1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.33:
3 * enhancement: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST now also works on most
5 * enhancement: Errors during compile-time-too processing (i.e. EVAL-WHEN)
6 are now caught and reported just like errors during macroexpansion.
7 * fixes and improvements related to Unicode and external formats:
8 ** bug fix: error handling and restart usage in the ucs-2 external format
10 ** there is now an implementation of the ucs-4 external format.
11 ** the utf-16 and utf-32 external formats are supported.
12 * bug fix: #p"\\\\" can now be read without error on Win32. (reported by
13 Willem Broekema; launchpad bug lp#489698).
14 * bug fix: some minor code rearrangements to reenable warning-free building
15 from CMUCL (reported by xme@gmx.net; launchpad bug lp#491104)
16 * bug fix: PRINT-OBJECT for clos instances respects the right margin when
18 * bug fix: FIND-PACKAGE & DEFPACKAGE were not thread safe. (reported by
21 changes in sbcl-1.0.33 relative to sbcl-1.0.32:
22 * new port: support added for x86-64 NetBSD. (thanks to Aymeric Vincent)
23 * improvement: support O_LARGEFILE access to files larger than 2GB on
24 x86-64/linux. (thanks to Daniel Janus; launchpad bug lp#453080)
25 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-DIRECTLY to get a list of
26 definitions for methods specializing on the passed class itself.
27 * new feature: SB-INTROSPECT:WHO-SPECIALIZES-GENERALLY to get a list of
28 definitions for methods specializing on the passed class itself, or on
30 * new build flag: :sb-xref-for-internals; SBCL will collect xref information
31 about itself during the build (e.g. for M-? in Slime), if this flag is
32 enabled in customize-target-features.lisp. This will increase the core
33 size by about 5-6mb, though, so it's mostly interesting to SBCL
35 * new feature: various GENCGC tuning parameters have been experimentally
36 documented and exported from SB-EXT. See documentation for details.
37 * fixes and improvements related to Unicode and external formats:
38 ** the Unicode character database has been upgraded to the
39 Unicode 5.2 standard, giving names and properties to a number of new
40 characters, and providing a few extra characters with case
42 ** improvement: restarts for providing replacement input/output on coding
43 errors for fd-stream external formats.
44 ** improvement: where :<encoding> is a keyword corresponding to an
45 external format the system supports, it is now possible to specify
46 (:<encoding> :replacement <character>) as an external format which will
47 automatically substitute <character> on encoding or decoding errors for
48 streams and for STRING-TO-OCTETS and its inverse. (launchpad bug
50 ** improvement: the file streams underlying the standard streams (such as
51 *STANDARD-INPUT*, *TERMINAL-IO*) are opened with an external format
52 which uses the replacement mechanism to handle encoding errors,
53 preventing various infinite error chains and unrecoverable I/O
55 ** minor incompatible change: the utf-8 external format now correctly
56 refuses to encode Lisp characters in the surrogate range (char-codes
57 between #xd800 and #xdfff).
58 ** fix a typo preventing conversion of strings into octet vectors
59 in the latin-2 encoding. (reported by Attila Lendvai; launchpad bug
61 ** fix a bug in the octet multibyte handling of decoding errors and the
62 USE-VALUE restart. (launchpad bug lp#314939)
63 ** fix the bug underlying the expected failure in the FORCE-END-OF-FILE
64 restart on fd-stream decoding errors.
65 ** fix a bug in the ATTEMPT-RESYNC fd-stream decoding restart when the
66 error is near the end of file.
67 ** fix a double-error case in unibyte octet conversions, when the first
68 use of USE-VALUE is ignored.
69 ** fix bugs in handling of undefined code points in unibyte encodings.
70 ** fix LISTEN (and consequent hangs in READ-CHAR-NO-HANG) on bivalent
71 streams after an UNREAD-CHAR.
72 * enhancement: SB-INTROSPECT:ALLOCATION-INFORMATION also reports if the
73 object is allocated in a boxed region of dynamic space.
74 * enhancement: SB-POSIX:FORK now signals an error if an attempt to
75 fork with multiple Lisp threads running is made, instead of going
76 ahead with unpredictable consequences. (reported by Leslie Polzer)
77 * bug fix: uses of slot accessors on specialized method parameters within
78 the bodies of SLOT-VALUE-USING-CLASS methods no longer triggers a type
79 error while finalizing the class. This fix may cause classes with slot
80 accessors to be finalized later than previously. (reported by Lars Rune
81 Nøstdal; launchpad bug lp#473699)
82 * bug fix: restore buildability on the MIPS platform. (regression from
83 1.0.30.38, reported by Samium Gromoff)
84 * bug fix: inspecting closures is less likely to fail with a type error.
85 * bug fix: no timer starvation when setting the system clock back.
86 (launchpad bug lp#460283)
87 * bug fix: WITH-STANDARD-IO-SYNTAX now binds *PRINT-PPRINT-DISPATCH* to the
88 standard pprint dispatch table as specified by CLHS.
89 * bug fix: give CLISP a hint about a type declaration to enable it to build
90 the cross-compiler without warnings. (thanks to Josh Elasser; launchpad
92 * bug fix: correctly dump literal objects in defaulting forms of arglists.
93 (reported by Attila Lendvai; launchpad bug lp#310132)
94 * bug fix: distinguish in type specifiers between arrays that might be
95 complex and arrays that are definitely complex. (launchpad bug lp#309129)
96 * bug fix: SUBTYPEP knows that the SYMBOL type is not SUBTYPEP the KEYWORD
97 type. (reported by Levente Mészáros; launchpad bug lp#485972)
98 * bug fix: setting the value of a symbol-macro within a method in the
99 presence of type declarations works properly again. (reported by Iban
100 Hatchondo; launchpad bug lp#485019)
102 changes in sbcl-1.0.32 relative to sbcl-1.0.31:
103 * optimization: faster FIND and POSITION on strings of unknown element type
104 in high SPEED policies. (thanks to Karol Swietlicki)
105 * optimization: faster CONCATENATE 'STRING in low SPEED policies (reported
107 * improvement: better error signalling for bogus parameter specializer names
108 in DEFMETHOD forms (reported by Pluijzer)
109 * bug fix: DELETE-FILE once again works on logical pathnames (regression
111 * bug fix: LOGICAL-PATHNAME signals a TYPE-ERROR if pathspec is specified
113 * bug fix: redefinition of a class via DEFCLASS without :DEFAULT-INITARGS
114 removes previous default initargs (reported by Lars Rune Nøstdal and
116 * bug fix: correct WHO-CALLS information for inlined lambdas with complex
117 lambda-lists. (reported by Peter Seibel)
118 * bug fix: SAVE-LISP-AND-DIE option :SAVE-RUNTIME-OPTIONS did not work
119 correctly when starting from an executable core without saved runtime
120 options (reported by Faré Rideau, thanks to Zach Beane)
121 * bug fix: (SETF SLOT-VALUE) signalled a warning which should have been
122 an optimization note instead. (reported by Martin Cracauer)
123 * bug fix: WITH-SLOTS did not work with THE forms. (thanks to David Tolpin)
124 * bug fix: Have RUN-PROGRAM with :INPUT T only run the subprocess in a
125 new process group if it doesn't need to share stdin with the sbcl
126 process. (thanks to Leslie Polzer)
127 * bug fix: SATISFIES could be misoptimized to refer to a local function.
128 (reported by Stanislaw Halik)
130 changes in sbcl-1.0.31 relative to sbcl-1.0.30:
131 * improvement: stack allocation is should now be possible in all nested
132 inlining cases: failure to stack allocate when equivalent code is manually
133 open coded is now considered a bug.
134 * improvements related to Unicode:
135 ** the Unicode character database has been upgraded to the
136 Unicode 5.1 standard, giving names and properties to a number of new
137 characters, and providing a few extra characters with case
139 ** the system now recognizes and produces names for Unicode Hangul
141 ** the EBCDIC-US external-format is now supported for octet operations
142 (as well as for stream operations).
143 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
144 be used to output toplevel compile-time effects into a separate .CFASL
146 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
147 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
149 * optimization: TRUNCATE on known single- and double-floats is upto 25%
151 * optimization: division of floating point numbers by constants uses
152 multiplication by reciprocal when an exact reciprocal exists.
153 * optimization: multiplication of single- and double-floats floats by
154 constant two has been optimized.
155 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
156 sufficient type information is available. (thanks to Leslie Polzer)
157 * optimization: SLOT-VALUE and (SETF SLOT-VALUE) with constant slot names on
158 known structure objects are as efficient as defstruct generated accessors.
159 * optimization: unused vector creation can now be optimized away.
160 * improvement: ASDF systems can now depends on SB-INTROSPECT.
161 * improvement: a STYLE-WARNING is signalled when a generic function
162 clobbers an earlier FTYPE proclamation.
163 * improvement: the compiler is able to track the effective type of
164 generic function across method addition and removal even in the
165 absence of an explicit DEFGENERIC.
166 * improvement: DESCRIBE now reports on symbols naming undefined
167 but assumed or declared function as well.
168 * improvement: recompilation of systems using SB-GROVEL now works
169 (thanks to Leslie Polzer)
170 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
171 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
173 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
174 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
175 well as user defined declaration names.
176 ** VARIABLE-INFORMATION is now aware of alien variables.
177 * improvement: improved address space layout on OpenBSD (thanks to Josh
179 * improvement: pretty-printing of various Lisp forms has been improved
180 (thanks to Tobias Rittweiler)
181 * bug fix: calls to DECODE-FLOAT and INTEGER-DECODE-FLOAT whose value was
182 unused were deleted in safe code. (reported by John Fremlin)
183 * bug fix: a failing AVER compiling certain MAKE-ARRAY forms. (reported
185 * bug fix: some out-of-line array predicates were missing (reported by
187 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
189 * bug fix: a failing AVER in %ALLOCATE-CLOSURES conversion has been fixed
190 (thanks to Larry D'Anna)
191 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
192 as well. (reported by Josh Elsasser)
193 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
194 leading to possible attempts to delete the same file twice. See docstring
195 on DELETE-FILE for details. (reported by John Fremlin)
196 * bug fix: DELETE-FILE once again deletes the file named by the pathname
197 designator argument, rather than its truename. (reported by Luis
199 * bug fix: the low-level debugger had 32-bit assumptions and was missing
200 information about some array types. (thanks to Luis Oliveira)
201 * bug fix: moderately complex combinations of inline expansions could
202 be miscompiled if the result was declared to be dynamic extent.
203 * bug fix: on x86, SAP-REF of sizes greater than 8 bits with offsets of the
204 form (+ <variable> <integer>) were miscompiled under certain
206 * bug fix: in some cases no compiler note about failure to stack allocate
207 was emitted, even if the objects were in fact heap allocated.
208 * bug fix: minor violation of "otherwise inaccessible" rule for stack
209 allocation could cause objects users might reasonably expect to
210 be heap allocated to be stack allocated.
211 * bug fix: DESCRIBE signalled an error for generic functions under
212 certain circumstances. (thanks to Leslie Polzer)
213 * bug fix: Fixed spelling of an error message.
215 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
216 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
217 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
218 of SB-THREAD:THREAD-ERROR-THREAD.
219 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
220 * new feature: docstrings for local and anonymous functions are no longer
221 discarded. (thanks to Leslie Polzer)
222 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
223 values in other threads.
224 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
225 about object allocation.
226 * optimization: division of a real float by a complex float is implemented
227 with a specialised code sequence.
228 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
229 keywords is an order of magnitude faster.
230 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
231 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
232 methods, and similarly for non-standard metaclass classes as long as there
233 are no methods additional on MAKE-INSTANCE.
234 * optimization: more efficient type-checks for FIXNUMs when the value
235 is known to be a signed word on x86 and x86-64.
236 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
237 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
239 * optimization: compiler is smarter about delegating argument type checks to
241 * optimization: several character functions are now compiled somewhat more
242 efficiently. (reported by Lynn Quam)
243 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
244 * improvement: less unsafe constant folding in floating point arithmetic,
245 especially for mixed complex/real -float operations.
246 * optimization: constant double and single floats are stored in native
247 unboxed format on x86[-64].
248 * optimization: smarter code for arithmetic operations with constant floats,
249 complex floats, or integers on x86[-64].
250 * optimization: smarter code for conjugate/multiplication of float complexes
251 and abs/negate of floats on x86-64.
252 * optimization: more efficient complex float and real float operations on
254 * improvement: complex float division is slightly more stable.
255 * improvement: DESCRIBE output has been reworked to be easier to read and
256 contains more pertinent information.
257 * improvement: failure to provide requested stack allocation compiler notes
258 provided in all cases (requested stack allocation not happening without a
259 note being issued is now considered a bug.)
260 * bug fix: SB-POSIX exports the documented types and functions
261 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
262 types. (reported by "abhi")
263 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
264 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
265 (thanks to Paul Khuong)
266 * bug fix: looping around HANDLER-CASE could silently consume stack space
267 on each iteration. (reported by "foobar")
268 * bug fix: better error signalling when calls to functions seeking elements
269 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
270 (reported by Tobias Rittweiler)
271 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
272 directory components now works as it used to. (various prolems reported by
273 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
274 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
275 (reported by Elliot Slaughter, thanks to Stas Boukarev)
276 * bug fix: bogus undefined variable warnings from fopcompiled references to
277 global variables. (thanks to Lars Rune Nøstdal)
278 * bug fix: foreign function names should now appear in backtraces on
279 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
280 * bug fix: SETF compiler macro documentation strings are not discarded
282 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
283 (thanks to Alex Plotnick)
284 * bug fix: SLEEP accepts large integer arguments, truncating them to
285 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
286 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
287 * bug fix: compiler notes for expensive slot type checks could be emitted
288 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
290 changes in sbcl-1.0.29 relative to 1.0.28:
291 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
292 https://bugs.launchpad.net/sbcl
293 Bugs can be reported directly there, or by sending email to
294 sbcl-bugs@lists.sourceforge.net
295 (no subscription required.)
296 * minor incompatible change: under weak type checking policy integer
297 types are weakened less aggressively.
298 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
299 allowed to return, which causes SBCL to quit with exit status 0. Previously
300 if the function returned with a small integer return value, that value
301 was accidentally reused as the exit status.
302 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
303 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
304 :HASH-FUNCTION argument. Refer to user manual for details.
305 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
307 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
308 microseconds since the Unix epoch on all platforms.
309 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
310 allows the compiler to safely elide boundedness checks for special
312 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
313 the symbol, prohibits both lexical and dynamic binding. This is mainly an
314 efficiency measure for threaded platforms, but also valueable in
316 * new feature: UNC pathnames are now understood by the system on Windows.
317 * optimization: the compiler uses a specialized version of FILL when the
318 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
320 * optimization: accesses to potentially non-simple arrays where element type
321 is known are 50% faster.
322 * optimization: compiler now generates faster array typechecking code.
323 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
325 * optimization: multidimensional array accesses in the absence of type
326 information regarding array rank are approximately 10% faster due to
327 open coding of ARRAY-RANK.
328 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
329 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
331 * optimization: result of call to VECTOR can now be stack allocated.
332 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
333 as long as the resulting array is one-dimensional and has a known
334 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
335 length of the list matches the known length of the vector does not
336 allocate the list as an intermediate step. Ditto for VECTOR and simple
338 * optimization: MAKE-ARRAY can now stack allocate in the presence of
339 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
340 known element type, and is known to be simple and one dimensional.
341 * improvement: SBCL now emits a compiler note where stack allocation was
342 requested but could not be provided (not in all cases, unfortunately)
343 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
344 * improvement: pretty-printing loop has been implemented properly. (thanks
345 to Tobias Rittweiler)
346 * documentation: CLOS slot typechecing policy has been documented.
347 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
348 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
349 pointer, leading to memory faults. (thanks to Bart Botta)
350 * bug fix: better floating point exception handling on x86/OpenBSD.
351 (thanks to Josh Elsasser)
352 * bug fix: exit status from QUIT when called under --script was lost
353 (reported by Hubert Kauker)
354 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
355 same implementation of FILL to initialize the array, even if a more
356 efficient one was available (reported by Stas Boukarev, thanks to
358 * bug fix: potential miscompilation of array stack allocation on x86 and
359 x86-64. (reported by Time Tossavainen)
360 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
361 result in their subforms being treated as top level forms. (reported by
363 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
364 than a machine register could leave garbage in the high bits of the
365 result register (bug 316325).
366 * bug fix: disable address space randomization Linux/x86-64 as well,
367 not just x86-64. (reported by Ken Olum)
368 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
369 generate incorrect code.
370 * bug fix: #201; type inference for CONS and ARRAY types could derive
371 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
372 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
373 could not be assigned to or bound with PROGV. (reported by Lorenz
375 * bug fix: the value of CL:- in the inspector was the previous expression
376 evaluated rather than the expression being evaluated.
377 * bug fix: constants can no longer be locally declared special.
378 * bug fix: signals delivered to threads started from foreign land (read:
379 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
380 thread by blocking all signals and resignalling.
381 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
382 from :INITFORM, if any.
384 changes in sbcl-1.0.28 relative to 1.0.27:
385 * a number of bugs in cross-compilation have been fixed, with the ultimate
386 result that building under (at least) clisp should be much more reliable.
387 * minor incompatible changes: echo-streams now propagate unread-char to the
388 underlying input stream, and no longer permit unreading more than one
390 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
391 frames, allowing for instance more reliable backtraces.
392 * improvement: the debugger REPL can now reference lexical variables
393 by name directly for code compiled with (DEBUG 3).
394 * improvement: errors from malformed declarations now have better source
395 paths associated with them. (thanks to Tobias Rittweiler)
396 * optimization: faster local calls on x86/x86-64
397 * bug fix: some error messages for out-of-bound array indexes confused the
398 index and the bound. (thanks to Stas Boukarev)
399 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
401 * bug fix: running regressions tests in shells without OSTYPE set now works.
402 (reported by Harald Hanche-Olsen)
403 * bug fix: more robust static space exhaustion signalling from
404 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
405 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
406 docstring away instead of storing it under names such as (LAMBDA (X)).
407 (reported by Leslie Polzer)
408 * bug fix: timers could go off in the wrong order, be delayed indefinitely
409 (thanks to Ole Arndt for the patch)
410 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
411 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
412 potentially-invalid effective methods in its cache.
413 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
414 instances as well (thanks to Paul Khuong)
415 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
416 SB-POSIX:WAITPID and their C equivalents.
417 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
419 changes in sbcl-1.0.27 relative to 1.0.26:
420 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
421 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
422 * improvement: the system either recovers from stack exhaustion or dies
423 properly as opposed to leaving the user uncertain of whether the handler
424 trampled on some random memory next to the stack or having to rely on
425 --lose-on-corruption (which is still a good idea to use in production
426 because stack exhaustion can happen in signal handlers which will likely
428 * bug fix: fix gc related interrupt handling bug on ppc (regression from
429 1.0.25.37, reported by Harald Hanche-Olsen)
430 * bug fix: work around signal delivery bug in darwin (regression from
431 1.0.25.44, reported by Sidney Markowitz)
432 * bug fix: fix ERROR leaking memory (reported by David Thompson)
434 changes in sbcl-1.0.26 relative to 1.0.25:
435 * incompatible change: an interruption (be it a function passed to
436 INTERRUPT-THREAD or a timer function) runs in an environment where
437 interrupts can be enabled. The interruption can use
438 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
439 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
440 potentially running out of stack. Keep in mind that in the absence
441 of WITHOUT-INTERRUPTS some potentially blocking operation such as
442 acquiring a lock can enable interrupts.
443 * incompatible change: GC-OFF and GC-ON are removed, as they were
444 always unsafe. Use WITHOUT-GCING instead.
445 * new feature: runtime option --disable-ldb
446 * new feature: runtime option --lose-on-corruption to die at the
447 slightest hint of possibly non-recoverable errors: running out of
448 memory, stack, alien stack, binding stack, encountering a memory
449 fault, etc. In the absence of --lose-on-corruption a warning is
451 * enhancement: detect binding stack exhaustion
452 * enhancement: detect alien stack exhaustion on x86/x86-64
453 * improvement: generally more stable and reliable interrupt handling
454 * improvement: there is a per thread interruption queue,
455 interruptions are executed in order of arrival
456 * improvement: a repeating timer reschedules itself when the it has
457 finished, but expiration times are spaced equally. If an
458 expiration time is in the past it will trigger after a short grace
459 period that may give a chance to other things to run.
460 * optimization: slightly faster gc on multithreaded builds
461 * optimization: faster WITHOUT-GCING
462 * bug fix: when JOIN-THREAD signals an error, do it when not holding
463 important locks so that the debugger/handler doesn't produce
464 recursive errors or deadlock.
465 * bug fix: real-time signals are not used anymore, so no more
466 hanging when the system wide real-time signal queue gets full.
467 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
469 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
470 * bug fix: fix random memory faults related to interrupts on alpha
471 * bug fix: fix random memory faults related to interrupts on sparc
472 * bug fix: fix deadlocks related to starting threads
473 * bug fix: fix deadlines on locks on futex platforms
474 * bug fix: restore errno in signal handlers
475 * bug fix: fix deadlocks related to hash tables
476 * bug fix: fix deadlocks in pcl
478 changes in sbcl-1.0.25 relative to 1.0.24:
479 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
480 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
481 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
482 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
483 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
484 better name and does not return values so stale on multiprocessor systems.
485 Also, HOLDING-MUTEX-P was added for about the only sane usage of
487 * improvement: unithread builds keep track of MUTEX-VALUE.
488 * improvement: reading from a TWO-WAY-STREAM does not touch the output
489 stream anymore making it thread safe to have a concurrent reader and
490 a writer, for instance, in a pipe.
491 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
492 arguments, making compiler-macros for SETF-functions able to inspect
493 their constant arguments.
494 * improvement: COMPILE-FILE reports times with millisecond accuracy
495 (thanks to Luis Oliveira)
496 * optimization: CHAR-CODE type derivation has been improved, making
497 TYPEP elimination on subtypes of CHARACTER work better. (reported
498 by Tobias Rittweiler, patch by Paul Khuong)
499 * bug fix: setting alien structure fields of type struct by value now
500 computes the right offset for the memory copy.
501 * bug fix: compilation problem involving inlined calls to aliens with
502 result type VOID. (reported by Ken Olum)
503 * bug fix: #235a; sequential inline expasion in different policies no
504 longer reuses the functional from the previous expansion site.
505 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
506 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
507 * bug fix: named ENUMs in multiply-referenced alien record types no longer
508 cause an implied type redefinition. Regression from 1.0.21.29.
509 * bug fix: sign-extension of small signed return values now works with C
510 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
511 * improvements to the Windows port:
512 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
513 Westfold, thanks to Rudi Schlatte)
515 changes in sbcl-1.0.24 relative to 1.0.23:
516 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
517 vector of a multidimensional SIMPLE-ARRAY.
518 * new feature: the system now signals a continuable error if standard
519 readtable modification is attempted. (thanks to Tobias Rittweiler)
520 * new feature: DIRECTORY has been extended with a non-standard keyword
521 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
522 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
523 has been robustified and documented better. (thanks to Robert Goldman)
524 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
525 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
526 * tradeoff: constant FORMAT control strings are now compiled unless
527 SPACE > SPEED (previously only when SPEED > SPACE.)
528 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
530 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
532 * bug fix: lexical type declarations are now correctly reported by
533 SB-CLTL2. (reported by Larry D'Anna)
534 * bug fix: STRING-TO-OCTETS did not handle :START properly when
535 using UTF-8 as external format. (thanks to Luis Oliveira)
536 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
537 are signalled correctly. (thanks to Stas Boukarev)
538 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
539 designator. (thanks to Tobias Rittweiler)
540 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
541 designator, and returns T instead of the function. (thanks to
543 * bug fix: direct superclasses of STANDARD-CLASS and
544 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
545 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
546 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
547 when no class with the named slot yet exists no longer causes a
548 compile-time style-warning.
549 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
550 in safe code. (reported by Didier Verna)
551 * bug fix: #430; nested structure constructors can stack allocate.
552 * bug fix: on some 64-bit platforms dynamic space size was truncated
553 to #xffff0000 bytes. (reported by Benjamin Lambert)
554 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
555 REPL unusable. (reported by Daniel Herring)
556 * bug fix: return values of READ-SEQUENCE did not take :START into
557 account on file streams, regressions since 1.0.12.22. (reported by
558 Thomas Russ, patch by Paul Khuong)
559 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
560 method specializer used to confuse permuation vector optimization.
561 * bug fix: system inserted bogus implicit type declarations for local
562 special variables in DEFMETHOD bodies.
563 * bug fix: #354; duplicated frames in backtraces due to
564 non-tail-call-optimized XEPs to functions with return type NIL
566 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
567 initializes structure object slots according to DEFSTRUCT initforms,
568 and DEFSTRUCT forms :INCLUDEind structure classes defined using
569 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
570 (reported by Bruno Haible and Stephen Wilson)
571 * bug fix: #395; fill-pointer output streams used now support
572 element-type BASE-CHAR as well.
573 * bug fix: compiler error when attempting to derive return value of
574 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
576 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
578 * bug fix: Mac OS X binaries should now be portable between Leopard
581 changes in sbcl-1.0.23 relative to 1.0.22:
582 * enhancement: when disassembling method functions, disassembly
583 for the associated fast function is also produced.
584 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
586 * optimization: printing with *PRINT-PRETTY* true is now more
587 efficient as long as the object being printed doesn't require
588 special handling by the pretty printer.
589 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
590 now interact correctly with type declarations.
591 * partial bug fix: PCL detects infinite recursion during wrapper
592 validation. (thanks to Attila Lendvai)
593 * bug fix: #426; nested function calls are inlined properly.
594 Previously if FOO was an inline function, in calls of the form
595 (FOO (FOO ...)) the outer call was not inlined.
596 * bug fix: long long arguments passed in registers to alien
597 functions on PowerPC Linux and NetBSD are now handled correctly
598 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
600 changes in sbcl-1.0.22 relative to 1.0.21:
601 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
602 for the shared object in the current directory, but passes the native
603 namestring of the designated pathname to the operation system's shared
604 object loading function as-it.
605 * minor incompatible change: --disable-debugger toplevel option now takes
606 effect before processing of initialization files and --eval or --load
608 * new feature: new commandline argument: --script, which supports
609 shebang lines. See documentation for details. (based on work by
611 * new feature: SAVE-LISP-AND-DIE can save current values of
612 --dynamic-space-size and --control-stack-size in the executable core,
613 causing it to skip normal runtime option processing. See documentation
614 for details. (thanks to Zach Beane)
615 * enhancement: inoccous calls to EVAL or generic functions dispatching
616 on subclasses of eg. STREAM no longer cause compiler notes to appear.
617 * enhancement: the system no longer resignals errors from --load and
618 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
619 associated with the original error to be lost. (thanks to Ariel
621 * enhancement: :DONT-SAVE keyword argument has been added to
622 LOAD-SHARED-OBJECT for controlling interaction with
623 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
624 the effects of an earlier LOAD-SHARED-OBJECT call.
625 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
626 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
627 provided. (reported by Cedric St-Jean)
628 * bug fix: circularity handling in the reader did not treat raw
629 structure slots correctly. (reported by Cedric St-Jean)
630 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
631 file descriptors when there were none.
