** The revised compiler is still a little unsteady on its feet.
4649 *** The debugging information it produces (particularly the names
4650 of FUNCTION objects) is sometimes much less useful than what
4651 the old compiler produced.
4652 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4653 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4654 within a MACROLET) has been temporarily weakened.
4655 ** There are new compiler optimizations for various functions:
4656 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4657 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4658 *** the math functions TRUNCATE, FLOOR, and CEILING
4659 *** the function-of-all-trades COERCE
4660 Mostly these should be transparent, but there's one
4661 potentially-annoying problem (bug 117): when the compiler
4662 inline-expands a function and does type analysis on the result,
4663 it can create control paths which have type mismatches, and
4664 when it can't prove that those control paths aren't taken,
4665 it will issue WARNINGs about the type mismatches. This is
4666 a particular problem in practice for the new sequence functions.
4667 It's not clear how this should be fixed, and for now, a
4668 workaround is given in the entry for 117 in the BUGS file.
4669 ** (Because of the interaction between the two previous items --
4670 occasional inlining problems and new inline expansions -- some
4671 of the new sequence function optimizations won't really kick in
4672 completely until debugging information, and then inlining, are
4673 straightened out in some future version.)
4674 * minor incompatible changes:
4675 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4676 DIRECTORY no longer implicitly promotes NIL slots of its
4677 pathname argument to :WILD. In particular, when you ask for the
4678 contents of a directory (which you used to be able to do without
4679 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4680 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4681 ** changes in behavior that ANSI explicitly defines to be
4682 implementation dependent:
4683 *** The new compiler-only implementation still conforms with ANSI,
4684 but acts a little different than before. Besides the obvious
4685 changes in performance tradeoffs (that the cost per form passed
4686 to EVAL has gone up, and the cost per form executed by EVAL
4687 has gone down), the behavior of the system changes a little
4688 because there are no longer any interpreted function objects.
4689 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4690 e.g. doing COMPILE on the output of interactive DEFUN is
4692 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4694 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4695 has changed from (:CL) to NIL.
4696 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4697 e.g. #\Newline and #\Tab, don't have names specified in the
4698 ANSI Common Lisp standard, is now based on their ASCII symbolic
4699 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4700 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4701 are no longer used for output.
4702 ** changes in internal implementation constants:
4703 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4704 4 million. (If your application spends a lot of time GCing and
4705 you have a lot of RAM, you might want to experiment with
4706 increasing it even more.)
4707 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4708 However, almost all old code should still continue to work without
4709 immediate update, as SB-C-CALL is now a (deprecated) nickname
4711 ** Old operator names in the style DEF-FOO are now deprecated in
4712 favor of new corresponding names DEFINE-FOO, for consistency with
4713 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4714 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4715 internal symbols, but a few supported extensions like
4716 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4717 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4718 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4719 etc. as you get deeper into recursive calls to the debugger
4720 command loop, instead of the old "5]", "5]]", "5]]]"
4721 sequence. (I was motivated to do this when squabbles between
4722 ILISP and SBCL left me very deeply nested in the debugger. In the
4723 short term, this change will probably provoke more ILISP/SBCL
4724 squabbles, but hopefully it will be an improvement in the long run.)
4725 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4726 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4727 function, since declaiming return types involving aliens is
4728 (1) annoyingly messy to do by hand and (2) vital to efficient
4729 compilation of code which calls such functions.
4730 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4731 longer reexported by the SB-EXT package. They're solely useful
4732 for alien code, so it seems more logical that you should get
4733 them from the SB-ALIEN package, not in SB-EXT.
4734 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4735 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4736 features. Instead, the code that they used to control is always
4737 built into the system.
4738 * many other bug fixes
4739 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4740 advantage of the new EVAL-WHEN stuff and to clean them up in
4741 general, and they are now more ANSI-compliant in a number of
4742 ways. Martin Atzmueller is responsible for a lot of this.
4743 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4745 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4746 *** correct ERROR type for various file operations
4747 *** some fixes for Lisp streams
4748 *** DEFMETHOD syntax checking
4749 *** changing old weird representation of debug information as
4750 strings (which, among their other deficiencies, don't transform
4751 correctly when you rename packages, and don't change their
4752 print representation when you change things like *PACKAGE*
4753 and *PRINT-LENGTH*) to symbols and lists of symbols
4754 He also made several improvements and fixed several bugs in DESCRIBE.
4755 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4757 *** misbehavior of WRITE-STRING/WRITE-LINE
4758 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4759 and several other LOOP problems as well
4760 *** DIRECTORY when similar filenames are present
4761 *** DEFGENERIC with :METHOD options
4762 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4763 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4764 *** argument ordering in FIND with :TEST option
4765 *** mishandled package designator argument in APROPOS-LIST
4766 *** various problems in the backquote readmacro
4767 *** a bug in APROPOS
4768 *** probably some others that I'm not describing very well here,
4769 since the CVS log documents them by reference to sbcl-devel
4770 messages, and the SourceForge archives aren't working well.:-(
4771 ** Dan Barlow improved the Alpha port (and is making progress on the
4772 PPC port, for those of you who think different).
4773 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4774 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4775 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4776 continues to work on the SPARC port (for those of you in a position
4777 to look down upon our little PC-compatible boxes from a great height).
