1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes relative to sbcl-1.0.30:
3 * improvement: stack allocation is should now be possible in all nested
4 inlining cases: failure to stack allocate when equivalent code is manually
5 open coded is now considered a bug.
6 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
7 be used to output toplevel compile-time effects into a separate .CFASL
9 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
10 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
12 * optimization: division of floating point numbers by constants uses
13 multiplication by reciprocal when an exact reciprocal exists.
14 * optimization: multiplication of single- and double-floats floats by
15 constant two has been optimized.
16 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
17 sufficient type information is available. (thanks to Leslie Polzer)
18 * improvement: a STYLE-WARNING is signalled when a generic function
19 clobbers an earlier FTYPE proclamation.
20 * improvement: the compiler is able to track the effective type of
21 generic function across method addition and removal even in the
22 absence of an explicit DEFGENERIC.
23 * improvement: DESCRIBE now reports on symbols naming undefined
24 but assumed or declared function as well.
25 * improvement: recompilation of systems using SB-GROVEL now works
26 (thanks to Leslie Polzer)
27 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
28 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
30 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
31 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
32 well as user defined declaration names.
33 ** VARIABLE-INFORMATION is now aware of alien variables.
34 * improvement: improved address space layout on OpenBSD (thanks to Josh
36 * improvement: pretty-printing of various Lisp forms has been improved
37 (thanks to Tobias Rittweiler)
38 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
40 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
41 as well. (reported by Josh Elsasser)
42 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
43 leading to possible attempts to delete the same file twice. See docstring
44 on DELETE-FILE for details. (reported by John Fremlin)
45 * bug fix: the low-level debugger had 32-bit assumptions and was missing
46 information about some array types. (thanks to Luis Oliveira)
47 * bug fix: moderately complex combinations of inline expansions could
48 be miscompiled if the result was declared to be dynamic extent.
49 * bug fix: in some cases no compiler note about failure to stack allocate
50 was emitted, even if the objects were in fact heap allocated.
51 * bug fix: minor violation of "otherwise inaccessible" rule for stack
52 allocation could cause objects users might reasonably expect to
53 be heap allocated to be stack allocated.
54 * bug fix: DESCRIBE signalled an error for generic functions under
55 certain circumstances. (thanks to Leslie Polzer)
56 * bug fix: Fixed spelling of an error message.
58 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
59 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
60 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
61 of SB-THREAD:THREAD-ERROR-THREAD.
62 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
63 * new feature: docstrings for local and anonymous functions are no longer
64 discarded. (thanks to Leslie Polzer)
65 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
66 values in other threads.
67 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
68 about object allocation.
69 * optimization: division of a real float by a complex float is implemented
70 with a specialised code sequence.
71 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
72 keywords is an order of magnitude faster.
73 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
74 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
75 methods, and similarly for non-standard metaclass classes as long as there
76 are no methods additional on MAKE-INSTANCE.
77 * optimization: more efficient type-checks for FIXNUMs when the value
78 is known to be a signed word on x86 and x86-64.
79 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
80 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
82 * optimization: compiler is smarter about delegating argument type checks to
84 * optimization: several character functions are now compiled somewhat more
85 efficiently. (reported by Lynn Quam)
86 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
87 * improvement: less unsafe constant folding in floating point arithmetic,
88 especially for mixed complex/real -float operations.
89 * optimization: constant double and single floats are stored in native
90 unboxed format on x86[-64].
91 * optimization: smarter code for arithmetic operations with constant floats,
92 complex floats, or integers on x86[-64].
93 * optimization: smarter code for conjugate/multiplication of float complexes
94 and abs/negate of floats on x86-64.
95 * optimization: more efficient complex float and real float operations on
97 * improvement: complex float division is slightly more stable.
98 * improvement: DESCRIBE output has been reworked to be easier to read and
99 contains more pertinent information.
100 * improvement: failure to provide requested stack allocation compiler notes
101 provided in all cases (requested stack allocation not happening without a
102 note being issued is now considered a bug.)
103 * bug fix: SB-POSIX exports the documented types and functions
104 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
105 types. (reported by "abhi")
106 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
107 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
108 (thanks to Paul Khuong)
109 * bug fix: looping around HANDLER-CASE could silently consume stack space
110 on each iteration. (reported by "foobar")
111 * bug fix: better error signalling when calls to functions seeking elements
112 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
113 (reported by Tobias Rittweiler)
114 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
115 directory components now works as it used to. (various prolems reported by
116 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
117 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
118 (reported by Elliot Slaughter, thanks to Stas Boukarev)
119 * bug fix: bogus undefined variable warnings from fopcompiled references to
120 global variables. (thanks to Lars Rune Nøstdal)
121 * bug fix: foreign function names should now appear in backtraces on
122 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
123 * bug fix: SETF compiler macro documentation strings are not discarded
125 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
126 (thanks to Alex Plotnick)
127 * bug fix: SLEEP accepts large integer arguments, truncating them to
128 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
129 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
130 * bug fix: compiler notes for expensive slot type checks could be emitted
131 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
133 changes in sbcl-1.0.29 relative to 1.0.28:
134 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
135 https://bugs.launchpad.net/sbcl
136 Bugs can be reported directly there, or by sending email to
137 sbcl-bugs@lists.sourceforge.net
138 (no subscription required.)
139 * minor incompatible change: under weak type checking policy integer
140 types are weakened less aggressively.
141 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
142 allowed to return, which causes SBCL to quit with exit status 0. Previously
143 if the function returned with a small integer return value, that value
144 was accidentally reused as the exit status.
145 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
146 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
147 :HASH-FUNCTION argument. Refer to user manual for details.
148 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
150 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
151 microseconds since the Unix epoch on all platforms.
152 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
153 allows the compiler to safely elide boundedness checks for special
155 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
156 the symbol, prohibits both lexical and dynamic binding. This is mainly an
157 efficiency measure for threaded platforms, but also valueable in
159 * new feature: UNC pathnames are now understood by the system on Windows.
160 * optimization: the compiler uses a specialized version of FILL when the
161 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
163 * optimization: accesses to potentially non-simple arrays where element type
164 is known are 50% faster.
165 * optimization: compiler now generates faster array typechecking code.
166 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
168 * optimization: multidimensional array accesses in the absence of type
169 information regarding array rank are approximately 10% faster due to
170 open coding of ARRAY-RANK.
171 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
172 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
174 * optimization: result of call to VECTOR can now be stack allocated.
175 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
176 as long as the resulting array is one-dimensional and has a known
177 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
178 length of the list matches the known length of the vector does not
179 allocate the list as an intermediate step. Ditto for VECTOR and simple
181 * optimization: MAKE-ARRAY can now stack allocate in the presence of
182 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
183 known element type, and is known to be simple and one dimensional.
184 * improvement: SBCL now emits a compiler note where stack allocation was
185 requested but could not be provided (not in all cases, unfortunately)
186 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
187 * improvement: pretty-printing loop has been implemented properly. (thanks
188 to Tobias Rittweiler)
189 * documentation: CLOS slot typechecing policy has been documented.
190 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
191 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
192 pointer, leading to memory faults. (thanks to Bart Botta)
193 * bug fix: better floating point exception handling on x86/OpenBSD.
194 (thanks to Josh Elsasser)
195 * bug fix: exit status from QUIT when called under --script was lost
196 (reported by Hubert Kauker)
197 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
198 same implementation of FILL to initialize the array, even if a more
199 efficient one was available (reported by Stas Boukarev, thanks to
201 * bug fix: potential miscompilation of array stack allocation on x86 and
202 x86-64. (reported by Time Tossavainen)
203 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
204 result in their subforms being treated as top level forms. (reported by
206 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
207 than a machine register could leave garbage in the high bits of the
208 result register (bug 316325).
209 * bug fix: disable address space randomization Linux/x86-64 as well,
210 not just x86-64. (reported by Ken Olum)
211 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
212 generate incorrect code.
213 * bug fix: #201; type inference for CONS and ARRAY types could derive
214 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
215 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
216 could not be assigned to or bound with PROGV. (reported by Lorenz
218 * bug fix: the value of CL:- in the inspector was the previous expression
219 evaluated rather than the expression being evaluated.
220 * bug fix: constants can no longer be locally declared special.
221 * bug fix: signals delivered to threads started from foreign land (read:
222 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
223 thread by blocking all signals and resignalling.
224 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
225 from :INITFORM, if any.
227 changes in sbcl-1.0.28 relative to 1.0.27:
228 * a number of bugs in cross-compilation have been fixed, with the ultimate
229 result that building under (at least) clisp should be much more reliable.
230 * minor incompatible changes: echo-streams now propagate unread-char to the
231 underlying input stream, and no longer permit unreading more than one
233 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
234 frames, allowing for instance more reliable backtraces.
235 * improvement: the debugger REPL can now reference lexical variables
236 by name directly for code compiled with (DEBUG 3).
237 * improvement: errors from malformed declarations now have better source
238 paths associated with them. (thanks to Tobias Rittweiler)
239 * optimization: faster local calls on x86/x86-64
240 * bug fix: some error messages for out-of-bound array indexes confused the
241 index and the bound. (thanks to Stas Boukarev)
242 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
244 * bug fix: running regressions tests in shells without OSTYPE set now works.
245 (reported by Harald Hanche-Olsen)
246 * bug fix: more robust static space exhaustion signalling from
247 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
248 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
249 docstring away instead of storing it under names such as (LAMBDA (X)).
250 (reported by Leslie Polzer)
251 * bug fix: timers could go off in the wrong order, be delayed indefinitely
252 (thanks to Ole Arndt for the patch)
253 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
254 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
255 potentially-invalid effective methods in its cache.
256 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
257 instances as well (thanks to Paul Khuong)
258 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
259 SB-POSIX:WAITPID and their C equivalents.
260 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
262 changes in sbcl-1.0.27 relative to 1.0.26:
263 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
264 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
265 * improvement: the system either recovers from stack exhaustion or dies
266 properly as opposed to leaving the user uncertain of whether the handler
267 trampled on some random memory next to the stack or having to rely on
268 --lose-on-corruption (which is still a good idea to use in production
269 because stack exhaustion can happen in signal handlers which will likely
271 * bug fix: fix gc related interrupt handling bug on ppc (regression from
272 1.0.25.37, reported by Harald Hanche-Olsen)
273 * bug fix: work around signal delivery bug in darwin (regression from
274 1.0.25.44, reported by Sidney Markowitz)
275 * bug fix: fix ERROR leaking memory (reported by David Thompson)
277 changes in sbcl-1.0.26 relative to 1.0.25:
278 * incompatible change: an interruption (be it a function passed to
279 INTERRUPT-THREAD or a timer function) runs in an environment where
280 interrupts can be enabled. The interruption can use
281 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
282 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
283 potentially running out of stack. Keep in mind that in the absence
284 of WITHOUT-INTERRUPTS some potentially blocking operation such as
285 acquiring a lock can enable interrupts.