632 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
633 was broken. (reporteed by Willem Broekema)
634 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
635 pathnames without a directory.
636 * bug fix: COMPILE-FILE did not warn about undefined variable
637 references at toplevel, and LOAD of the resulting fasl did
639 * bug fix: functions with non-required arguments used to end up with
640 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
641 * bug fix: redefining a function with non-required arguments didn't
642 update the system's knowledge about its call signature properly.
643 * bug fix: fixed #431; incompatible alien record type redefinitions
644 are detected and handled. (reported by Neil Haven)
645 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
646 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
648 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
649 after alien stack frames.
650 * bug fix: strings passed to mk{s,d}temp lacked null termination
652 changes in sbcl-1.0.21 relative to 1.0.20:
653 * new feature: the compiler is able to track the effective type of a
654 generic function across method addition and removal.
655 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
656 appropriately typed structure slots without locking.
657 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
658 information like those gathered by TIME using a programming-friendly
660 * new feature: TIME reports time taken even if the form performs a
661 non-local transfer of control.
662 * enhancement: reduced conservativism on GENCGC platforms: on
663 average 45% less pages pinned (measured from SBCL self build).
664 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
665 mutate constant symbols or violate declaimed type of the symbol.
666 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
667 sleeping on the semaphore in heavily contested semaphores.
668 * bug fix: semaphores and condition variables were not interrupt
670 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
671 owned by other threads anymore.
672 * bug fix: FIND on lists called KEY outside the specified
673 subsequence. (reported by budden)
674 * bug fix: LOG doesn't use single-float intermediate results when
675 given mixed integer and double-float arguments, leading to better
676 precision. (reported by Bob Felts)
677 * bug fix: LOG with base zero returned values of inconsistent type.
679 changes in sbcl-1.0.20 relative to 1.0.19:
680 * minor incompatible change: OPTIMIZE qualities
681 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
682 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
683 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
684 * documentation: some slot access efficiency guidelines have been
685 added to the user manual.
686 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
687 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
688 as ASSOC and MEMEBER.
689 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
690 to more efficient EQ-comparison versions more often.
691 * optimization: enhanced derivation of DOLIST iteration variable type
693 * optimization: constant folding of simple (LIST ...) forms as DOLIST
695 * optimization: runtime lookup of function definitions can be
696 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
697 * optimization: compiler is able to derive the return type of
698 (AREF (THE STRING X) Y) as being CHARACTER.
699 * optimization: CLRHASH on empty hash-tables no longer does pointless
700 work. (thanks to Alec Berryman)
701 * optimization: REPLACE deftransforms don't punt when :START1 or
703 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
704 no longer emits a code deletion note. (reported by Andrew
706 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
707 functions with an odd number of &OPTIONAL arguments, a &REST
708 argument, and one or more &KEY arguments at the call site.
709 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
710 arguments appeared at call sites as well.
711 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
712 READ-LINE &co can no longer read from them afterwards. (reported
714 * bug fix: fixed #427: unused local aliens no longer cause compiler
715 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
717 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
718 alien-stack leakage. (reported by Andy Hefner)
719 * bug fix: PROGV signals an error when an attempt to violate declared
720 type of a variable or bind a constant is made.
721 * bug fix: SET signals an error when an attempt to violate declared
722 type of a variable is made.
723 * bug fix: restart computation during the execution of a restart
724 test function no longer causes infinite recursion. (reported by
726 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
727 signals an error. (thanks to Pierre Mai)
728 * bug fix: EXPORT left symbol unexported in conflict situations.
729 (thanks to Michael Weber)
730 * bug fix: correctly handle name conflicts involving CL:NIL.
731 (thanks to Michael Weber)
732 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
733 conflicts arising from USEing package with conflicting symbols
734 correctly. (thanks to Michael Weber)
736 changes in sbcl-1.0.19 relative to 1.0.18:
737 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
738 warnings that go otherwise unhandled will be muffled if they are
739 of the type that's the value of this variable.
740 * optimization: stack allocation is slightly more efficient on x86
742 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
743 where a raw slot always is initialized using the initform whose
744 type is not know sufficiently well a compile-time are now compiled
745 correctly. (reported by John Morrison)
746 * bug fix: compiler no longer makes erronous assumptions in the
747 presence of non-foldable SATISFIES types.
748 * bug fix: stack analysis missed cleanups of dynamic-extent
749 arguments in non-let-converted calls to local functions.
750 * improvements to the Windows port:
751 ** adjusted address spaces for building on both Win32 and
752 Win64. (thanks for John Connors)
753 * fixed some bugs revealed by Paul Dietz' test suite:
754 ** interval arithmetic during type derivation used inexact integer
755 to single-float coercions.
756 ** arithmetic operations involving large integers and single
757 floats give the same results in compiled and interpreted code.
758 ** deriving the result type of COERCE no longer signals an error
759 if the derived type of the second argument is a MEMBER type
760 containing invalid type specifiers.
761 ** ADJOIN with constant NIL as second argument works correctly.
762 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
764 changes in sbcl-1.0.18 relative to 1.0.17:
765 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
766 profiles only the current thread.
767 * minor incompatible change: changes to SYMBOL-VALUE of constants
768 defined with DEFCONSTANT now signal an error.
769 * enhancement: SB-SPROF now has support for wallclock profiling,
770 and is also able to profile specific threads. REPORT output
771 has also additional sorting options.
772 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
774 * optimization: structure allocation has been improved
775 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
776 ** out of line constructors are ~10% faster.
777 ** inline constructors are ~15% faster.
778 ** inline constructors are capable of dynamic extent allocation
779 (generally on x86 and x86-64, in some cases on other platforms
781 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
783 * optimization: file compiler is now able to coalesce non-circular
784 lists, non-base strings, and bit-vectors. Additionally, constants
785 are never referenced through SYMBOL-VALUE at runtime.
786 * optimization: code defining methods on PRINT-OBJECT (and other
787 generic functions in the COMMON-LISP package) now loads faster.
788 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
790 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
791 fasl is now deleted. (reported by Attila Lendvai)
792 * bug fix: READ-LINE always returned NIL for the last line in files.
793 (reported by Yoshinori Tahara)
794 * bug fix: more accurate disassembly annotations of foreign function
795 calls. (thanks to Andy Hefner)
796 * bug fix: trimming non-simple strings and non-string string
797 designators when the there is nothing to trim works properly.
798 (thanks to James Knight)
799 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
800 (thanks to Travis Cross)
801 * fixed some bugs revealed by Paul Dietz' test suite:
802 ** NIL is a valid function name (regression at 1.0.13.38)
803 ** FILL on lists was missing its return value (regression at 1.0.12.27)
804 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
805 fill pointers properly (regression at 1.0.12.23)
806 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
807 displacement indices properly (regression at 1.0.12.23)
809 changes in sbcl-1.0.17 relative to 1.0.16:
810 * temporary regression: user code can no longer allocate closure
811 variable storage on stack, due to bug 419 without explicitly
812 requesting it. Please consult sbcl-devel for advice if you need to
813 use this feature in the meanwhile.
814 * new feature: runtime argument --control-stack-size can be used to
815 adjust thread default control stack size.
816 * enhancement: improved TIME output
817 ** all times are reported using the measured accuracy (milliseconds
818 for real and GC times, microseconds for everything else.)
819 ** processor cycle counts on x86 and x86-64.
820 ** interpreted forms are counted for both evaluator modes.
821 ** number of lambdas converted by the compiler is reported.
822 ** CPU percentage report (computed from real and total run time.)
823 ** more comprehensive run time reporting, using a condenced format
824 ** interperted form, lambda, and page fault counts are omitted
826 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
828 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
829 * optimization: two argument forms of LAST are upto ~10% faster
830 in normal SPEED policies.
831 * optimization: NCONC no longer needs to heap cons its &REST list
832 in normal SPEED policies.
833 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
835 * bug fix: UNION and NUNION work with :TEST-NOT once more,
836 regression since 1.0.9.1. (thanks to Eric Marsden)
837 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
838 since 1.0.15.36. (thanks to Paul Khuong)
839 * bug fix: LAST when always returned the whole list when given a bignum
840 as the second argument.
841 * bug fix: dynamic extent allocation of nested lists and vectors
842 could leak to otherwise accessible parts.
843 * bug fix: invalid optimization of heap-allocated alien variable
845 * bug fix: fasl header checking is less vulnerable to different
846 platform word lengths.
847 * bug fix: more correct assembler syntax for GNU binutils
848 2.18.50.0.4 support. (thanks to Marijn Schouten)
849 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
851 * bug fix: the fopcompiler can handle LOCALLY forms (with no
852 declarations) successfully. (reported by Attila Lendvai)
854 changes in sbcl-1.0.16 relative to 1.0.15:
855 * minor incompatible change: revert the changes to sb-posix's error
856 signaling added in 1.0.14.
857 * minor incompatible change: change PROBE-FILE back to returning
858 NIL whenever we can't get a truename, as was the case before 1.0.14.
859 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
860 inherits from ERROR instead of just CONDITION.
861 * new feature: SB-INTROSPECT can provide source locations for instances
862 as well. (thanks to Tobias Rittweiler)
863 * optimization: binding special variables now generates smaller code
864 on threaded platforms.
865 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
866 and cases where no :TEST is given but the compiler can infer that
867 the element to search is of type (OR FIXNUM (NOT NUMBER)).
868 * optimization: better LOGNOT on fixnums.
869 * optimization: modular arithmetic for a particular requested width
870 is implemented using a tagged representation unless a better
871 representation is available.
872 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
873 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
874 for wellformedness and returns a specialized vector. (reported by
875 Francois-Rene Rideau)
876 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
877 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
878 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
879 unscheduling timer at the same time as another timer fires.
880 * bug fix: don't reschedule timers for dead threads.
881 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
882 * bug fix: copying output from RUN-PROGRAM to a stream signalled
883 bogus errors if select() was interrupted.
884 * enhancement: add support for fcntl's struct flock to SB-POSIX.
886 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
887 * enhancement: cleaner backtraces for interactive interrupts, as
888 well as other cases where the interesting frames used to be
889 obscured by interrupt handling frames.
890 * enhancement: untracing a whole package using (UNTRACE "FOO") is
891 now supported, and tracing a whole package using (TRACE "FOO") now
892 traces SETF-functions as well.
893 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
894 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
895 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
896 * unreadably printed representation of hash-tables now includes
898 * bug fix: partially fixed #188: type propagation from assignments
899 is now more efficient.
900 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
901 'DOUBLE-FLOAT) are not flushable.
902 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
903 full address of the object, and none of the tag bits.
904 * bug fix: readably printing hash-tables now respects other printer
905 control variables. (reported by Cedric St-Jean)
906 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
907 keyword with MAKE-HASH-TABLE.
908 * bug fix: export SB-POSIX:MKSTEMP.
909 * bug fix: SORT was not interrupt safe.
910 * bug fix: XREF accounts for the last node of each basic-block as
912 * bug fix: MAKE-INSTANCE optimizations interacted badly with
913 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
914 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
915 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
916 create obsolete instances in the case of redefinition or
917 obsoletion of a superclass. (thanks to Andy Hefner)
918 * bug fix: Support for the Alpha architecture has been revived; it had
919 suffered somewhat from lack of maintenance since sbcl-1.0.
920 * improvements to the Windows port:
921 ** The system detects the codepage to use at startup, instead of
922 using the value from the saved core. (thanks to Kei Suzuki)
924 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
925 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
926 (see documentation for details.)
927 * revived support for OpenBSD (contributed by Josh Elsasser)
928 * partially fixed bug #108: ROOM no longer suffers from occasional
929 (AVER (SAP= CURRENT END)) failures.
930 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
931 bodies no longer cause a WARNING to be signalled. (reported by
933 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
934 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
935 the bug in TRUNCATE.) (reported by Michael Weber)
936 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
937 no samples. (reported by Andy Hefner)
938 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
939 no longer appear as (NIL ...) frames in backtraces.
940 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
941 is now actually exported from SB-EXT as documented. (reported by
943 * bug fix: sb-aclrepl now correctly understands how to inspect
944 single-floats on 64-bit platforms where single-floats are not boxed.
945 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
946 yet been finalized. (reported by Levente Meszaros)
947 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
948 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
949 SBCL-specific optimize qualities.
951 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
952 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
953 an executable in the search path, and does so in the child
954 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
955 been removed; it can be found in the CVS history, for somebody who
956 needs that search behavior (see the manual).
957 * minor incompatible change: compiler policy re. weakening type
958 checks has changed: now type checks are weakened only if SAFETY < 2
960 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
961 unparsing of directory pathnames as files. Analogously,
962 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
963 filename to parse into a directory pathname.
964 * enhancement: implicit generic function creation now signals a
965 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
966 users can bind handlers for and muffle around calls to LOAD.
967 * enhancement: RUN-PROGRAM allows unicode arguments and environments
968 to be used (using the default stream external format), and allows
969 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
970 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
971 strings and vectors whose element-type or simplicity is not fully
972 known at compile-time.
973 * optimization: STRING-TRIM and related functions no longer allocate
974 a new string when no trimming needs to be performed. These functions
975 are also faster than before when the input string has been declared
977 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
978 * optimization: READ-LINE is significantly faster for files containing
980 * optimization: non-open coded uses of character comparison operators
981 (e.g. char=) no longer cons when called with more than one parameter
982 on platforms supporting dynamic-extent allocation.
983 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
984 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
985 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
987 * bug fix: some sequence functions elided bounds checking when
989 * bug fix: too liberal weakening of union-type checks when SPEED >
991 * bug fix: more bogus fixnum declarations in ROOM implementation
994 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
995 * new feature: MAKE-HASH-TABLE now experimentally accepts a
996 :SYNCHRONIZED argument, which makes the hash-table safe for
997 concurrent accesses (but not iteration.) See also:
998 SB-EXT:WITH-LOCKED-HASH-TABLE, and
999 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
1000 * optimization: CONCATENATE on strings is an order of magnitude faster
1001 in code compiled with (> SPEED SPACE).
1002 * optimization: SUBSEQ is ~50% faster on lists.
1003 * bug fix: bug 417 fixed -- source location reporting is now more robust.
1004 * bug fix: SUBSEQ on a list will now correctly signal an error if if
1005 END is smaller then START.
1006 * bug fix: SB-PROFILE will no longer report extra consing for nested
1007 calls to profiled functions.
1008 * bug fix: ROOM implementation had bogus fixnum declarations which
1009 could cause type-errors when calling ROOM on large images.
1010 * bug fix: if file compilation is aborted, the partial fasl is now
1011 deleted, and COMPILE-FILE returns NIL as the primary value.
1012 * bug fix: number of thread safety issues relating to SBCL's internal
1013 hash-table usage have been fixed.
1014 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
1015 be returned from its body when the values were being returned
1016 using unknown-values return convection and the W-P-O was wrapped
1017 inside an UNWIND-PROTECT.
1018 * bug fix: sb-posix should now compile again under Windows, enabling
1019 slime to work again.
1021 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
1022 * incompatible change: hash-table accessor functions are no longer
1023 automatically protected by locks. Concurrent accesses on the same hash-table
1024 from multiple threads can give inconsistent results or even corrupt the
1025 hash-table completely. Multi-threaded applications should do their own
1026 locking at the correct granularity. In the current implementation it is
1027 still safe to have multiple readers access the same table, but it's not
1028 guaranteed that this property will be maintained in future releases.
1029 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
1030 and will signal an error at runtime.
1031 * enhancement: SB-THREAD package now exports a semaphore interface.
1032 * enhancement: CONS can now stack-allocate on x86 and
1033 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
1035 * enhancement: nested lists can now be stack allocated on
1036 platforms providing stack allocation support.
1037 * enhancement: dynamic-extent support has been extended to support
1038 cases where there are multiple possible sources for the stack
1040 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
1041 if the mutex is uncontested on Linux.
1042 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
1043 as specified by the CLHS. (thanks to Tobias Rittweiler)
1044 * bug fix: NaN comparison now works on x86-64.
1045 * bug fix: CLOSE :ABORT T on a stream with pending output now
1047 * bug fix: instances of non-standard metaclasses using standard
1048 instance structure protocol sometimes missed the slot type checks
1050 * bug fix: known functions can be inlined in high-debug code.
1052 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
1053 * minor incompatible change: the MSI installer on Windows no longer
1054 associates .lisp and .fasl files with the installed SBCL.
1055 * minor incompatible change: :UNIX is no longer present in *FEATURES*
1056 on Windows. (thanks to Luis Oliviera)
1057 * new feature: SB-CLTL2 contrib module now implements
1058 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
1059 * optimization: scavenging weak pointers is now more efficient,
1060 requiring O(1) instead of O(N) per weak pointer to identify
1061 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
1062 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
1063 method bodies using constant slot-names when the first argument is
1064 a specializer parameter for the method.
1065 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
1066 names, when the class of the instance is a direct instance
1067 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
1068 as slow as the constant slot-name case.)
1069 * optimization: member type construction is now O(N) instead
1071 * optimization: UNION and NUNION are now O(N+M) for large
1072 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
1074 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
1075 now more readable in environments like Slime which display it.
1076 (thanks to Tobias C. Rittweiler)
1077 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
1078 was able to infer the correct argument type for the object on which
1079 the CAS operation was being performed.
1080 * bug fix: copy propagation interfered with parallel assignment
1081 semantics in local calls. (reported by Paul Khuong)
1082 * bug fix: the signed modular fixnum shift compiled to wrong code on
1083 x86 and x86-64. (spotted by a slight modification to some of
1086 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
1087 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
1088 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
1089 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
1091 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
1092 and generic function now print the object identity for anonymous
1093 instances. (thanks to Attila Lendvai)
1094 * enhancement: as an extension to MOP, SBCL now supports SETF of
1095 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
1096 (thanks to Attila Lendvai)
1097 * enhancement: sb-cover has an alternate annotation mode which puts
1098 the form annotations at the car of the form, instead of annotating
1099 it completely. The benefit of this mode is that it shows better
1100 which forms were instrumented by the compiler.
1101 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
1102 by printer control variables. (reported by Dan Corkill)
1103 * bug fix: system leaked memory when delayed output was performed by
1104 the OS in smaller chunks then expected. (thanks to David Smith)
1105 * bug fix: system leaked memory when file streams were not closed
1107 * bug fix: large objects written to slow streams that were modified
1108 after the write could end up with the modified state written to
1109 the underlying file descriptor.
1110 * bug fix: multiple threads operating in parallel on the same stream
1111 could cause buffer-overflows.
1112 * bug fix: source location information is stored correctly
1113 (broken since 1.0.6). This bug would generally show up as the
1114 Slime debugger highlighting the wrong form when the sldb-show-source
1116 * bug fix: Branch forms are again annotated as branches in the sb-cover
1118 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
1119 on x86-64. (reported by Gregory Vanuxem)
1120 * bug fix: EQUALP could return wrong results for structures with raw
1121 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
1122 word). (reported by Vjacheslav Fyodorov)
1123 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
1126 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
1127 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
1128 atomic compare-and-swap operations on threaded platforms.
1129 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
1130 allows assigning a global minimum value to optimization qualities
1131 (overriding proclamations and declarations).
1132 * enhancement: closed over variables can be stack-allocated on x86
1134 * performance bug fix: GETHASH and (SETF GETHASH) are once again
1136 * optimization: slot definition lookup is now O(1). This speeds up
1137 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
1138 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
1139 * optimization: ASSOC and MEMBER can now be open-coded for all
1140 combinations of keyword arguments when second argument is constant
1141 and SPEED >= SPACE. In other cases a specialized version is
1143 * bug fix: using obsoleted structure instances with TYPEP and
1144 generic functions now signals a sensible error.
1145 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
1146 (reported by Kristoffer Kvello)
1147 * bug fix: backtrace construction is now more careful when making
1148 lisp-objects from pointers on the stack, to avoid creating bogus
1149 objects that can be seen by the GC.
1150 * bug fix: defaulting of values in contexts expecting more than 7
1151 variables now works on x86-64. (reported by Christopher Laux)
1152 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1154 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1155 as the property-list of a symbol.
1156 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1157 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1158 situations requested, are once again file-compileable. (reported
1161 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1162 * MOP improvement: support for user-defined subclasses of
1163 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1164 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1165 the expansion of DEFMETHOD, is responsible for generating a form
1166 which creates a list of specializers when evaluated. Additional
1167 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1168 debugging and introspective support.
1169 * minor incompatible change: the (unsupported) spinlock interface
1170 has changed: free spinlock now has the value NIL, and a held spinlock
1171 has the owning thread as its value.
1172 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1173 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1175 * enhancement: name of a socket-stream is now "a socket" instead of
1176 "a constant string".
1177 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1178 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1179 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1180 for generic functions. (Thanks to Tobias C. Rittweiler)
1181 * optimization: bignum printing speed has been improved by 20-40%
1182 (depending on the bignum size.)
1183 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1185 * bug fix: the cache used by the CLOS to store precomputed effective
1186 methods, slot offsets, and constant return values is now thread and
1188 * bug fix: generic function dispatch function updating is now thread
1189 and interrupt safe (in the sense that the known issues have been
1191 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1192 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1193 * bug fix: asdf systems with dependencies to the SB-POSIX or
1194 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1195 * bug fix: interrupt safety of applicable method computation has been
1198 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1199 * new contrib: sb-cover, an experimental code coverage tool, is included
1200 as a contrib module.
1201 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1202 significantly faster.
1203 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1204 produce more efficient code on x86-64 if the compiler can determine
1205 that (+ INDEX <constant>) does not require a bounds check and FOO
1206 has an element type at least 8 bits wide.
1207 * enhancement: a new, experimental synchronous timeout facility is
1208 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1209 * enhancement: when a symbol name conflict error arises, the
1210 conflicting symbols are always printed with a package prefix.
1211 (thanks to Kevin Reid)
1212 * enhancement: stepping is now once again supported on the SPARC and
1213 MIPS platforms. (It is also now more likely to work on CheneyGC
1215 * enhancement: sb-sprof can now also track and report accurate call
1217 * bug fixes: the treatment of non-standard subclasses of
1218 SB-MOP:SPECIALIZER is more correct.
1219 * incompatible change: PURIFY no longer copies the data from the
1220 dynamic space into the static and read-only spaces on platforms
1221 that use the generational garbage collector
1222 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1224 * bug fix: GC race condition occasionally resulting in crashes with
1225 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1227 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1229 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1230 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1231 * bug fix: an asynchronous interrupt could previously leave the
1232 system running with GC inhibited.
1233 * bug fix: a DECLARE form evaluated at top-level now causes an error
1234 rather than silently (or verbosely) returning NIL.
1235 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1236 result in a error being signalled due to source form lookup errors.
1237 (reported by Peter Graves)
1239 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1240 * incompatible change: removed writer methods for host-ent-name,
1241 host-ent-addresses -- changing the values did not update the DNS
1243 * minor incompatible change: changed experimental JOIN-THREAD interface
1244 * documentation: the manual now lists reader and writer methods
1245 in class slot documentation sections. (thanks to Richard M Kreuter)
1246 * documentation: unwinding from asyncronous events has been
1247 documented as unsafe.
1248 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1249 in multithreaded application code.
1250 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1251 platforms. (thanks to James Anderson for the optimization hint)
1252 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1254 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1255 variants no longer cons.
1256 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1257 their NOT- variants no longer cons.
1258 * optimization: EQUAL hash tables no longer use SXHASH for objects
1259 of all data types, but instead use an EQL hash for types for which
1260 EQUAL is the same as EQL.
1261 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1262 are significantly faster.
1263 * optimization: new STRING-OUTPUT-STREAM implementation is much
1264 faster for large outputs and conses 30% less on average.
1265 * enhancement: XREF information is now collected for references made
1266 to global variables using SYMBOL-VALUE and a constant argument.
1267 * enhancement: SIGINT now causes a specific condition
1268 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1269 * bug fix: FILE-POSITION used to signal an error in some cases where
1270 ANSI requires it to return NIL.
1271 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1272 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1273 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1275 * bug fix: bad type declaration in the CLOS implementation has
1276 been fixed. (reported by James Anderson)
1277 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1278 (thanks to Lutz Euler)
1279 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1280 required by the CLHS. (reported by Samium Gromoff)
1281 * bug fix: dead unbound variable references now signal an error.
1282 (reported by Marco Monteiro)
1283 * bug fix: / with an unused value was being deleted in safe code.
1284 (thanks to Marco Monteiro and Kevin Reid)
1285 * bug fix: number of characters that can be written onto a single
1286 line in a file is unlimited.
1287 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1288 been fixed by inhibiting interrupts for when GC is disabled.
1289 * bug fix: some interrupt safety issues with GETHASH, (SETF
1290 GETHASH), CLRHASH and REMHASH have been fixed.
1291 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1292 specifier no longer causes infinite recursion.
1293 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1294 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1295 * bug fix: an &environment argument with the correct variable information
1296 is passed to macros that are expanded during byte compilation.
1297 (reported by Samium Gromoff)
1298 * bug fix: a logic error was causing the hash function for bignums to
1299 have excessive amounts of collisions. (reported by Faré Rideau)
1300 * bug fix: modifying the contents of an array could change the return
1301 value of SXHASH on that array, which is only allowed for strings
1302 and bit vectors. (bug introduced in 0.9.16)
1303 * improvement: the x86-64/darwin port now passes all tests (except
1304 for the debugger tests) but should still be considered
1305 experimental until this is fixed.
1306 * improvement: a style-warning is signaled for CASE (etc) clauses with
1307 duplicate keys. (thanks to Kevin Reid)
1308 * improvement: macos/x86 and macos/x86-64 now use mach exception
1309 handlers for dealing with illegal instructions (for trapping and
1310 error handling) and memory protection violations (for GC).
1312 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1313 * new platform: experimental support for x86-64/darwin (MacOS).
1314 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1315 and getnameinfo sockets functions are used instead of gethostbyaddr
1316 and gethostbyname, on platforms where the newer functions are available.
1317 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1319 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1320 don't print the startup banner, but behave as if --noinform was passed
1321 as a command line argument. (thanks to Kevin Reid)
1322 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1323 * optimization: code using alien values with undeclared types is much faster.
1324 * optimization: the compiler is now able to open code SEARCH in more cases.
1325 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1326 * bug fix: using standardized COMMON-LISP special variables as loop
1327 variables no longer signals bogus package lock violations. (reported
1329 * bug fix: declaring local loop variables to be of a range-limited type
1330 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1331 (reported by Andras Simon)
1332 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1333 bugs remain on x86-64.)
1334 * bug fix: the #= and ## reader macros now interact reasonably with
1335 funcallable instances.
1336 * bug fix: type-checks for function arguments were compiled using the
1337 compiler policy settings of the wrong lexical environment (in 1.0.2
1339 * bug fix: SHADOW accepts characters as string designators, as required
1340 by the spec (thanks to Eric Marsden)
1341 * bug fix: fixed GC safety issues when foreign functions are called with
1342 non-base strings as arguments
1343 * bug fix: more consistent error output for fatal-compiler-warnings, like
1345 * bug fix: functions with proper names were showing up as NIL in
1346 backtraces (reported by Edi Weitz, regression in 1.0.2)
1348 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1349 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1350 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1351 produce more efficient code on the x86 if the compiler can determine
1352 that (+ INDEX <constant>) does not require a bounds check and FOO
1353 has an element type at least 8 bits wide.