4778 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4779 and uses that copy, rather than the current dynamic value, when
4780 it's trying to decide whether to truncate output. Thus e.g.
4781 (let ((*print-lines* 50))
4782 (pprint-logical-block (stream nil)
4784 (let ((*print-lines* 8))
4785 (print (aref possiblybigthings i) stream)))))
4786 should now truncate the logical block only at 50 lines, instead of
4787 often truncating it at 8 lines, as it did before.
4788 * The doc/cmucl/ directory, containing old CMU CL documentation
4789 from the time of the fork, is no longer part of the base system.
4790 SourceForge has shut down its anonymous FTP service, and with it
4791 my original plan for distributing the old CMU CL documentation
4792 there. For now, if you need these files you can download an old
4793 SBCL source release and extract them from it.
4794 * The fasl file version number changed again, for dozens of reasons,
4795 some of which are apparent above.
4797 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4798 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4799 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4800 patch, so that DISASSEMBLE output is much nicer.
4801 * The code in the SB-PROFILE package now seems reasonably stable.
4802 I still haven't decided what the final interface should look like
4803 (I'd like PROFILE to interact cleanly with TRACE, since both
4804 facilities use function encapsulation) but if you have a need
4805 for profiling now, you can probably use it successfully with
4806 the current CMU-CL-style interface.
4807 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4808 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4809 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4810 link now returns the dangling link itself, and for similar
4811 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4812 link itself. (In these cases the old code signalled an error and
4813 looped endlessly, respectively.) Thus, DIRECTORY now works even
4814 in the presence of dangling and cyclic symbolic links.
4815 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4816 is now a supported extension again, since the consensus on
4817 sbcl-devel was that it can be useful for ordinary development
4818 work, not just for debugging SBCL itself.
4819 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4820 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4821 that currently-defined functions might be redefined later with
4822 different return types.
4823 * Hash tables can be printed readably, as inspired by CMU CL code
4824 of Eric Marsden and SBCL code of Martin Atzmueller.
4825 * better error handling in CLOS method combination, thanks to
4826 Martin Atzmueller porting Pierre Mai's CMU CL patches
4827 * more overflow fixes for >16Mbyte I/O buffers
4828 * A bug in READ has been fixed, so that now a single Ctrl-D
4829 character suffices to cause end-of-file on character streams.
4830 In particular, now you only need one Ctrl-D at the command
4831 line (not two) to exit SBCL.
4832 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4834 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4835 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4836 does the right thing.
4837 * The implementation of some type tests, especially for CONDITION
4838 types, is now tidier and maybe faster, due to CMU CL code
4839 originally by Douglas Crosher, ported by Martin Atzmueller.
4840 * Some math functions have been fixed, and there are new
4841 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4842 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4843 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4844 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4845 likely to be built into sbcl-0.7.0.)
4846 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4847 CL, ported to SBCL by Martin Atzmueller)
4848 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4849 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4850 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4851 * There's a new slam.sh hack to shorten the edit/compile/debug
4852 cycle for low-level changes to SBCL itself, and a new
4853 :SB-AFTER-XC-CORE target feature to control the generation of
4854 the after-xc.core file needed by slam.sh.
4855 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4856 COMPILE-FILE is no longer supported, so that now every function
4857 gets an entry point, so that block compilation looks a little
4858 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4859 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4860 returns the number of bytes consed since the system started,
4861 rather than the number consed since the first time the function
4862 was called. (The new definition parallels ANSI functions like
4863 CL:GET-INTERNAL-RUN-TIME.)
4864 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4865 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4866 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4867 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4868 * Fasl file version numbers are now independent of the target CPU,
4869 since historically most system changes which required version
4870 number changes have affected all CPUs equally. Similarly,
4871 the byte fasl file version is now equal to the ordinary
4874 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4875 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4876 is no longer recognized. I apologize for this, because it was
4877 listed in SB-EXT as a supported extension, but I found that
4878 its existing behavior was poorly specified, as well as incorrectly
4879 specified, and it looked like too much of a mess to straighten it
4880 out. I have enough on my hands trying to get ANSI stuff to work..
4881 * many patches ported from CMU CL by Martin Atzmueller, with
4882 half a dozen bug fixes in pretty-printing and the debugger, and
4883 half a dozen others elsewhere
4884 * fixed bug 13: Floating point infinities are now supported again.
4885 They might still be a little bit flaky, but thanks to bug reports
4886 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4887 as flaky as they were.
4888 * The --noprogrammer command line option is now supported. (Its
4889 behavior is slightly different in detail from what the old man
4890 page claimed it would do, but it's still appropriate under the
4891 same circumstances that the man page talks about.)
4892 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4893 are now supported, and enabled by default. Thus, the compiler can
4894 handle many floating point and complex operations much less
4895 inefficiently. (Thus e.g. you can implement a complex FFT
4897 * The compiler now detects type mismatches between DECLAIM FTYPE
4898 and DEFUN better, and implements CHECK-TYPE more correctly, and
4899 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4900 thanks to patches from Martin Atzmueller.