286 * incompatible change: GC-OFF and GC-ON are removed, as they were
287 always unsafe. Use WITHOUT-GCING instead.
288 * new feature: runtime option --disable-ldb
289 * new feature: runtime option --lose-on-corruption to die at the
290 slightest hint of possibly non-recoverable errors: running out of
291 memory, stack, alien stack, binding stack, encountering a memory
292 fault, etc. In the absence of --lose-on-corruption a warning is
294 * enhancement: detect binding stack exhaustion
295 * enhancement: detect alien stack exhaustion on x86/x86-64
296 * improvement: generally more stable and reliable interrupt handling
297 * improvement: there is a per thread interruption queue,
298 interruptions are executed in order of arrival
299 * improvement: a repeating timer reschedules itself when the it has
300 finished, but expiration times are spaced equally. If an
301 expiration time is in the past it will trigger after a short grace
302 period that may give a chance to other things to run.
303 * optimization: slightly faster gc on multithreaded builds
304 * optimization: faster WITHOUT-GCING
305 * bug fix: when JOIN-THREAD signals an error, do it when not holding
306 important locks so that the debugger/handler doesn't produce
307 recursive errors or deadlock.
308 * bug fix: real-time signals are not used anymore, so no more
309 hanging when the system wide real-time signal queue gets full.
310 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
312 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
313 * bug fix: fix random memory faults related to interrupts on alpha
314 * bug fix: fix random memory faults related to interrupts on sparc
315 * bug fix: fix deadlocks related to starting threads
316 * bug fix: fix deadlines on locks on futex platforms
317 * bug fix: restore errno in signal handlers
318 * bug fix: fix deadlocks related to hash tables
319 * bug fix: fix deadlocks in pcl
321 changes in sbcl-1.0.25 relative to 1.0.24:
322 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
323 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
324 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
325 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
326 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
327 better name and does not return values so stale on multiprocessor systems.
328 Also, HOLDING-MUTEX-P was added for about the only sane usage of
330 * improvement: unithread builds keep track of MUTEX-VALUE.
331 * improvement: reading from a TWO-WAY-STREAM does not touch the output
332 stream anymore making it thread safe to have a concurrent reader and
333 a writer, for instance, in a pipe.
334 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
335 arguments, making compiler-macros for SETF-functions able to inspect
336 their constant arguments.
337 * improvement: COMPILE-FILE reports times with millisecond accuracy
338 (thanks to Luis Oliveira)
339 * optimization: CHAR-CODE type derivation has been improved, making
340 TYPEP elimination on subtypes of CHARACTER work better. (reported
341 by Tobias Rittweiler, patch by Paul Khuong)
342 * bug fix: setting alien structure fields of type struct by value now
343 computes the right offset for the memory copy.
344 * bug fix: compilation problem involving inlined calls to aliens with
345 result type VOID. (reported by Ken Olum)
346 * bug fix: #235a; sequential inline expasion in different policies no
347 longer reuses the functional from the previous expansion site.
348 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
349 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
350 * bug fix: named ENUMs in multiply-referenced alien record types no longer
351 cause an implied type redefinition. Regression from 1.0.21.29.
352 * bug fix: sign-extension of small signed return values now works with C
353 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
354 * improvements to the Windows port:
355 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
356 Westfold, thanks to Rudi Schlatte)
358 changes in sbcl-1.0.24 relative to 1.0.23:
359 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
360 vector of a multidimensional SIMPLE-ARRAY.
361 * new feature: the system now signals a continuable error if standard
362 readtable modification is attempted. (thanks to Tobias Rittweiler)
363 * new feature: DIRECTORY has been extended with a non-standard keyword
364 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
365 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
366 has been robustified and documented better. (thanks to Robert Goldman)
367 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
368 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
369 * tradeoff: constant FORMAT control strings are now compiled unless
370 SPACE > SPEED (previously only when SPEED > SPACE.)
371 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
373 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
375 * bug fix: lexical type declarations are now correctly reported by
376 SB-CLTL2. (reported by Larry D'Anna)
377 * bug fix: STRING-TO-OCTETS did not handle :START properly when
378 using UTF-8 as external format. (thanks to Luis Oliveira)
379 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
380 are signalled correctly. (thanks to Stas Boukarev)
381 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
382 designator. (thanks to Tobias Rittweiler)
383 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
384 designator, and returns T instead of the function. (thanks to
386 * bug fix: direct superclasses of STANDARD-CLASS and
387 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
388 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
389 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
390 when no class with the named slot yet exists no longer causes a
391 compile-time style-warning.
392 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
393 in safe code. (reported by Didier Verna)
394 * bug fix: #430; nested structure constructors can stack allocate.
395 * bug fix: on some 64-bit platforms dynamic space size was truncated
396 to #xffff0000 bytes. (reported by Benjamin Lambert)
397 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
398 REPL unusable. (reported by Daniel Herring)
399 * bug fix: return values of READ-SEQUENCE did not take :START into
400 account on file streams, regressions since 1.0.12.22. (reported by
401 Thomas Russ, patch by Paul Khuong)
402 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
403 method specializer used to confuse permuation vector optimization.
404 * bug fix: system inserted bogus implicit type declarations for local
405 special variables in DEFMETHOD bodies.
406 * bug fix: #354; duplicated frames in backtraces due to
407 non-tail-call-optimized XEPs to functions with return type NIL
409 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
410 initializes structure object slots according to DEFSTRUCT initforms,
411 and DEFSTRUCT forms :INCLUDEind structure classes defined using
412 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
413 (reported by Bruno Haible and Stephen Wilson)
414 * bug fix: #395; fill-pointer output streams used now support
415 element-type BASE-CHAR as well.
416 * bug fix: compiler error when attempting to derive return value of
417 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
419 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
421 * bug fix: Mac OS X binaries should now be portable between Leopard
424 changes in sbcl-1.0.23 relative to 1.0.22:
425 * enhancement: when disassembling method functions, disassembly
426 for the associated fast function is also produced.
427 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
429 * optimization: printing with *PRINT-PRETTY* true is now more
430 efficient as long as the object being printed doesn't require
431 special handling by the pretty printer.
432 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
433 now interact correctly with type declarations.
434 * partial bug fix: PCL detects infinite recursion during wrapper
435 validation. (thanks to Attila Lendvai)
436 * bug fix: #426; nested function calls are inlined properly.
437 Previously if FOO was an inline function, in calls of the form
438 (FOO (FOO ...)) the outer call was not inlined.
439 * bug fix: long long arguments passed in registers to alien
440 functions on PowerPC Linux and NetBSD are now handled correctly
441 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
443 changes in sbcl-1.0.22 relative to 1.0.21:
444 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
445 for the shared object in the current directory, but passes the native
446 namestring of the designated pathname to the operation system's shared
447 object loading function as-it.
448 * minor incompatible change: --disable-debugger toplevel option now takes
449 effect before processing of initialization files and --eval or --load
451 * new feature: new commandline argument: --script, which supports
452 shebang lines. See documentation for details. (based on work by
454 * new feature: SAVE-LISP-AND-DIE can save current values of
455 --dynamic-space-size and --control-stack-size in the executable core,
456 causing it to skip normal runtime option processing. See documentation
457 for details. (thanks to Zach Beane)
458 * enhancement: inoccous calls to EVAL or generic functions dispatching
459 on subclasses of eg. STREAM no longer cause compiler notes to appear.
460 * enhancement: the system no longer resignals errors from --load and
461 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
462 associated with the original error to be lost. (thanks to Ariel
464 * enhancement: :DONT-SAVE keyword argument has been added to
465 LOAD-SHARED-OBJECT for controlling interaction with
466 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
467 the effects of an earlier LOAD-SHARED-OBJECT call.
468 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
469 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
470 provided. (reported by Cedric St-Jean)
471 * bug fix: circularity handling in the reader did not treat raw
472 structure slots correctly. (reported by Cedric St-Jean)
473 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
474 file descriptors when there were none.
475 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
476 was broken. (reporteed by Willem Broekema)
477 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
478 pathnames without a directory.
479 * bug fix: COMPILE-FILE did not warn about undefined variable
480 references at toplevel, and LOAD of the resulting fasl did
482 * bug fix: functions with non-required arguments used to end up with
483 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
484 * bug fix: redefining a function with non-required arguments didn't
485 update the system's knowledge about its call signature properly.
486 * bug fix: fixed #431; incompatible alien record type redefinitions
487 are detected and handled. (reported by Neil Haven)
488 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
489 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
491 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
492 after alien stack frames.
493 * bug fix: strings passed to mk{s,d}temp lacked null termination
495 changes in sbcl-1.0.21 relative to 1.0.20:
496 * new feature: the compiler is able to track the effective type of a
497 generic function across method addition and removal.
498 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
499 appropriately typed structure slots without locking.
500 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
501 information like those gathered by TIME using a programming-friendly
503 * new feature: TIME reports time taken even if the form performs a
504 non-local transfer of control.
505 * enhancement: reduced conservativism on GENCGC platforms: on
506 average 45% less pages pinned (measured from SBCL self build).
507 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
508 mutate constant symbols or violate declaimed type of the symbol.
509 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
510 sleeping on the semaphore in heavily contested semaphores.
511 * bug fix: semaphores and condition variables were not interrupt
513 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
514 owned by other threads anymore.
515 * bug fix: FIND on lists called KEY outside the specified
516 subsequence. (reported by budden)
517 * bug fix: LOG doesn't use single-float intermediate results when
518 given mixed integer and double-float arguments, leading to better
519 precision. (reported by Bob Felts)
520 * bug fix: LOG with base zero returned values of inconsistent type.
522 changes in sbcl-1.0.20 relative to 1.0.19:
523 * minor incompatible change: OPTIMIZE qualities
524 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
525 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
526 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
527 * documentation: some slot access efficiency guidelines have been
528 added to the user manual.
529 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
530 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
531 as ASSOC and MEMEBER.
532 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
533 to more efficient EQ-comparison versions more often.