1354 * bug fix: references to undefined variables are handled the same way
1355 in toplevel forms as in the normal compiler
1356 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1358 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1359 (thanks to Jon Buller)
1360 * improvement: faster compilation times for complex functions
1361 * improvement: added readlink support to SB-POSIX (thanks to Richard
1364 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1365 * improvement: experimental support for mach exception handling on
1366 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1368 * improvement: support for GBK external format.
1369 (thanks to Chun Tian (binghe))
1370 * improvement: the debugger now displays variables that have been closed
1371 over, in code compiled with (DEBUG 2) or higher
1372 * improvement: support for executable cores on NetBSD (thanks to
1374 * new feature: added a RESTART-FRAME debugger command
1375 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1376 be used to provide an implementation for FILE-POSITION on Gray streams
1377 (thanks to Eric Marsden)
1378 * new feature: add syslog, openlog and closelog support to SB-POSIX
1379 (thanks to Richard Kreuter)
1380 * optimization: the function call overhead in code compiled with
1381 a high DEBUG optimization setting is significantly
1382 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1383 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1384 (thanks to Tony Martinez)
1385 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1386 evaluated once (reported by Kevin Reid)
1387 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1388 works again (thanks to Kilian Sprotte)
1389 * bug fix: an error is signaled for tagbodies with duplicate tags
1390 (thanks to Stephen Wilson)
1391 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1392 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1393 stack frames from alien callbacks.
1394 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1395 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1396 65k pages (thanks to David Woodhouse)
1397 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1399 changes in sbcl-1.0.1 relative to sbcl-1.0:
1400 * new platform: FreeBSD/x86-64, including support for threading.
1401 * new feature: the compiler stores cross-referencing information
1402 about function calls (who-calls), macroexpansion (who-macroexpands)
1403 and special variables (who-binds, who-sets, who-references) for code
1404 compiled with (< SPACE 3). This information is available through the
1405 sb-introspect contrib.
1406 * new feature: users may subclass SEQUENCE, and have instances of
1407 these classes interoperate with standard Common Lisp functions if
1408 a number of methods are defined. (This feature is experimental
1409 and the interface subject to change based on feedback from SBCL
1410 users and the general community)
1411 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1412 x86/x86-64, rather than the previous fixed depth of 8
1413 * improvement: another pthread back-end of mutex "pthread-futex".
1414 * bug fix: non-ascii command-line arguments are processed correctly
1415 (thanks to Yaroslav Kavenchuk)
1416 * bug fix: non-required arguments were not passed correctly when a method
1417 defined using DEFMETHOD was called from a mop-generated method using
1418 CALL-NEXT-METHOD (reported by Pascal Costanza)
1419 * bug fix: recursion is now permitted in accessors through
1420 SLOT-UNBOUND. (reported by Pascal Costanza)
1421 * bug fix: an error was signaled at startup if the HOME environment
1422 variable was defined, but had an empty value (reported by Peter Van Eynde)
1423 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1424 signal a PROGRAM-ERROR, not a BUG.
1425 * bug fix: SB-POSIX:READDIR works when built with large file support.
1426 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1428 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1429 for better automatic indentation support. (thanks to Matt Pillsbury)
1430 * optimization: loading generic functions no longer takes O(n^2) time,
1431 proportional to the amount of methods in the generic function
1432 (reported by Todd Sabin and Jeremy Brown)
1433 * optimization: the FIND and POSITION family of sequence functions
1434 are significantly faster on arrays whose element types have been
1436 * improvements to the Windows port:
1437 ** Intermittent heap corruption problems have been fixed. (thanks
1438 to Alastair Bridgewater)
1439 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1441 ** Lisp is able to unwind foreign exception frames from alien
1442 callbacks. (thanks to Alastair Bridgewater)
1444 changes in sbcl-1.0 relative to sbcl-0.9.18:
1445 * improvement: experimental support for threading on FreeBSD/x86.
1446 (thanks to NIIMI Satoshi)
1447 * improvement: runtime option --dynamic-space-size can be used
1448 to set the size of the dynamic space reserved on startup.
1449 * improvement: floating point modes in effect are now saved in
1450 core, and restored on startup.
1451 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1452 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1453 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1454 threads are not running after *SAVE-HOOKS* have run.
1455 * improvement: writes to CLOS instance slots are type-checked in code
1456 compiled with (SAFETY 3)
1457 * improvement: floating-point exception handling on FreeBSD (thanks to
1459 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1460 (thanks to Zach Beane)
1461 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1463 * improvement: added support for the Shift-JIS external format.
1464 (contributed by NIIMI Satoshi)
1465 * improvement: callbacks are supported on Linux/PPC. (thanks to
1467 * bug fix: compiler bug triggered by a (non-standard) VALUES
1468 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1469 * bug fix: file compiler no longer confuses validated and already
1470 dumped structurres. (reported by Kaersten Poeck)
1471 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1472 fill-pointer signals a type-error as required. (thanks to
1474 * bug fix: disassemly of funcallable instances works.
1475 * bug fix: single stepping on PPC.
1476 * bug fix: fix thread-safety problems in the type system (generally
1477 manifesting as nonsensical errors like "STRING is a bad type specifier
1478 for sequences" or "The value 1 is not of type FIXNUM")
1479 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1480 (reported by Josip Gracin)
1481 * bug fix: an error is signaled for attempts to displace arrays with
1482 incompatible element types (thanks to Mario Mommer)
1483 * bug fix: more correct handling of wide characters in debug info
1484 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1485 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1486 and don't cause extra consing
1487 * optimization: MAP and MAP-INTO are significantly faster on vectors
1488 whose elements types have been declared.
1489 * Improvements to SB-SPROF:
1490 ** Support for allocation profiling
1491 ** Reduced profiling overhead, especially for long profiling runs
1492 * Improvements to the Windows port:
1493 ** floating point exceptions are now reported correctly.
1494 ** stack exhaustion detection works partially.
1495 ** more accurate GET-INTERNAL-REAL-TIME.
1496 ** better RUN-PROGRAM behaviour in case of errors in the child
1498 ** PROBE-FILE now simplifies pathnames correctly.
1499 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1500 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1501 1st 1970 (although time timezone data will be unreliable).
1502 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1504 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1505 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1506 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1508 * bug fix: two potential GC deadlocks affecting threaded builds.
1509 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1510 returns T (reported by Anton Kazennikov)
1511 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1512 with non-variable places
1513 * bug fix: remove a race condition in the setting of
1514 funcallable-instance functions, this should make threaded CLOS
1515 code more stable against memory faults.
1516 * bug fix: corruption of specials when unbinding is interrupted by an
1517 asynchronous unwind (reported by Hannu Koivisto)
1518 * improvement: the debugger will now also display local variables that
1519 are only used once, for code compiled with a DEBUG optimization quality
1522 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1523 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1524 * incompatible change: External-format support for FFI calls. The
1525 SB-ALIEN:C-STRING no longer implies an ASCII
1526 external-format. Instead the string is subject to external-format
1527 conversion. Additionally return values of type SB-ALIEN:C-STRING
1528 are by default of type (SIMPLE-ARRAY CHARACTER), not
1529 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1530 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1531 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1532 * incompatible change: SB-EXT package no longer contains the
1533 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1534 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1536 * incompatible change: the single-stepper is no longer available
1537 on Alpha, Mips and Sparc platforms.
1538 * minor incompatible change: the direct superclasses of
1539 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1540 not (STANDARD-OBJECT FUNCTION). This makes the
1541 class-precedence-lists of GENERIC-FUNCTION and
1542 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1544 * minor incompatible change: the default stream external format on
1545 non-unicode SBCL is no longer detected from the locale, but is always
1547 * new feature: Add a version of evaluator that uses an interpreter instead
1548 of the compiler. EVAL still uses the compiler by default, to switch it
1549 to use the interpreter, set the value of the variable
1550 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1551 * minor incompatible change: the single-stepper REPL has been merged
1552 with the normal debugger (see the "Stepping" heading of the debugger help
1553 for more details). The debugger command STEP will no longer switch
1554 to the single-stepper REPL.
1555 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1556 for a type now works.
1557 * bug fix: loading of default sysinit file works. (thanks to Leonid
1559 * bug fix: better detection of circularities in the file-compiler.
1560 (reported by Marco Monteiro)
1561 * bug fix: the CL pathname functions now work with files that have
1562 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1563 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1564 systems with Unix98 pty semantics.
1565 * bug fix: ASDF-INSTALL will now work with bsd tar.
1566 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1568 * bug fix: timers expiring in dead threads no longer cause a
1569 type-error (reported by Paul "Nonny Mouse").
1570 * bug fix: thanks to more lightweight single-stepper instrumentation,
1571 code compiled with (DEBUG 3) will compile and execute significantly faster,
1572 and will have more accurate type-inferencing than before
1573 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1574 whose bindings are modified
1575 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1576 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1577 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1578 CL:SOFTWARE-VERSION with proper version
1580 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1581 * feature: implemented the READER-METHOD-CLASS and
1582 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1583 as specified by AMOP.
1584 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1586 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1587 (thanks to Lutz Euler)
1588 * optimization: hashing of general arrays and vectors has been
1589 improved. (reported by Any Fingerhut)
1590 * enhancement: SB-INTROSPECT is now able to find definitions of
1591 profiled functions. (thanks to Troels Henriksen)
1592 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1593 as well, allowing compiler-macros for SETF-functions to expand.
1594 * enhancement: step-instrumentation no longer wraps known
1595 single-value functions with multiple-value context, allowing
1596 better type inference.
1597 * fixed bug #337: use of MAKE-METHOD in method combination now works
1598 even in the presence of user-defined method classes. (reported by
1599 Bruno Haible and Pascal Costanza)
1600 * fixed bug #339(c): if there are applicable methods not part of any
1601 long-form method-combination group, call INVALID-METHOD-ERROR.
1602 (reported by Bruno Haible)
1603 * fixed bug #361: the :FUNCTION initarg in the protocol for
1604 initialization of methods can now be used to override
1605 internally-produced optimized functions. (reported by Bruno
1607 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1608 system-provided lambda expression no longer cause warnings about
1609 unbound #:|pv-table| symbols.
1610 * bug fix: improved the handling of type declarations and the
1611 detection of violations for keyword arguments with non-constant
1613 * bug fix: potentially erronous calls to PATHNAME and
1614 MERGE-PATHNAMES were being flushed in some cases.
1615 (reported by Richard Kreuter)
1616 * bug fix: compiled calls to TYPEP were mishandling obsolete
1617 instances. (reported by James Bielman and Attila Lendvai)
1618 * bug fix: format strings with ~> without matching ~< no longer
1619 trigger an AVER, but signal an understandable error instead.
1620 (reported by Antonio Martinez)
1621 * bug fix: specifying an output-file without a pathname-type for
1622 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1623 of input-file instead of "fasl". (reported by Robert Dodier)
1624 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1625 (reported by James Y Knight).
1626 * bug fix: compiler-macros-function did not consider the environment
1627 argument for shadowing by local functions.
1628 * bug fix: compiler-macros expansion was inhibited by local INLINE
1630 * bug fix: inline expansions of known functions were subject to
1631 step-instrumentation in high debug policies, leading to problems
1632 with type-inference.
1633 * bug fix: compiler failed to differentiate between different CONS
1634 types in some cases.
1635 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1636 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1637 * bug fix: #368: incorrect use of expressed vs. upgraded array
1639 * bug fix: #306a: more precise unions of array types.
1640 * thread-safety improvements:
1641 ** CONDITION-WAIT could return early on Linux, if the thread was
1642 interrupted and subsequently continued with SIGCONT.
1643 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1645 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1646 * added support for the ucs-2 external format. (contributed by Ivan
1648 * minor incompatible change: pretty printing of objects of type
1649 (cons symbol) is, in the default pprint-dispatch-table, now
1650 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1652 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1653 called later in a class's lifetime, possibly as late as when the
1654 first instance of the class is created. Previously,
1655 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1656 class became finalizeable.
1657 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1658 for multibyte external-format streams. (thanks to "vbzoli")
1659 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1660 with :ALLOCATION :CLASS inherited from superclasses of the
1662 * fixed bug: anonymous classes can now be created using the :NAME
1663 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1664 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1665 * fixed bug: core-files saved with :EXECUTABLE T can again be
1666 executed when SBCL_HOME isn't set. (reported by James Knight)
1667 * fixed bug: toplevel LOCALLY forms with declarations could
1668 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1669 * fixed bug: printing from several different threads using different
1670 values of *print-case* could cause invalid output, due to
1671 some internal special variables of the printer not being bound
1672 thread-locally (reported by Max Mikhanosha)
1673 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1674 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1675 (reported by Pascal Costanza)
1676 * fixed bug: FINALIZE-INHERITANCE is called from
1677 REINITIALIZE-INSTANCE on classes when the class has previously
1678 been finalized, as required by AMOP.
1679 * minor code generation optimizations:
1680 ** better register allocation in CLOS dispatching functions
1681 ** overflow detection when coercing signed bytes to fixnums on x86-64
1682 ** is now implemented with one IMUL instruction instead of three shifts
1683 ** more efficient bit-vector access on x86 and x86-64
1684 ** more efficient access to raw structure slots on x86 and x86-64
1685 * fixed some bugs revealed by Paul Dietz' test suite:
1686 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1687 return its argument.
1689 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1690 * feature: thread support on Solaris/x86, and experimental thread support
1692 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1694 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1695 default method for SLOT-UNBOUND.
1696 * minor incompatible change: prevent the user from specializing the
1697 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1698 somewhat counter to the intent of the protocol, I (CSR) think, and
1699 additionally it just doesn't work in SBCL as currently
1700 implemented, thanks to optimizations (that are always valid for
1701 the other three SLOT-VALUEish functions, but not for the setter).
1702 * minor incompatible change: the :SB-LDB feature is now enabled by
1703 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1704 the low-level debugger.
1705 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1706 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1707 down to an SB-SYS:FD-STREAM, allowing direct communication in
1708 more cases. Temporary files and pipes are still used as a fallback
1710 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1711 on the non-reentrant C functions localtime(3) and gmtime(3).
1712 * bug fix: global optimization policy was not visible in LOCALLY and
1714 * bug fix: class objects can be used as specializers in methods.
1715 (reported by Pascal Costanza)
1716 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1717 with a type error. (reported by blitz_ on #lisp)
1718 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1719 failed with a type error. (reported by Pascal Bourguignon)
1720 * bug fix: merging pathnames against defaults with :DIRECTORY
1721 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1722 (reported by James Y Knight)
1723 * bug fix: saving large (>2GB) cores on x86-64 now works
1724 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1725 constant index and a (simple-array (signed-byte 32)) array
1726 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1727 error (patch by Robert J. Macomber)
1728 * feature: TIME now displays the amount of run-time spent in GC
1729 * bug fix: The debugger now does a better job of respecting
1730 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1731 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1732 workaround for bug 403.)
1733 * bug fix: floating point exception handling now works on FreeBSD
1734 (thanks to NIIMI Satoshi)
1735 * fixed some bugs revealed by Paul Dietz' test suite:
1736 ** MISC.641: LET-conversion were not supposed to work in late
1738 * improvements to the Win32/x86 port:
1739 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1740 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1742 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1743 * new feature: source path information is generated for macro-expansion
1744 errors for use in IDE's like Slime (thanks to Helmut Eller)
1745 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1746 * bug fix: compiler does not loop forever on an invalid type in
1748 * improvement: compilation of most CLOS applications is significantly
1750 * optimization: added a limited bytecode compiler for simple toplevel
1751 forms, speeding up compilation and FASL loading
1752 * bug fix: the statistical profiler now properly distinguishes anonymous
1755 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1756 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1757 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1758 into subclasses of the finalized class. Now user calls to
1759 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1760 system return before any subclasses are finalized.
1761 * minor incompatible change: The reader no longer ignores errors
1762 regarding non-existent packages in #+ and #- feature tests.
1763 * new feature: command line options --no-sysinit, --no-userinit to
1764 inhibit loading the corresponding init files
1765 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1766 for sending data through UDP sockets (thanks to François-René Rideau)
1767 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1768 error is signalled instead (thanks to François-René Rideau)
1769 * minor incompatible change: Error signalling behaviour of lexical
1770 operations violating package locks has changed slightly. Refer to
1771 documentation on package locks for details.
1772 * enhancement: EVAL can process IF-expressions without resorting to the
1774 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1775 constant-folded properly due to missing out-of-line functions.
1776 (reported by Mika Pihlajamäki)
1777 * bug fix: LISTEN sometimes returned T even in cases where no data was
1778 immediately available from the stream
1779 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1780 were reversed. (reported by Levente Mészáros)
1781 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1782 when given a vector displaced to an adjustable array. (reported
1784 * bug fix: derivation of float boundaries from numbers outside the
1785 appropriate float range (reported by John Wiseman)
1786 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1788 * bug fix: precomputing cache entries for generic functions with
1789 some subclasses of specializers as yet invalid does not attempt to
1790 fill a cache line with a negative offset. (reported by Levente
1792 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1793 allow condition class objects as arguments to DOCUMENTATION and
1794 (SETF DOCUMENTATION); only find and set documentation for
1795 structure names for the STRUCTURE doc-type. (suggested by Gary
1797 * improvements to the Win32/x86 port:
1798 ** (user-homedir-pathname) and default initialization file
1799 locations now know about the user's "Documents and Settings"
1800 directory (thanks to Yaroslav Kavenchuk)
1801 ** run-program is implemented (thanks to Mike Thomas)
1802 ** sockets support (thanks to Timothy Ritchey)
1803 ** better backtrace support (thanks to Alastair Bridgewater)
1804 ** sb-grovel supported
1805 ** asdf-install and sb-posix work somewhat
1806 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1807 * improvements to the Solaris/x86 port:
1808 ** works on Solaris 11/Solaris Express
1809 ** floating-point exception handling support
1810 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1811 * fixed some bugs revealed by Paul Dietz' test suite:
1812 ** REMOVE-METHOD returns its generic function argument even when
1813 the method is not one of the generic functions' methods.
1814 ** objects declared as MEMBER types can be admissible arguments to
1815 structure accessors.
1816 ** printing characters should simply be printed by the FORMAT ~:C
1818 ** compiler failure when compiling functions with hairy constant
1819 defaults for optional parameters.
1820 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1821 function, which is already optimized.
1823 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1824 * new platform: experimental support for SBCL x86/Darwin, including
1825 MacOS X 10.4.5 on Intel.
1826 * new feature: Unicode character names are now known to the system
1827 (through CHAR-NAME and NAME-CHAR).
1828 * new feature: the filesystem location of SBCL's core file is
1829 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1830 * minor incompatible change: the contrib modules SB-POSIX and
1831 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1832 this change is to make it easier to distribute
1833 location-independent binaries.
1834 * bug fix: as implied by AMOP, standardized classes no longer have
1835 slots named by external symbols of public packages. (reported by
1837 * optimization: calling functions via constant symbols -- as in
1838 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1839 via the function object as in (FUNCALL #'FOO).
1840 * optimization: on x86 and x86-64, the calling convention for the
1841 case of dealing with an unknown number of values has been altered
1842 to be friendlier to the prediction heuristics implemented,
1843 particularly with reference to CALL and RET pairing. (thanks to
1844 Alastair Bridgewater)
1845 * optimization: on x86, the code for access to thread-local symbol
1846 values and binding/unbinding of thread-local symbols is smaller.
1847 (thanks to Alastair Bridgewater)
1848 * enhancement: CONSTANTP is now able to determine constantness of
1849 more complex forms, including calls to constant-foldable standardized
1850 functions and some special forms beyond QUOTE.
1851 * fixed bug: occasional GC crashes on Solaris/x86
1852 * optimization: x86-64 supports stack allocation of results of simple
1853 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1854 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1855 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1856 (thanks to James Knight)
1857 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1858 by the standard (thanks to Stephen Compall)
1860 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1861 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1862 be used for bundling the runtime and the core file into one
1863 executable binary. This feature is not currently supported on all SBCL
1864 platforms. (thanks to James Bielman and NIIMI Satoshi)
1865 * new feature: a generational or ephemeral garbage collector is now
1866 the default on the PowerPC platform (both Linux and Darwin). The
1867 old Cheney (stop and copy) collector is a build-time option.
1868 (thanks to Cyrus Harmon, after Raymond Toy)
1869 * minor incompatible change: the method by which SBCL finds its
1870 contributed modules has changed; it no longer relies on symbolic
1871 links from an $SBCL_HOME/systems directory, but searches directly
1872 in the subdirectories of $SBCL_HOME.
1873 * enhancement: the dynamic heap size on the Linux/PPC platform is
1874 markedly larger, even using the older Cheney garbage collector.
1875 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1876 certain complicated string types. (reported by Gary King)
1877 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1878 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1879 * fixed bug: DOCUMENTATION on structures no longer signals an
1880 error if no structure type of the right name exists. (reported by
1882 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1883 applicable non-standard primary or :AROUND method on
1884 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1885 no longer cause unbound variable errors. (reported by Kevin Reid)
1886 * fixed bug: metacircle resolution in cases where methods have slots
1887 added before the slots from STANDARD-METHOD. (reported by Jean
1889 * fixed bug: the Power PC assembly code for calling into lisp
1890 sometimes computed its return address wrongly (depending on the
1891 15th bit of the address, fixed at link time). This used to
1892 manifest itself in a segmentation violation while building PCL.
1893 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1894 many others over the years)
1895 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1896 the stack pointer (treated as a signed integer) is negative.
1897 (thanks to Peter van Eynde)
1899 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1900 * new platform: experimental support for the Windows operating
1901 system has been added. (thanks to Alastair Bridgewater)
1902 * fixed several bugs in and robustified the PPC FFI (including
1903 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1904 * bug fix: allow non-simple string symbol names (reported by Paul
1906 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1908 * bug fix: saving a core corrupted callbacks on x86/x86-64
1909 * bug fix: closed a loophole in metacircularity detection and
1910 grounding in the PCL implementation of CLOS.
1911 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1912 (reported by tomppa on #lisp)
1913 * bug fix: clear the direction flag on Lisp -> C transitions, as
1914 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1915 (reported by Andrej Grozin and Hendrik Maryns)
1916 * optimization: major improvements to GC efficiency on GENCGC platforms
1917 * optimization: faster implementation of EQUAL
1918 * optimization: emit more efficient opcodes for some common
1919 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1920 * optimization: several other minor code-generation improvements on x86-64
1921 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1923 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1924 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1925 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1926 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1927 by ANSI as a generic function, and for consistency (SETF
1928 GENERIC-FUNCTION-NAME) is treated likewise.
1929 * fixed bug #233.b: make constraint propagation notice when a variable
1930 value is changed after it is referenced but before it is used
1931 * fixed bug #296: no more arbitrary behaviour from filenames with
1932 odd characters as --load arguments. (reported by Adam Warner)
1933 * fixed bug #390: :CHARACTER-SET pathname components now work as
1934 expected. (reported by Tim Daly Jr)
1935 * fixed bug #391: complicated :TYPE intersections in slot
1936 definitions no longer cause an error in PCL internals.
1937 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1938 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1939 returning the number of octets which would be written to the
1940 file-stream. (thanks to Robert J. Macomber)
1941 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1942 arguments to the internal stream functions. (thanks to David
1944 * bug fix: stack exhaustion is now handled on the x86/SunOS
1945 platform. (thanks to Daisuke Homma)
1946 * bug fix: more accurate ROOM results on GENCGC platforms
1947 * optimization: improved type inference for arithmetic-for
1948 index variables in LOOP
1949 * optimization: faster floating-point SQRT on x86-64
1950 * fixed some bugs revealed by Paul Dietz' test suite:
1951 ** DOCUMENTATION returns NIL instead of "" for method combinations
1952 that don't have a docstring
1954 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1955 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1956 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1957 therefore are not customizeable by user code (as seems to be at
1958 least permitted and maybe required by AMOP). As a consolation,
1959 however, the SBCL implementation of these functions now calls
1960 REINITIALIZE-INSTANCE as specified by AMOP.
1961 * bug fix: slot-definition documentation is propagated as per ANSI
1962 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1963 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1964 argument, as do eslotd initialization methods. (from Pascal
1965 Costanza's "Closer" project)
1966 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1967 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1969 * bug fix: it is now possible to have more than one subclass of
1970 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1971 (reported by Bruno Haible, Pascal Costanza and others)
1972 * bug fix: the dependent update protocol now works for generic
1973 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1974 and Pascal Costanza)
1975 * bug fix: condition-class instances corresponding to
1976 DEFINE-CONDITION forms are now created eagerly. (reported by
1977 Kalle Olavi Niemitalo on comp.lang.lisp)
1978 * bug fix: floating point printing is more accurate in some
1979 circumstances. (thanks to Simon Alexander)
1980 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1981 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1982 * bug fix: callbacks on OS X now preserve stack-alignment.
1983 (thanks to Cyrus Harmon)
1984 * enhancement: the x86-64 disassembler is much better at
1985 disassembling SSE instructions. (thanks to Lutz Euler)
1986 * enhancement: CHECK-TYPE on a variable now allows type-inference
1987 (thanks to Vincent Arkesteijn)
1988 * optimization: improved performance of EUC-JP external format.
1989 (thanks to NIIMI Satoshi)
1990 * optimization: performance improvements to IO on file streams of
1991 :ELEMENT-TYPE CHARACTER
1992 * optimization: much faster memory allocation on x86-64
1993 * optimization: faster garbage collections (latency and throughput)
1995 * optimization: faster bignum subtraction and division on x86 and x86-64
1996 (thanks to Lutz Euler)
1997 * optimization: more accurate type-inference for loops that use a
1998 floating point index variable or a negative step.
2000 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
2001 * bug fix: add a workaround to SBCL looping infinitely at startup on
2002 Linux kernels with apparently buggy implementations of personality().
2003 (thanks to Svein Ove Aas)
2004 * bug fix: Unicode symbols are correctly printed in LDB backtraces
2005 (thanks to David Lichteblau)
2006 * bug fix: local bindings shadow global symbol macros for
2007 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
2008 * optimization: non-open coded uses of numeric comparison operators
2009 (e.g. >) no longer cons when called with more than one parameter
2010 on platforms supporting dynamic-extent allocation.
2011 * enhancement: saving cores with foreign code loaded is now supported
2012 on MIPS/Linux in addition to the previously supported platforms.
2013 * bug fix: FIND-RESTART now tests for activity, not applicability when given
2014 a restart object as identifier. (reported by Helmut Eller for CMUCL)
2015 * bug fix: division by zero in sb-sprof when no samples were collected
2016 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
2017 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
2018 (thanks to Alexander Kjeldaas)
2019 * new feature: ignoring the return values of destructive functions
2020 like NREVERSE signals a compile-time style-warning.