4901 * various fixes to make the cross-compiler more portable to
4902 ANSI-conforming-but-different cross-compilation hosts (notably
4903 Lispworks for Windows, following bug reports from Arthur Lemmens)
4904 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4905 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4906 fixed thanks to Pierre Mai's CMU CL patches.
4907 * fixes to keep the system from overflowing internal counters when
4908 it tries to use i/o buffers larger than 16M bytes
4909 * fixed bug 45a: Various internal functions required to support
4910 complex special functions have been merged from CMU CL sources.
4911 (When I was first setting up SBCL, I misunderstood a compile-time
4912 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4913 * improved support for type intersection and union, fixing bug 12
4914 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4915 more obscure bugs as well
4916 * some steps toward byte-compiling non-performance-critical
4917 parts of the system, courtesy of patches from Martin Atzmueller
4918 * Christophe Rhodes has made some debian packages of sbcl at
4919 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4920 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4921 stable, but are nonetheless usable. When he's ready, I'd be happy
4922 to add them to the SourceForge "File Releases" section. (And if
4923 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4924 * new fasl file format version number (because of changes in
4925 internal representation of (OR ..) types to accommodate the new
4926 support for (AND ..) types, among other things)
4928 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4929 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4931 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4933 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4934 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4935 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4936 * DESCRIBE now works on structure objects again.
4937 * Most function call argument type mismatches are now handled as
4938 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4939 know whether the function will be redefined before the call is
4940 executed. (The compiler could flag local calls with full WARNINGs,
4941 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4942 it doesn't keep track of enough information to know whether calls
4943 are local in this sense.)
4944 * Compiler output is now more verbose, with messages truncated
4945 later than before. (There should be some supported way for users
4946 to override the default verbosity, but I haven't decided how to
4947 provide it yet, so this behavior is still controlled by the internal
4948 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4949 src/compiler/ir1util.lisp.)
4950 * Fasl file format version numbers have increased again, because
4951 support for the Gray streams extension changes the layout of the
4952 system's STREAM objects.
4953 * The Gray subclassable streams extension now works, thanks to a
4954 patch from Martin Atzmueller.
4955 * The full LOAD-FOREIGN extension (not just the primitive
4956 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4957 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4958 but like most other programs, it defaults to copying the Unix
4959 environment from the original process instead of starting the
4960 new process in an empty environment.
4961 * Extensions which manipulate the Unix environment now support
4962 an :ENVIRONMENT keyword option which doesn't smash case or
4963 do other bad things. The CMU-CL-style :ENV option is retained
4964 for porting convenience.
4965 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4966 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4968 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4970 * A patch from Martin Atzmueller seems to have solved the SIGINT
4971 problem, and as far as we know, signal-handling now works cleanly.
4972 (If you find any new bugs, please report them!)
4973 * The system no longer defaults Lisp source file names to types
4974 ".l", ".cl", or ".lsp", but only to ".lisp".
4975 * The compiler no longer uses special default file extensions for
4976 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4977 seems to expect a single default extension for all compiled code,
4978 and there's no compelling reason to try to stretch the standard
4979 to allow two different extensions.) Instead, byte-compiled files
4980 default to the same extension as native-compiled files.
4981 * Fasl file format version numbers have increased again, because
4982 a rearrangement of internal implementation packages made some
4983 dumped symbols in old fasl files unreadable in new cores.
4984 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4985 many fewer weird special cases.
4986 * Bug #17 (differing COMPILE-FILE behavior between logical and
4987 physical pathnames) has been fixed, and some related misbehavior too,
4988 thanks to a patch from Martin Atzmueller.
4989 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4990 by Tim Moore, ported to SBCL by Martin Atzmueller.
4991 * Martin Atzmueller fixed several filesystem-related problems,
4992 including bug #36, in part by porting CMU CL patches, which were
4993 written in part by Paul Werkowski.
4994 * More compiler warnings in src/runtime/ are gone, thanks to
4995 more patches from Martin Atzmueller.
4996 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4999 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
5001 * DESCRIBE now works on CONDITION objects.
5002 * The debugger now handles errors which arise when trying to print
5003 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
5005 * The build system now uses an additional file, customize-target-features.lisp,
5006 to allow local modifications to the target *FEATURES* list. (The point of
5007 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
5008 debugging features enabled, without having to worry about propagating it
5009 into everyone's system when I do a "cvs update".) When no
5010 customize-target-features.lisp file exists, the target *FEATURES* list
5011 should be constructed the same way as before.
5012 * fixed bugs in DEFCONSTANT ANSI-compatibility:
5013 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
5014 warn about behavior which is undefined under the ANSI spec. Note: This
5015 is specified by ANSI, but it's not very popular with programmers.
5016 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
5017 macro in the SBCL sources for an example of a workaround which you
5018 might use to make portable ANSI-standard code which does what you want.
5019 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
5020 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
5021 when it's used as a non-toplevel form. (This is required in order
5022 to implement the DEFCONSTANT-EQX macro.)
5023 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
5024 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
5025 funny things, due to bugs in the implementation of EVAL-WHEN
5026 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
5027 non-toplevel DEFCONSTANTs are uncommon.)