534 * optimization: enhanced derivation of DOLIST iteration variable type
536 * optimization: constant folding of simple (LIST ...) forms as DOLIST
538 * optimization: runtime lookup of function definitions can be
539 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
540 * optimization: compiler is able to derive the return type of
541 (AREF (THE STRING X) Y) as being CHARACTER.
542 * optimization: CLRHASH on empty hash-tables no longer does pointless
543 work. (thanks to Alec Berryman)
544 * optimization: REPLACE deftransforms don't punt when :START1 or
546 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
547 no longer emits a code deletion note. (reported by Andrew
549 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
550 functions with an odd number of &OPTIONAL arguments, a &REST
551 argument, and one or more &KEY arguments at the call site.
552 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
553 arguments appeared at call sites as well.
554 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
555 READ-LINE &co can no longer read from them afterwards. (reported
557 * bug fix: fixed #427: unused local aliens no longer cause compiler
558 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
560 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
561 alien-stack leakage. (reported by Andy Hefner)
562 * bug fix: PROGV signals an error when an attempt to violate declared
563 type of a variable or bind a constant is made.
564 * bug fix: SET signals an error when an attempt to violate declared
565 type of a variable is made.
566 * bug fix: restart computation during the execution of a restart
567 test function no longer causes infinite recursion. (reported by
569 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
570 signals an error. (thanks to Pierre Mai)
571 * bug fix: EXPORT left symbol unexported in conflict situations.
572 (thanks to Michael Weber)
573 * bug fix: correctly handle name conflicts involving CL:NIL.
574 (thanks to Michael Weber)
575 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
576 conflicts arising from USEing package with conflicting symbols
577 correctly. (thanks to Michael Weber)
579 changes in sbcl-1.0.19 relative to 1.0.18:
580 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
581 warnings that go otherwise unhandled will be muffled if they are
582 of the type that's the value of this variable.
583 * optimization: stack allocation is slightly more efficient on x86
585 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
586 where a raw slot always is initialized using the initform whose
587 type is not know sufficiently well a compile-time are now compiled
588 correctly. (reported by John Morrison)
589 * bug fix: compiler no longer makes erronous assumptions in the
590 presence of non-foldable SATISFIES types.
591 * bug fix: stack analysis missed cleanups of dynamic-extent
592 arguments in non-let-converted calls to local functions.
593 * improvements to the Windows port:
594 ** adjusted address spaces for building on both Win32 and
595 Win64. (thanks for John Connors)
596 * fixed some bugs revealed by Paul Dietz' test suite:
597 ** interval arithmetic during type derivation used inexact integer
598 to single-float coercions.
599 ** arithmetic operations involving large integers and single
600 floats give the same results in compiled and interpreted code.
601 ** deriving the result type of COERCE no longer signals an error
602 if the derived type of the second argument is a MEMBER type
603 containing invalid type specifiers.
604 ** ADJOIN with constant NIL as second argument works correctly.
605 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
607 changes in sbcl-1.0.18 relative to 1.0.17:
608 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
609 profiles only the current thread.
610 * minor incompatible change: changes to SYMBOL-VALUE of constants
611 defined with DEFCONSTANT now signal an error.
612 * enhancement: SB-SPROF now has support for wallclock profiling,
613 and is also able to profile specific threads. REPORT output
614 has also additional sorting options.
615 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
617 * optimization: structure allocation has been improved
618 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
619 ** out of line constructors are ~10% faster.
620 ** inline constructors are ~15% faster.
621 ** inline constructors are capable of dynamic extent allocation
622 (generally on x86 and x86-64, in some cases on other platforms
624 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
626 * optimization: file compiler is now able to coalesce non-circular
627 lists, non-base strings, and bit-vectors. Additionally, constants
628 are never referenced through SYMBOL-VALUE at runtime.
629 * optimization: code defining methods on PRINT-OBJECT (and other
630 generic functions in the COMMON-LISP package) now loads faster.
631 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
633 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
634 fasl is now deleted. (reported by Attila Lendvai)
635 * bug fix: READ-LINE always returned NIL for the last line in files.
636 (reported by Yoshinori Tahara)
637 * bug fix: more accurate disassembly annotations of foreign function
638 calls. (thanks to Andy Hefner)
639 * bug fix: trimming non-simple strings and non-string string
640 designators when the there is nothing to trim works properly.
641 (thanks to James Knight)
642 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
643 (thanks to Travis Cross)
644 * fixed some bugs revealed by Paul Dietz' test suite:
645 ** NIL is a valid function name (regression at 1.0.13.38)
646 ** FILL on lists was missing its return value (regression at 1.0.12.27)
647 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
648 fill pointers properly (regression at 1.0.12.23)
649 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
650 displacement indices properly (regression at 1.0.12.23)
652 changes in sbcl-1.0.17 relative to 1.0.16:
653 * temporary regression: user code can no longer allocate closure
654 variable storage on stack, due to bug 419 without explicitly
655 requesting it. Please consult sbcl-devel for advice if you need to
656 use this feature in the meanwhile.
657 * new feature: runtime argument --control-stack-size can be used to
658 adjust thread default control stack size.
659 * enhancement: improved TIME output
660 ** all times are reported using the measured accuracy (milliseconds
661 for real and GC times, microseconds for everything else.)
662 ** processor cycle counts on x86 and x86-64.
663 ** interpreted forms are counted for both evaluator modes.
664 ** number of lambdas converted by the compiler is reported.
665 ** CPU percentage report (computed from real and total run time.)
666 ** more comprehensive run time reporting, using a condenced format
667 ** interperted form, lambda, and page fault counts are omitted
669 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
671 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
672 * optimization: two argument forms of LAST are upto ~10% faster
673 in normal SPEED policies.
674 * optimization: NCONC no longer needs to heap cons its &REST list
675 in normal SPEED policies.
676 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
678 * bug fix: UNION and NUNION work with :TEST-NOT once more,
679 regression since 1.0.9.1. (thanks to Eric Marsden)
680 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
681 since 1.0.15.36. (thanks to Paul Khuong)
682 * bug fix: LAST when always returned the whole list when given a bignum
683 as the second argument.
684 * bug fix: dynamic extent allocation of nested lists and vectors
685 could leak to otherwise accessible parts.
686 * bug fix: invalid optimization of heap-allocated alien variable
688 * bug fix: fasl header checking is less vulnerable to different
689 platform word lengths.
690 * bug fix: more correct assembler syntax for GNU binutils
691 2.18.50.0.4 support. (thanks to Marijn Schouten)
692 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
694 * bug fix: the fopcompiler can handle LOCALLY forms (with no
695 declarations) successfully. (reported by Attila Lendvai)
697 changes in sbcl-1.0.16 relative to 1.0.15:
698 * minor incompatible change: revert the changes to sb-posix's error
699 signaling added in 1.0.14.
700 * minor incompatible change: change PROBE-FILE back to returning
701 NIL whenever we can't get a truename, as was the case before 1.0.14.
702 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
703 inherits from ERROR instead of just CONDITION.
704 * new feature: SB-INTROSPECT can provide source locations for instances
705 as well. (thanks to Tobias Rittweiler)
706 * optimization: binding special variables now generates smaller code
707 on threaded platforms.
708 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
709 and cases where no :TEST is given but the compiler can infer that
710 the element to search is of type (OR FIXNUM (NOT NUMBER)).
711 * optimization: better LOGNOT on fixnums.
712 * optimization: modular arithmetic for a particular requested width
713 is implemented using a tagged representation unless a better
714 representation is available.
715 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
716 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
717 for wellformedness and returns a specialized vector. (reported by
718 Francois-Rene Rideau)
719 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
720 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
721 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
722 unscheduling timer at the same time as another timer fires.
723 * bug fix: don't reschedule timers for dead threads.
724 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
725 * bug fix: copying output from RUN-PROGRAM to a stream signalled
726 bogus errors if select() was interrupted.
727 * enhancement: add support for fcntl's struct flock to SB-POSIX.
729 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
730 * enhancement: cleaner backtraces for interactive interrupts, as
731 well as other cases where the interesting frames used to be
732 obscured by interrupt handling frames.
733 * enhancement: untracing a whole package using (UNTRACE "FOO") is
734 now supported, and tracing a whole package using (TRACE "FOO") now
735 traces SETF-functions as well.
736 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
737 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
738 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
739 * unreadably printed representation of hash-tables now includes
741 * bug fix: partially fixed #188: type propagation from assignments
742 is now more efficient.
743 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
744 'DOUBLE-FLOAT) are not flushable.
745 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
746 full address of the object, and none of the tag bits.
747 * bug fix: readably printing hash-tables now respects other printer
748 control variables. (reported by Cedric St-Jean)
749 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
750 keyword with MAKE-HASH-TABLE.
751 * bug fix: export SB-POSIX:MKSTEMP.
752 * bug fix: SORT was not interrupt safe.
753 * bug fix: XREF accounts for the last node of each basic-block as
755 * bug fix: MAKE-INSTANCE optimizations interacted badly with
756 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
757 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
758 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
759 create obsolete instances in the case of redefinition or
760 obsoletion of a superclass. (thanks to Andy Hefner)
761 * bug fix: Support for the Alpha architecture has been revived; it had
762 suffered somewhat from lack of maintenance since sbcl-1.0.
763 * improvements to the Windows port:
764 ** The system detects the codepage to use at startup, instead of
765 using the value from the saved core. (thanks to Kei Suzuki)
767 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
768 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
769 (see documentation for details.)
770 * revived support for OpenBSD (contributed by Josh Elsasser)
771 * partially fixed bug #108: ROOM no longer suffers from occasional
772 (AVER (SAP= CURRENT END)) failures.
773 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
774 bodies no longer cause a WARNING to be signalled. (reported by
776 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
777 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
778 the bug in TRUNCATE.) (reported by Michael Weber)
779 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
780 no samples. (reported by Andy Hefner)
781 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
782 no longer appear as (NIL ...) frames in backtraces.
783 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
784 is now actually exported from SB-EXT as documented. (reported by
786 * bug fix: sb-aclrepl now correctly understands how to inspect
787 single-floats on 64-bit platforms where single-floats are not boxed.
788 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
789 yet been finalized. (reported by Levente Meszaros)
790 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
791 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
792 SBCL-specific optimize qualities.
794 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
795 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
796 an executable in the search path, and does so in the child
797 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
798 been removed; it can be found in the CVS history, for somebody who
799 needs that search behavior (see the manual).
800 * minor incompatible change: compiler policy re. weakening type
801 checks has changed: now type checks are weakened only if SAFETY < 2
803 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
804 unparsing of directory pathnames as files. Analogously,
805 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
806 filename to parse into a directory pathname.