2021 (thanks to Kevin Reid)
2022 * new feature: the alignment of alien structure fields can be
2023 explicitly specified. (thanks to Cyrus Harmon)
2024 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
2025 and dump core on SIGQUIT
2027 ** incompatible change: threads do not inherit values of specials
2028 from their parents (see manual)
2029 ** bug fix: threads stacks belonging to dead threads are freed by the
2030 next exiting thread, no need to gc to collect thread stacks anymore
2031 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
2032 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
2033 ** bug fix: dynamic variable and thread start related gc lossage
2034 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
2036 * fixed some bugs revealed by Paul Dietz' test suite:
2037 ** SUBTYPEP is slightly more accurate on heinously complicated
2038 CONS types where some of the members have uncertain (in the
2039 NIL, NIL sense) type relationships to each other.
2041 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
2042 no longer purified unless :PURIFY T is explicitly specified.
2043 ** Non-purified cores are significantly smaller than before
2045 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
2046 * new feature: timers based on Zach Beane's excellent timer package
2047 * added support for the following external formats: koi8-u,
2048 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
2049 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
2050 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
2051 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
2052 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
2053 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
2054 * incompatible change: a threaded SBCL will no longer revert to
2055 non-threaded mode on non-NPTL systems, but refuse to start entirely.
2056 * bug fix: interrupts are disabled until startup is complete; no
2057 more sigsegvs when receiving a signal to soon
2058 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
2060 * bug fix: add a workaround for the memory randomization features in
2061 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
2062 workaround will only be in effect on systems with the proc filesystem
2064 * bug fix: printing objects of type HASH-TABLE signals a
2065 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
2067 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
2068 have been running for over 50 days. (reported by Gilbert Baumann)
2069 * bug fix: the logic for getting names of functions gets less
2070 confused when confronded with alternate-metaclass
2071 funcallable-instances. (reported by Cyrus Harmon)
2072 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
2073 consistent, even on internal alternate-metaclass objects.
2074 * bug fix: flush closure information collected by physical
2075 environment analysis prepass before the main pass. (bug reported
2077 * bug fix: compiler pack phase does not modify a hashtable, which is
2078 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
2079 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
2080 may be expanded. (reported by Alexander <malishev> on c.l.l)
2081 * bug fix: no more style-warnings when compiling code using LAST
2082 under high-SPEED low-SPACE settings. (reported by David Wallin
2084 * fixed some incompatibilities between SBCL's MOP and the MOP
2086 ** the METAOBJECT class is now implemented;
2087 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
2088 STANDARD-OBJECT, as required;
2089 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
2090 now compatible (as required by VALIDATE-SUPERCLASS); there
2091 remains a consistency requirement over the presence or absence
2092 of the FUNCTION class in the superclass list of finalized
2093 classes; see the manual for more details;
2094 ** the :AROUND method for COMPUTE-SLOTS on
2095 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
2096 requested slot ordering.
2098 ** bug fix: parent thread now can be gc'ed even with a live
2100 ** bug fix: binding a special with PROGV to no value is not
2101 the same as that symbol not having been bound (thanks to
2103 ** bug fix: binding specials is thread safe (thanks to
2105 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
2106 and SB-SPROF do not die with 'no handler for signal XX in
2107 interrupt_handle_now(..)' anymore
2108 ** bug fix: WITH-TIMEOUT works with multiple running threads
2109 * fixed some bugs revealed by Paul Dietz' test suite:
2110 ** ENSURE-GENERIC-FUNCTION should take a method class object for
2111 the :method-class keyword argument.
2113 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
2114 * new port: the Solaris operating system on x86 processors is now
2115 mostly supported, though some rough edges in the environment
2116 remain. (thanks to Daisuke Homma)
2117 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
2118 space for its heap. (thanks to Thiemo Seufer)
2119 * optimization: SBCL on MIPS platforms now supports dynamic-extent
2120 closures. (thanks to Thiemo Seufer)
2121 * minor incompatible change: eof selects abort in the debugger.
2122 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
2123 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
2125 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
2126 deprecated, and will go away in a future revision of SBCL.
2127 * minor incompatible change: GC-ON and GC-OFF are no longer
2128 implemented with a counter, it does not matter how many times gc
2129 is switched on or off
2130 * bug fix: discriminating functions for generic function classes
2131 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
2132 make invalid assumptions about method precedence order. (reported
2134 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
2135 platforms. (thanks to Thiemo Seufer)
2136 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
2137 * bug fix: correct pathname printing with printer escaping is on.
2138 (thanks to Kevin Reid)
2139 * bug fix: complex VOP definitions in "user-space" no longer trigger
2140 package locks. (reported by Zach Beane)
2141 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
2142 for funcallable-instance functions, and is no different from
2144 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
2145 characters in the current readtable. (reported by Nicholas Neuss)
2146 * bug fix: SUBTYPEP on various CONS types returns more a more
2147 accurate acknowledgment of its certainty.
2148 * optimizations: REMOVE-DUPLICATES now runs in linear time on
2149 lists in some cases. This partially fixes bug 384.
2150 * flush all standard streams before prompting in the REPL and the
2152 * bug fix: signal handling and triggering gc do not conflict
2153 directly anymore, in particular a high frequency sb-sprof does
2154 not prevent gc from running
2155 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2156 approximation for timezone and DST information between the
2157 universal time epoch and the smallest negative 32-bit time_t.
2158 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2159 year 1899 so long as the encoded time is non-negative.
2160 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2161 the correct answer for non-constant rotate amounts when performing
2162 an inline 32-bit rotation.
2164 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2165 there is only one thread in the session
2166 ** bug fix: memory leak for streams created in one thread and
2167 written to in another
2168 ** bug fix: lockup when compiled with gcc4
2169 ** bug fix: race that allows the gc to be triggered when gc is
2171 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2172 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2174 ** bug fix: finished threads are gc'ed properly
2175 * fixed some bugs revealed by Paul Dietz' test suite:
2176 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2177 has arguments with a different set of applicable methods from
2178 the orignal arguments.
2179 ** The type error thrown by MAP now has a correct expected-type
2181 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2182 name a compiled function.
2183 ** Three MISC tests where a large bignum was improperly coerced to
2184 a float that couldn't represent that bignum during type
2185 derivation were fixed.
2186 ** SUBTYPEP can now handle types involving the negation of a
2187 list-form FUNCTION type.
2188 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2189 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2190 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2192 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2193 * New feature: Experimental support for bivalent streams: streams
2194 opened with :element-type :default now allow character and binary
2195 (unsigned-byte 8) I/O
2196 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2197 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2199 * bug fix: no more highly sporadic "couldn't check whether ~S is
2200 readable" when reading a stream and an interrupt hits in the middle
2201 of a select system call
2202 * compiler better recognizes complex arrays (reported by Eduardo
2204 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2205 for variables on SBCL built with linkage-tables. (reported by Luis
2207 * various error reporting improvements.
2208 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2209 (Thanks to Thiemo Seufer)
2210 * bug fix: nested reader invokations work correctly
2211 * bug fix: it is possible to have simultaneous references to foreign
2212 code and foreign data with the same name.
2214 ** added x86-64 support
2215 ** incompatible change: the threading api now works with thread
2216 objects instead of thread ids
2217 ** bug fix: threads are protected from signals and interruption when
2218 starting up or going down
2219 ** bug fix: a race where an exiting thread could lose its stack to gc
2220 ** bug fix: don't halt on infinite error in threads if possible
2221 ** fixed numerous gc deadlocks introduced in the pthread merge
2222 ** bug fix: fixed thread safety issues in read and print
2223 ** bug fix: debugger doesn't hang on session lock if interrupted at
2224 an inappropriate moment
2225 ** bug fix: run-program is now thread safe(r)
2226 ** bug fix: inner with-recursive-lock no longer releases the mutex
2227 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2228 * fixed some bugs revealed by Paul Dietz' test suite:
2229 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2230 EXPECTED-TYPE slots filled.
2231 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2233 ** MAKE-CONDITION accepts classes as type-designators.
2234 ** COMPILE may never return NIL.
2235 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2236 range before calling Unix time functions
2238 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2239 * numerous signal handling fixes to increase stability
2240 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2241 * minor incompatible change: we now correctly canonize default
2242 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2243 as per the MOP, rather than the historical (INITARG INITFUNCTION
2244 INITFORM). (reported by Bruno Haible)
2245 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2246 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2247 for more information.
2248 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2249 pathname is a directory pathname.
2250 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2251 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2253 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2254 :METACLASS argument in addition to a class name. (reported by
2255 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2256 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2257 (reported by Bruno Haible; patch from Gerd Moellmann)
2258 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2260 * bug fix: more cleanups to the floating point exception handling on
2261 x86-64 (thanks to James Knight)
2262 * bug fix: the compiler does not try to constant fold calls of
2263 COERCE and BIT-* functions when they return freshly constructed
2264 objects. (reported by Nikodemus Siivola and Paul Dietz)
2265 * optimization: improved the MIPS versions of generic (in the
2266 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2267 * optimization: direct conversion of (unsigned-byte 32) to floats on
2268 the PowerPC platform.
2269 * optimization: structure instances with raw slots now use less
2270 memory, and probably show better memory locality. (thanks to
2272 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2273 are treated as requests for stack allocation on the x86-64,
2274 Alpha-32, PPC and SPARC platforms.
2275 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2276 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2277 (thanks to Tony Martinez)
2279 ** gcing a dead thread can no longer lead to lockups
2280 ** threads block signals until they are set up properly
2281 ** errno is no longer shared by threads
2282 ** interrupt-thread restores the eflags register on x86
2283 ** fixed some lockups due to gc/thread interaction
2284 * fixed some bugs revealed by Paul Dietz' test suite:
2285 ** invalid dotted lists no longer raise a read error when
2286 *READ-SUPPRESS* is T
2287 ** signal an error if a symbol that names a declaration is used
2288 as the name of a type, or vice versa
2289 ** allow using the (declare (typespec var*)) abbreviation for
2290 (declare (type typespec var*)) with all type specifiers
2291 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2292 ** always return NIL from PPRINT-POP when OBJECT is NIL
2293 ** don't signal errors when pretty-printing malformed LABELS,
2294 FLET or MACROLET forms
2295 ** declarations in a DOLIST body are also in scope for the
2297 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2299 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2302 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2303 * fixed cross-compiler leakages that prevented building a 32-bit
2304 target with a 64-bit host compiler.
2305 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2306 opened with :IF-EXISTS :OVERWRITE.
2307 * fixed bug 281: error for an invalid qualifier in a short-form method
2308 combination method is not signalled until the faulty method is called.
2309 * bug fix: iteration variable type inferrer failed to deal with open
2310 intervals. (reported by Alan Shields)
2311 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2312 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2313 or more runtime options were provided to the sbcl binary.
2314 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2315 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2316 * optimization: there's now a fast-path for fixnum arguments in the
2317 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2318 * optimization: the code generated on x86-64 is more compact thanks
2319 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2320 * bug fix: floating-point exception handling now partly works on
2321 x86-64. (Thanks to James Knight)
2322 * improvement to the MIPS backend from Thiemo Seufer: C-style
2323 64-bit long long arguments and return values to alien functions
2325 * fixed some bugs revealed by Paul Dietz' test suite:
2326 ** the type-error signalled from WARN has a filled-in DATUM slot.
2327 ** the type-error required when a stream is not associated with
2328 a file has the stream as its datum.
2329 ** type-errors on single-floats on x86-64 no longer have
2330 :INVALID-OBJECT as the datum
2331 ** the type-errors signalled for invalid function names now have
2332 a correct expected type
2333 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2334 for structures defined with a :TYPE. Documentation strings for
2335 typed structures are no longer immediately discarded
2336 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2337 broadcast streams and synonym streams. FILE-LENGTH now also works
2338 on broadcast streams.
2340 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2341 * incompatible change: the --noprogrammer option, deprecated since
2342 version 0.7.5, has been removed. Please use the equivalent
2343 --disable-debugger option instead.
2344 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2346 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2347 inoperational for a while now) has been completely removed.
2348 * null lexical environments are now printed as #<NULL-LEXENV>,
2349 significantly reducing the amount of clutter in typical backtraces.
2350 * documentation on weak pointers, finalization, and after GC hooks
2351 has been added to the manual.
2352 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2353 as other specialized array types, is much faster. SUBSEQ and
2354 COPY-SEQ on such arrays have also been sped up.
2355 * optimization: EQL is now more efficient when at least other argument
2356 is known to be of type (OR FIXNUM (NOT NUMBER)).
2357 * fixed bug: the runtime is now less vulnerable to changes in the
2358 size of the SBCL object on OS X, and virtual memory is reserved for
2359 all spaces that need to be at a fixed address.
2360 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2361 * fixed bug: finalizers and after GC hooks that cause consing are now
2363 * fixed bug: compiler error messages and summaries are now printed to
2364 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2365 * fixed inference of the upper bound of an iteration variable.
2366 (reported by Rajat Datta).
2367 * fixed bug 211e: calling local functions with duplicated constant
2368 keyword argument no longer causes a bogus style warning about an
2370 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2371 ftype declaration to be disregarded. (reported by Dave Roberts)
2372 * fixed bug 373: caused by erronous compilation of references to alien
2373 variables in the runtime on ppc/darwin.
2374 * fixed bug 376: CONJUGATE type deriver.
2375 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2376 (reported by Baughn on #lisp)
2377 * a cross-compiler bug on non-x86 platforms has been identified and
2378 fixed. (thanks to Bruno Haible)
2379 * improvements to the MIPS runtime code for increased stability.
2380 (thanks to Thiemo Seufer)
2381 * increased the maximimum compact environment size to allow
2382 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2383 * partial workaround for bug 135: don't name the function we're
2384 calling for hairy cases of EVAL, so as not to accumulate one environment
2385 entry per call to EVAL. (reported by Kevin Reid)
2386 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2387 * optimization: type testing for non-vector arrays should be faster.
2388 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2390 * bug fix: setting 31st element of a bit vector to zero did not work
2392 * bug fix: redefining a class definition which failed due to a
2393 previous accessor / function clash now works (but see BUGS entry
2394 #380 for more problems in this area). (thanks to Zach Beane)
2395 * the long form of DEFINE-METHOD-COMBINATION disables method group
2396 checking when given a single method group with pattern *. (thanks
2398 * on x86 compiler supports stack allocation of results of simple
2399 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2400 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2401 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2402 * fixed some bugs related to Unicode integration:
2403 ** the restarts for recovering from input and output encoding
2404 errors only appear when there is in fact such an error to
2406 * fixed some bugs revealed by Paul Dietz' test suite:
2407 ** MISC.549 and similar: late transformation of unsafe type
2408 assertions into derived types caused unexpected code
2410 ** SCALE-FLOAT type deriver is less wrong.
2411 ** type derivers for EXP, LOG and similar functions compute result
2412 types for complex arguments better.
2413 ** (MISC.563) CONJUGATE type deriver works for very restricted
2415 ** out-of-line type testers for character strings are available.
2416 ** EQUAL compiler transform understands specialness of objects
2418 ** accessing double-floats stored on the stack now works on x86-64.
2419 ** debugger internals could sometimes create invalid lispobjs,
2420 resulting in GC crashes.
2421 ** MISC.548: type check weakening can convert required type into
2423 ** initialization forms for bindings are not in scope of free special
2426 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2427 * incompatible change: thread support for non-NPTL systems has
2428 been removed - locking is buggy and unreliable. A threaded
2429 SBCL build will now warn at startup and refuse to create
2430 new threads, unless futex support is detected in the kernel
2431 * incompatible change: the top level REPL now has only an ABORT
2432 restart associated with it, not TOPLEVEL and ABORT as it used to.
2433 TOP and TOPLEVEL are now available as debugger commands for
2434 returning to the top level.
2435 * incompatible change: forms evaluated in the REPL now use the
2436 global optimization policy.
2437 * incompatible change: user- and system-initialization files are
2438 no longer processed with LOAD, but by READ and EVAL; hence the
2439 global optimization policy, startup package, readtable, etc,
2441 * The .fasl file format number has been incremented because of
2442 various incompatible changes.
2443 * internal entry point details and argument counts no longer appear
2444 in backtraces unless explicitly requested by setting
2445 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2446 * built-in and standard functions no longer have names like "top
2447 level local call to FOO".
2448 * fixed bug 32: functions defined in non-null lexical environments
2449 now have more legible printed representation
2450 * fixed bug 33: functions defined in non-null lexical environemnts
2451 are now more amenable to inspection by INSPECT.
2452 * workaround for bug 354: XEPs no longer appear in backtraces unless
2453 explicitly requested.
2454 * fixed bug: receiving the signal which results from attempting to
2455 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2456 SIGBUS on other *BSDs) on architectures where the C stack is also
2457 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2458 to debugger confusion or wild instability. (reported by Cheuksan
2460 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2461 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2462 (reported by Lutz Euler)
2463 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2464 compiled code. (reported by Paul Dietz)
2465 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2466 specialized parameter to a method to something that is not TYPEP
2467 the specializer is now possible.
2468 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2469 face of package deletion.
2470 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2471 pathname. (reported by David Tolpin on c.l.l)
2472 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2473 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2474 STRING-STREAM and FILE-STREAM, respectively.
2475 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2476 than previously; in addition, its test suite is now run on build.
2477 (thanks to Luke Gorrie)
2478 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2479 * optimization: added a immediate representation for single-floats
2481 * fixed some bugs revealed by Paul Dietz' test suite:
2482 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2483 ** attempting to create a package with a colliding nickname causes
2484 correctable errors to be signalled.
2485 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2486 ** several x86-64 backend bugs related to sign-extension of immediate
2489 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2490 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2491 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2493 * fixed loading of multiply forward-referenced layouts.
2494 (thanks to Cheuksan Wang)
2495 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2496 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2497 (see COMPILE-FILE documentation for details of :PRINT option)
2498 * fixed bugs 19 and 317: fixed-format floating point printing is
2499 more accurate. This also fixes a bug reported by Adam Warner
2500 related to the ~@F format directive.
2501 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2503 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2504 dispatch table if the from-char is a dispatch macro character.
2505 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2506 by the wordsize no longer ignore the last word. (reported by Lutz
2508 * fixed bug in type checking of dynamic-extent variables. (reported
2510 * optimization: sequence traversal functions use their freedom to
2511 coerce function designators to functions.
2512 * optimization: code with many calls to CLOS methods specialized on
2513 CLOS classes has had redundant type checks removed.
2514 * optimization: added declarations to speed up operations that access
2515 the internal character database (for example STRING-UPCASE)
2516 * optimization: comparison operations between floats and sufficiently small
2517 fixnums no longer create extra rationals
2518 * fixed some bugs related to Unicode integration:
2519 ** portions of multibyte characters at the end of buffers for
2520 character-based file input are correctly transferred to the
2521 start of the buffer at the next read.
2522 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2523 passing it through to OPEN.
2524 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2525 argument given to internal calls to OPEN.
2526 ** the built-in comment readers (introduced by character sequences
2527 ";" and "#|") are more forgiving to encoding errors; they will
2528 STYLE-WARN and then attempt to resync the stream at a character
2529 boundary. (thanks to Teemu Kalvas)
2530 * fixed some bugs in the x86-64 port:
2531 ** Negative short int return values from c-calls are sign-extended
2533 ** The stack is aligned to 16-bytes for c-calls, as required by
2534 the ABI. (thanks to Cheuksan Wang)
2535 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2536 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2537 for enums. (thanks to Vincent Arkesteijn)
2538 ** Multiple small optimizations and bugfixes for floating point
2540 * fixed some bugs revealed by Paul Dietz' test suite:
2541 ** Space, Tab, Linefeed, Return and Page have the invalid
2542 secondary constituent character trait.
2543 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2545 ** WITH-INPUT-FROM-STRING should only update the index place on
2547 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2548 works more reliably.
2549 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2550 with one-dimensional arrays with fill pointers.
2551 ** TYPE-OF failed on a complex with an integer realpart and a
2553 ** compiler failure during type inference for the code of form
2554 (IF (EQL X (THE ...)) ...) (MISC.535).
2556 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2557 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2558 platform. The port supports 61-bit fixnums, large memory spaces
2559 and reloading shared object files.
2560 * enhancement: saving cores with foreign code loaded is now
2561 supported on ppc/Darwin in addition to the previously supported
2563 * enhancement: the statistical profiler now walks deeper into the
2564 call stack for more meaningful call-graphs and accrued time
2565 reports (x86/x86-64 only). It also now reports time spent in
2567 * enhancement: it is now possible to trace most individual methods
2568 of a generic function in addition to tracing the generic function
2570 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2572 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2573 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2574 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2575 produces an error. (thanks to Vincent Arkesteijn)
2576 * bug fix: NAMESTRING on pathnames with :WILD components in their
2577 directories works correctly. (thanks to Artem V. Andreev)
2578 * fixed bug 125: compiler preserves identity of closures. (reported
2580 * bug fixed: functions with &REST arguments sometimes failed with
2581 "Undefined function" when compiled with (DEBUG 3). (reported by
2583 * bug fix: overflow during compiling of setting element of a bit
2584 vector with constant index and value. (reported by Timmy Douglas)
2585 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2586 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2587 * on x86 compiler partially supports stack allocation of dynamic-extent
2589 * GO and RETURN-FROM do not check the extent of their exit points
2590 when compiled with SAFETY 0.
2591 * fixed some bugs related to Unicode integration:
2592 ** encoding and decoding errors are now much more robustly
2593 handled; it should now be possible to recover even from invalid
2594 input or output to the terminal. (thanks to Teemu Kalvas)
2595 ** provided a first cut at implementing STRING-TO-OCTETS and
2596 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2597 ** altered the SB-MD5 contributed module slightly, changing the
2598 interface just enough for it to be supportable for builds where
2599 lisp characters are not eight bits.
2600 * fixed some bugs revealed by Paul Dietz' test suite:
2601 ** the FORMATTER-generated functions for ~V[ conditionals require
2602 the correct number of arguments.
2603 ** READ-FROM-STRING returns the mandated second value when applied
2604 to displaced strings.
2605 ** the #\Rubout and #\Backspace characters are treated as invalid
2606 constituent characters by the tokenizer.
2608 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2609 * new feature: reloading changed shared object files with
2610 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2611 * new feature: references to foreign variables and functions
2612 can now be compiled and loaded before the corresponding shared
2613 object file is loaded, as long as the foreign definitions are
2614 available at runtime.
2615 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2616 supported. (thanks to Dan Debertin)
2617 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2618 just keywords. (thanks to Vincent Arkesteijn)
2619 * fixed bug #331: structure-class instances corresponding to
2620 DEFSTRUCT forms are now created eagerly.
2621 * fixed bug #345: backtraces from calls to undefined functions work
2622 on x86 as well. Related bug #61 is now also partially fixed on x86
2623 (backtraces from throws to unknown catch tags.)
2624 * bug fix: lambda-list parsing is now stricter vrt. order and number
2625 of lambda-list keywords.
2626 * bug fix: as specified by AMOP, an error is signalled if a
2627 class-option appears multiple times in a DEFCLASS form. (reported
2629 * bug fix: RANDOM can be compiled when the compiler derives the type
2630 of its numeric argument as a disjoint set of small integers.
2631 (reported by Paul Dietz)
2632 * bug fix: starting a core saved with shared objects loaded when
2633 those objects are not available no longer causes threaded SBCL to
2634 hang. (reported by Sean Ross)
2635 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2637 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2638 argument, which must always be NIL. (reported by Kalle Niemitalo)
2639 * bug fix: printing 1.0d+23 no longer results in an error.
2640 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2641 * bug fix: structure slot setters preserve evaluation order. (thanks
2643 * fixed some bugs related to Unicode integration:
2644 ** RUN-PROGRAM can allow its child to take input from a Lisp
2645 stream. (reported by Stefan Scholl)
2646 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2647 locales. (reported by Ken Causey)
2648 * fixed some bugs revealed by Paul Dietz' test suite:
2649 ** INCF, DECF and REMF evaluate their place form as specified in
2651 ** FORMATTER expands ~{ ~} iteration directives with V or #
2652 parameters correctly.
2653 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2654 consequent uses no arguments correctly.
2655 ** the system has a partial understanding of the (COMPLEX RATIO)
2657 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2660 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2661 * new feature: a build-time option (controlled by the :SB-UNICODE
2662 keyword feature, enabled by default) for building the system with
2663 support for the entire 21-bit character space defined by the
2665 * new feature: the system now has rudimentary external-format
2666 support; the primary user-visible change at this time is that
2667 characters with the high bit set (such as lower-case-e-acute) will
2668 print correctly to a terminal in a UTF-8 environment.
2669 * minor incompatible change: BASE-CHAR no longer names a class;
2670 however, CHARACTER continues to do so, as required by ANSI.
2671 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2672 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2673 has been moved to the SB-EXT package (temporarily re-exported from
2675 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2676 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2678 * fixed bug #351: better error-handlind and reporting for malformed
2680 * fixed bug #350: bignum-printing is now more memory-efficient,
2681 allowing printing of very large bignums, eg. (expt 2 10000000).
2682 (reported by Bruno Haible)
2683 * fixed bug #302: better primitive-type selection for intersection
2685 * fixed bug #308: non-graphic characters now all have names, as
2686 required. (reported by Bruno Haible)
2687 * bug fix: redefining a class with different superclasses now correctly
2688 removes it from the direct-subclasses of its previous superclasses.
2689 (reported by David Morse)
2690 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2691 new value now works. (reported by Bruno Haible)
2692 * bug fix: correct canonicalization of multiple non-standard slot
2693 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2694 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2695 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2696 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2697 now exists, an signals an error.
2698 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2699 during macroexpansion for non-iteration variables as well. (reported
2700 by Bruno Haible for CMUCL)
2701 * bug fix: Cyclic structures and unprintable objects in compiler
2702 messages no longer cause errors. (reported by Bruno Haible)
2703 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2704 and READ-FROM-STRING all now return a primary value of NIL if
2705 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2706 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2707 (reported by Bruno Haible for CMUCL)
2708 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2709 do not designate a condition. (reported by Bruno Haible for
2711 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2712 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2713 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2714 permitting resolution in favour of any of the conflicting symbols.
2715 (reported by Bruno Haible for CMUCL)
2716 * bug fix: EQUAL compiler optimizations is less aggressive on
2717 strings which can potentially compare true despite having distinct
2718 specialized array element types.
2719 * bug fix: unit enumerations can be defined without dividing by
2720 zero. (thanks to Vincent Arkesteijn)
2721 * FORMAT compile-time argument count checking has been enhanced.
2722 (report from Bruno Haible for CMUCL)
2723 * a partial workaround for the bug 262: the compiler does not try to
2724 inline-expand a local function doing RETURN-FROM from a deleted
2725 BLOCK. (thanks to Peter Denno for the bug report and to David
2726 Wragg for the simple test case)
2727 * fixed some bugs revealed by Paul Dietz' test suite:
2728 ** RENAME-PACKAGE allows all package designators as new package
2730 ** constraint propagation and lambda variable substitution are
2731 more cautious in dealing with partially deleted code.
2732 ** compiler sometimes lost reoptimization passes.
2733 ** CERROR, given a condition as condition designator, passes its
2734 remaining arguments to the continue format control without
2736 ** Case-altering FORMAT directives work correctly on non-ASCII
2738 ** The REAL type specifier handles bounds outside the floating
2739 point ranges without signalling FLOATING-POINT-OVERFLOW.