5028 * The core file version number and fasl file version number have been
5029 incremented, because the old noncompliant DEFCONSTANT behavior involved
5030 calling functions which no longer exist, and because I also took the
5031 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
5032 * fixed bug 1 (error handling before read-eval-print loop starts), and
5033 redid debugger restarts and related debugger commands somewhat while
5035 ** The QUIT debugger command is gone, since it did something
5036 rather different than the SB-EXT:QUIT command, and since it never
5037 worked properly outside the main toplevel read/eval/print loop.
5038 Invoking the new TOPLEVEL restart provides the same functionality.
5039 ** The GO debugger command is also gone, since you can just invoke
5040 the CONTINUE restart directly instead.
5041 ** The TOP debugger command is also gone, since it's redundant with the
5042 FRAME 0 command, and since it interfered with abbreviations for the
5044 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
5046 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
5047 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
5048 some porting work by Martin Atzmueller)
5049 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
5050 with a boatload of patches which he ported to SBCL. Now that those
5052 ** The system tries to make sure that its low-priority messages
5053 are prefixed by semicolons, to help people who like to use
5054 syntax highlighting in their ILISP buffer. (This patch
5055 was originally due to Raymond Toy.)
5056 ** The system now optimizes INTEGER-LENGTH better, thanks to more
5057 patches originally written by Raymond Toy.
5058 ** The compiler understands coercion between single-value and
5059 multiple-VALUES type expressions better, getting rid of some very
5060 weird behavior, thanks to patches originally by Robert MacLachlan
5061 and Douglas Crosher.
5062 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
5063 lambda lists, thanks to a patch originally by Pierre Mai.
5064 ** The system no longer bogusly warns about "abbreviated type
5066 ** The compiler gets less confused by inlining and RETURN-FROM,
5067 thanks to some patches originally by Tim Moore.
5068 ** The system no longer hangs when dumping circular lists to fasl
5069 files, thanks to a patch originally from Douglas Crosher.
5070 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
5071 undefined function error.
5072 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
5073 in structure slot initforms) for now, documented workaround instead:-|
5074 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
5075 * fixed bug 5: added stubs for various Gray stream functions called
5076 in the not-a-CL:STREAM case, so that even when Gray streams aren't
5077 installed, at least appropriate type errors are generated
5078 * fixed bug 8: better reporting of various PROGRAM-ERRORs
5079 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
5080 consistently in DEFMETHOD forms.
5081 * removed bug 21 from BUGS, since Martin Atzmueller points out that
5082 it doesn't seem to affect SBCL after all
5083 * The C runtime system now builds with better optimization and many
5084 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
5086 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
5088 * The system is now under CVS at SourceForge (instead of the
5089 CVS repository on my home machine).
5090 * The new signal handling code has been tweaked to treat register
5091 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
5092 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
5093 downstream, e.g. in debug-int.lisp, has implicit dependencies
5094 on the unsignedness of integer representation of machine words,
5095 and that caused the system to bomb out with infinite regress
5096 when trying to recover from type errors involving signed values,
5097 e.g. (BUTLAST '(1 2 3) -1).)
5098 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
5099 (This was one of the bugs Peter Van Eynde reported back in July.)
5100 * The system now uses code inspired by Colin Walters' O(N)
5101 implementation of MAP (from the cmucl-imp@cons.org mailing
5102 list, 2 September 2000) when it can't use a DEFTRANSFORM to
5103 inline the MAP operation, and there is more than one
5104 sequence argument to the MAP call (so that it can't just
5105 do ETYPECASE once and for all based on the type of the
5106 single sequence argument). (The old non-inline implementation
5107 of the general M-argument sequence-of-length-N case required
5108 O(M*N*N) time when any of the sequence arguments were LISTs.)
5109 * The QUIT :UNIX-CODE keyword argument has been renamed to
5110 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5111 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5113 * Raymond Wiker's patches to port dynamic loading from Linux to
5114 FreeBSD have been added.
5115 * The BUGS file is now more nearly up to date, thanks in large part
5116 to Martin Atzmueller's review of it.
5117 * The debugger now flushes standard output streams before it begins
5118 its output ("debugger invoked" and so forth).
5119 * The core version number and fasl file version number have both
5120 been incremented, because of incompatible changes in the layout
5122 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5123 used to not be called for a saved Lisp image.)
5124 * Martin Atzmueller's version of a patch to fix a compiler crash,
5125 as posted on sbcl-devel 13 September 2000, has been installed.
5126 * Instead of installing Martin Atzmueller's patch for the
5127 compiler transform for SUBSEQ, I deleted the compiler transform,
5128 and transforms for some similar consing operations.
5129 * A bug in signal handling which kept TRACE from working on OpenBSD
5131 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5132 away by constant folding
5133 * The system now defines its address space constants in one place
5134 (in the Lisp sources), and propagates them automatically elsewhere
5135 (through GENESIS and the sbcl.h file). Therefore, patching the
5136 address map is less unnecessarily tedious and error-prone. The
5137 Lisp names of address space constants have also been systematized.
5138 * CVS tags like dollar-Header-dollar have been removed from
5139 the sources, because they have never saved me trouble and
5140 they've been source of trouble working with patches and other
5141 diff-related operations.