807 * enhancement: implicit generic function creation now signals a
808 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
809 users can bind handlers for and muffle around calls to LOAD.
810 * enhancement: RUN-PROGRAM allows unicode arguments and environments
811 to be used (using the default stream external format), and allows
812 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
813 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
814 strings and vectors whose element-type or simplicity is not fully
815 known at compile-time.
816 * optimization: STRING-TRIM and related functions no longer allocate
817 a new string when no trimming needs to be performed. These functions
818 are also faster than before when the input string has been declared
820 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
821 * optimization: READ-LINE is significantly faster for files containing
823 * optimization: non-open coded uses of character comparison operators
824 (e.g. char=) no longer cons when called with more than one parameter
825 on platforms supporting dynamic-extent allocation.
826 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
827 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
828 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
830 * bug fix: some sequence functions elided bounds checking when
832 * bug fix: too liberal weakening of union-type checks when SPEED >
834 * bug fix: more bogus fixnum declarations in ROOM implementation
837 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
838 * new feature: MAKE-HASH-TABLE now experimentally accepts a
839 :SYNCHRONIZED argument, which makes the hash-table safe for
840 concurrent accesses (but not iteration.) See also:
841 SB-EXT:WITH-LOCKED-HASH-TABLE, and
842 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
843 * optimization: CONCATENATE on strings is an order of magnitude faster
844 in code compiled with (> SPEED SPACE).
845 * optimization: SUBSEQ is ~50% faster on lists.
846 * bug fix: bug 417 fixed -- source location reporting is now more robust.
847 * bug fix: SUBSEQ on a list will now correctly signal an error if if
848 END is smaller then START.
849 * bug fix: SB-PROFILE will no longer report extra consing for nested
850 calls to profiled functions.
851 * bug fix: ROOM implementation had bogus fixnum declarations which
852 could cause type-errors when calling ROOM on large images.
853 * bug fix: if file compilation is aborted, the partial fasl is now
854 deleted, and COMPILE-FILE returns NIL as the primary value.
855 * bug fix: number of thread safety issues relating to SBCL's internal
856 hash-table usage have been fixed.
857 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
858 be returned from its body when the values were being returned
859 using unknown-values return convection and the W-P-O was wrapped
860 inside an UNWIND-PROTECT.
861 * bug fix: sb-posix should now compile again under Windows, enabling
864 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
865 * incompatible change: hash-table accessor functions are no longer
866 automatically protected by locks. Concurrent accesses on the same hash-table
867 from multiple threads can give inconsistent results or even corrupt the
868 hash-table completely. Multi-threaded applications should do their own
869 locking at the correct granularity. In the current implementation it is
870 still safe to have multiple readers access the same table, but it's not
871 guaranteed that this property will be maintained in future releases.
872 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
873 and will signal an error at runtime.
874 * enhancement: SB-THREAD package now exports a semaphore interface.
875 * enhancement: CONS can now stack-allocate on x86 and
876 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
878 * enhancement: nested lists can now be stack allocated on
879 platforms providing stack allocation support.
880 * enhancement: dynamic-extent support has been extended to support
881 cases where there are multiple possible sources for the stack
883 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
884 if the mutex is uncontested on Linux.
885 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
886 as specified by the CLHS. (thanks to Tobias Rittweiler)
887 * bug fix: NaN comparison now works on x86-64.
888 * bug fix: CLOSE :ABORT T on a stream with pending output now
890 * bug fix: instances of non-standard metaclasses using standard
891 instance structure protocol sometimes missed the slot type checks
893 * bug fix: known functions can be inlined in high-debug code.
895 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
896 * minor incompatible change: the MSI installer on Windows no longer
897 associates .lisp and .fasl files with the installed SBCL.
898 * minor incompatible change: :UNIX is no longer present in *FEATURES*
899 on Windows. (thanks to Luis Oliviera)
900 * new feature: SB-CLTL2 contrib module now implements
901 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
902 * optimization: scavenging weak pointers is now more efficient,
903 requiring O(1) instead of O(N) per weak pointer to identify
904 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
905 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
906 method bodies using constant slot-names when the first argument is
907 a specializer parameter for the method.
908 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
909 names, when the class of the instance is a direct instance
910 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
911 as slow as the constant slot-name case.)
912 * optimization: member type construction is now O(N) instead
914 * optimization: UNION and NUNION are now O(N+M) for large
915 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
917 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
918 now more readable in environments like Slime which display it.
919 (thanks to Tobias C. Rittweiler)
920 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
921 was able to infer the correct argument type for the object on which
922 the CAS operation was being performed.
923 * bug fix: copy propagation interfered with parallel assignment
924 semantics in local calls. (reported by Paul Khuong)
925 * bug fix: the signed modular fixnum shift compiled to wrong code on
926 x86 and x86-64. (spotted by a slight modification to some of
929 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
930 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
931 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
932 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
934 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
935 and generic function now print the object identity for anonymous
936 instances. (thanks to Attila Lendvai)
937 * enhancement: as an extension to MOP, SBCL now supports SETF of
938 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
939 (thanks to Attila Lendvai)
940 * enhancement: sb-cover has an alternate annotation mode which puts
941 the form annotations at the car of the form, instead of annotating
942 it completely. The benefit of this mode is that it shows better
943 which forms were instrumented by the compiler.
944 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
945 by printer control variables. (reported by Dan Corkill)
946 * bug fix: system leaked memory when delayed output was performed by
947 the OS in smaller chunks then expected. (thanks to David Smith)
948 * bug fix: system leaked memory when file streams were not closed
950 * bug fix: large objects written to slow streams that were modified
951 after the write could end up with the modified state written to
952 the underlying file descriptor.
953 * bug fix: multiple threads operating in parallel on the same stream
954 could cause buffer-overflows.
955 * bug fix: source location information is stored correctly
956 (broken since 1.0.6). This bug would generally show up as the
957 Slime debugger highlighting the wrong form when the sldb-show-source
959 * bug fix: Branch forms are again annotated as branches in the sb-cover
961 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
962 on x86-64. (reported by Gregory Vanuxem)
963 * bug fix: EQUALP could return wrong results for structures with raw
964 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
965 word). (reported by Vjacheslav Fyodorov)
966 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
969 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
970 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
971 atomic compare-and-swap operations on threaded platforms.
972 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
973 allows assigning a global minimum value to optimization qualities
974 (overriding proclamations and declarations).
975 * enhancement: closed over variables can be stack-allocated on x86
977 * performance bug fix: GETHASH and (SETF GETHASH) are once again
979 * optimization: slot definition lookup is now O(1). This speeds up
980 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
981 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
982 * optimization: ASSOC and MEMBER can now be open-coded for all
983 combinations of keyword arguments when second argument is constant
984 and SPEED >= SPACE. In other cases a specialized version is
986 * bug fix: using obsoleted structure instances with TYPEP and
987 generic functions now signals a sensible error.
988 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
989 (reported by Kristoffer Kvello)
990 * bug fix: backtrace construction is now more careful when making
991 lisp-objects from pointers on the stack, to avoid creating bogus
992 objects that can be seen by the GC.
993 * bug fix: defaulting of values in contexts expecting more than 7
994 variables now works on x86-64. (reported by Christopher Laux)
995 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
997 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
998 as the property-list of a symbol.
999 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1000 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1001 situations requested, are once again file-compileable. (reported
1004 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1005 * MOP improvement: support for user-defined subclasses of
1006 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1007 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1008 the expansion of DEFMETHOD, is responsible for generating a form
1009 which creates a list of specializers when evaluated. Additional
1010 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1011 debugging and introspective support.
1012 * minor incompatible change: the (unsupported) spinlock interface
1013 has changed: free spinlock now has the value NIL, and a held spinlock
1014 has the owning thread as its value.
1015 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1016 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1018 * enhancement: name of a socket-stream is now "a socket" instead of
1019 "a constant string".
1020 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1021 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1022 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1023 for generic functions. (Thanks to Tobias C. Rittweiler)
1024 * optimization: bignum printing speed has been improved by 20-40%
1025 (depending on the bignum size.)
1026 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1028 * bug fix: the cache used by the CLOS to store precomputed effective
1029 methods, slot offsets, and constant return values is now thread and
1031 * bug fix: generic function dispatch function updating is now thread
1032 and interrupt safe (in the sense that the known issues have been
1034 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1035 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1036 * bug fix: asdf systems with dependencies to the SB-POSIX or
1037 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1038 * bug fix: interrupt safety of applicable method computation has been
1041 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1042 * new contrib: sb-cover, an experimental code coverage tool, is included
1043 as a contrib module.
1044 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1045 significantly faster.
1046 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1047 produce more efficient code on x86-64 if the compiler can determine
1048 that (+ INDEX <constant>) does not require a bounds check and FOO
1049 has an element type at least 8 bits wide.
1050 * enhancement: a new, experimental synchronous timeout facility is
1051 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1052 * enhancement: when a symbol name conflict error arises, the
1053 conflicting symbols are always printed with a package prefix.
1054 (thanks to Kevin Reid)
1055 * enhancement: stepping is now once again supported on the SPARC and
1056 MIPS platforms. (It is also now more likely to work on CheneyGC
1058 * enhancement: sb-sprof can now also track and report accurate call
1060 * bug fixes: the treatment of non-standard subclasses of
1061 SB-MOP:SPECIALIZER is more correct.
1062 * incompatible change: PURIFY no longer copies the data from the
1063 dynamic space into the static and read-only spaces on platforms
1064 that use the generational garbage collector
1065 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1067 * bug fix: GC race condition occasionally resulting in crashes with
1068 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1070 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1072 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1073 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1074 * bug fix: an asynchronous interrupt could previously leave the
1075 system running with GC inhibited.
1076 * bug fix: a DECLARE form evaluated at top-level now causes an error
1077 rather than silently (or verbosely) returning NIL.
1078 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1079 result in a error being signalled due to source form lookup errors.
1080 (reported by Peter Graves)
1082 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1083 * incompatible change: removed writer methods for host-ent-name,
1084 host-ent-addresses -- changing the values did not update the DNS
1086 * minor incompatible change: changed experimental JOIN-THREAD interface
1087 * documentation: the manual now lists reader and writer methods
1088 in class slot documentation sections. (thanks to Richard M Kreuter)
1089 * documentation: unwinding from asyncronous events has been
1090 documented as unsafe.
1091 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1092 in multithreaded application code.
1093 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1094 platforms. (thanks to James Anderson for the optimization hint)
1095 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1097 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1098 variants no longer cons.