2740 ** Functions with IR1-transformations can create intercomponent
2741 references to global functions.
2742 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2744 ** FORMAT ~R treats a nil value for its first parameter correctly.
2746 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2747 * enhancement: saving cores with foreign code loaded is now
2748 supported on x86/NetBSD and sparc/Linux in addition to the previously
2749 supported platforms.
2750 * bug fix: on some platforms repeated installations caused multiple
2751 copies of HTML documentation to be installed -- should not happen
2752 any more. (reported by Stefan Scholl)
2753 * bug fix: parsing self-recursive alien record types multiple times
2754 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2755 original patch by Helmut Eller for CMUCL)
2756 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2757 (thanks to Richard Kreuter)
2758 * bug fix: defining classes whose accessors are methods on existing
2759 generic functions in other (locked) packages no longer signals
2760 bogus package lock violations. (reported by François-René Rideau)
2761 * bug fix: special variables as DEFMETHOD parameters no longer have
2762 associated bogus type declarations. (reported by David Wragg and
2764 * bug fix: read-write consistency on streams of element-type
2765 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2766 * bug fix: redefiniton of the only method of a generic function with
2767 no DEFGENERIC no longer emits a full WARNING. In addition,
2768 redefinition of generic functions with no DEFGENERIC to an
2769 incompatible lambda list now signals an error. (thanks to Zach
2771 * bug fix: DEFGENERIC now works even when there's a function of the
2772 same name in an enclosing lexical environment. (thanks to Zach
2774 * fixed compiler failure, caused by instrumenting code during
2775 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2776 * optimization: added loop analysis and improved register allocation
2777 * fixed some bugs revealed by Paul Dietz' test suite:
2778 ** POSITION on displaced vectors with non-zero displacement
2779 returns the right answer.
2780 ** (SIMPLE-STRING) is a valid type specifier for sequence
2782 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2784 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2785 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2787 ** The justification version of the FORMAT ~< directive treats
2788 non-zero minpad parameter correctly.
2790 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2791 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2792 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2793 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2794 the supported interface.
2795 * new feature: Single-stepping of code compiled with DEBUG 2 or
2796 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2797 * new feature: saving cores with foreign code loaded is now
2798 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2799 Timothy Moore's work for CMUCL)
2800 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2801 parameters to * instead of NIL if no initform is supplied.
2802 (reported by Johan Bockgaard)
2803 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2804 correctly when FUNCALL appears as the car of the form. Note:
2805 despite this FUNCALL forms are not currently subject to
2806 compiler-macro expansion. (port of Raymond Toy's fix for the
2807 same from CMUCL, reported by Johan Bockgaard)
2808 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2809 (thanks for Teemu Kalvas)
2810 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2811 case instead of terminating the iteration (thanks for Julian
2812 Squires, Sean Champ and Raymond Toy)
2813 * bug fix: incorrect expansion of defgeneric that caused a style
2814 warning. (thanks for Zach Beane)
2815 * bug fix: slot accessor effective method computation works properly
2816 for classes with multiple non-standard applicable methods on
2817 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2818 * on x86 compiler supports stack allocation of results of LIST and
2819 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2820 CMUCL implementation by Gerd Moellmann)
2821 * fixed some bugs revealed by Paul Dietz' test suite:
2822 ** FORMAT strings with both the ~<~:;~> form of the justification
2823 directive and pretty-printing directives cause an error.
2825 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2826 * incompatible change: the internal functions
2827 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2828 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2829 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2830 instead of the old functions.
2831 * new feature: on platforms where "dladdr" is available foreign
2832 function names now appear in backtraces. (based on Helmut Eller's
2834 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2835 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2837 * documentation: ASDF manual is now distributed with SBCL.
2838 * bug fix: SBCL can now load its contributed modules with REQUIRE
2839 even if the system-provided entries have been removed from
2840 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2842 * bug fix: backtraces involving undefined functions or assembly
2843 routines are more informative. (thanks to Brian Downing)
2844 * bug fix: mutually referent alien structures now work correctly.
2845 (reported by Rick Taube)
2846 * bug fix: structures defined by WITH-ALIEN can be referred to
2847 within other definitions in the same WITH-ALIEN.
2848 * bug fix: division operators (MOD, TRUNCATE and the like) with
2849 constant zero divisors and integer dividends no longer generate
2851 * bug fix: provide default methods for INPUT-STREAM-P and
2852 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2853 * bug fix: improve the detection and resolution of MOP metacycles
2854 (where effective method computation depends on the generic function
2855 having its effective method computed). (reported by Bruno Haible)
2856 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2857 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2858 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2859 * bug fix: pathnames with relative directory components can be
2860 represented relative to default pathnames.
2861 * optimization: in taking the GCD of bignums, reduce the two bignums
2862 to approximately the same size (using Euclid's algorithm) before
2863 applying the more sophisticated binary GCD. (thanks to Juho
2865 * optimization: COUNT on bitvectors now operates word-at-a-time.
2866 * optimization: ASH with a positive, but not necessarily constant,
2867 (leftwards) shift, when in a modular context, is compiled to a
2869 * fixed some bugs revealed by Paul Dietz' test suite:
2870 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2871 if the corresponding argument is NIL.
2872 ** FORMAT directives accepting a minpad parameter treat negative
2874 ** literal commas as character parameters to FORMAT directives are
2876 ** literal spaces directly after ~<Newline> directives within a
2877 format-logical-block (~:< ~@:>) do not induce :FILL-style
2878 conditional newlines.
2879 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2880 as specified: it no longer includes conditional newlines).
2881 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2883 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2884 :PER-LINE-PREFIX argument does not evaluate to a string.
2886 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2887 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2888 the manual for details; for now, package locks can be disabled by
2889 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2890 if no major problems are found then it is likely that they will be
2891 compiled in unconditionally.
2892 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2893 unsupported operators on all platforms. To load a shared library
2894 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2895 object file, link it into a shared library outside of SBCL and
2896 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2898 * bug workaround: ROOM T and the GC were not working together
2899 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2900 reporting facility aren't preserved. That reporting has been
2901 disabled, so now until and unless someone figures out how to make it
2902 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2903 * minor incompatible change: as threatened around sbcl-0.8.0, the
2904 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2905 The symbols are also exported from SB-PCL for backwards
2906 compatibility, but more so than before SB-PCL should be treated as
2907 an implementation-internal package.
2908 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2910 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2911 It is known as of this release not to work on the Alpha, however.
2912 * fixed bug #167: errors signalled due to illegal syntax in method
2913 bodies are now more legible.
2914 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2915 designators and can hence be used with TYPEP.
2916 * fixed bug #333: CHECK-TYPE now ensures that the type error
2917 signalled, if any, has the right object to be accessed by
2918 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2919 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2920 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2922 * fixed bug #334: programmatic addition of slots using specialized
2923 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2924 and :ALLOCATION :CLASS slots.
2925 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2926 float exponents. (rereported by Peter Seibel)
2927 * fixed a bug: #\Space (and other whitespace characters) are no
2928 longer considered to be macro characters in standard syntax by
2929 GET-MACRO-CHARACTER.
2930 * fixed bug: initialization of condition class metaobjects no longer
2931 causes an instance of the condition to be created. (reported by
2933 * fixed bug: it is now possible to have slots such that
2934 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2935 description is neither :INSTANCE nor :CLASS.
2936 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2937 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2938 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2939 non-standard SLOT-DEFINITION-ALLOCATION values.
2940 * fixed bugs: various race conditions handling exiting threads.
2941 CL-PPCRE's thread torture test now passes.
2942 * fixed arguably-a-bug: GC time proportional to number of threads in
2943 system even when most of them are idle
2944 * optimization: improved performance of BIT and SBIT on bit-vectors.
2945 * .fasl file incompatibility: The fasl file version number has
2946 been incremented because of changes associated with package locks.
2948 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2949 * minor incompatible change: the system no longer provides
2950 optimization hints (conditions which are TYPEP
2951 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2952 mode; these hints are still emitted when higher SPEED optimization
2954 * new contrib module: a sampling profiler (profiling by statistical
2955 sampling, rather than by instrumenting functions) is available as
2956 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2957 * the behaviour of the standard function ED is now customizeable by
2958 third parties through a hook variable: see ED's documentation
2959 string for information on the protocol.
2960 * the compiler no longer emits efficiency notes for (FUNCALL X)
2961 when the type of X is uncertain under default optimization
2963 * fixed bug 276: mutating a binding of a specialized parameter to a
2964 method to something that is not TYPEP the specializer is now
2966 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2967 exhibit the required behaviour on the x86 platform. (thanks to
2968 Peter van Eynde, Eric Marsden and Bruno Haible)
2969 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2970 even for difficult arguments. (reported by Peter Graves)
2971 * fixed bug 141a: the backquote printer now descends quoted
2973 * fixed another bug in backquote printing: no more destructive
2974 modification of the form's list structure. (reported by Brian
2976 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2977 move between its address being taken and the call to
2978 interrupt_thread, fixing a crashing race condition.
2979 * the SB-POSIX contrib implementation has been adjusted so that it
2980 no longer exhibits ridiculously poor performance when constructing
2981 instances corresponding to C structs.
2983 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2984 * minor incompatible change: the sb-grovel contrib now treats C
2985 structures as alien (in the sense of SB-ALIEN) objects rather than
2986 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2987 has implications for memory management of client code
2988 (sb-grovel-returned objects must now be manually managed) and for
2989 type safety (alien objects now have full types).
2990 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2991 used to control emission of compiler diagnostics, rather than the
2992 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2993 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2994 quality should be considered deprecated.
2995 * install.sh now installs the user manual as well
2996 * (not quite a new documentable feature, but worth considering in
2997 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2998 beginnings of a semantically meaningful condition hierarchy is
2999 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
3001 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
3002 designator as the defaults argument.
3003 * fixed bug: Displaced arrays whose displaced-to array has become
3004 too small now cause ARRAY-DIMENSION to signal an error, providing
3005 for safer bounds-checking. (reported by Bruno Haible)
3006 * fixed bug: DEFCLASS slot definitions with identical :READER and
3007 :WRITER names now signal a reasonable error. (reported by Thomas
3009 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
3011 * fixed bug: Invalid slot specification errors now print correctly.
3012 (thanks to Zach Beane)
3013 * fixed bug 320: Shared to local slot value transfers in class
3014 redefinitions now happen corrently from superclasses as
3015 well. (reported by Bruno Haible)
3016 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
3018 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
3019 improper lists correctly. (reported by Bruno Haible)
3020 * fixed bug 313: source-transform for <fun-name> was erroneously
3021 applied to a call of a value of a variable with name <fun-name>.
3022 (reported by Antonio Menezes Leitao)
3023 * fixed bug 307: The obsolete instance protocol ensures that
3024 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
3025 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
3026 remove dead unknown-values globs from the middle of the stack.
3027 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
3028 easier to resume long computations after using *BREAK-ON-SIGNALS*
3029 to diagnose and fix failures. (thanks to Nikodemus Siivola)
3030 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
3031 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
3032 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
3033 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
3034 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
3036 * fixed a (fairly theoretical) bug in string printing: if
3037 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
3038 does not have array-element-type equal to the most general string
3040 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
3041 function name. (thanks to Nikodemus Siivola)
3042 * fixed bug: compile-time format string checker failed on
3043 non-closed ~{. (reported by Thomas F Burdick)
3044 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
3045 don't warn on legal loop code involving more than one aggregate
3046 boolean. (thanks to Nikodemus Siivola)
3047 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
3048 in keyword position to MAKE-PACKAGE was not being reported as
3050 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
3051 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
3053 * fixed bugs 280 and 312: the checking for multiple definitions in a
3054 file is less likely to become confused by uses of inline
3056 * fixed bug: the #S reader macro performs the keyword coercion
3057 specified for slot names. (reported by Kalle Niemitalo)
3058 * fixed bug: lambda lists may contain symbols whose names start with
3059 & but are not lambda-list-keywords; their occurrence triggers a
3061 * fixed bug 321: define-method-combination argument lambda lists do
3062 not cause invalid code to be generated when &optional or &aux
3063 variables have default values. (reported by Bruno Haible)
3064 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
3065 have CLOS classes; this appears also to have fixed reported
3066 instability in stack exhaustion detection.
3067 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
3068 MUFFLE-WARNING when no associated restart is present is now
3070 * optimization: rearranged the expansion of various defining macros
3071 so that each expands into only one top-level form in a
3072 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
3074 * optimization: used a previously little-used slot in symbols to
3075 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
3077 * fixed some bugs revealed by Paul Dietz' test suite:
3078 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
3080 ** Fixed an optimization bug related to inheritance of initforms
3081 from local to shared slots.
3082 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
3083 ** CAST optimizer forgot to flush argument derived type.
3084 ** print/read consistency on floats is now orders of magnitude
3085 more likely. (thanks also to Bruno Haible for a similar report
3087 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
3088 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
3089 for objects of type REAL. Make it so.
3090 ** SXHASH is specified (infelicitously) to respect similarity,
3091 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
3092 it so. (thanks to Markus Ziegler)
3093 ** on the Alpha, the compiler succeeds in compiling functions
3094 returning a known number of arguments greater than 63.
3095 ** fixed handling of invalid NIL arguments in keyword position
3097 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
3099 ** provided a readably-printable representation for RANDOM-STATE
3101 ** ensured that pathnames, where they have a namestring, always
3102 print using #P"..." syntax.
3104 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
3105 * Support for the forthcoming 2.0 version of the NetBSD kernel
3106 running on x86 hardware has been added. (thanks to Perry
3107 E. Metzger most immediately, and others for their past work)
3108 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
3109 current release version; SBCL's OpenBSD support had been broken
3110 since about the time of OpenBSD's switch to ELF binary format.)
3111 * [placeholder for DX summary]
3112 ** user code with &REST lists declared dynamic-extent, under high
3113 speed or space and low safety and debug optimization policy.
3114 * The manual has been converted to Texinfo format and the debugger
3115 chapter from the cmucl manual has been added.
3116 * A facility has been added to extract documentation strings from
3117 sbcl and store them as Texinfo-formatted snippets for inclusion in
3118 the manual (via Texinfo's @include directive)
3119 * bug fix: compiler emitted division in optimized DEREF. (thanks for
3120 the test case to Dave Roberts)
3121 * bug fix: multidimensional simple arrays loaded from FASLs had fill
3122 pointers. (reported by Sean Ross)
3123 * bug fix: PROFILE output is printed nicely even for large numerical
3124 values. (thanks to Zach Beane)
3125 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
3126 greater than 32 handle EOF correctly.
3127 * bug fix: on X86 an immediate argument of the IMUL instruction is
3128 correctly printed in disassembly. (thanks to Lutz Euler)
3129 * bug fix: class slots in redefined classes preserve their old
3130 values. (thanks to Bruno Haible and Nikodemus Siivola)
3131 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
3132 succeeds. (reported by Marco Baringer)
3133 * fixed some bugs revealed by Paul Dietz' test suite:
3134 ** READ-SEQUENCE now works on ECHO-STREAMs.
3135 ** RATIONALIZE works more according to its specification. (thanks
3138 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
3139 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
3140 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
3141 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
3142 should matter to you only if you rebind the printer control
3143 variables and then find you want different bindings in the
3144 debugger than in the ordinary execution of your program.)
3145 * The runtime build system has been tweaked to support building
3146 (on SPARC/SunOS) using a C compiler which invokes Sun's own
3147 assembler and linker. (thanks to Nikodemus Siivola)
3148 * Unbound, undefined, undeclared variables now trigger full
3149 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
3150 more often programmer error than deliberate exploitation of undefined
3152 * optimization: the hash algorithm for strings has changed to one
3153 that is less vulnerable to spurious collisions. (thanks to Juho
3155 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3156 needless bounds checking. (thanks to Juho Snellman)
3157 * optimization: implemented multiplication as a modular
3158 (UNSIGNED-BYTE 32) operation on the PPC backend.
3159 * fixed some bugs revealed by Paul Dietz' test suite:
3160 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3162 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3163 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3164 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3166 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3168 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3170 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3171 their output stream on EOF from read.
3172 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3173 have been read to end-of-file.
3174 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3176 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3177 description of determination of which consecutive characters
3179 ** Printing the "Space" character with escaping on now yields "#\\ ",
3180 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3181 ** Reading floating-point numbers with *READ-BASE* set to a number
3182 less than 10 works correctly.
3183 ** Reading floating-point numbers with *READ-BASE* set to a number
3184 more than 10 works correctly.
3185 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3186 the readtable currently in effect.
3188 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3189 * minor incompatible change: parsing of namestrings on a physical
3190 (Unix) host has changed; numbers after the final #\. in a
3191 namestring are no longer interpreted as a version field. This is
3192 intented to be largely invisible to the user, except that the
3193 meaning of the namestring "*.*.*" has changed: it now refers to a
3194 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3195 should usually be replaced by
3196 (make-pathname :name :wild :type :wild :version :wild)
3197 with the added benefit that this is more likely to be portable.
3198 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3199 option to OPEN now signals an error if the file being opened
3200 exists; this may have an impact on existing code.
3201 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3202 signals received, on the PowerPC platforms (both Linux and
3203 Darwin). (thanks to Pierre Mai for pointing out the location of
3205 * several fixes on OS X: The system now builds and runs cleanly on
3206 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3207 (thanks to Brian Mastenbrook)
3208 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3209 second-resolution: integer multiples of 1/3600 between -24 and 24.
3210 (thanks to Vincent Arkesteijn)
3211 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3212 type when called with 1 argument; PEEK-CHAR checked type of
3213 PEEK-TYPE only after having read first character from a
3214 stream. (reported by Peter Graves)
3215 * bug fix: the garbage collector now has much better locality
3216 behaviour, and in particular no longer treats all memory as being
3217 exhausted when many small objects point to each other in a deeply
3219 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3220 recognized as being TYPEP their class.
3221 * bug fix: the PUSHNEW documentation string has been corrected.
3222 (thanks to Vincent Arkesteijn)
3223 * bug fix: defaulting of the value for the last of an atypically
3224 large number of multiple values being bound was not being
3225 performed correctly on the Alpha or PPC platforms
3226 * optimization: implemented multiplication as a modular
3227 (UNSIGNED-BYTE 32) operation on the x86 backend.
3228 * optimization: SEARCH on simple-base-strings can now be open-coded.
3229 (see also contrib/compiler-extras.lisp for inspiration for
3230 teaching the compiler about the Boyer-Moore algorithm).
3231 * value, returned by MAX (and MIN) called with several EQUALP, but
3232 not EQL, arguments now does not depend on compiler settings.
3233 * fixed some bugs revealed by Paul Dietz' test suite:
3234 ** in stack analysis liveness information is propagated from
3235 non-local entry points.
3236 ** pathwise CAST removing failed when the CAST node did not start
3238 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3239 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3241 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3242 host is already defined.
3243 ** RENAME-FILE works on streams instead of signalling an internal
3245 ** PEEK-CHAR uses the current readtable when determining whether
3246 or not a character is whitespace.
3247 ** MERGE-PATHNAMES handles the case when the pathname does not
3248 specify a name while the default-pathname specifies a version.
3249 ** Pathnames now stand a better chance of respecting print/read
3251 ** Attempting to use standardized file system operators with a
3252 pathname with invalid :DIRECTORY components signals a
3254 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3256 ** DIRECTORY on logical pathnames is more correct.
3257 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3258 signal a TYPE-ERROR if their argument is not a stream.
3259 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3260 designator argument does not designate a stream.
3261 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3262 examining the synonym.
3263 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3265 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3266 element-type larger than ([UN]SIGNED-BYTE 32).
3268 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3269 * When built with the :SB-FUTEX feature, threaded builds now take
3270 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3271 for faster/more reliable mutex and condition variable support.
3272 * Incompatible change (but one you probably shouldn't have been using
3273 anyway): the interface and code for arbitrating between multiple
3274 threads in the same user session has been redesigned.
3275 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3276 error when called without an explicit environment argument.
3277 (thanks to Nikodemus Siivola)
3278 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3279 confused about their position. (thanks to Adam Warner and Gerd
3281 * bug fix: Pretty printing backquoted forms with unquotations in the
3282 argument list position of various code constructs such as LAMBDA
3283 now works correctly. (reported by Paul Dietz)
3284 * bug fix: Pretty printing unquotations no longer loses all
3285 stream position information.
3286 * optimization: performance of string output streams is now less
3287 poor for multiple small sequence writes.
3288 * optimization: performance of CSUBTYPEP in the presence of complex
3289 expressions involving CONS and NOT many times has been improved.
3290 (reported by Paul Dietz)
3291 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3293 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3294 streams for the REPL, for the debugger, and so on. (thanks to
3296 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3297 * fixed some bugs revealed by Paul Dietz' test suite:
3298 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3299 is ignored if it is too small, rather than propagating through
3300 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3301 ** extremely complex negations of CONS types were not being
3302 sufficiently canonicalized, leading to inconsistencies in
3304 ** VALUES tranformer lost derived type.
3306 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3307 * fixed a bootstrapping bug: the build process no longer assumes
3308 that the various BOOLE-related constants have the same value in
3309 host and target lisps. (noted by Paul Dietz' test suite on an
3310 SBCL binary built from CLISP)
3311 * The system can now be dynamically linked on the MIPS platform,
3312 which enables dynamic loading of foreign code from Lisp. (thanks
3313 to Ralf Baechle for discussions on the MIPS ABI)
3314 * The system now records debugging information for its own source
3315 files in a filesystem-position-independent manner, relative to
3316 the "SYS" logical host.
3317 * fixed a compiler bug: MV-LET convertion did not check references
3318 to the "max args" entry point. (reported by Brian Downing)
3319 * tweaked disassembly notes to be less confident about proclaiming
3320 some instruction as an LRA. (thanks to Brian Downing)
3321 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3322 now each have their own history, command character, and other
3323 characteristics. (thanks to David Lichteblau)
3324 * fixed some bugs revealed by Paul Dietz' test suite:
3325 ** compiler failure in compiling LOGAND expressions including a
3327 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3328 shift greater than 32.
3329 ** FLUSH-DEST did not mark blocks for type check regeneration.
3330 ** HANDLER-CASE failed to accept declarations in handler clauses
3331 in some circumstances.
3333 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3334 * New code in contrib/sb-introspect (still probably not entirely
3335 stable yet) provides some support for smart Lisp development
3336 environments like SLIME.
3337 * The conditions signalled for errors occurring when loading .fasl
3338 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3339 in a way which should help ASDF recover gracefully.
3340 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3341 stuff has been cleaned up. If you code contrib/ stuff, this might
3342 affect you, and you can look at contrib/README, contrib/STANDARDS,
3343 and/or the 0.8.4.27 diff to check.
3344 * In full calls the compiler now does not generate checks for declared
3345 argument types for all arguments.
3346 * various threading fixes
3347 ** and some experimental patches which didn't make it into
3348 the main tree for this release, but which are shipped in
3349 contrib/experimental-thread.patch as a possible fix for some
3350 failures (deadlock, spinning...) in GC-intensive multithreaded
3352 * fixed PPC build problem (source code incompatibility of different
3353 library versions): added offsetof() hackery which attempts to divine
3354 where glibc maintainers put uc_mcontext today
3355 * fixed bug 282: compiler does not trust type assertions while passing
3356 arguments to a full call.
3357 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3358 &OPTIONAL VALUES type parameter.
3359 * fix bug 214: algorithm for noting rejected templates is now more
3360 similar to that of template seletion. (also reported by rydis on
3362 * fixed bug 141b: printing backquoted information readably and prettily
3363 inserts a space where necessary.
3364 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3365 (CHARACTER 10) are now reported as errors, rather than propagated
3366 as unknown types. (reported by piso on #lisp)
3367 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3368 with values NIL and :ERROR. (thanks to Milan Zamazal)
3369 * fixed bug 191c: CLOS now does proper keyword argument checking as
3370 described in CLHS 7.6.5 and 7.6.5.1.
3371 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3372 longer raise an error; further, using a list as a for-as-arithmetic
3373 counter now raises a meaningful error.
3374 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3375 now understood by sequence creation functions such as MAKE-SEQUENCE
3377 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3378 a STRING-INPUT-STREAM.
3379 * compiler enhancement: SIGNUM is now better able to derive the type
3381 * type declarations inside WITH-SLOTS are checked. (reported by
3383 * fixed some bugs revealed by Paul Dietz' test suite:
3384 ** incorrect optimization of TRUNCATE for a positive first
3385 argument and negative second.
3386 ** compiler failure in let-convertion during flushing dead code.
3387 ** compiler failure while deriving type of TRUNCATE on an
3388 interval, containing 0.
3389 ** ASH of a negative bignum by a negative bignum count now returns
3391 ** intersection of CONS types now canonicalizes properly, fixing
3392 inconsistencies in SUBTYPEP.
3394 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3395 * incompatible change: The --disable-debugger command line
3396 option now clobbers the debugger at a more fundamental
3397 level, by redefining #'INVOKE-DEBUGGER instead of by
3398 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3399 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3400 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3401 enter the debugger REPL and then suffer recursive errors
3402 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3403 while under the new system, BREAK in --disable-debugger mode
3404 terminates the system just as an unhandled error would.
3405 * fixed compiler performance when processing loops with a step >1;
3406 * bug fix: DOCUMENTATION now retrieves generic function
3407 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3408 support has been systematized, and now supports the methods
3409 specified by ANSI, along with a default method and a method for
3410 slot documentation. (reported by Nathan Froyd)
3411 * bug fix: effective methods associated with a generic function are
3412 no longer cached over a change of that generic function's method
3413 combination. (reported by Andreas Fuchs)
3414 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3415 names a non-existent directory. (thanks to Andreas Fuchs)
3416 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3417 the PPC platform, lead to stack corruption; this has been fixed.
3418 (reported by Rainer Joswig)
3419 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3420 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3421 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3422 arithmetic in general, is now much more reliable on the PPC
3424 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3425 platform now returns the right answer.
3426 * optimization: restored some effective method precomputation in
3427 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3428 precomputation is now tunable.
3429 * optimization: compiler-internal data structure use has been
3430 reviewed, and changes have been made that should improve the
3431 performance of the compiler by about 20%.
3432 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3433 simple-base-strings and simple-bit-vectors is improved.
3434 * optimization: the optimization of 32-bit logical and arithmetic
3435 functions introduced in version 0.8.3 on the x86 has been
3436 implemented on the mips, ppc and sparc platforms; an
3437 implementation of the same facility, but for 64-bit arithmetic,
3438 has been added for the alpha.
3439 * microoptimization: the compiler is better able to make use of the
3440 x86 LEA instruction for multiplication by constants.
3441 * bug fix: in some situations compiler did not report usage of
3442 generic arithmetic in (SPEED 3) policy.
3443 * bug 145b fix: compiler used wrong type specifier while converting
3444 MEMBER-types to numeric.
3445 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3447 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3448 index argument now works properly on the Alpha platform.
3449 * bug fix: floating point exception treatment on the Alpha platform
3451 * bug fix: FILE-POSITION works much better on string input and
3452 output streams. (thanks to Nikodemus Siivola)
3453 * bug fix: many threading/garbage collection symptoms sorted.
3454 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3455 might be pseudo-atomic.