5142 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5143 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5145 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5147 * The system has been ported to OpenBSD.
5148 * The system now compiles with a simple "sh make.sh" on the systems
5149 that it's supported on. I.e., now you no longer need to tweak
5150 text in the target-features.lisp-expr and symlinks in src/runtime/
5151 by hand, the make.sh takes care of it for you.
5152 * The system is no longer so grossly inefficient when compiling code
5153 involving vectors implemented as general (not simple) vectors (VECTOR T),
5154 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5155 which dares to use the various sequence functions on non-simple
5156 vectors, takes less of a performance hit.
5157 * There is now a primitive type predicate VECTOR-T-P
5158 to test for the (VECTOR T) type, so that e.g.
5159 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5160 can now be compiled with some semblance of efficiency. (The old code
5161 turned the type declaration into a full call to %TYPEP at runtime!)
5162 * AREF on (VECTOR T) is still not fast, since it's still compiled
5163 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5164 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5165 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5166 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5167 without making full calls to SUBTYPEP at runtime.
5168 (Some analogous efficiency issues for non-simple vectors specialized to
5169 element types other than T, or for non-simple multidimensional arrays,
5170 have not been addressed. They could almost certainly be handled the
5171 same way if anyone is motivated to do so.)
5172 * The changes in array handling break binary compatibility, so
5173 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5174 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5175 instead of (VALUES T T).
5176 * By following the instructions that Dan Barlow posted to sbcl-devel
5177 on 2 July 2000, I was able to enable primitive dynamic object
5178 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5179 functionality is not implemented (since it calls ld to resolve
5180 library references automatically, requiring RUN-PROGRAM for its
5181 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5182 try to resolve library references) is now supported.
5183 * The system now flushes the standard output streams when it terminates,
5184 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5185 them at several other probably-convenient times, e.g. in each pass of
5186 the toplevel read-eval-print loop, and after evaluating a form given
5187 as an "--eval" command-line option. (These changes were motivated by a
5188 discussion of stream flushing issues on cmucl-imp in August 2000.)
5189 * The source transform for TYPEP of array types no longer assumes
5190 that an array whose element type is a not-yet-defined type
5191 is implemented as an array of T, but instead punts, so that the
5192 type will be interpreted at runtime.
5193 * There is now some support for cross-compiling in make.sh: each of
5194 the phases of make.sh has its own script. (This should be transparent
5195 to people doing ordinary, non-cross-compile builds.)
5196 * Since my laptop doesn't have hundreds of megabytes of memory like
5197 my desktop machine, I became more motivated to do some items on
5198 my to-do list in order to reduce the size of the system a little:
5199 ** Arrange for various needed-only-at-cold-init things to be
5200 uninterned after cold init. To support this, those things have
5201 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5202 symbols with such names are now uninterned after cold init).
5203 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5204 fasl files for cold load.
5205 ** Remove the old compiler structure pooling code (which used to
5206 be conditional on the target feature :SB-ALLOC) completely.
5207 ** Redo the representation of some data in cold init to be more compact.
5208 (I also looked into supporting byte compiled code at bootstrap time,
5209 which would probably reduce the size of the system a lot, but that
5210 looked too complicated, so I punted for now.)
5211 * The maximum signal nesting depth in the src/runtime/ support code has
5212 been reduced from 4096 to 256. (I don't know any reason for the very
5213 large old value. If the new smaller value turns out to break something,
5214 I'll probably just bump it back up.)
5215 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5217 * Many, many bugs reported by Peter Van Eynde have been added to
5218 the BUGS list; some have even been fixed.
5219 * While enabling dynamic object file loading, I tried to make the
5220 code easier to understand, renaming various functions and variables
5221 with less ambiguous names, and changing some function calling
5222 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5223 * While trying to figure out how to do the OpenBSD port, I tried to
5224 clean up some of the code in src/runtime/. In particular, I dropped
5225 support for non-POSIX signal handling, added various comments,
5226 tweaked the code to reduce the number of compilation warnings, and
5227 renamed some files to increase consistency.
5228 * To support the new automatic configuration functionality in make.sh,
5229 the source file target-features.lisp-expr has been replaced with the
5230 source file base-target-features.lisp-expr and the machine-generated
5231 file local-target-features.lisp-expr.
5232 * fixed a stupid quoting error in make.sh so that using CMU CL
5233 "lisp -batch" as cross-compilation host works again
5235 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5237 * DESCRIBE no longer tries to call itself recursively to describe
5238 bound/fbound values, so that it no longer fails on symbols which are
5239 bound to themselves (like keywords, T, and NIL).
5240 * DESCRIBE now works on generic functions.
5241 * The printer now prints less-screwed-up representations of closures
5242 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5243 * A private symbol is used instead of the :EMPTY keyword previously
5244 used to mark empty slots in hash tables. Thus
5245 (DEFVAR *HT* (MAKE-HASH-TABLE))
5246 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5247 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5248 now does what ANSI says that it should. (You can still get
5249 similar noncompliant behavior if bang on the hash table
5250 implementation with all the symbols you get back from
5251 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5252 This breaks binary compatibility, since tests for equality to
5253 :EMPTY are wired into things like the macroexpansion of
5254 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5256 * There's now a minimal placeholder implementation for CL:STEP,
5257 as required by ANSI.