1099 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1100 their NOT- variants no longer cons.
1101 * optimization: EQUAL hash tables no longer use SXHASH for objects
1102 of all data types, but instead use an EQL hash for types for which
1103 EQUAL is the same as EQL.
1104 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1105 are significantly faster.
1106 * optimization: new STRING-OUTPUT-STREAM implementation is much
1107 faster for large outputs and conses 30% less on average.
1108 * enhancement: XREF information is now collected for references made
1109 to global variables using SYMBOL-VALUE and a constant argument.
1110 * enhancement: SIGINT now causes a specific condition
1111 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1112 * bug fix: FILE-POSITION used to signal an error in some cases where
1113 ANSI requires it to return NIL.
1114 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1115 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1116 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1118 * bug fix: bad type declaration in the CLOS implementation has
1119 been fixed. (reported by James Anderson)
1120 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1121 (thanks to Lutz Euler)
1122 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1123 required by the CLHS. (reported by Samium Gromoff)
1124 * bug fix: dead unbound variable references now signal an error.
1125 (reported by Marco Monteiro)
1126 * bug fix: / with an unused value was being deleted in safe code.
1127 (thanks to Marco Monteiro and Kevin Reid)
1128 * bug fix: number of characters that can be written onto a single
1129 line in a file is unlimited.
1130 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1131 been fixed by inhibiting interrupts for when GC is disabled.
1132 * bug fix: some interrupt safety issues with GETHASH, (SETF
1133 GETHASH), CLRHASH and REMHASH have been fixed.
1134 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1135 specifier no longer causes infinite recursion.
1136 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1137 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1138 * bug fix: an &environment argument with the correct variable information
1139 is passed to macros that are expanded during byte compilation.
1140 (reported by Samium Gromoff)
1141 * bug fix: a logic error was causing the hash function for bignums to
1142 have excessive amounts of collisions. (reported by Faré Rideau)
1143 * bug fix: modifying the contents of an array could change the return
1144 value of SXHASH on that array, which is only allowed for strings
1145 and bit vectors. (bug introduced in 0.9.16)
1146 * improvement: the x86-64/darwin port now passes all tests (except
1147 for the debugger tests) but should still be considered
1148 experimental until this is fixed.
1149 * improvement: a style-warning is signaled for CASE (etc) clauses with
1150 duplicate keys. (thanks to Kevin Reid)
1151 * improvement: macos/x86 and macos/x86-64 now use mach exception
1152 handlers for dealing with illegal instructions (for trapping and
1153 error handling) and memory protection violations (for GC).
1155 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1156 * new platform: experimental support for x86-64/darwin (MacOS).
1157 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1158 and getnameinfo sockets functions are used instead of gethostbyaddr
1159 and gethostbyname, on platforms where the newer functions are available.
1160 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1162 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1163 don't print the startup banner, but behave as if --noinform was passed
1164 as a command line argument. (thanks to Kevin Reid)
1165 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1166 * optimization: code using alien values with undeclared types is much faster.
1167 * optimization: the compiler is now able to open code SEARCH in more cases.
1168 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1169 * bug fix: using standardized COMMON-LISP special variables as loop
1170 variables no longer signals bogus package lock violations. (reported
1172 * bug fix: declaring local loop variables to be of a range-limited type
1173 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1174 (reported by Andras Simon)
1175 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1176 bugs remain on x86-64.)
1177 * bug fix: the #= and ## reader macros now interact reasonably with
1178 funcallable instances.
1179 * bug fix: type-checks for function arguments were compiled using the
1180 compiler policy settings of the wrong lexical environment (in 1.0.2
1182 * bug fix: SHADOW accepts characters as string designators, as required
1183 by the spec (thanks to Eric Marsden)
1184 * bug fix: fixed GC safety issues when foreign functions are called with
1185 non-base strings as arguments
1186 * bug fix: more consistent error output for fatal-compiler-warnings, like
1188 * bug fix: functions with proper names were showing up as NIL in
1189 backtraces (reported by Edi Weitz, regression in 1.0.2)
1191 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1192 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1193 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1194 produce more efficient code on the x86 if the compiler can determine
1195 that (+ INDEX <constant>) does not require a bounds check and FOO
1196 has an element type at least 8 bits wide.
1197 * bug fix: references to undefined variables are handled the same way
1198 in toplevel forms as in the normal compiler
1199 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1201 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1202 (thanks to Jon Buller)
1203 * improvement: faster compilation times for complex functions
1204 * improvement: added readlink support to SB-POSIX (thanks to Richard
1207 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1208 * improvement: experimental support for mach exception handling on
1209 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1211 * improvement: support for GBK external format.
1212 (thanks to Chun Tian (binghe))
1213 * improvement: the debugger now displays variables that have been closed
1214 over, in code compiled with (DEBUG 2) or higher
1215 * improvement: support for executable cores on NetBSD (thanks to
1217 * new feature: added a RESTART-FRAME debugger command
1218 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1219 be used to provide an implementation for FILE-POSITION on Gray streams
1220 (thanks to Eric Marsden)
1221 * new feature: add syslog, openlog and closelog support to SB-POSIX
1222 (thanks to Richard Kreuter)
1223 * optimization: the function call overhead in code compiled with
1224 a high DEBUG optimization setting is significantly
1225 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1226 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1227 (thanks to Tony Martinez)
1228 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1229 evaluated once (reported by Kevin Reid)
1230 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1231 works again (thanks to Kilian Sprotte)
1232 * bug fix: an error is signaled for tagbodies with duplicate tags
1233 (thanks to Stephen Wilson)
1234 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1235 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1236 stack frames from alien callbacks.
1237 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1238 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1239 65k pages (thanks to David Woodhouse)
1240 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1242 changes in sbcl-1.0.1 relative to sbcl-1.0:
1243 * new platform: FreeBSD/x86-64, including support for threading.
1244 * new feature: the compiler stores cross-referencing information
1245 about function calls (who-calls), macroexpansion (who-macroexpands)
1246 and special variables (who-binds, who-sets, who-references) for code
1247 compiled with (< SPACE 3). This information is available through the
1248 sb-introspect contrib.
1249 * new feature: users may subclass SEQUENCE, and have instances of
1250 these classes interoperate with standard Common Lisp functions if
1251 a number of methods are defined. (This feature is experimental
1252 and the interface subject to change based on feedback from SBCL
1253 users and the general community)
1254 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1255 x86/x86-64, rather than the previous fixed depth of 8
1256 * improvement: another pthread back-end of mutex "pthread-futex".
1257 * bug fix: non-ascii command-line arguments are processed correctly
1258 (thanks to Yaroslav Kavenchuk)
1259 * bug fix: non-required arguments were not passed correctly when a method
1260 defined using DEFMETHOD was called from a mop-generated method using
1261 CALL-NEXT-METHOD (reported by Pascal Costanza)
1262 * bug fix: recursion is now permitted in accessors through
1263 SLOT-UNBOUND. (reported by Pascal Costanza)
1264 * bug fix: an error was signaled at startup if the HOME environment
1265 variable was defined, but had an empty value (reported by Peter Van Eynde)
1266 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1267 signal a PROGRAM-ERROR, not a BUG.
1268 * bug fix: SB-POSIX:READDIR works when built with large file support.
1269 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1271 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1272 for better automatic indentation support. (thanks to Matt Pillsbury)
1273 * optimization: loading generic functions no longer takes O(n^2) time,
1274 proportional to the amount of methods in the generic function
1275 (reported by Todd Sabin and Jeremy Brown)
1276 * optimization: the FIND and POSITION family of sequence functions
1277 are significantly faster on arrays whose element types have been
1279 * improvements to the Windows port:
1280 ** Intermittent heap corruption problems have been fixed. (thanks
1281 to Alastair Bridgewater)
1282 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1284 ** Lisp is able to unwind foreign exception frames from alien
1285 callbacks. (thanks to Alastair Bridgewater)
1287 changes in sbcl-1.0 relative to sbcl-0.9.18:
1288 * improvement: experimental support for threading on FreeBSD/x86.
1289 (thanks to NIIMI Satoshi)
1290 * improvement: runtime option --dynamic-space-size can be used
1291 to set the size of the dynamic space reserved on startup.
1292 * improvement: floating point modes in effect are now saved in
1293 core, and restored on startup.
1294 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1295 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1296 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1297 threads are not running after *SAVE-HOOKS* have run.
1298 * improvement: writes to CLOS instance slots are type-checked in code
1299 compiled with (SAFETY 3)
1300 * improvement: floating-point exception handling on FreeBSD (thanks to
1302 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1303 (thanks to Zach Beane)
1304 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1306 * improvement: added support for the Shift-JIS external format.
1307 (contributed by NIIMI Satoshi)
1308 * improvement: callbacks are supported on Linux/PPC. (thanks to
1310 * bug fix: compiler bug triggered by a (non-standard) VALUES
1311 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1312 * bug fix: file compiler no longer confuses validated and already
1313 dumped structurres. (reported by Kaersten Poeck)
1314 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1315 fill-pointer signals a type-error as required. (thanks to
1317 * bug fix: disassemly of funcallable instances works.
1318 * bug fix: single stepping on PPC.
1319 * bug fix: fix thread-safety problems in the type system (generally
1320 manifesting as nonsensical errors like "STRING is a bad type specifier
1321 for sequences" or "The value 1 is not of type FIXNUM")
1322 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1323 (reported by Josip Gracin)
1324 * bug fix: an error is signaled for attempts to displace arrays with
1325 incompatible element types (thanks to Mario Mommer)
1326 * bug fix: more correct handling of wide characters in debug info
1327 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1328 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1329 and don't cause extra consing
1330 * optimization: MAP and MAP-INTO are significantly faster on vectors
1331 whose elements types have been declared.
1332 * Improvements to SB-SPROF:
1333 ** Support for allocation profiling
1334 ** Reduced profiling overhead, especially for long profiling runs
1335 * Improvements to the Windows port:
1336 ** floating point exceptions are now reported correctly.
1337 ** stack exhaustion detection works partially.
1338 ** more accurate GET-INTERNAL-REAL-TIME.
1339 ** better RUN-PROGRAM behaviour in case of errors in the child
1341 ** PROBE-FILE now simplifies pathnames correctly.
1342 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1343 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1344 1st 1970 (although time timezone data will be unreliable).
1345 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1347 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1348 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1349 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1351 * bug fix: two potential GC deadlocks affecting threaded builds.