3456 * internal change: Stopping for GC is now done with signals not
3457 ptrace. GC is now done in whichever thread wanted it, instead of
3459 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3461 * bug fix: result form in DO is not contained in the implicit
3463 * incompatible change: ICR structure is changed; the value part of
3464 CONTINUATION is now called LVAR; corresponding functions are
3465 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3466 * added type deriver for ISQRT (thanks to Robert E. Brown).
3467 * bug fix: better support for loading from the command line when an
3468 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3470 * fixed some bugs revealed by Paul Dietz' test suite:
3471 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3472 ** ROUND and FROUND now give the right answer when given very
3473 small float arguments.
3474 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3476 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3477 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3478 ** LCM should return a non-negative integer.
3479 ** PARSE-INTEGER returned the index of a terminator instead of the
3480 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3481 ** PARSE-INTEGER returned an incorrect index being applied to a
3483 ** LCM with two arguments of 0 returns 0 rather than signalling
3485 ** unsigned addition of a 32-bit constant with the high bit set no
3486 longer causes an internal compiler error on the x86.
3487 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3488 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3489 with negative last argument.
3490 ** byte specifiers with zero size and position no longer cause
3491 an error during type derivation.
3492 ** bignum multiplication on the Alpha platform now returns the
3494 * porting: The system now builds on SuSE AMD64, although it still
3495 generates a 32-bit binary.
3496 * .fasl file incompatibility: The fasl file version number has
3497 been incremented (because of the changes to internal compiler
3498 data structures referred to above).
3500 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3501 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3502 more accurately, on the Darwin kernel running on PowerPC hardware.
3503 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3504 * Compiler code deletion notes now signal a condition of type
3505 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3506 an associated MUFFLE-WARNING restart.
3507 * The compiler now performs limited argument count validation of
3508 constant format strings in FORMAT, and where appropriate in ERROR,
3509 CERROR and WARN. (thanks to Gerd Moellmann)
3510 * New ASDF-INSTALL contrib can be used for automatic download and
3511 installation of third-party Lisp code from CCLAN or other sites
3513 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3514 which forces another thread to execute a function supplied by the
3516 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3517 now accept and act upon their :ELEMENT-TYPE keyword argument.
3518 (reported by Martin Atzmueller, Edi Weitz)
3519 * bug fix: FILE-POSITION now accepts position designators up to
3520 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3521 is the greater. (thanks to Patrik Nordebo)
3522 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3523 Kalle Olavi Niemitalo)
3524 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3525 printable. (reported by Eric Marsden)
3526 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3527 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3528 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3529 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3530 (thanks to Stig Sandoe)
3531 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3532 UNREAD-CHAR now correctly includes the unread character in the
3533 target sequence. (thanks to Gerd Moellmann)
3534 * bug fix in threaded builds: the system can now be suspended and
3535 resumed by shell job control with minimal disruption.
3536 * bug fixes in times and timezones >2038AD
3537 * better handling of "where is GNU make?" problem in build scripts
3538 (thanks to Nikodemus Siivola)
3539 * new optimization: inside a named function any reference to a
3540 function with the same name is considered to be a self-reference;
3541 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3542 optimization quality.
3543 * new optimization on x86: logical functions and + now have
3544 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3545 used when the result is truncated to 32 bits.
3546 * VALUES declaration is partially enabled.
3547 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3548 * fixed some bugs revealed by Paul Dietz' test suite:
3549 ** The system now obeys the constraint imposed by
3550 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3551 types form a lattice under type intersection.
3552 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3553 ** ASSOC now ignores NIL elements in an alist.
3554 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3555 and (1+ MOST-POSITIVE-FIXNUM) answers.
3556 ** The addition of a method with invalid qualifiers to a generic
3557 function does not cause an error to be signalled immediately;
3558 a warning is signalled, and the error is generated only on
3559 calling the generic function.
3560 * changed .fasl file version number, in part to add type codes for
3561 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3562 obscure ANSI requirements
3564 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3565 * fixed bug 148: failure to inline-expand a local function left
3566 garbage, confusing the compiler.
3567 * fixed bugs 3cd: structure slot readers perform type check if the
3568 slot can have an invalid value (i.e. it is either not initialized
3569 or can be written with a less specific slot writer).
3570 * bug fix: the compiler now traps array references to elements off
3571 the end of an array; previously, the bounds checking in some
3572 circumstances could go off-by-one.
3573 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3575 * type declarations for array element types now obey the description
3576 on the CLHS page "Declaration TYPE", as per discussions on
3577 sbcl-help around 2003-05-08. This means that a declaration
3578 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3579 declaration, all references to BAR will be asserted or assumed
3580 (with THE, so dependent on compiler policy) to involve objects of
3581 type FOO. Note that no such declaration is implied in
3582 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3583 * declared types of functions from the "Conditions"
3584 chapter. (reported by Paul Dietz)
3585 * bug fix: CERROR accepts a function as its first argument.
3586 * bug fix: NTH an NTHCDR accept a bignum as index
3587 arguments. (reported by Adam Warner)
3588 * optimization: character compare routines now optimize comparing
3589 against a constant character. (reported by Gilbert Baumann)
3590 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3591 argument now works properly on the MIPS platform.
3592 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3593 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3594 declared to have a functional type, does not cause an error
3596 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3597 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3598 * fixed bug: if last continuation of a deleted block has a
3599 destination, this destination should be deleted too. (reported by
3601 * fixed a bug in the bootstrap process: the host compiler's values
3602 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3603 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3604 test case from Patrik Nordebo)
3605 * improved the ability of the disassembler on the PPC platform to
3606 provide helpful disassembly notes.
3607 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3608 the class in more cases than previously.
3609 * bug fix: FILE-POSITION now understands :START and :END for
3610 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3611 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3612 8) data. (thanks to David Lichteblau)
3613 * bug fix: it is possible to add a method to a generic function
3614 without lambda list.
3615 * bug fix: reader failed to signal END-OF-FILE inside an
3616 object representation. (reported by Nikodemus Siivola)
3617 * fixed some bugs revealed by Paul Dietz' test suite:
3618 ** LAST and [N]BUTLAST should accept a bignum.
3619 ** condition slot accessors are methods.
3620 ** (VECTOR NIL) is a subtype of STRING.
3622 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3623 * minor incompatible change: some nonsensical specialized lambda
3624 lists (used in DEFMETHOD) which were previously ignored now signal
3626 * minor incompatible change: the system is now aware of the types of
3627 variables in the COMMON-LISP package, and will signal errors for
3628 most violations of these type constraints (where previously they
3629 were silently accepted).
3630 * minor incompatible change: COMPILE-FILE now uses the freedom
3631 afforded (ANSI 3.2.2.3) to use derived function types for
3632 functions defined in the same file. This also permits the system
3633 to warn on static type mismatches and function
3634 redefinition. (Currently it does not work with high DEBUG level.)
3635 * minor incompatible change: VALUES declaration is disabled.
3636 * When issuing notes, the compiler now signals a condition of type
3637 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3638 restart for use in user handlers. It is expected that the
3639 COMPILER-NOTE condition will eventually become a condition
3640 supertype to a hierarchy of note types, which will then be
3641 handleable in a similar fashion. However, at the moment, no such
3642 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3643 but it's an implementation detail, not a classification for the
3645 * Changes in type checking closed the following bugs:
3646 ** type checking of unused values (192b, 194d, 203);
3647 ** template selection based on unsafe type assertions (192c, 236);
3648 ** type checking in branches (194bc).
3649 * A short form of VALUES type specifier has ANSI meaning (it has
3650 increased the number of situations when SBCL cannot perform type
3652 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3653 keywords or constants is permissible.
3654 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3655 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3656 classes. (thanks to Antonio Martinez)
3657 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3658 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3659 argument to a pretty-print stream. Instead, it leaves any such
3660 operations to DESCRIBE-OBJECT methods.
3661 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3663 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3664 lambda lists. (thanks to David Lichteblau)
3665 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3666 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3667 (thanks to Gerd Moellmann)
3668 * fixed bug 246: increased compilation speed of long
3669 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3671 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3673 * DEFCONSTANT now throws a condition of type
3674 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3675 constant to a non-EQL value; CONTINUE and ABORT restarts
3676 respectively change and preserve the value.
3677 * fixed bug 63: The code walker, part of the implementation of CLOS,
3678 is now better at handling symbol macros.
3679 * bug fix: There is no longer an internal implementation type named
3680 CL:LENGTH. (reported by Raymond Toy)
3681 * bug fix: In macro-like defining macros/special operators the
3682 implicit block does not enclose the lambda list.
3683 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3684 enclosed in parentheses) are not suitable as type specifiers, and
3685 their use properly signals an error now.
3686 * bug fix: An argument count mismatch for a type specifier in code
3687 being compiled no longer causes an unhandled error at compile
3688 time, but signals a compile-time warning.
3689 * fixed simple vector readable printing
3690 * bug fix: DESCRIBE takes more care over whether the class
3691 precedence list slot of a class is bound before accessing it.
3692 (reported by Markus Krummenacker)
3693 * bug fix: FORMATTER can successfully compile pretty-printer format
3694 strings which use variants of the ~* directive inside.
3695 * bug fix: SEARCH now applies its TEST predicate to the elements of
3696 the arguments in the correct order. (thanks to Wolfhard Buss)
3697 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3698 in type inference. (thanks to Robert E. Brown)
3699 * bug fix: Reading in symbols with an explicit package name of ""
3700 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3701 * fixed some bugs revealed by Paul Dietz' test suite:
3702 ** NIL is now allowed as a structure slot name.
3703 ** Arbitrary numbers, not just REALs, are allowed in certain
3704 circumstances in LOOP for-as-arithmetic clauses.
3705 ** Multiple class redefinitions before slot access no longer
3706 causes a type error.
3707 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3708 association between the name and a class.
3709 ** Generic functions with non-standard method-combination and over
3710 five methods all of which return constants no longer return NIL
3711 after the first few invocations. (thanks to Gerd Moellmann)
3712 ** CALL-NEXT-METHOD with no arguments now passes the original
3713 values of the arguments, even in the presence of assignment.
3714 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3716 ** DISASSEMBLE works with closures and funcallable instances.
3717 ** ADD-METHOD now returns the generic function, not the new method.
3718 ** FIND-METHOD signals an error if the lengths of the specializers
3719 is incompatible with the generic function, even if the ERRORP
3721 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3722 which its argument is a member.
3723 ** DEFCLASS only redefines the class named by its class-name
3724 argument if that name is the proper name of the class;
3725 otherwise, it creates a new class.
3726 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3727 of the UNBOUND-SLOT condition to the name of the slot.
3728 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3729 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3730 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3731 SLOT-MAKUNBOUND in the specified fashion.
3733 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3734 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3735 cross-compilation host. As a consequence, we can now bootstrap our
3736 way up to SBCL starting with a bare gcc toolchain and human-readable
3737 source code (first the source to CLISP, then the source to SBCL).
3738 * A contributed module containing a partial implementation of the
3739 simple-streams interface has been included. (thanks to Rudi
3741 * A contributed module implementing the RFC1321 Message Digest
3742 Algorithm, known as MD5, has been included.
3743 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3744 no longer has any effect, as the code controlled by this feature
3745 has been deleted. (As far as we know, no-one has ever built using
3746 this feature, and its semantics were confused in any case).
3747 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3748 work on conditions (as required by the ANSI specification),
3749 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3750 expected behaviour on conditions. Users should note, however,
3751 that such behaviour is not required by the ANSI specification,
3752 and so use of this behaviour may render their code unportable.
3753 * fixed some bugs revealed by Paul Dietz' test suite:
3754 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3756 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3757 ** SXHASH on condition objects no longer returns NIL.
3758 ** :ALLOCATION :CLASS slots are better treated; their values are
3759 updated on class redefinition, and initforms inherited from
3760 superclasses are applied.
3761 ** REMOVE-METHOD returns its generic function argument even when
3762 no method was removed.
3763 ** SHARED-INITIALIZE now initializes the values of the requested
3764 slots, including those with :ALLOCATION :CLASS.
3765 ** ALLOCATE-INSTANCE now works on structure classes defined via
3766 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3768 ** SLOT-EXISTS-P now works on conditions, as well as structures
3770 ** MAKE-LOAD-FORM now has the required methods on
3771 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3772 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3773 keyword, and now implements the SLOT-NAMES argument.
3774 ** methods with &OPTIONAL arguments no longer allow too many
3775 arguments to be passed in the call without error.
3776 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3777 option is consistent with the required arguments of the generic
3778 function lambda list.
3779 * bug fix: REQUIRE accepts a string designator. (Thanks to
3781 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3782 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3783 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3784 * bug fix: The compiler now checks for duplicated variables in macro
3786 * bug fix: SETQ on globals returns the correct value.
3787 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3788 not a COMPILER-ERROR (followed by some other strange error on
3789 choosing the CONTINUE restart).
3790 * bug fix: make.sh and friends are now more consistent in the way that
3791 they look for GNU "make".
3793 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3794 * experimental native threads support (on x86 Linux >=2.4 only).
3795 This is not compiled in by default: you need to add :SB-THREAD to
3796 the target features. See the "Beyond ANSI" chapter of the manual
3798 * fix for longstanding nonANSIism: The old distinction between
3799 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3800 The return value from CL:FIND-CLASS is now a CLOS class, and
3801 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3802 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3803 * An interface to the MetaObject Protocol, as described in Kiczales,
3804 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3805 MIT Press, 1991, is available from the SB-MOP package.
3806 * incompatible change: the SB-PCL package should now be considered
3807 a private implementation detail, and no longer a semi-private MOP
3809 * minor incompatible change: due to rearrangement for threads, the
3810 control stack and binding stack are now allocated at arbitrary
3811 addresses instead of being hardcoded per-port. Users affected by
3812 this probably have to be doing advanced things with shared
3813 libraries, and will know who they are.
3814 * minor incompatible change: Previously, all --eval forms used were
3815 processed with READ before any of them were processed with EVAL.
3816 Now each --eval form is processed with both READ and EVAL before
3817 the next --eval form is processed. (Thus package operations like
3818 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3819 work as the user might reasonably expect.)
3820 * minor incompatible change: *STANDARD-INPUT* is now only an
3821 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3823 * minor incompatible change: Y-OR-N-P is now character-oriented,
3824 not line oriented. Also, YES-OR-NO-P now works without errors.
3825 (thanks to Antonio Martinez)
3826 * sb-aclrepl module improvements: an integrated inspector, added
3827 repl features, and a bug fix to :trace command.
3828 * Known functions, which cannot be open coded by the backend, are
3829 considered to be able to check types of their arguments. (fixing
3830 a bug report by Nathan J. Froyd)
3831 * fixed a bug in computing method discriminating functions: It is
3832 now possible to define methods specialized on classes which have
3833 forward-referenced superclasses. (thanks to Gerd Moellmann)
3834 * fixed evaluation order in optional entries (reported by Gilbert
3836 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3837 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3838 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3839 required-by-AMOP NAME argument, as well as CLASS and
3840 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3841 * fixed bug 20: DEFMETHOD can define methods using names that are
3842 not the proper names of classes to designate class specializers.
3843 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3844 Unix file descriptors, instead of blowing up. (thanks to Antonio
3846 * Garbage collection refactoring: user-visible change is that a
3847 call to the GC function during WITHOUT-GCING will not do garbage
3848 collection until the end of the WITHOUT-GCING. If you were doing
3849 this you were probably losing anyway.
3850 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3851 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3852 TYPEP the latter but not the former.
3853 * The compiler issues a full WARNING for calls to undefined functions
3854 with names from the CL package.
3855 * MAP-INTO for a vector destination is open coded. (reported by
3856 Brian Downing on c.l.l)
3857 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3858 documentation string.
3859 * fixed some bugs revealed by Paul Dietz' test suite:
3860 ** COPY-ALIST now signals an error if its argument is a dotted
3862 ** Condition slots are now accessed more correctly in the presence
3863 of multiple initargs for a given slot.
3864 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3865 exclude restarts of the same name associated with a different
3867 ** DEFCLASS of forward-referenced classes with another
3868 forward-referenced class in the superclasses list no longer
3870 ** Condition slots are now initialized once each, not multiple
3871 times. (thanks to Gerd Moellmann)
3872 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3874 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3875 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3876 arguments contain duplicated elements.
3877 ** RESTART-CASE understands local macros.
3878 ** RESTART-CASE associates exactly its own restarts with a condition.
3879 ** ENDP in safe mode checks its argument to be of type LIST.
3880 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3881 in question is unbound.
3882 ** Optimized MAKE-INSTANCE functions no longer cause internal
3883 assertion failures in the presence of duplicate initargs.
3884 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3885 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3886 propagated even in the ctor optimized implementation of
3888 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3890 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3891 * a better implementation of SXHASH on (simple) bit vectors,
3892 measured both in execution speed and in distribution of results
3893 over the positive fixnums, has been installed. Likewise, a better
3894 implementation of EQUAL for simple bit vectors is now available.
3895 * fixed CEILING optimization for a divisor of form 2^k.
3896 * fixed bug 240 (emitting extra style warnings "using the lexical
3897 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3898 by Antonio Martinez)
3899 * fixed SXHASH, giving different results for NIL depending on type
3900 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3901 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3902 variable. (found by Rolf Wester)
3903 * SBCL does not ignore type declarations for special
3904 variables. (reported by rif on c.l.l 2003-03-05)
3905 * some bug fixes in contrib/sb-aclrepl/
3906 * fixed some bugs revealed by Paul Dietz' test suite:
3907 ** a bug in the CONS type specifier, whereby the CAR and CDR
3908 types got intertwined, has been fixed;
3909 ** the type system is now able to reason about the interaction
3910 between INTEGER and RATIO types more completely;
3911 ** APPEND, [N]REVERSE and NRECONC check that those their
3912 arguments, which must be proper lists, are really so;
3913 ** An array specialized to be unable to hold elements has been
3914 implemented, as required -- yes, really -- by ANSI;
3915 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3916 on malformed property lists;
3918 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3919 * incompatible packaging change: in line with Unix convention,
3920 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3921 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3922 * REQUIRE and PROVIDE are now optionally capable of doing something
3923 useful. See the documentation string for REQUIRE.
3924 * infrastructure for a managed SBCL contrib system: contributed
3925 modules in this release include:
3926 ** the ASDF system definition facility;
3927 ** an interface to the BSD Sockets API;
3928 ** an ACL-like convenience interface to the repl;
3929 (thanks to Kevin Rosenberg)
3930 ** an implementation of ROTATE-BYTE, with efficient implementation
3932 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3933 gives a closer approximation to the right answer than previously.
3934 (thanks to Raymond Toy)
3935 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3936 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3937 optional environment arguments, as required by ANSI.
3938 * fixed bugs in other functions taking environment objects, allowing
3939 calls with an explicit NIL environment argument to be compiled
3941 * fixed bug 228: primary return values from
3942 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3943 COMPILE or FUNCTION.
3944 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3945 :TYPE will no longer signal errors on innocuous objects.
3946 * fixed bug 231b: SETQ is better at respecting type declarations in
3947 the lexical environment.
3948 * fixed a bug in DEFCLASS: classes named by symbols with no or
3949 unprintable packages can now be defined.
3950 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3951 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3952 * fixed compiler failure related to checking types of functions.
3953 (reported by Robert E. Brown)
3954 * the compiler is now much more consistent in its error-checking
3955 treatment of bounding index arguments to sequence functions: in
3956 (SAFETY 3) code, errors will be signalled in almost all cases if
3957 invalid sequence bounding indices are passed to functions defined
3958 by ANSI to operate on sequences.
3959 * fixed a bug in the build procedure: documentation of SBCL-specific
3960 packages is now preserved and available in the final Lisp image.
3961 * lifted FDEFINITION lookup out of loops in the implementation of
3962 many list operations. (thanks to Robert E. Brown)
3963 * fixed a bug in the reader: the #n# reader macro now works for
3964 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3965 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3966 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3967 E. Brown and Tony Martinez respectively)
3968 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3969 SLOT-BOUNDP: in particular, it is now able to optimize them much
3970 better, and is now not vulnerable to having packages renamed.
3971 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3972 always called when a slot is not present in an instance. (thanks
3974 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3975 longer signal an error on unreading a character following EOF on
3976 the previous constituent stream. (thanks to Tony Martinez)
3977 * fixed some bugs revealed by Paul Dietz' test suite:
3978 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3979 not just nonnegative fixnums;
3980 ** the logical bit-array operators such as BIT-AND now accept an
3981 explicit NIL for their "opt-arg" argument (to indicate a
3982 freshly-consed result bit-array);
3983 ** ELT now signals an error on an invalid sequence index in safe
3985 ** the type system is now cleverer about negations of numeric
3986 types, and consequently understands the BIGNUM and RATIO types
3988 ** the type system is now cleverer about the interaction between
3989 INTEGER and RATIO types: while bugs still remain, many more
3990 cases are accurately computed;
3991 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3992 if it is in the last clause;
3993 ** CONSTANTLY now correctly returns a side-effect-free function in
3995 ** DECLARE is no longer treated as a special-operator; in
3996 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3997 * incremented fasl file version number due to the change in the
3998 DEFSTRUCT-SLOT-DESCRIPTION structure.
4000 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
4001 * minor incompatible change: code processed by the "interpreter" or
4002 EVAL now has a compilation optimization policy of (DEBUG 2)
4003 (changed from (DEBUG 1)) to improve debuggability of interactive
4004 development, and to allow the use of the debug RETURN command in
4006 * an experimental implementation of the RETURN command for the
4007 debugger has been included. (thanks to Frederik Kuivinen)
4008 * fixed bug 62: constraints were not propagated into a loop.
4009 * fixed bug in embedded calls of SORT (reported and investigated by
4011 * fixed some bugs revealed by Paul Dietz' test suite:
4012 ** printing and reading of arrays with some dimensions having
4013 length 0 (thanks to Gerd Moellmann);
4014 ** BOA constructor with &AUX argument without a default value does
4015 not cause a type error;
4016 ** CONSTANTP now returns true for all self-evaluating objects.
4018 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
4019 * fixed bug 127: DEFSTRUCT now does not clobber old structure
4020 accessors that are related by inheritance, as specified in the
4021 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
4022 Valtteri Vuorikoski)
4023 * The compiler is now able to inline functions that were defined in
4024 a complex lexical environment (e.g. inside a MACROLET).
4025 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
4027 * Support for the upcoming FreeBSD-5.0 release has been included.
4028 (thanks to Dag-Erling Smorgrav)
4029 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
4030 effect when it is not in a toplevel context.
4031 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
4032 stand a better chance of being correct. (thanks to Gerd
4034 * fixed bug in COERCE, which now signals an error on coercing a
4035 rational to a bounded real type which excludes the expected
4037 * The compiler is now able to derive types more accurately from the
4038 COERCE and COMPILE functions.
4039 * fixed bug 223: functional binding is considered to be constant
4040 only for symbols in the CL package.
4041 * fixed bug 231: SETQ did not check the type of a variable being set
4042 (reported by Robert E. Brown)
4043 * A new optimization for MAKE-INSTANCE has been included, fixing
4044 various bugs (including relating to :ALLOCATION :CLASS slots and
4045 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
4047 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
4048 ** As required by ANSI, LOOP now disallows anonymous collection
4049 clauses such as COLLECT I in conjunction with aggregate boolean
4050 clauses such as THEREIS (= I 1);
4051 ** LOOP now signals an error when any variable is reused in the
4052 same loop (including the potentially useful construct analogous
4053 to WITH A = 1 WITH A = (1+ A);
4054 ** IT is only a special loop symbol within the first clause of a
4055 conditional loop clause;
4056 ** LOOP with a typed iteration variable over a hashtable now
4057 signals a type error iff it should.
4058 * fixed some other bugs revealed by Paul Dietz' test suite:
4059 ** FILE-STREAM now names the class previously known as FD-STREAM;
4060 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
4061 argument) no longer signals an error;
4062 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
4063 of the slot symbol, rather than using the current package
4064 ((:CONC-NAME "") continues to intern the slot's name in the
4066 * incremented fasl file version number, because of the incompatible
4067 change to the DEFSTRUCT-DESCRIPTION structure, and again because
4068 of the new implementation of DEFINE-COMPILER-MACRO.
4070 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
4071 * Support for building SBCL for MIPS platforms running in
4072 little-endian mode has now been checked in, and basic
4073 functionality on said platforms verified.
4074 * minor incompatible change: PCL now records the pathname of a file
4075 in which methods and the like are defined, rather than its
4077 * minor incompatible change: TRUENAME now considers the truename of
4078 a file naming a directory to be the pathname with :DIRECTORY
4079 component indicating that directory.
4080 * minor incompatible change: a NAMED clause in the extended form of
4081 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
4082 reason for the previous behaviour is unclear.
4083 * more systematization and improvement of CLOS and MOP conformance
4084 in PCL (thanks to Gerd Moellman and Pierre Mai):
4085 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
4087 ** DEFINE-METHOD-COMBINATION no longer signals an error for
4088 primary methods with no specializers;
4089 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
4091 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
4092 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
4093 FORWARD-REFERENCED-CLASSes; error reporting on
4094 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
4096 ** SXHASH on CLOS instances now uses a slot internal to the
4097 instance to return different numbers on distinct instances,
4098 while preserving the same return value through invocations of
4100 ** DEFMETHOD signals errors when methods with longer incongruent
4101 lambda lists are added to generic functions;
4102 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
4103 CLASS, as specified in AMOP;
4104 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
4105 on the order returned by the primary method for classes of
4106 class STANDARD-CLASS;
4107 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
4108 * fixed some bugs shown by Paul Dietz' test suite:
4109 ** DOLIST puts its body in TAGBODY;
4110 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
4112 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
4113 value producing form;
4114 ** if more variables are given to PROGV than values, extra
4115 variables are bound and made to have no value;
4116 ** NSUBSTITUTE on list arguments gets the right answer with
4118 ** ELT signals an error of type TYPE-ERROR when the index argument
4119 is not a valid sequence index;
4120 ** LOOP signals (at macroexpansion time) an error of type
4121 PROGRAM-ERROR when duplicate variable names are found;
4122 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
4123 ** FUNCALL of special-operators now cause an error of type
4125 ** PSETQ now works as required in the presence of side-effecting
4126 symbol-macro places;
4127 ** NCONC accepts any object as its last argument;
4128 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
4130 ** loop-for-as-package does not require a package to be explicitely
4132 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
4134 * fixed bug 166: compiler preserves "there is a way to go"
4135 invariant when deleting code.
4136 * fixed bug 172: macro lambda lists with required arguments after
4137 &REST arguments now cause an error to be signalled. (thanks to
4139 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
4140 bug. (thanks to Matthew Danish)
4141 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
4143 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
4144 arguments are lexically rebound. (thanks to Gerd Moellmann and
4146 * fixed bug 194: error messages are now more informative when there
4147 is no primary method applicable in a call to a generic
4148 function. (thanks to Gerd Moellmann)
4149 * fixed bug in command line argument checking (thanks to Julian
4151 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
4152 (thanks to Matthew Danish)
4153 * incremented fasl file version number, because of the
4154 SXHASH-related changes in the layout of CLOS data structures
4156 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4157 * minor incompatible change: The runtime (the Unix executable named
4158 "sbcl") is now much pickier about the .core files it will load.