5258 * An obscure bug in the interaction of the normal compiler, the byte
5259 compiler, inlining, and structure predicates has been patched
5260 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5261 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5263 * Missing ordinary arguments in a macro call are now detected even
5264 when the macro lambda list contains &KEY or &REST.
5265 * The debugger no longer complains about encountering the top of the
5266 stack when you type "FRAME 0" to explicitly instruct it to go to
5267 the top of the stack. And it now prints the frame you request even
5268 if it's the current frame (instead of saying "You are here.").
5269 * As specified by ANSI, the system now always prints keywords
5270 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5272 * The default initial SIZE of HASH-TABLEs is now smaller.
5273 * Type information from CLOS class dispatch is now propagated
5274 into DEFMETHOD bodies, so that e.g.
5275 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5277 is now basically equivalent to
5278 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5279 (DECLARE (TYPE SINGLE-FLOAT X))
5281 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5282 operation, without having to do run-time type dispatch.
5283 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5284 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5285 * Since I don't seem to be making big file reorganizations very often
5286 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5287 is overflowing my ability to conveniently back them up), I've finally
5288 checked the system into CVS. (The CVS repository is on my home system,
5289 not at SourceForge -- putting it on SourceForge might come later.)
5290 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5291 high-level GC-NOTIFY-FOO functions send their output. (There's
5292 still very little control of where low-level verbose GC functions
5293 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5294 less than it used to -- the GC-NOTIFY-FOO functions are now under
5295 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5296 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5297 in only one place in the source code, and propagates it automatically
5298 everywhere that it's needed. Thus e.g. when I bump the version from
5299 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5300 * The C source files now include boilerplate legalese and documentation
5301 at the head of each file (just as the Lisp source files already did).
5302 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5303 to his page will be replaced with a link to his new CLiki service.
5305 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5307 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5308 * The build process now looks for GNU make under the default name "gmake",
5309 instead of "make" as it used to. If GNU make is not available as "gmake"
5310 on your system, you can change this default behavior by setting the
5311 GNUMAKE environment variable.
5312 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5313 can build without error under CMU CL.
5315 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5317 * There is now a partial SBCL user manual (with some new text and some
5318 text cribbed from the CMU CL manual).
5319 * The beginnings of a profiler have been added (starting with the
5320 CMU CL profiler and simplifying and cleaning up). Eventually the
5321 main interface should be through the TRACE macro, but for now,
5322 it's still accessed through vaguely CMU-CL-style functions and macros
5323 exported from the package SB-PROFILE.
5324 * Some problems left over from porting CMU CL to the new
5325 cross-compilation bootstrap process have been cleaned up:
5326 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5327 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5328 part when getting rid of the compiler *BACKEND* stuff.)
5329 ** The value of *NULL-TYPE* was screwed up, because it was
5330 being initialized before the type system knew the final
5331 definition of the 'NULL type. This screwed up several key
5332 optimizations in the compiler, causing inefficiency in all sorts
5333 of places. (I found it because I wanted to understand why
5334 GET-INTERNAL-RUN-TIME was consing.)
5335 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5336 PROCLAIM FTYPE information. Unfortunately this broke binary
5337 compatibility again, since now the forms output by DEFGENERIC
5338 to refer to functions which didn't exist in 0.6.3.
5339 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5340 can use the new (as of 0.6.3) transform for SOME into MAP into
5342 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5343 values to (INTEGER 0 1000000), so that the time code will no longer
5344 occasionally get blown up by Linux returning 1000000 microseconds
5345 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5346 its output conform to the ANSI spec. (Alas, this makes its output
5347 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5348 * A full call to MAP NIL with a single sequence argument no longer conses.
5349 * fixes to problems pointed out by Martin Atzmueller:
5350 * The manual page no longer talks about multiprocessing as though
5351 it were currently supported.
5352 * The ILISP support patches have been removed from the distribution,
5353 because as of version 5.10.1, ILISP now supports SBCL without us
5354 having to maintain patches.
5355 * added a modified version of Raymond Toy's recent CMU CL patch for
5356 EQUALP comparison of HASH-TABLE
5358 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5360 * The system still can't cross-compile itself with
5361 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5362 entails), but at least it can compile more of itself that way
5363 than it used to be able to, and various buglets which were uncovered
5364 by trying to cross-compile itself that way have now been fixed.
5365 * This release breaks binary compatibility again. This time
5366 at least I've incremented the FASL file format version to 2, so that the
5367 problem can be detected reliably instead of just causing weird errors.
5368 * various new style warnings:
5369 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5370 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5371 ** using the lexical binding of a variable named in the *FOO* style
5372 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5373 as specified by ANSI.
5374 * *RANDOM-STATE* is no longer automatically initialized from
5375 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5376 default behavior of the system is to repeat its behavior every time
5377 it's run. If you'd like to change this behavior, you can always
5378 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5379 old convention there was no comparably easy way to get the system to
5380 repeat its behavior every time it was run.
5381 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5382 FTYPE declarations has been removed, in favor of their ANSI
5383 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5384 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5385 the types of their sequence arguments can be determined at compile time.