1352 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1353 returns T (reported by Anton Kazennikov)
1354 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1355 with non-variable places
1356 * bug fix: remove a race condition in the setting of
1357 funcallable-instance functions, this should make threaded CLOS
1358 code more stable against memory faults.
1359 * bug fix: corruption of specials when unbinding is interrupted by an
1360 asynchronous unwind (reported by Hannu Koivisto)
1361 * improvement: the debugger will now also display local variables that
1362 are only used once, for code compiled with a DEBUG optimization quality
1365 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1366 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1367 * incompatible change: External-format support for FFI calls. The
1368 SB-ALIEN:C-STRING no longer implies an ASCII
1369 external-format. Instead the string is subject to external-format
1370 conversion. Additionally return values of type SB-ALIEN:C-STRING
1371 are by default of type (SIMPLE-ARRAY CHARACTER), not
1372 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1373 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1374 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1375 * incompatible change: SB-EXT package no longer contains the
1376 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1377 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1379 * incompatible change: the single-stepper is no longer available
1380 on Alpha, Mips and Sparc platforms.
1381 * minor incompatible change: the direct superclasses of
1382 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1383 not (STANDARD-OBJECT FUNCTION). This makes the
1384 class-precedence-lists of GENERIC-FUNCTION and
1385 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1387 * minor incompatible change: the default stream external format on
1388 non-unicode SBCL is no longer detected from the locale, but is always
1390 * new feature: Add a version of evaluator that uses an interpreter instead
1391 of the compiler. EVAL still uses the compiler by default, to switch it
1392 to use the interpreter, set the value of the variable
1393 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1394 * minor incompatible change: the single-stepper REPL has been merged
1395 with the normal debugger (see the "Stepping" heading of the debugger help
1396 for more details). The debugger command STEP will no longer switch
1397 to the single-stepper REPL.
1398 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1399 for a type now works.
1400 * bug fix: loading of default sysinit file works. (thanks to Leonid
1402 * bug fix: better detection of circularities in the file-compiler.
1403 (reported by Marco Monteiro)
1404 * bug fix: the CL pathname functions now work with files that have
1405 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1406 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1407 systems with Unix98 pty semantics.
1408 * bug fix: ASDF-INSTALL will now work with bsd tar.
1409 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1411 * bug fix: timers expiring in dead threads no longer cause a
1412 type-error (reported by Paul "Nonny Mouse").
1413 * bug fix: thanks to more lightweight single-stepper instrumentation,
1414 code compiled with (DEBUG 3) will compile and execute significantly faster,
1415 and will have more accurate type-inferencing than before
1416 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1417 whose bindings are modified
1418 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1419 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1420 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1421 CL:SOFTWARE-VERSION with proper version
1423 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1424 * feature: implemented the READER-METHOD-CLASS and
1425 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1426 as specified by AMOP.
1427 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1429 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1430 (thanks to Lutz Euler)
1431 * optimization: hashing of general arrays and vectors has been
1432 improved. (reported by Any Fingerhut)
1433 * enhancement: SB-INTROSPECT is now able to find definitions of
1434 profiled functions. (thanks to Troels Henriksen)
1435 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1436 as well, allowing compiler-macros for SETF-functions to expand.
1437 * enhancement: step-instrumentation no longer wraps known
1438 single-value functions with multiple-value context, allowing
1439 better type inference.
1440 * fixed bug #337: use of MAKE-METHOD in method combination now works
1441 even in the presence of user-defined method classes. (reported by
1442 Bruno Haible and Pascal Costanza)
1443 * fixed bug #339(c): if there are applicable methods not part of any
1444 long-form method-combination group, call INVALID-METHOD-ERROR.
1445 (reported by Bruno Haible)
1446 * fixed bug #361: the :FUNCTION initarg in the protocol for
1447 initialization of methods can now be used to override
1448 internally-produced optimized functions. (reported by Bruno
1450 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1451 system-provided lambda expression no longer cause warnings about
1452 unbound #:|pv-table| symbols.
1453 * bug fix: improved the handling of type declarations and the
1454 detection of violations for keyword arguments with non-constant
1456 * bug fix: potentially erronous calls to PATHNAME and
1457 MERGE-PATHNAMES were being flushed in some cases.
1458 (reported by Richard Kreuter)
1459 * bug fix: compiled calls to TYPEP were mishandling obsolete
1460 instances. (reported by James Bielman and Attila Lendvai)
1461 * bug fix: format strings with ~> without matching ~< no longer
1462 trigger an AVER, but signal an understandable error instead.
1463 (reported by Antonio Martinez)
1464 * bug fix: specifying an output-file without a pathname-type for
1465 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1466 of input-file instead of "fasl". (reported by Robert Dodier)
1467 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1468 (reported by James Y Knight).
1469 * bug fix: compiler-macros-function did not consider the environment
1470 argument for shadowing by local functions.
1471 * bug fix: compiler-macros expansion was inhibited by local INLINE
1473 * bug fix: inline expansions of known functions were subject to
1474 step-instrumentation in high debug policies, leading to problems
1475 with type-inference.
1476 * bug fix: compiler failed to differentiate between different CONS
1477 types in some cases.
1478 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1479 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1480 * bug fix: #368: incorrect use of expressed vs. upgraded array
1482 * bug fix: #306a: more precise unions of array types.
1483 * thread-safety improvements:
1484 ** CONDITION-WAIT could return early on Linux, if the thread was
1485 interrupted and subsequently continued with SIGCONT.
1486 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1488 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1489 * added support for the ucs-2 external format. (contributed by Ivan
1491 * minor incompatible change: pretty printing of objects of type
1492 (cons symbol) is, in the default pprint-dispatch-table, now
1493 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1495 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1496 called later in a class's lifetime, possibly as late as when the
1497 first instance of the class is created. Previously,
1498 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1499 class became finalizeable.
1500 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1501 for multibyte external-format streams. (thanks to "vbzoli")
1502 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1503 with :ALLOCATION :CLASS inherited from superclasses of the
1505 * fixed bug: anonymous classes can now be created using the :NAME
1506 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1507 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1508 * fixed bug: core-files saved with :EXECUTABLE T can again be
1509 executed when SBCL_HOME isn't set. (reported by James Knight)
1510 * fixed bug: toplevel LOCALLY forms with declarations could
1511 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1512 * fixed bug: printing from several different threads using different
1513 values of *print-case* could cause invalid output, due to
1514 some internal special variables of the printer not being bound
1515 thread-locally (reported by Max Mikhanosha)
1516 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1517 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1518 (reported by Pascal Costanza)
1519 * fixed bug: FINALIZE-INHERITANCE is called from
1520 REINITIALIZE-INSTANCE on classes when the class has previously
1521 been finalized, as required by AMOP.
1522 * minor code generation optimizations:
1523 ** better register allocation in CLOS dispatching functions
1524 ** overflow detection when coercing signed bytes to fixnums on x86-64
1525 ** is now implemented with one IMUL instruction instead of three shifts
1526 ** more efficient bit-vector access on x86 and x86-64
1527 ** more efficient access to raw structure slots on x86 and x86-64
1528 * fixed some bugs revealed by Paul Dietz' test suite:
1529 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1530 return its argument.
1532 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1533 * feature: thread support on Solaris/x86, and experimental thread support
1535 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1537 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1538 default method for SLOT-UNBOUND.
1539 * minor incompatible change: prevent the user from specializing the
1540 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1541 somewhat counter to the intent of the protocol, I (CSR) think, and
1542 additionally it just doesn't work in SBCL as currently
1543 implemented, thanks to optimizations (that are always valid for
1544 the other three SLOT-VALUEish functions, but not for the setter).
1545 * minor incompatible change: the :SB-LDB feature is now enabled by
1546 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1547 the low-level debugger.
1548 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1549 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1550 down to an SB-SYS:FD-STREAM, allowing direct communication in
1551 more cases. Temporary files and pipes are still used as a fallback
1553 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1554 on the non-reentrant C functions localtime(3) and gmtime(3).
1555 * bug fix: global optimization policy was not visible in LOCALLY and
1557 * bug fix: class objects can be used as specializers in methods.
1558 (reported by Pascal Costanza)
1559 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1560 with a type error. (reported by blitz_ on #lisp)
1561 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1562 failed with a type error. (reported by Pascal Bourguignon)
1563 * bug fix: merging pathnames against defaults with :DIRECTORY
1564 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1565 (reported by James Y Knight)
1566 * bug fix: saving large (>2GB) cores on x86-64 now works
1567 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1568 constant index and a (simple-array (signed-byte 32)) array
1569 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1570 error (patch by Robert J. Macomber)
1571 * feature: TIME now displays the amount of run-time spent in GC
1572 * bug fix: The debugger now does a better job of respecting
1573 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1574 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1575 workaround for bug 403.)
1576 * bug fix: floating point exception handling now works on FreeBSD
1577 (thanks to NIIMI Satoshi)
1578 * fixed some bugs revealed by Paul Dietz' test suite:
1579 ** MISC.641: LET-conversion were not supposed to work in late
1581 * improvements to the Win32/x86 port:
1582 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1583 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1585 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1586 * new feature: source path information is generated for macro-expansion
1587 errors for use in IDE's like Slime (thanks to Helmut Eller)
1588 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1589 * bug fix: compiler does not loop forever on an invalid type in
1591 * improvement: compilation of most CLOS applications is significantly
1593 * optimization: added a limited bytecode compiler for simple toplevel
1594 forms, speeding up compilation and FASL loading
1595 * bug fix: the statistical profiler now properly distinguishes anonymous
1598 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1599 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1600 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1601 into subclasses of the finalized class. Now user calls to
1602 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1603 system return before any subclasses are finalized.
1604 * minor incompatible change: The reader no longer ignores errors
1605 regarding non-existent packages in #+ and #- feature tests.
1606 * new feature: command line options --no-sysinit, --no-userinit to
1607 inhibit loading the corresponding init files
1608 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1609 for sending data through UDP sockets (thanks to François-René Rideau)
1610 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1611 error is signalled instead (thanks to François-René Rideau)
1612 * minor incompatible change: Error signalling behaviour of lexical
1613 operations violating package locks has changed slightly. Refer to
1614 documentation on package locks for details.
1615 * enhancement: EVAL can process IF-expressions without resorting to the
1617 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1618 constant-folded properly due to missing out-of-line functions.