4159 Essentially it now requires .core files to descend from the same
4160 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4161 as the runtime does. (The intent is to prevent the crashes which
4162 can occur, and which can even be reported as mysterious failures,
4163 when people patch the sources or change the build parameters
4164 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4165 sbcl and .core files.)
4166 * fixed bug: VALUES-LIST is no longer optimized away.
4167 * fixed bug 142: The FFI conversion of C string values to Lisp
4168 string values no longer conses excessively. (thanks to Nathan
4169 Froyd porting Raymond Toy's fix to CMU CL)
4170 * began to systematize and improve MOP conformance in PCL (thanks to
4171 Nathan Froyd, Gerd Moellman and Pierre Mai):
4172 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4174 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4175 ** FINALIZE-INHERITANCE is now called on class finalization;
4176 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4177 argument precedence order.
4178 * fixed bug 202: The compiler no longer fails on functions whose
4179 derived types contradict their declared type.
4180 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4181 so it can be non-toplevel.
4182 * The fasl file version number has changed (because of the new
4183 implementation of DEFMACRO).
4184 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4185 safe code, that any length requirement by their type-specifier
4186 argument is valid. The exceptions to this are described in bug
4188 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4189 functions now check the types of their inputs as required by ANSI.
4190 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4191 introduced symbol is DECLAREd to be SPECIAL.
4192 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4193 * fixed bug: :COUNT argument to sequence functions may be negative
4194 * fixed bug: body of DO-SYMBOLS may contain declarations
4195 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4196 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4197 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4198 symbol macro only once
4199 * fixed printing of call frame when argument list is unavailable
4200 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4201 * compiler no longer signals WARNING on unknown keyword
4204 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4205 * A beta-quality port to the mips architecture running Linux,
4206 based on the old CMUCL backend, has been made. It has been tested
4207 on a big-endian kernel, and works sufficiently well to be able to
4208 rebuild itself; it has not been tested in little-endian mode.
4209 * fixed an inconsistency between gencgc.c and purify.c which made
4210 dumping/loading .core files unreliable
4211 * fixed bug 120a: The compiler now deals correctly with IFs where
4212 the consequent is the same as the alternative, instead of
4213 misderiving the return type. (thanks to Alexey Dejneka)
4214 * fixed bug 113: Logical pathnames are now dumpable (the logical
4215 host is resolved at load-time, throwing an error if it is not
4217 * fixed bug 174: FORMAT's error message is slightly clearer when a
4218 non-printing character is used in a format directive.
4219 * fixed several bugs in compiler checking of type declarations, i.e.
4220 violations of the Python "declarations are assertions" principle
4221 (thanks to Alexey Dejneka)
4222 * fixed several bugs in PCL's error checking (thanks to Gerd
4224 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4226 * fixed bug in compilation of functions as first class values
4227 (thanks to Antonio Martinez-Shotton)
4228 * The compiler's handling TYPE-ERRORs which it can prove will
4229 inevitably happen at runtime has been cleaned up and corrected
4230 in several ways. (thanks to Alexey Dejneka)
4231 * improved argument type checking for various basic arithmetic
4232 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4233 lavished on them in the past that they can be compiled in many
4234 ways in different special cases
4235 * fixed bug 181: compiler checks validity of user supplied type
4237 * cleaned up code flushing in optimization: Function calls which
4238 should signal errors for safety purposes (e.g. which ANSI says
4239 should signal errors when their arguments are of incorrect type)
4240 are no longer optimized away.
4241 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4242 * incremented fasl file version number, because changes in the
4243 implementation of sequence functions like COERCE caused
4244 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4245 in old inline expansions) to become undefined. (Actually these
4246 changes were later undone, so we might very well be binary
4247 compatible with 0.7.7 after all, but leaving the version number
4248 incremented seemed like the simplest and most conservative
4251 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4252 * An alpha-quality port to the parisc architecture running Linux,
4253 based on the old CMUCL backend, has been made. This, even more so
4254 than the other backends, should be considered still a work in
4255 progress; known problems include that the Linux kernel in 64-bit
4256 mode does not propagate the correct sigcontext structure to
4257 userspace, and consequently SBCL on a parisc64 kernel will not
4259 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4260 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4261 suppressed.) Also now that the compiler is looking at declarations
4262 in the environment, it checks optimization declarations as well,
4263 and suppresses inlining when (> DEBUG SPEED).
4264 * More fixes have been made to treatment of floating point exception
4265 treatment and other Unix signals. In particular, floating point
4266 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4267 * The detection and handling of control stack exhaustion (infinite
4268 or very deeply nested recursion) has changed. Stack exhaustion
4269 detection is now done by write-protecting pages at the OS level
4270 and applies at all optimization settings; when found, a
4271 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4272 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4273 be caught using IGNORE-ERRORS.
4274 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4275 correctly when transforming recursive LABELS functions to LETs.
4276 (thanks to Alexey Dejneka)
4277 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4278 that are names of constants or global variables.
4279 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4280 alien routines with docstrings.
4281 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4282 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4284 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4285 (thanks to Eric Marsden)
4286 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4287 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4288 * Bug fix: The SPARC backend can now compile functions involving
4289 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4290 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4291 to a functions expecting a PATHNAME-DESIGNATOR.
4292 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4293 lambda lists. (thanks to Alexey Dejneka)
4294 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4295 functions so that EXPORTing the name of the class doesn't cause
4296 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4297 to Antonio Martinez for reporting this)
4298 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4299 constant-folded without failing an assertion. (thanks to Einar
4300 Floystad Dorum for reporting this)
4301 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4302 types are now dealt with more correctly by the compiler.
4303 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4304 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4305 behaviour analogous to RENAME-FILE. This puts its behaviour more
4306 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4308 * The fasl file version number has changed again. (because of the
4309 bug fix involving the names of PCL MAKE-INSTANCE functions)
4311 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4312 * bug fix: Floating point exceptions are treated much more
4313 consistently on the x86/Linux and PPC/Linux platforms.
4314 * Array initialization with :INITIAL-ELEMENT is now much faster for
4315 cases when the compiler cannot open code the array creation, but
4316 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4317 array accesses have also seen a speed increase.
4318 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4319 specified. (This was at the root of some bad interactions between
4320 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4322 * bug fix: Internal error arguments for undefined functions are now
4323 computed correctly on the PPC/Linux platform.
4324 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4325 Raymond Toy's patch for CMU CL)
4326 * Support for the Solaris 9 operating environment has been included
4327 (thanks to Daniel Merritt)
4328 * A very ugly but hopefully complete draft of the missing FFI chapter
4329 of the manual has been created by reformatting the corresponding
4330 CMU CL manual chapter into (currently very ugly and incoherent)
4331 DocBook and bringing it up to date for SBCL behavior. Thus, the
4332 manual is now essentially complete, at least by my extreme
4333 once-and-only-once standards, whereby it's acceptable to refer to
4334 the doc strings of SB-EXT functions as the primary documentation.
4335 * The fasl file version number has changed again, due to cleanup of
4336 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4337 is no longer a static symbol.)
4339 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4340 * SBCL now builds with OpenMCL (version 0.12) as the
4341 cross-compilation host; also, more progress has been made toward
4342 bootstrapping under CLISP.
4343 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4345 * bug 158 fixed: The compiler can now deal with integer loop
4346 increments different from 1; fixing this turned out also to fix
4348 * bug 169 fixed: no more bogus warnings about using lexical bindings
4349 despite the presence of perfectly good SPECIAL declarations (thanks
4350 to David Lichteblau)
4351 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4352 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4353 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4355 * bug 180 fixed: Method combination specifications no longer ignore
4356 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4357 * bug fix: Structure type predicate functions now check their argument
4358 count as they should.
4359 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4360 correctly. (thanks to Pierre Mai)
4361 * minor incompatible change: The --noprogrammer option is deprecated
4362 in favor of the new --disable-debugger option, which is very similar.
4363 (The major difference is that it takes effect at a slightly different
4364 time at startup, causing handling of errors in --sysinit and
4365 --userinit files will be affected differently.) The
4366 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4367 been added to allow this functionality to be controlled from ordinary
4368 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4369 maintainers, who might want to run non-interactive scripts to
4370 build SBCL cores which will later be used interactively.)
4371 * minor incompatible change: The LOAD function no longer, when given
4372 a wild pathname to load, loads all files matching that pathname.
4373 Instead, an error of type FILE-ERROR is signalled.
4375 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4376 * bug 147 fixed: The compiler preserves its block link/count
4377 invariants more correctly now so that it doesn't crash. (thanks
4379 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4381 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4382 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4383 no-value-for-this-character as NIL (as specified by ANSI).
4384 * HOST-NAMESTRING on physical pathnames now returns a string that is
4385 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4386 (thanks to Christophe Rhodes)
4387 * The Alpha port handles icache flushing more correctly. (thanks to
4389 * More progress has been made toward bootstrapping under CLISP. (thanks
4390 to Christophe Rhodes)
4391 * The fasl file format has changed again, because dynamic loading
4392 on OpenBSD (which has non-ELF object files) motivated some cleanups
4393 in the way that foreign symbols are transformed and passed around.
4394 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4395 is no longer treated as whitespace by the reader, but instead as
4396 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4397 |A
\7fB|, instead of A as it used to.
4399 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4400 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4401 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4402 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4403 this as of 0.7.1.45, but was left out of the previous news section
4404 (thanks to Dan Barlow)
4405 * SBCL now runs on the Solaris operating system on SPARC architectures
4406 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4407 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4408 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4409 * SPARC backend cleanups, allowing builds of cores optimized for V8
4410 and V9 SPARCS, and also emission of code targeted to a particular
4411 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4413 * SBCL is closer to bootstrapping under CLISP, thanks to various
4414 fixes by Christophe Rhodes.
4415 * The fasl file format has changed again, to allow the compiler's
4416 INFO database to support symbol macros.
4417 * The user manual (in doc/) is formatted into HTML more nicely.
4418 (thanks to coreythomas)
4419 * The system is smarter about SUBTYPEP relationships, especially
4420 those involving NOT types (including types such as ATOM which are
4421 represented internally using NOT types). Thus SUBTYPEP is less
4422 likely to return (VALUES NIL NIL) in general, and in particular
4423 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4425 * The fasl file format has changed again, because the internal
4426 representation of types now includes a new slot to support the new
4427 SUBTYPEP-of-NOT-types logic.
4428 * (not a change in the main branch of SBCL, but a related prototype
4429 which can hopefully be merged into the main branch of SBCL in the
4430 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4431 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4432 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4433 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4434 dispatch macro character combinations. (thanks to Alexey Dejneka)
4435 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4436 that superficially look like logical namestrings correctly.
4437 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4439 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4440 default supplied-p) for &optional and &key arguments. (thanks to
4443 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4444 * incompatible change: The compiler is now less aggressive about
4445 tail call optimization, doing it only when (> SPACE DEBUG) or
4446 (> SPEED DEBUG). (This is an incompatible change because there are
4447 programs which relied on the old CMU-CL-style behavior to optimize
4448 away their unbounded recursion which will now die of stack overflow.)
4449 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4450 for non-GENCGC systems has been increased to 20M (since that
4451 seems much closer to the likely performance optimum for modern
4452 systems than the old 4M value was)
4453 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4454 and *DEBUG-PRINT-LEVEL*
4455 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4456 of CMU CL's support for SPARC, and various endianness and other
4457 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4458 * new syntactic sugar for the Unix command line: --load foo.bar is now
4459 an alternate notation for --eval '(load "foo.bar")'.
4461 ** The system now detects stack overflow and handles it gracefully,
4462 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4463 optimization settings. (This is a good thing in general, and
4464 its introduction in this version should be particularly timely
4465 for anyone whose code fails because of suppression of tail
4467 ** The system now hunts for the C variable "environ" in a more
4468 devious way, to avoid segfaults when the C library version differs
4469 between compile time and run time. (thanks to Christophe Rhodes)
4470 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4471 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4472 ** The compiler no longer issues bogus style warnings for undefined
4473 classes in the same source file as the DEFCLASSes which defined
4474 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4476 ** fixes in CONDITION class precedence list for undefined function
4477 errors (thanks to Alexei Dejneka)
4478 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4479 correctly. (thanks to Dan Barlow)
4480 ** portability fixes aiming at bootstrapping under CLISP (thanks
4481 to Dave McDonald and Christophe Rhodes)
4482 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4483 ** fixes in type translation and and type inference (thanks to
4485 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4486 ** various fixes in the new ports (thanks to Dan Barlow)
4487 * several changes related to debugging:
4488 ** suppression of tail recursion, as noted above
4489 ** stack overflow detection, as noted above
4490 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4491 is now the default. (For some time encapsulation has been more
4492 reliable than the breakpoint-based :ENCAPSULATE NIL
4493 implementation, at least on X86 systems; and I just noticed that
4494 encapsulation also seems closer to the spirit of the ANSI
4497 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4499 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4500 up properly again. (There was a packaging bug in 0.7.0 which
4501 left their definitions in SB-SYS::LOAD-FOREIGN and
4502 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4503 vital for most things which interface to C-level interfaces,
4504 like extensions working with sockets or databases or
4505 Perl-compatible regexes or whatever, and the need to fix
4506 this bug is the main reason that 0.7.1 was released so
4508 ** DEFGENERIC is now choosier about the methods it redefines, so that
4509 reLOADing a previously-LOADed file containing DEFGENERICs does
4510 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4511 cycle now works as it should. (thanks to Alexey Dejneka)
4512 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4513 fixed by Christophe Rhodes. (He actually submitted this patch
4514 months ago, and I delayed until after 0.7.0.)
4515 ** Bug 111 (internal compiler confusion about runtime checks on
4516 FUNCTION types) was fixed by Alexey Dejneka.
4517 * Some internal cleanups (getting rid of variables which aren't
4518 needed now that the byte interpreter is gone) caused the fasl
4519 file format number to change again.
4521 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4522 * major incompatible change: The default fasl file extension, i.e. the
4523 default extension for files produced by COMPILE-FILE, has changed
4524 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4526 ** There are many changes in the implementation of the compiler.
4527 SBCL is now essentially a compiler-only implementation of ANSI
4528 Common Lisp. EVAL still "interprets" a few special cases, but
4529 almost all the interesting cases are handled by creating
4530 a LAMBDA expression, calling COMPILE on it, then calling
4531 FUNCALL on the result.
4532 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4533 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4534 Since the code is newer, there might still be some new bugs
4535 (though not as many as before Martin Atzmueller's fixes:-). But
4536 the new code is substantially simpler and clearer, and hopefully
4537 any remaining bugs will be simpler, less fundamental, and more
4538 fixable then the bugs in the old code.
4539 ** The revised compiler is still a little unsteady on its feet.
4541 *** The debugging information it produces (particularly the names
4542 of FUNCTION objects) is sometimes much less useful than what
4543 the old compiler produced.
4544 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4545 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4546 within a MACROLET) has been temporarily weakened.
4547 ** There are new compiler optimizations for various functions:
4548 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4549 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4550 *** the math functions TRUNCATE, FLOOR, and CEILING
4551 *** the function-of-all-trades COERCE
4552 Mostly these should be transparent, but there's one
4553 potentially-annoying problem (bug 117): when the compiler
4554 inline-expands a function and does type analysis on the result,
4555 it can create control paths which have type mismatches, and
4556 when it can't prove that those control paths aren't taken,
4557 it will issue WARNINGs about the type mismatches. This is
4558 a particular problem in practice for the new sequence functions.
4559 It's not clear how this should be fixed, and for now, a
4560 workaround is given in the entry for 117 in the BUGS file.
4561 ** (Because of the interaction between the two previous items --
4562 occasional inlining problems and new inline expansions -- some
4563 of the new sequence function optimizations won't really kick in
4564 completely until debugging information, and then inlining, are
4565 straightened out in some future version.)
4566 * minor incompatible changes:
4567 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4568 DIRECTORY no longer implicitly promotes NIL slots of its
4569 pathname argument to :WILD. In particular, when you ask for the
4570 contents of a directory (which you used to be able to do without
4571 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4572 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4573 ** changes in behavior that ANSI explicitly defines to be
4574 implementation dependent:
4575 *** The new compiler-only implementation still conforms with ANSI,
4576 but acts a little different than before. Besides the obvious
4577 changes in performance tradeoffs (that the cost per form passed
4578 to EVAL has gone up, and the cost per form executed by EVAL
4579 has gone down), the behavior of the system changes a little
4580 because there are no longer any interpreted function objects.
4581 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4582 e.g. doing COMPILE on the output of interactive DEFUN is
4584 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4586 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4587 has changed from (:CL) to NIL.
4588 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4589 e.g. #\Newline and #\Tab, don't have names specified in the
4590 ANSI Common Lisp standard, is now based on their ASCII symbolic
4591 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4592 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4593 are no longer used for output.
4594 ** changes in internal implementation constants:
4595 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4596 4 million. (If your application spends a lot of time GCing and
4597 you have a lot of RAM, you might want to experiment with
4598 increasing it even more.)
4599 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4600 However, almost all old code should still continue to work without
4601 immediate update, as SB-C-CALL is now a (deprecated) nickname
4603 ** Old operator names in the style DEF-FOO are now deprecated in
4604 favor of new corresponding names DEFINE-FOO, for consistency with
4605 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4606 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4607 internal symbols, but a few supported extensions like
4608 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4609 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4610 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4611 etc. as you get deeper into recursive calls to the debugger
4612 command loop, instead of the old "5]", "5]]", "5]]]"
4613 sequence. (I was motivated to do this when squabbles between
4614 ILISP and SBCL left me very deeply nested in the debugger. In the
4615 short term, this change will probably provoke more ILISP/SBCL
4616 squabbles, but hopefully it will be an improvement in the long run.)
4617 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4618 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4619 function, since declaiming return types involving aliens is
4620 (1) annoyingly messy to do by hand and (2) vital to efficient
4621 compilation of code which calls such functions.
4622 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4623 longer reexported by the SB-EXT package. They're solely useful
4624 for alien code, so it seems more logical that you should get
4625 them from the SB-ALIEN package, not in SB-EXT.
4626 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4627 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4628 features. Instead, the code that they used to control is always
4629 built into the system.
4630 * many other bug fixes
4631 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4632 advantage of the new EVAL-WHEN stuff and to clean them up in
4633 general, and they are now more ANSI-compliant in a number of
4634 ways. Martin Atzmueller is responsible for a lot of this.
4635 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4637 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4638 *** correct ERROR type for various file operations
4639 *** some fixes for Lisp streams
4640 *** DEFMETHOD syntax checking
4641 *** changing old weird representation of debug information as
4642 strings (which, among their other deficiencies, don't transform
4643 correctly when you rename packages, and don't change their
4644 print representation when you change things like *PACKAGE*
4645 and *PRINT-LENGTH*) to symbols and lists of symbols
4646 He also made several improvements and fixed several bugs in DESCRIBE.
4647 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4649 *** misbehavior of WRITE-STRING/WRITE-LINE
4650 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4651 and several other LOOP problems as well
4652 *** DIRECTORY when similar filenames are present
4653 *** DEFGENERIC with :METHOD options
4654 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4655 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4656 *** argument ordering in FIND with :TEST option
4657 *** mishandled package designator argument in APROPOS-LIST
4658 *** various problems in the backquote readmacro
4659 *** a bug in APROPOS
4660 *** probably some others that I'm not describing very well here,
4661 since the CVS log documents them by reference to sbcl-devel
4662 messages, and the SourceForge archives aren't working well.:-(
4663 ** Dan Barlow improved the Alpha port (and is making progress on the
4664 PPC port, for those of you who think different).
4665 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4666 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4667 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4668 continues to work on the SPARC port (for those of you in a position
4669 to look down upon our little PC-compatible boxes from a great height).
4670 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4671 and uses that copy, rather than the current dynamic value, when
4672 it's trying to decide whether to truncate output. Thus e.g.
4673 (let ((*print-lines* 50))
4674 (pprint-logical-block (stream nil)
4676 (let ((*print-lines* 8))
4677 (print (aref possiblybigthings i) stream)))))
4678 should now truncate the logical block only at 50 lines, instead of
4679 often truncating it at 8 lines, as it did before.
4680 * The doc/cmucl/ directory, containing old CMU CL documentation
4681 from the time of the fork, is no longer part of the base system.
4682 SourceForge has shut down its anonymous FTP service, and with it
4683 my original plan for distributing the old CMU CL documentation
4684 there. For now, if you need these files you can download an old
4685 SBCL source release and extract them from it.
4686 * The fasl file version number changed again, for dozens of reasons,
4687 some of which are apparent above.
4689 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4690 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4691 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4692 patch, so that DISASSEMBLE output is much nicer.
4693 * The code in the SB-PROFILE package now seems reasonably stable.
4694 I still haven't decided what the final interface should look like
4695 (I'd like PROFILE to interact cleanly with TRACE, since both
4696 facilities use function encapsulation) but if you have a need
4697 for profiling now, you can probably use it successfully with
4698 the current CMU-CL-style interface.
4699 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4700 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4701 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4702 link now returns the dangling link itself, and for similar
4703 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4704 link itself. (In these cases the old code signalled an error and
4705 looped endlessly, respectively.) Thus, DIRECTORY now works even
4706 in the presence of dangling and cyclic symbolic links.
4707 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4708 is now a supported extension again, since the consensus on
4709 sbcl-devel was that it can be useful for ordinary development
4710 work, not just for debugging SBCL itself.
4711 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4712 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4713 that currently-defined functions might be redefined later with
4714 different return types.
4715 * Hash tables can be printed readably, as inspired by CMU CL code
4716 of Eric Marsden and SBCL code of Martin Atzmueller.
4717 * better error handling in CLOS method combination, thanks to
4718 Martin Atzmueller porting Pierre Mai's CMU CL patches
4719 * more overflow fixes for >16Mbyte I/O buffers
4720 * A bug in READ has been fixed, so that now a single Ctrl-D
4721 character suffices to cause end-of-file on character streams.
4722 In particular, now you only need one Ctrl-D at the command
4723 line (not two) to exit SBCL.
4724 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4726 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4727 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4728 does the right thing.
4729 * The implementation of some type tests, especially for CONDITION
4730 types, is now tidier and maybe faster, due to CMU CL code
4731 originally by Douglas Crosher, ported by Martin Atzmueller.
4732 * Some math functions have been fixed, and there are new
4733 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4734 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4735 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4736 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4737 likely to be built into sbcl-0.7.0.)
4738 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4739 CL, ported to SBCL by Martin Atzmueller)
4740 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4741 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4742 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4743 * There's a new slam.sh hack to shorten the edit/compile/debug
4744 cycle for low-level changes to SBCL itself, and a new
4745 :SB-AFTER-XC-CORE target feature to control the generation of
4746 the after-xc.core file needed by slam.sh.
4747 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4748 COMPILE-FILE is no longer supported, so that now every function
4749 gets an entry point, so that block compilation looks a little
4750 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4751 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4752 returns the number of bytes consed since the system started,
4753 rather than the number consed since the first time the function
4754 was called. (The new definition parallels ANSI functions like
4755 CL:GET-INTERNAL-RUN-TIME.)
4756 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4757 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4758 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4759 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4760 * Fasl file version numbers are now independent of the target CPU,
4761 since historically most system changes which required version
4762 number changes have affected all CPUs equally. Similarly,
4763 the byte fasl file version is now equal to the ordinary
4766 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4767 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4768 is no longer recognized. I apologize for this, because it was
4769 listed in SB-EXT as a supported extension, but I found that
4770 its existing behavior was poorly specified, as well as incorrectly
4771 specified, and it looked like too much of a mess to straighten it
4772 out. I have enough on my hands trying to get ANSI stuff to work..
4773 * many patches ported from CMU CL by Martin Atzmueller, with
4774 half a dozen bug fixes in pretty-printing and the debugger, and
4775 half a dozen others elsewhere
4776 * fixed bug 13: Floating point infinities are now supported again.
4777 They might still be a little bit flaky, but thanks to bug reports
4778 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4779 as flaky as they were.
4780 * The --noprogrammer command line option is now supported. (Its
4781 behavior is slightly different in detail from what the old man
4782 page claimed it would do, but it's still appropriate under the
4783 same circumstances that the man page talks about.)
4784 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4785 are now supported, and enabled by default. Thus, the compiler can
4786 handle many floating point and complex operations much less
4787 inefficiently. (Thus e.g. you can implement a complex FFT
4789 * The compiler now detects type mismatches between DECLAIM FTYPE
4790 and DEFUN better, and implements CHECK-TYPE more correctly, and
4791 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4792 thanks to patches from Martin Atzmueller.
4793 * various fixes to make the cross-compiler more portable to
4794 ANSI-conforming-but-different cross-compilation hosts (notably
4795 Lispworks for Windows, following bug reports from Arthur Lemmens)
4796 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4797 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4798 fixed thanks to Pierre Mai's CMU CL patches.
4799 * fixes to keep the system from overflowing internal counters when
4800 it tries to use i/o buffers larger than 16M bytes
4801 * fixed bug 45a: Various internal functions required to support
4802 complex special functions have been merged from CMU CL sources.
4803 (When I was first setting up SBCL, I misunderstood a compile-time
4804 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4805 * improved support for type intersection and union, fixing bug 12
4806 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4807 more obscure bugs as well
4808 * some steps toward byte-compiling non-performance-critical
4809 parts of the system, courtesy of patches from Martin Atzmueller
4810 * Christophe Rhodes has made some debian packages of sbcl at
4811 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4812 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4813 stable, but are nonetheless usable. When he's ready, I'd be happy
4814 to add them to the SourceForge "File Releases" section. (And if
4815 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4816 * new fasl file format version number (because of changes in
4817 internal representation of (OR ..) types to accommodate the new
4818 support for (AND ..) types, among other things)
4820 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4821 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4823 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4825 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4826 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4827 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4828 * DESCRIBE now works on structure objects again.
4829 * Most function call argument type mismatches are now handled as
4830 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4831 know whether the function will be redefined before the call is
4832 executed. (The compiler could flag local calls with full WARNINGs,
4833 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4834 it doesn't keep track of enough information to know whether calls
4835 are local in this sense.)
4836 * Compiler output is now more verbose, with messages truncated
4837 later than before. (There should be some supported way for users
4838 to override the default verbosity, but I haven't decided how to
4839 provide it yet, so this behavior is still controlled by the internal
4840 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4841 src/compiler/ir1util.lisp.)
4842 * Fasl file format version numbers have increased again, because
4843 support for the Gray streams extension changes the layout of the
4844 system's STREAM objects.
4845 * The Gray subclassable streams extension now works, thanks to a
4846 patch from Martin Atzmueller.
4847 * The full LOAD-FOREIGN extension (not just the primitive
4848 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4849 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4850 but like most other programs, it defaults to copying the Unix
4851 environment from the original process instead of starting the
4852 new process in an empty environment.
4853 * Extensions which manipulate the Unix environment now support
4854 an :ENVIRONMENT keyword option which doesn't smash case or
4855 do other bad things. The CMU-CL-style :ENV option is retained
4856 for porting convenience.
4857 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4858 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4860 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4862 * A patch from Martin Atzmueller seems to have solved the SIGINT
4863 problem, and as far as we know, signal-handling now works cleanly.