5386 This is done through a new open code expansion for MAP which eliminates
5387 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5388 argument types can be determined at compile time.
5389 * The optimizer now transforms COERCE into an identity operation when it
5390 can prove that the coerced object is already of the correct type. (This
5391 can be a win for machine generated code, including the output of other
5392 optimization transforms, such as the MAP transform above.)
5393 * Credit information has been moved from source file headers into CREDITS.
5394 * Source file headers have been made more standard.
5395 * The CASE macro now compiles without complaining even when it has
5398 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5400 * (Note that the way that the PCL macroexpansions were rewritten
5401 to accommodate the change in DEFGENERIC below breaks binary
5402 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5403 not run under sbcl-0.6.2. Once we get out of alpha releases,
5404 i.e. hit release 1.0.0, we'll probably try to maintain binary
5405 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5406 and sbcl-1.4.4. Until then, however, it might be fairly common
5407 for maintenance releases to break binary compatibility.)
5408 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5410 * The reporting of unhandled signals has been changed to print some
5411 explanatory text as well as the report form. (Previously only
5412 the report form was printed.)
5413 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5414 it defines, so that the compiler no longer issues undefined function
5415 warnings for compiled-but-not-yet-loaded generic functions.
5416 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5417 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5418 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5419 section of the ANSI spec are supported.
5420 * The "" nickname for the "KEYWORD" package has been removed.
5421 The reader still handles symbol tokens which begin with a package marker
5422 as keywords, but it doesn't expose its mechanism for doing so in the
5423 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5424 * The system now issues STYLE-WARNINGs for contradictory TYPE
5425 proclamations. (Warnings for contradictory FTYPE proclamations would
5426 be nice too, but those can't be done usefully unless the type system
5427 is made smarter about FUNCTION types.)
5428 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5429 systematized, so that "*target-*.lisp is supposed to exist only on the
5430 target and imply that there's a related file which exists on the
5431 host, and *host-*.lisp is supposed to exist only on the host and imply
5432 that there's a related file which exists on the target. This involves a
5433 lot of renaming. Hopefully the acute confusion caused by the renaming
5434 will be justified by the reduction in chronic confusion..
5435 ** runtime-type.lisp -> early-target-type.lisp
5436 ** target-type.lisp -> late-target-type.lisp
5437 ** early-host-format.lisp -> early-format.lisp
5438 ** late-host-format.lisp -> late-format.lisp
5439 ** host-error.lisp -> misc-error.lisp
5440 ** early-error.lisp -> early-target-error.lisp
5441 ** late-error.lisp -> late-target-error.lisp
5442 ** host-defboot.lisp -> early-defboot.lisp
5443 ** code/misc.lisp -> code/target-misc.lisp
5444 ** code/host-misc.lisp -> code/misc.lisp
5445 ** code/numbers.lisp -> code/target-numbers.lisp
5446 ** code/early-numbers.lisp -> numbers.lisp
5447 ** early-host-type.lisp -> early-type.lisp
5448 ** late-host-type.lisp -> late-type.lisp
5449 ** host-typep.lisp -> typep.lisp
5450 ** load.lisp -> target-load.lisp
5451 ** host-load.lisp -> load.lisp
5452 ** host-disassem.lisp -> disassem.lisp
5453 ** host-insts.lisp -> insts.lisp
5454 ** byte-comp.lisp -> target-byte-comp.lisp
5455 ** host-byte-comp.lisp -> byte-comp.lisp
5456 ** host-signal.lisp -> signal.lisp
5457 ** host-defstruct.lisp -> defstruct.lisp
5458 ** late-target-type.lisp -> deftypes-for-target.lisp
5459 Furthermore, several other previously target-only files foo.lisp (e.g.
5460 hash-table.lisp and random.lisp) have been split into a target-and-host
5461 foo.lisp file and a target-only target-foo.lisp file, with their key type
5462 definitions in the target-and-host part, so that the cross-compiler will
5463 know more about target types.
5464 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5465 gone away. In their place are various *BACKEND-FOO* variables
5466 corresponding to the slots of the old structure.
5467 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5468 into the target SBCL has been fixed.
5469 * The chill.lisp system for loading cold code into a running SBCL
5471 * Support for the CMU CL "scavenger hook" extension has been removed.
5472 (It was undocumented and unused in the CMU CL sources that SBCL was
5473 derived from, and stale in sbcl-0.6.1.)
5474 * Various errors in the cross-compiler type system were detected
5475 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5476 (enabling various consistency checks). Many of them were fixed,
5477 but some hard problems remain, so the compiler is back to
5478 running without *TYPE-SYSTEM-INITIALIZED* for now.
5479 * As part of the cross-compiler type system cleanup, I implemented
5480 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5481 * I have started adding UNCROSS calls throughout the type system
5482 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5483 on cross-compiler input files will be able to go away, and various
5485 * CONSTANTP now returns true for quoted forms (as explicitly required
5488 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5490 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5491 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5492 (MAKE-LIST -1). In the longer term, it ought to become true
5493 that declarations are assertions even at SAFETY 1. For now, it's not
5494 quite true even at SAFETY 3, but it's at least more nearly true..