1619 (reported by Mika Pihlajamäki)
1620 * bug fix: LISTEN sometimes returned T even in cases where no data was
1621 immediately available from the stream
1622 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1623 were reversed. (reported by Levente Mészáros)
1624 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1625 when given a vector displaced to an adjustable array. (reported
1627 * bug fix: derivation of float boundaries from numbers outside the
1628 appropriate float range (reported by John Wiseman)
1629 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1631 * bug fix: precomputing cache entries for generic functions with
1632 some subclasses of specializers as yet invalid does not attempt to
1633 fill a cache line with a negative offset. (reported by Levente
1635 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1636 allow condition class objects as arguments to DOCUMENTATION and
1637 (SETF DOCUMENTATION); only find and set documentation for
1638 structure names for the STRUCTURE doc-type. (suggested by Gary
1640 * improvements to the Win32/x86 port:
1641 ** (user-homedir-pathname) and default initialization file
1642 locations now know about the user's "Documents and Settings"
1643 directory (thanks to Yaroslav Kavenchuk)
1644 ** run-program is implemented (thanks to Mike Thomas)
1645 ** sockets support (thanks to Timothy Ritchey)
1646 ** better backtrace support (thanks to Alastair Bridgewater)
1647 ** sb-grovel supported
1648 ** asdf-install and sb-posix work somewhat
1649 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1650 * improvements to the Solaris/x86 port:
1651 ** works on Solaris 11/Solaris Express
1652 ** floating-point exception handling support
1653 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1654 * fixed some bugs revealed by Paul Dietz' test suite:
1655 ** REMOVE-METHOD returns its generic function argument even when
1656 the method is not one of the generic functions' methods.
1657 ** objects declared as MEMBER types can be admissible arguments to
1658 structure accessors.
1659 ** printing characters should simply be printed by the FORMAT ~:C
1661 ** compiler failure when compiling functions with hairy constant
1662 defaults for optional parameters.
1663 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1664 function, which is already optimized.
1666 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1667 * new platform: experimental support for SBCL x86/Darwin, including
1668 MacOS X 10.4.5 on Intel.
1669 * new feature: Unicode character names are now known to the system
1670 (through CHAR-NAME and NAME-CHAR).
1671 * new feature: the filesystem location of SBCL's core file is
1672 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1673 * minor incompatible change: the contrib modules SB-POSIX and
1674 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1675 this change is to make it easier to distribute
1676 location-independent binaries.
1677 * bug fix: as implied by AMOP, standardized classes no longer have
1678 slots named by external symbols of public packages. (reported by
1680 * optimization: calling functions via constant symbols -- as in
1681 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1682 via the function object as in (FUNCALL #'FOO).
1683 * optimization: on x86 and x86-64, the calling convention for the
1684 case of dealing with an unknown number of values has been altered
1685 to be friendlier to the prediction heuristics implemented,
1686 particularly with reference to CALL and RET pairing. (thanks to
1687 Alastair Bridgewater)
1688 * optimization: on x86, the code for access to thread-local symbol
1689 values and binding/unbinding of thread-local symbols is smaller.
1690 (thanks to Alastair Bridgewater)
1691 * enhancement: CONSTANTP is now able to determine constantness of
1692 more complex forms, including calls to constant-foldable standardized
1693 functions and some special forms beyond QUOTE.
1694 * fixed bug: occasional GC crashes on Solaris/x86
1695 * optimization: x86-64 supports stack allocation of results of simple
1696 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1697 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1698 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1699 (thanks to James Knight)
1700 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1701 by the standard (thanks to Stephen Compall)
1703 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1704 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1705 be used for bundling the runtime and the core file into one
1706 executable binary. This feature is not currently supported on all SBCL
1707 platforms. (thanks to James Bielman and NIIMI Satoshi)
1708 * new feature: a generational or ephemeral garbage collector is now
1709 the default on the PowerPC platform (both Linux and Darwin). The
1710 old Cheney (stop and copy) collector is a build-time option.
1711 (thanks to Cyrus Harmon, after Raymond Toy)
1712 * minor incompatible change: the method by which SBCL finds its
1713 contributed modules has changed; it no longer relies on symbolic
1714 links from an $SBCL_HOME/systems directory, but searches directly
1715 in the subdirectories of $SBCL_HOME.
1716 * enhancement: the dynamic heap size on the Linux/PPC platform is
1717 markedly larger, even using the older Cheney garbage collector.
1718 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1719 certain complicated string types. (reported by Gary King)
1720 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1721 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1722 * fixed bug: DOCUMENTATION on structures no longer signals an
1723 error if no structure type of the right name exists. (reported by
1725 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1726 applicable non-standard primary or :AROUND method on
1727 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1728 no longer cause unbound variable errors. (reported by Kevin Reid)
1729 * fixed bug: metacircle resolution in cases where methods have slots
1730 added before the slots from STANDARD-METHOD. (reported by Jean
1732 * fixed bug: the Power PC assembly code for calling into lisp
1733 sometimes computed its return address wrongly (depending on the
1734 15th bit of the address, fixed at link time). This used to
1735 manifest itself in a segmentation violation while building PCL.
1736 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1737 many others over the years)
1738 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1739 the stack pointer (treated as a signed integer) is negative.
1740 (thanks to Peter van Eynde)
1742 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1743 * new platform: experimental support for the Windows operating
1744 system has been added. (thanks to Alastair Bridgewater)
1745 * fixed several bugs in and robustified the PPC FFI (including
1746 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1747 * bug fix: allow non-simple string symbol names (reported by Paul
1749 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1751 * bug fix: saving a core corrupted callbacks on x86/x86-64
1752 * bug fix: closed a loophole in metacircularity detection and
1753 grounding in the PCL implementation of CLOS.
1754 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1755 (reported by tomppa on #lisp)
1756 * bug fix: clear the direction flag on Lisp -> C transitions, as
1757 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1758 (reported by Andrej Grozin and Hendrik Maryns)
1759 * optimization: major improvements to GC efficiency on GENCGC platforms
1760 * optimization: faster implementation of EQUAL
1761 * optimization: emit more efficient opcodes for some common
1762 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1763 * optimization: several other minor code-generation improvements on x86-64
1764 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1766 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1767 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1768 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1769 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1770 by ANSI as a generic function, and for consistency (SETF
1771 GENERIC-FUNCTION-NAME) is treated likewise.
1772 * fixed bug #233.b: make constraint propagation notice when a variable
1773 value is changed after it is referenced but before it is used
1774 * fixed bug #296: no more arbitrary behaviour from filenames with
1775 odd characters as --load arguments. (reported by Adam Warner)
1776 * fixed bug #390: :CHARACTER-SET pathname components now work as
1777 expected. (reported by Tim Daly Jr)
1778 * fixed bug #391: complicated :TYPE intersections in slot
1779 definitions no longer cause an error in PCL internals.
1780 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1781 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1782 returning the number of octets which would be written to the
1783 file-stream. (thanks to Robert J. Macomber)
1784 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1785 arguments to the internal stream functions. (thanks to David
1787 * bug fix: stack exhaustion is now handled on the x86/SunOS
1788 platform. (thanks to Daisuke Homma)
1789 * bug fix: more accurate ROOM results on GENCGC platforms
1790 * optimization: improved type inference for arithmetic-for
1791 index variables in LOOP
1792 * optimization: faster floating-point SQRT on x86-64
1793 * fixed some bugs revealed by Paul Dietz' test suite:
1794 ** DOCUMENTATION returns NIL instead of "" for method combinations
1795 that don't have a docstring
1797 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1798 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1799 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1800 therefore are not customizeable by user code (as seems to be at
1801 least permitted and maybe required by AMOP). As a consolation,
1802 however, the SBCL implementation of these functions now calls
1803 REINITIALIZE-INSTANCE as specified by AMOP.
1804 * bug fix: slot-definition documentation is propagated as per ANSI
1805 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1806 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1807 argument, as do eslotd initialization methods. (from Pascal
1808 Costanza's "Closer" project)
1809 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1810 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1812 * bug fix: it is now possible to have more than one subclass of
1813 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1814 (reported by Bruno Haible, Pascal Costanza and others)
1815 * bug fix: the dependent update protocol now works for generic
1816 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1817 and Pascal Costanza)
1818 * bug fix: condition-class instances corresponding to
1819 DEFINE-CONDITION forms are now created eagerly. (reported by
1820 Kalle Olavi Niemitalo on comp.lang.lisp)
1821 * bug fix: floating point printing is more accurate in some
1822 circumstances. (thanks to Simon Alexander)
1823 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1824 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1825 * bug fix: callbacks on OS X now preserve stack-alignment.
1826 (thanks to Cyrus Harmon)
1827 * enhancement: the x86-64 disassembler is much better at
1828 disassembling SSE instructions. (thanks to Lutz Euler)
1829 * enhancement: CHECK-TYPE on a variable now allows type-inference
1830 (thanks to Vincent Arkesteijn)
1831 * optimization: improved performance of EUC-JP external format.
1832 (thanks to NIIMI Satoshi)
1833 * optimization: performance improvements to IO on file streams of
1834 :ELEMENT-TYPE CHARACTER
1835 * optimization: much faster memory allocation on x86-64
1836 * optimization: faster garbage collections (latency and throughput)
1838 * optimization: faster bignum subtraction and division on x86 and x86-64
1839 (thanks to Lutz Euler)
1840 * optimization: more accurate type-inference for loops that use a
1841 floating point index variable or a negative step.
1843 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1844 * bug fix: add a workaround to SBCL looping infinitely at startup on
1845 Linux kernels with apparently buggy implementations of personality().
1846 (thanks to Svein Ove Aas)
1847 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1848 (thanks to David Lichteblau)
1849 * bug fix: local bindings shadow global symbol macros for
1850 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1851 * optimization: non-open coded uses of numeric comparison operators
1852 (e.g. >) no longer cons when called with more than one parameter
1853 on platforms supporting dynamic-extent allocation.
1854 * enhancement: saving cores with foreign code loaded is now supported
1855 on MIPS/Linux in addition to the previously supported platforms.
1856 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1857 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1858 * bug fix: division by zero in sb-sprof when no samples were collected
1859 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1860 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1861 (thanks to Alexander Kjeldaas)
1862 * new feature: ignoring the return values of destructive functions
1863 like NREVERSE signals a compile-time style-warning.
1864 (thanks to Kevin Reid)
1865 * new feature: the alignment of alien structure fields can be
1866 explicitly specified. (thanks to Cyrus Harmon)
1867 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1868 and dump core on SIGQUIT
1870 ** incompatible change: threads do not inherit values of specials
1871 from their parents (see manual)
1872 ** bug fix: threads stacks belonging to dead threads are freed by the
1873 next exiting thread, no need to gc to collect thread stacks anymore
1874 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1875 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1876 ** bug fix: dynamic variable and thread start related gc lossage
1877 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1879 * fixed some bugs revealed by Paul Dietz' test suite:
1880 ** SUBTYPEP is slightly more accurate on heinously complicated
1881 CONS types where some of the members have uncertain (in the
1882 NIL, NIL sense) type relationships to each other.
1884 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1885 no longer purified unless :PURIFY T is explicitly specified.