4864 (If you find any new bugs, please report them!)
4865 * The system no longer defaults Lisp source file names to types
4866 ".l", ".cl", or ".lsp", but only to ".lisp".
4867 * The compiler no longer uses special default file extensions for
4868 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4869 seems to expect a single default extension for all compiled code,
4870 and there's no compelling reason to try to stretch the standard
4871 to allow two different extensions.) Instead, byte-compiled files
4872 default to the same extension as native-compiled files.
4873 * Fasl file format version numbers have increased again, because
4874 a rearrangement of internal implementation packages made some
4875 dumped symbols in old fasl files unreadable in new cores.
4876 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4877 many fewer weird special cases.
4878 * Bug #17 (differing COMPILE-FILE behavior between logical and
4879 physical pathnames) has been fixed, and some related misbehavior too,
4880 thanks to a patch from Martin Atzmueller.
4881 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4882 by Tim Moore, ported to SBCL by Martin Atzmueller.
4883 * Martin Atzmueller fixed several filesystem-related problems,
4884 including bug #36, in part by porting CMU CL patches, which were
4885 written in part by Paul Werkowski.
4886 * More compiler warnings in src/runtime/ are gone, thanks to
4887 more patches from Martin Atzmueller.
4888 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4891 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4893 * DESCRIBE now works on CONDITION objects.
4894 * The debugger now handles errors which arise when trying to print
4895 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4897 * The build system now uses an additional file, customize-target-features.lisp,
4898 to allow local modifications to the target *FEATURES* list. (The point of
4899 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4900 debugging features enabled, without having to worry about propagating it
4901 into everyone's system when I do a "cvs update".) When no
4902 customize-target-features.lisp file exists, the target *FEATURES* list
4903 should be constructed the same way as before.
4904 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4905 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4906 warn about behavior which is undefined under the ANSI spec. Note: This
4907 is specified by ANSI, but it's not very popular with programmers.
4908 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4909 macro in the SBCL sources for an example of a workaround which you
4910 might use to make portable ANSI-standard code which does what you want.
4911 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4912 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4913 when it's used as a non-toplevel form. (This is required in order
4914 to implement the DEFCONSTANT-EQX macro.)
4915 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4916 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4917 funny things, due to bugs in the implementation of EVAL-WHEN
4918 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4919 non-toplevel DEFCONSTANTs are uncommon.)
4920 * The core file version number and fasl file version number have been
4921 incremented, because the old noncompliant DEFCONSTANT behavior involved
4922 calling functions which no longer exist, and because I also took the
4923 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4924 * fixed bug 1 (error handling before read-eval-print loop starts), and
4925 redid debugger restarts and related debugger commands somewhat while
4927 ** The QUIT debugger command is gone, since it did something
4928 rather different than the SB-EXT:QUIT command, and since it never
4929 worked properly outside the main toplevel read/eval/print loop.
4930 Invoking the new TOPLEVEL restart provides the same functionality.
4931 ** The GO debugger command is also gone, since you can just invoke
4932 the CONTINUE restart directly instead.
4933 ** The TOP debugger command is also gone, since it's redundant with the
4934 FRAME 0 command, and since it interfered with abbreviations for the
4936 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4938 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4939 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4940 some porting work by Martin Atzmueller)
4941 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4942 with a boatload of patches which he ported to SBCL. Now that those
4944 ** The system tries to make sure that its low-priority messages
4945 are prefixed by semicolons, to help people who like to use
4946 syntax highlighting in their ILISP buffer. (This patch
4947 was originally due to Raymond Toy.)
4948 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4949 patches originally written by Raymond Toy.
4950 ** The compiler understands coercion between single-value and
4951 multiple-VALUES type expressions better, getting rid of some very
4952 weird behavior, thanks to patches originally by Robert MacLachlan
4953 and Douglas Crosher.
4954 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4955 lambda lists, thanks to a patch originally by Pierre Mai.
4956 ** The system no longer bogusly warns about "abbreviated type
4958 ** The compiler gets less confused by inlining and RETURN-FROM,
4959 thanks to some patches originally by Tim Moore.
4960 ** The system no longer hangs when dumping circular lists to fasl
4961 files, thanks to a patch originally from Douglas Crosher.
4962 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4963 undefined function error.
4964 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4965 in structure slot initforms) for now, documented workaround instead:-|
4966 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4967 * fixed bug 5: added stubs for various Gray stream functions called
4968 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4969 installed, at least appropriate type errors are generated
4970 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4971 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4972 consistently in DEFMETHOD forms.
4973 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4974 it doesn't seem to affect SBCL after all
4975 * The C runtime system now builds with better optimization and many
4976 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4978 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4980 * The system is now under CVS at SourceForge (instead of the
4981 CVS repository on my home machine).
4982 * The new signal handling code has been tweaked to treat register
4983 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4984 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4985 downstream, e.g. in debug-int.lisp, has implicit dependencies
4986 on the unsignedness of integer representation of machine words,
4987 and that caused the system to bomb out with infinite regress
4988 when trying to recover from type errors involving signed values,
4989 e.g. (BUTLAST '(1 2 3) -1).)
4990 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4991 (This was one of the bugs Peter Van Eynde reported back in July.)
4992 * The system now uses code inspired by Colin Walters' O(N)
4993 implementation of MAP (from the cmucl-imp@cons.org mailing
4994 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4995 inline the MAP operation, and there is more than one
4996 sequence argument to the MAP call (so that it can't just
4997 do ETYPECASE once and for all based on the type of the
4998 single sequence argument). (The old non-inline implementation
4999 of the general M-argument sequence-of-length-N case required
5000 O(M*N*N) time when any of the sequence arguments were LISTs.)
5001 * The QUIT :UNIX-CODE keyword argument has been renamed to
5002 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
5003 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
5005 * Raymond Wiker's patches to port dynamic loading from Linux to
5006 FreeBSD have been added.
5007 * The BUGS file is now more nearly up to date, thanks in large part
5008 to Martin Atzmueller's review of it.
5009 * The debugger now flushes standard output streams before it begins
5010 its output ("debugger invoked" and so forth).
5011 * The core version number and fasl file version number have both
5012 been incremented, because of incompatible changes in the layout
5014 * FINISH-OUTPUT is now called more consistently on QUIT. (It
5015 used to not be called for a saved Lisp image.)
5016 * Martin Atzmueller's version of a patch to fix a compiler crash,
5017 as posted on sbcl-devel 13 September 2000, has been installed.
5018 * Instead of installing Martin Atzmueller's patch for the
5019 compiler transform for SUBSEQ, I deleted the compiler transform,
5020 and transforms for some similar consing operations.
5021 * A bug in signal handling which kept TRACE from working on OpenBSD
5023 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
5024 away by constant folding
5025 * The system now defines its address space constants in one place
5026 (in the Lisp sources), and propagates them automatically elsewhere
5027 (through GENESIS and the sbcl.h file). Therefore, patching the
5028 address map is less unnecessarily tedious and error-prone. The
5029 Lisp names of address space constants have also been systematized.
5030 * CVS tags like dollar-Header-dollar have been removed from
5031 the sources, because they have never saved me trouble and
5032 they've been source of trouble working with patches and other
5033 diff-related operations.
5034 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
5035 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
5037 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
5039 * The system has been ported to OpenBSD.
5040 * The system now compiles with a simple "sh make.sh" on the systems
5041 that it's supported on. I.e., now you no longer need to tweak
5042 text in the target-features.lisp-expr and symlinks in src/runtime/
5043 by hand, the make.sh takes care of it for you.
5044 * The system is no longer so grossly inefficient when compiling code
5045 involving vectors implemented as general (not simple) vectors (VECTOR T),
5046 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
5047 which dares to use the various sequence functions on non-simple
5048 vectors, takes less of a performance hit.
5049 * There is now a primitive type predicate VECTOR-T-P
5050 to test for the (VECTOR T) type, so that e.g.
5051 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
5052 can now be compiled with some semblance of efficiency. (The old code
5053 turned the type declaration into a full call to %TYPEP at runtime!)
5054 * AREF on (VECTOR T) is still not fast, since it's still compiled
5055 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
5056 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
5057 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
5058 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
5059 without making full calls to SUBTYPEP at runtime.
5060 (Some analogous efficiency issues for non-simple vectors specialized to
5061 element types other than T, or for non-simple multidimensional arrays,
5062 have not been addressed. They could almost certainly be handled the
5063 same way if anyone is motivated to do so.)
5064 * The changes in array handling break binary compatibility, so
5065 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
5066 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
5067 instead of (VALUES T T).
5068 * By following the instructions that Dan Barlow posted to sbcl-devel
5069 on 2 July 2000, I was able to enable primitive dynamic object
5070 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
5071 functionality is not implemented (since it calls ld to resolve
5072 library references automatically, requiring RUN-PROGRAM for its
5073 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
5074 try to resolve library references) is now supported.
5075 * The system now flushes the standard output streams when it terminates,
5076 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
5077 them at several other probably-convenient times, e.g. in each pass of
5078 the toplevel read-eval-print loop, and after evaluating a form given
5079 as an "--eval" command-line option. (These changes were motivated by a
5080 discussion of stream flushing issues on cmucl-imp in August 2000.)
5081 * The source transform for TYPEP of array types no longer assumes
5082 that an array whose element type is a not-yet-defined type
5083 is implemented as an array of T, but instead punts, so that the
5084 type will be interpreted at runtime.
5085 * There is now some support for cross-compiling in make.sh: each of
5086 the phases of make.sh has its own script. (This should be transparent
5087 to people doing ordinary, non-cross-compile builds.)
5088 * Since my laptop doesn't have hundreds of megabytes of memory like
5089 my desktop machine, I became more motivated to do some items on
5090 my to-do list in order to reduce the size of the system a little:
5091 ** Arrange for various needed-only-at-cold-init things to be
5092 uninterned after cold init. To support this, those things have
5093 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
5094 symbols with such names are now uninterned after cold init).
5095 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
5096 fasl files for cold load.
5097 ** Remove the old compiler structure pooling code (which used to
5098 be conditional on the target feature :SB-ALLOC) completely.
5099 ** Redo the representation of some data in cold init to be more compact.
5100 (I also looked into supporting byte compiled code at bootstrap time,
5101 which would probably reduce the size of the system a lot, but that
5102 looked too complicated, so I punted for now.)
5103 * The maximum signal nesting depth in the src/runtime/ support code has
5104 been reduced from 4096 to 256. (I don't know any reason for the very
5105 large old value. If the new smaller value turns out to break something,
5106 I'll probably just bump it back up.)
5107 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
5109 * Many, many bugs reported by Peter Van Eynde have been added to
5110 the BUGS list; some have even been fixed.
5111 * While enabling dynamic object file loading, I tried to make the
5112 code easier to understand, renaming various functions and variables
5113 with less ambiguous names, and changing some function calling
5114 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
5115 * While trying to figure out how to do the OpenBSD port, I tried to
5116 clean up some of the code in src/runtime/. In particular, I dropped
5117 support for non-POSIX signal handling, added various comments,
5118 tweaked the code to reduce the number of compilation warnings, and
5119 renamed some files to increase consistency.
5120 * To support the new automatic configuration functionality in make.sh,
5121 the source file target-features.lisp-expr has been replaced with the
5122 source file base-target-features.lisp-expr and the machine-generated
5123 file local-target-features.lisp-expr.
5124 * fixed a stupid quoting error in make.sh so that using CMU CL
5125 "lisp -batch" as cross-compilation host works again
5127 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
5129 * DESCRIBE no longer tries to call itself recursively to describe
5130 bound/fbound values, so that it no longer fails on symbols which are
5131 bound to themselves (like keywords, T, and NIL).
5132 * DESCRIBE now works on generic functions.
5133 * The printer now prints less-screwed-up representations of closures
5134 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
5135 * A private symbol is used instead of the :EMPTY keyword previously
5136 used to mark empty slots in hash tables. Thus
5137 (DEFVAR *HT* (MAKE-HASH-TABLE))
5138 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
5139 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
5140 now does what ANSI says that it should. (You can still get
5141 similar noncompliant behavior if bang on the hash table
5142 implementation with all the symbols you get back from
5143 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
5144 This breaks binary compatibility, since tests for equality to
5145 :EMPTY are wired into things like the macroexpansion of
5146 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
5148 * There's now a minimal placeholder implementation for CL:STEP,
5149 as required by ANSI.
5150 * An obscure bug in the interaction of the normal compiler, the byte
5151 compiler, inlining, and structure predicates has been patched
5152 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5153 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5155 * Missing ordinary arguments in a macro call are now detected even
5156 when the macro lambda list contains &KEY or &REST.
5157 * The debugger no longer complains about encountering the top of the
5158 stack when you type "FRAME 0" to explicitly instruct it to go to
5159 the top of the stack. And it now prints the frame you request even
5160 if it's the current frame (instead of saying "You are here.").
5161 * As specified by ANSI, the system now always prints keywords
5162 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5164 * The default initial SIZE of HASH-TABLEs is now smaller.
5165 * Type information from CLOS class dispatch is now propagated
5166 into DEFMETHOD bodies, so that e.g.
5167 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5169 is now basically equivalent to
5170 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5171 (DECLARE (TYPE SINGLE-FLOAT X))
5173 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5174 operation, without having to do run-time type dispatch.
5175 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5176 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5177 * Since I don't seem to be making big file reorganizations very often
5178 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5179 is overflowing my ability to conveniently back them up), I've finally
5180 checked the system into CVS. (The CVS repository is on my home system,
5181 not at SourceForge -- putting it on SourceForge might come later.)
5182 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5183 high-level GC-NOTIFY-FOO functions send their output. (There's
5184 still very little control of where low-level verbose GC functions
5185 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5186 less than it used to -- the GC-NOTIFY-FOO functions are now under
5187 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5188 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5189 in only one place in the source code, and propagates it automatically
5190 everywhere that it's needed. Thus e.g. when I bump the version from
5191 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5192 * The C source files now include boilerplate legalese and documentation
5193 at the head of each file (just as the Lisp source files already did).
5194 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5195 to his page will be replaced with a link to his new CLiki service.
5197 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5199 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5200 * The build process now looks for GNU make under the default name "gmake",
5201 instead of "make" as it used to. If GNU make is not available as "gmake"
5202 on your system, you can change this default behavior by setting the
5203 GNUMAKE environment variable.
5204 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5205 can build without error under CMU CL.
5207 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5209 * There is now a partial SBCL user manual (with some new text and some
5210 text cribbed from the CMU CL manual).
5211 * The beginnings of a profiler have been added (starting with the
5212 CMU CL profiler and simplifying and cleaning up). Eventually the
5213 main interface should be through the TRACE macro, but for now,
5214 it's still accessed through vaguely CMU-CL-style functions and macros
5215 exported from the package SB-PROFILE.
5216 * Some problems left over from porting CMU CL to the new
5217 cross-compilation bootstrap process have been cleaned up:
5218 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5219 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5220 part when getting rid of the compiler *BACKEND* stuff.)
5221 ** The value of *NULL-TYPE* was screwed up, because it was
5222 being initialized before the type system knew the final
5223 definition of the 'NULL type. This screwed up several key
5224 optimizations in the compiler, causing inefficiency in all sorts
5225 of places. (I found it because I wanted to understand why
5226 GET-INTERNAL-RUN-TIME was consing.)
5227 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5228 PROCLAIM FTYPE information. Unfortunately this broke binary
5229 compatibility again, since now the forms output by DEFGENERIC
5230 to refer to functions which didn't exist in 0.6.3.
5231 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5232 can use the new (as of 0.6.3) transform for SOME into MAP into
5234 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5235 values to (INTEGER 0 1000000), so that the time code will no longer
5236 occasionally get blown up by Linux returning 1000000 microseconds
5237 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5238 its output conform to the ANSI spec. (Alas, this makes its output
5239 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5240 * A full call to MAP NIL with a single sequence argument no longer conses.
5241 * fixes to problems pointed out by Martin Atzmueller:
5242 * The manual page no longer talks about multiprocessing as though
5243 it were currently supported.
5244 * The ILISP support patches have been removed from the distribution,
5245 because as of version 5.10.1, ILISP now supports SBCL without us
5246 having to maintain patches.
5247 * added a modified version of Raymond Toy's recent CMU CL patch for
5248 EQUALP comparison of HASH-TABLE
5250 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5252 * The system still can't cross-compile itself with
5253 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5254 entails), but at least it can compile more of itself that way
5255 than it used to be able to, and various buglets which were uncovered
5256 by trying to cross-compile itself that way have now been fixed.
5257 * This release breaks binary compatibility again. This time
5258 at least I've incremented the FASL file format version to 2, so that the
5259 problem can be detected reliably instead of just causing weird errors.
5260 * various new style warnings:
5261 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5262 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5263 ** using the lexical binding of a variable named in the *FOO* style
5264 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5265 as specified by ANSI.
5266 * *RANDOM-STATE* is no longer automatically initialized from
5267 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5268 default behavior of the system is to repeat its behavior every time
5269 it's run. If you'd like to change this behavior, you can always
5270 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5271 old convention there was no comparably easy way to get the system to
5272 repeat its behavior every time it was run.
5273 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5274 FTYPE declarations has been removed, in favor of their ANSI
5275 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5276 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5277 the types of their sequence arguments can be determined at compile time.
5278 This is done through a new open code expansion for MAP which eliminates
5279 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5280 argument types can be determined at compile time.
5281 * The optimizer now transforms COERCE into an identity operation when it
5282 can prove that the coerced object is already of the correct type. (This
5283 can be a win for machine generated code, including the output of other
5284 optimization transforms, such as the MAP transform above.)
5285 * Credit information has been moved from source file headers into CREDITS.
5286 * Source file headers have been made more standard.
5287 * The CASE macro now compiles without complaining even when it has
5290 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5292 * (Note that the way that the PCL macroexpansions were rewritten
5293 to accommodate the change in DEFGENERIC below breaks binary
5294 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5295 not run under sbcl-0.6.2. Once we get out of alpha releases,
5296 i.e. hit release 1.0.0, we'll probably try to maintain binary
5297 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5298 and sbcl-1.4.4. Until then, however, it might be fairly common
5299 for maintenance releases to break binary compatibility.)
5300 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5302 * The reporting of unhandled signals has been changed to print some
5303 explanatory text as well as the report form. (Previously only
5304 the report form was printed.)
5305 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5306 it defines, so that the compiler no longer issues undefined function
5307 warnings for compiled-but-not-yet-loaded generic functions.
5308 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5309 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5310 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5311 section of the ANSI spec are supported.
5312 * The "" nickname for the "KEYWORD" package has been removed.
5313 The reader still handles symbol tokens which begin with a package marker
5314 as keywords, but it doesn't expose its mechanism for doing so in the
5315 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5316 * The system now issues STYLE-WARNINGs for contradictory TYPE
5317 proclamations. (Warnings for contradictory FTYPE proclamations would
5318 be nice too, but those can't be done usefully unless the type system
5319 is made smarter about FUNCTION types.)
5320 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5321 systematized, so that "*target-*.lisp is supposed to exist only on the
5322 target and imply that there's a related file which exists on the
5323 host, and *host-*.lisp is supposed to exist only on the host and imply
5324 that there's a related file which exists on the target. This involves a
5325 lot of renaming. Hopefully the acute confusion caused by the renaming
5326 will be justified by the reduction in chronic confusion..
5327 ** runtime-type.lisp -> early-target-type.lisp
5328 ** target-type.lisp -> late-target-type.lisp
5329 ** early-host-format.lisp -> early-format.lisp
5330 ** late-host-format.lisp -> late-format.lisp
5331 ** host-error.lisp -> misc-error.lisp
5332 ** early-error.lisp -> early-target-error.lisp
5333 ** late-error.lisp -> late-target-error.lisp
5334 ** host-defboot.lisp -> early-defboot.lisp
5335 ** code/misc.lisp -> code/target-misc.lisp
5336 ** code/host-misc.lisp -> code/misc.lisp
5337 ** code/numbers.lisp -> code/target-numbers.lisp
5338 ** code/early-numbers.lisp -> numbers.lisp
5339 ** early-host-type.lisp -> early-type.lisp
5340 ** late-host-type.lisp -> late-type.lisp
5341 ** host-typep.lisp -> typep.lisp
5342 ** load.lisp -> target-load.lisp
5343 ** host-load.lisp -> load.lisp
5344 ** host-disassem.lisp -> disassem.lisp
5345 ** host-insts.lisp -> insts.lisp
5346 ** byte-comp.lisp -> target-byte-comp.lisp
5347 ** host-byte-comp.lisp -> byte-comp.lisp
5348 ** host-signal.lisp -> signal.lisp
5349 ** host-defstruct.lisp -> defstruct.lisp
5350 ** late-target-type.lisp -> deftypes-for-target.lisp
5351 Furthermore, several other previously target-only files foo.lisp (e.g.
5352 hash-table.lisp and random.lisp) have been split into a target-and-host
5353 foo.lisp file and a target-only target-foo.lisp file, with their key type
5354 definitions in the target-and-host part, so that the cross-compiler will
5355 know more about target types.
5356 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5357 gone away. In their place are various *BACKEND-FOO* variables
5358 corresponding to the slots of the old structure.
5359 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5360 into the target SBCL has been fixed.
5361 * The chill.lisp system for loading cold code into a running SBCL
5363 * Support for the CMU CL "scavenger hook" extension has been removed.
5364 (It was undocumented and unused in the CMU CL sources that SBCL was
5365 derived from, and stale in sbcl-0.6.1.)
5366 * Various errors in the cross-compiler type system were detected
5367 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5368 (enabling various consistency checks). Many of them were fixed,
5369 but some hard problems remain, so the compiler is back to
5370 running without *TYPE-SYSTEM-INITIALIZED* for now.
5371 * As part of the cross-compiler type system cleanup, I implemented
5372 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5373 * I have started adding UNCROSS calls throughout the type system
5374 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5375 on cross-compiler input files will be able to go away, and various
5377 * CONSTANTP now returns true for quoted forms (as explicitly required
5380 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5382 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5383 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5384 (MAKE-LIST -1). In the longer term, it ought to become true
5385 that declarations are assertions even at SAFETY 1. For now, it's not
5386 quite true even at SAFETY 3, but it's at least more nearly true..
5387 (Note that this change seems to increases the size of the system by
5388 O(5%) and to decrease the speed of the compiler by 20% or more.)
5389 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5390 for the problem of printing dozens of lines of distracting information
5391 about low-level system machinery as part of the top stack frame
5392 on entry to the debugger when an undefined function was called.
5393 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5394 is not reset to COMMON-LISP-USER.
5395 * Compilation of stuff related to dyncount.lisp has been made conditional
5396 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5397 smaller. The various dyncount-related symbols have been moved into
5398 a new "SB-DYNCOUNT" package.
5399 * tty-inspect.lisp has been renamed to inspect.lisp.
5400 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5401 feature has gone away. (When we eventually port to other flavors of
5402 libc and/or Unix, we'll try to make the differences between flavors
5403 invisible at the user level.)
5404 * Various other *FEATURES* tags, and/or their associated conditionals,
5405 have been removed if obsolescent, or given better documentation, or
5406 sometimes given more-mnemonic names.
5408 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5410 * tidied up "make.sh" script
5411 * tidied up system directory structure
5412 * better "clean.sh" behavior
5413 * added doc/FOR-CMUCL-DEVELOPERS
5414 * many many small tweaks to output format, e.g. removing possibly-confusing
5415 trailing #\. character in DESCRIBE-INSTANCE
5416 * (EQUALP #\A 'A) no longer signals an error.
5417 * new hashing code, including EQUALP hashing
5418 * tidied up Lisp initialization and toplevel
5419 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5420 * command line argument processing
5421 * added POSIX-GETENV function to deal with Unix-ish environment variables
5422 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5423 terminating SBCL on EOF
5424 * non-verbose GC by default
5425 * There is no more "sbcl" shell script; the sbcl file is now the C
5426 runtime executable (just like CMU CL).
5427 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5429 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5430 symbol and package information as Lisp native symbol and package objects
5431 instead of strings naming symbols and strings naming packages. This way,
5432 whenever packages are renamed (as in warm init), debug information is
5433 transformed along with everything else.
5434 * tweaked the optimization policy declarations which control the building
5435 of SBCL itself. Now, among other things, the system no longer saves
5436 source location debugging information. (This helps two problems at once
5437 by reducing SBCL size and by keeping SBCL from trying to look for its
5438 sources -- which may not exist -- when reporting errors.)
5439 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5440 debugging and testing purposes
5441 * cleaned up printing, making the printer call PRINT-OBJECT for
5442 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5443 methods, giving nearly-ANSI behavior
5444 * converted almost all special variables to use *FOO* naming convention
5445 * deleted PARSE-TIME functionality, since it can be done portably
5446 * moved some files out of cold init into warm init
5447 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5449 * regularized formatting of source files
5450 * added an install.sh script
5451 * fixed ridiculous memory usage of cross-compiler by making
5452 compiler/alloc.lisp not try to do pooling unless it can hook
5453 itself into the GC of the cross-compilation host. Now the system
5454 builds nicely on my old laptop.
5455 * added :SB-ALLOC in target-features.lisp-expr
5456 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5457 was not implemented)
5458 * re-did condition handling and note reporting in the compiler. Notes
5459 are no longer handled by signalling conditions. Style warnings
5460 and warnings are handled more correctly and reported in such a way
5461 that it's easy to find one or the other in your output (so that you
5462 can e.g. figure out which of many problems caused COMPILE-FILE to
5464 * changed the severity of several compiler warnings from full WARNING
5465 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5466 compiler note reporting so that it doesn't use the condition system
5467 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5468 COMPILE-FILE command)
5469 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5470 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5471 no longer supported, since their implementation was deeply intertwingled
5472 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5473 * removed bogus "support" for compiler macros named (SETF FOO), and
5474 removed the compiler macro for SETF INFO (but only after making a fool
5475 of myself on the cmucl-imp mailing list by posting a bogus patch for
5476 DEFINE-COMPILER-MACRO..)
5477 * Compiled files containing forms which have side effects on the Lisp
5478 reader (such as DEFPACKAGE forms) are now handled more correctly.
5479 (Compiler queuing of top level lambdas has been suppressed by setting
5480 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5481 * deleted various currently-unused source files, e.g. gengc.lisp. They
5482 may be added back at some point e.g. when porting to other architectures,
5483 but until they are it's distracting to distribute them and to try to
5485 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5486 there were so many of them they're just distractions, and UNCROSS is
5487 known to be able to handle the current sources
5488 * moved PROFILE functionality into TRACE, so that it will be clear
5489 how the wrapping and unwrapping of functions when you profile them
5490 interacts with the wrapping and unwrapping of functions when you
5491 trace them. (Actually, the functionality isn't there yet, but at least
5492 the interface specification is there. Hopefully, the functionality will
5493 arrive with some future maintenance release.)
5494 * removed host-oops.lisp
5495 * changed signature of QUIT function to allow UNIX-CODE argument
5496 * fixed READ-SEQUENCE bug
5497 * tweaked verbose GC output so that it looks more like the progress
5498 output that ANSI specifies for functions like LOAD
5499 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5500 * added <http://sbcl.sourceforge.com> to the banner information printed by