5495 (Note that this change seems to increases the size of the system by
5496 O(5%) and to decrease the speed of the compiler by 20% or more.)
5497 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5498 for the problem of printing dozens of lines of distracting information
5499 about low-level system machinery as part of the top stack frame
5500 on entry to the debugger when an undefined function was called.
5501 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5502 is not reset to COMMON-LISP-USER.
5503 * Compilation of stuff related to dyncount.lisp has been made conditional
5504 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5505 smaller. The various dyncount-related symbols have been moved into
5506 a new "SB-DYNCOUNT" package.
5507 * tty-inspect.lisp has been renamed to inspect.lisp.
5508 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5509 feature has gone away. (When we eventually port to other flavors of
5510 libc and/or Unix, we'll try to make the differences between flavors
5511 invisible at the user level.)
5512 * Various other *FEATURES* tags, and/or their associated conditionals,
5513 have been removed if obsolescent, or given better documentation, or
5514 sometimes given more-mnemonic names.
5516 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5518 * tidied up "make.sh" script
5519 * tidied up system directory structure
5520 * better "clean.sh" behavior
5521 * added doc/FOR-CMUCL-DEVELOPERS
5522 * many many small tweaks to output format, e.g. removing possibly-confusing
5523 trailing #\. character in DESCRIBE-INSTANCE
5524 * (EQUALP #\A 'A) no longer signals an error.
5525 * new hashing code, including EQUALP hashing
5526 * tidied up Lisp initialization and toplevel
5527 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5528 * command line argument processing
5529 * added POSIX-GETENV function to deal with Unix-ish environment variables
5530 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5531 terminating SBCL on EOF
5532 * non-verbose GC by default
5533 * There is no more "sbcl" shell script; the sbcl file is now the C
5534 runtime executable (just like CMU CL).
5535 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5537 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5538 symbol and package information as Lisp native symbol and package objects
5539 instead of strings naming symbols and strings naming packages. This way,
5540 whenever packages are renamed (as in warm init), debug information is
5541 transformed along with everything else.
5542 * tweaked the optimization policy declarations which control the building
5543 of SBCL itself. Now, among other things, the system no longer saves
5544 source location debugging information. (This helps two problems at once
5545 by reducing SBCL size and by keeping SBCL from trying to look for its
5546 sources -- which may not exist -- when reporting errors.)
5547 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5548 debugging and testing purposes
5549 * cleaned up printing, making the printer call PRINT-OBJECT for
5550 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5551 methods, giving nearly-ANSI behavior
5552 * converted almost all special variables to use *FOO* naming convention
5553 * deleted PARSE-TIME functionality, since it can be done portably
5554 * moved some files out of cold init into warm init
5555 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5557 * regularized formatting of source files
5558 * added an install.sh script
5559 * fixed ridiculous memory usage of cross-compiler by making
5560 compiler/alloc.lisp not try to do pooling unless it can hook
5561 itself into the GC of the cross-compilation host. Now the system
5562 builds nicely on my old laptop.
5563 * added :SB-ALLOC in target-features.lisp-expr
5564 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5565 was not implemented)
5566 * re-did condition handling and note reporting in the compiler. Notes
5567 are no longer handled by signalling conditions. Style warnings
5568 and warnings are handled more correctly and reported in such a way
5569 that it's easy to find one or the other in your output (so that you
5570 can e.g. figure out which of many problems caused COMPILE-FILE to
5572 * changed the severity of several compiler warnings from full WARNING
5573 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5574 compiler note reporting so that it doesn't use the condition system
5575 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5576 COMPILE-FILE command)
5577 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5578 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5579 no longer supported, since their implementation was deeply intertwingled
5580 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5581 * removed bogus "support" for compiler macros named (SETF FOO), and
5582 removed the compiler macro for SETF INFO (but only after making a fool
5583 of myself on the cmucl-imp mailing list by posting a bogus patch for
5584 DEFINE-COMPILER-MACRO..)
5585 * Compiled files containing forms which have side effects on the Lisp
5586 reader (such as DEFPACKAGE forms) are now handled more correctly.
5587 (Compiler queuing of top level lambdas has been suppressed by setting
5588 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5589 * deleted various currently-unused source files, e.g. gengc.lisp. They
5590 may be added back at some point e.g. when porting to other architectures,
5591 but until they are it's distracting to distribute them and to try to
5593 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5594 there were so many of them they're just distractions, and UNCROSS is
5595 known to be able to handle the current sources
5596 * moved PROFILE functionality into TRACE, so that it will be clear
5597 how the wrapping and unwrapping of functions when you profile them
5598 interacts with the wrapping and unwrapping of functions when you
5599 trace them. (Actually, the functionality isn't there yet, but at least
5600 the interface specification is there. Hopefully, the functionality will
5601 arrive with some future maintenance release.)
5602 * removed host-oops.lisp
5603 * changed signature of QUIT function to allow UNIX-CODE argument
5604 * fixed READ-SEQUENCE bug
5605 * tweaked verbose GC output so that it looks more like the progress
5606 output that ANSI specifies for functions like LOAD
5607 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5608 * added <http://sbcl.sourceforge.com> to the banner information printed by
projects / sbcl.git / blob