1886 ** Non-purified cores are significantly smaller than before
1888 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1889 * new feature: timers based on Zach Beane's excellent timer package
1890 * added support for the following external formats: koi8-u,
1891 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1892 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1893 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1894 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1895 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1896 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1897 * incompatible change: a threaded SBCL will no longer revert to
1898 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1899 * bug fix: interrupts are disabled until startup is complete; no
1900 more sigsegvs when receiving a signal to soon
1901 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1903 * bug fix: add a workaround for the memory randomization features in
1904 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1905 workaround will only be in effect on systems with the proc filesystem
1907 * bug fix: printing objects of type HASH-TABLE signals a
1908 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1910 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1911 have been running for over 50 days. (reported by Gilbert Baumann)
1912 * bug fix: the logic for getting names of functions gets less
1913 confused when confronded with alternate-metaclass
1914 funcallable-instances. (reported by Cyrus Harmon)
1915 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1916 consistent, even on internal alternate-metaclass objects.
1917 * bug fix: flush closure information collected by physical
1918 environment analysis prepass before the main pass. (bug reported
1920 * bug fix: compiler pack phase does not modify a hashtable, which is
1921 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1922 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1923 may be expanded. (reported by Alexander <malishev> on c.l.l)
1924 * bug fix: no more style-warnings when compiling code using LAST
1925 under high-SPEED low-SPACE settings. (reported by David Wallin
1927 * fixed some incompatibilities between SBCL's MOP and the MOP
1929 ** the METAOBJECT class is now implemented;
1930 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1931 STANDARD-OBJECT, as required;
1932 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1933 now compatible (as required by VALIDATE-SUPERCLASS); there
1934 remains a consistency requirement over the presence or absence
1935 of the FUNCTION class in the superclass list of finalized
1936 classes; see the manual for more details;
1937 ** the :AROUND method for COMPUTE-SLOTS on
1938 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1939 requested slot ordering.
1941 ** bug fix: parent thread now can be gc'ed even with a live
1943 ** bug fix: binding a special with PROGV to no value is not
1944 the same as that symbol not having been bound (thanks to
1946 ** bug fix: binding specials is thread safe (thanks to
1948 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1949 and SB-SPROF do not die with 'no handler for signal XX in
1950 interrupt_handle_now(..)' anymore
1951 ** bug fix: WITH-TIMEOUT works with multiple running threads
1952 * fixed some bugs revealed by Paul Dietz' test suite:
1953 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1954 the :method-class keyword argument.
1956 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1957 * new port: the Solaris operating system on x86 processors is now
1958 mostly supported, though some rough edges in the environment
1959 remain. (thanks to Daisuke Homma)
1960 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1961 space for its heap. (thanks to Thiemo Seufer)
1962 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1963 closures. (thanks to Thiemo Seufer)
1964 * minor incompatible change: eof selects abort in the debugger.
1965 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1966 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1968 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1969 deprecated, and will go away in a future revision of SBCL.
1970 * minor incompatible change: GC-ON and GC-OFF are no longer
1971 implemented with a counter, it does not matter how many times gc
1972 is switched on or off
1973 * bug fix: discriminating functions for generic function classes
1974 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1975 make invalid assumptions about method precedence order. (reported
1977 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1978 platforms. (thanks to Thiemo Seufer)
1979 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1980 * bug fix: correct pathname printing with printer escaping is on.
1981 (thanks to Kevin Reid)
1982 * bug fix: complex VOP definitions in "user-space" no longer trigger
1983 package locks. (reported by Zach Beane)
1984 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1985 for funcallable-instance functions, and is no different from
1987 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1988 characters in the current readtable. (reported by Nicholas Neuss)
1989 * bug fix: SUBTYPEP on various CONS types returns more a more
1990 accurate acknowledgment of its certainty.
1991 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1992 lists in some cases. This partially fixes bug 384.
1993 * flush all standard streams before prompting in the REPL and the
1995 * bug fix: signal handling and triggering gc do not conflict
1996 directly anymore, in particular a high frequency sb-sprof does
1997 not prevent gc from running
1998 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1999 approximation for timezone and DST information between the
2000 universal time epoch and the smallest negative 32-bit time_t.
2001 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2002 year 1899 so long as the encoded time is non-negative.
2003 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2004 the correct answer for non-constant rotate amounts when performing
2005 an inline 32-bit rotation.
2007 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2008 there is only one thread in the session
2009 ** bug fix: memory leak for streams created in one thread and
2010 written to in another
2011 ** bug fix: lockup when compiled with gcc4
2012 ** bug fix: race that allows the gc to be triggered when gc is
2014 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2015 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2017 ** bug fix: finished threads are gc'ed properly
2018 * fixed some bugs revealed by Paul Dietz' test suite:
2019 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2020 has arguments with a different set of applicable methods from
2021 the orignal arguments.
2022 ** The type error thrown by MAP now has a correct expected-type
2024 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2025 name a compiled function.
2026 ** Three MISC tests where a large bignum was improperly coerced to
2027 a float that couldn't represent that bignum during type
2028 derivation were fixed.
2029 ** SUBTYPEP can now handle types involving the negation of a
2030 list-form FUNCTION type.
2031 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2032 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2033 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2035 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2036 * New feature: Experimental support for bivalent streams: streams
2037 opened with :element-type :default now allow character and binary
2038 (unsigned-byte 8) I/O
2039 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2040 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2042 * bug fix: no more highly sporadic "couldn't check whether ~S is
2043 readable" when reading a stream and an interrupt hits in the middle
2044 of a select system call
2045 * compiler better recognizes complex arrays (reported by Eduardo
2047 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2048 for variables on SBCL built with linkage-tables. (reported by Luis
2050 * various error reporting improvements.
2051 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2052 (Thanks to Thiemo Seufer)
2053 * bug fix: nested reader invokations work correctly
2054 * bug fix: it is possible to have simultaneous references to foreign
2055 code and foreign data with the same name.
2057 ** added x86-64 support
2058 ** incompatible change: the threading api now works with thread
2059 objects instead of thread ids
2060 ** bug fix: threads are protected from signals and interruption when
2061 starting up or going down
2062 ** bug fix: a race where an exiting thread could lose its stack to gc
2063 ** bug fix: don't halt on infinite error in threads if possible
2064 ** fixed numerous gc deadlocks introduced in the pthread merge
2065 ** bug fix: fixed thread safety issues in read and print
2066 ** bug fix: debugger doesn't hang on session lock if interrupted at
2067 an inappropriate moment
2068 ** bug fix: run-program is now thread safe(r)
2069 ** bug fix: inner with-recursive-lock no longer releases the mutex
2070 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2071 * fixed some bugs revealed by Paul Dietz' test suite:
2072 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2073 EXPECTED-TYPE slots filled.
2074 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2076 ** MAKE-CONDITION accepts classes as type-designators.
2077 ** COMPILE may never return NIL.
2078 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2079 range before calling Unix time functions
2081 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2082 * numerous signal handling fixes to increase stability
2083 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2084 * minor incompatible change: we now correctly canonize default
2085 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2086 as per the MOP, rather than the historical (INITARG INITFUNCTION
2087 INITFORM). (reported by Bruno Haible)
2088 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2089 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2090 for more information.
2091 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2092 pathname is a directory pathname.
2093 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2094 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2096 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2097 :METACLASS argument in addition to a class name. (reported by
2098 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2099 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2100 (reported by Bruno Haible; patch from Gerd Moellmann)
2101 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2103 * bug fix: more cleanups to the floating point exception handling on
2104 x86-64 (thanks to James Knight)
2105 * bug fix: the compiler does not try to constant fold calls of
2106 COERCE and BIT-* functions when they return freshly constructed
2107 objects. (reported by Nikodemus Siivola and Paul Dietz)
2108 * optimization: improved the MIPS versions of generic (in the
2109 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2110 * optimization: direct conversion of (unsigned-byte 32) to floats on
2111 the PowerPC platform.
2112 * optimization: structure instances with raw slots now use less
2113 memory, and probably show better memory locality. (thanks to
2115 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2116 are treated as requests for stack allocation on the x86-64,
2117 Alpha-32, PPC and SPARC platforms.
2118 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2119 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2120 (thanks to Tony Martinez)
2122 ** gcing a dead thread can no longer lead to lockups
2123 ** threads block signals until they are set up properly
2124 ** errno is no longer shared by threads
2125 ** interrupt-thread restores the eflags register on x86
2126 ** fixed some lockups due to gc/thread interaction
2127 * fixed some bugs revealed by Paul Dietz' test suite:
2128 ** invalid dotted lists no longer raise a read error when
2129 *READ-SUPPRESS* is T
2130 ** signal an error if a symbol that names a declaration is used
2131 as the name of a type, or vice versa
2132 ** allow using the (declare (typespec var*)) abbreviation for
2133 (declare (type typespec var*)) with all type specifiers
2134 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2135 ** always return NIL from PPRINT-POP when OBJECT is NIL
2136 ** don't signal errors when pretty-printing malformed LABELS,
2137 FLET or MACROLET forms
2138 ** declarations in a DOLIST body are also in scope for the
2140 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2142 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2145 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2146 * fixed cross-compiler leakages that prevented building a 32-bit
2147 target with a 64-bit host compiler.
2148 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2149 opened with :IF-EXISTS :OVERWRITE.
2150 * fixed bug 281: error for an invalid qualifier in a short-form method
2151 combination method is not signalled until the faulty method is called.
2152 * bug fix: iteration variable type inferrer failed to deal with open
2153 intervals. (reported by Alan Shields)
2154 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2155 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2156 or more runtime options were provided to the sbcl binary.
2157 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2158 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2159 * optimization: there's now a fast-path for fixnum arguments in the
2160 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2161 * optimization: the code generated on x86-64 is more compact thanks
2162 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2163 * bug fix: floating-point exception handling now partly works on
2164 x86-64. (Thanks to James Knight)
2165 * improvement to the MIPS backend from Thiemo Seufer: C-style
2166 64-bit long long arguments and return values to alien functions
2168 * fixed some bugs revealed by Paul Dietz' test suite:
2169 ** the type-error signalled from WARN has a filled-in DATUM slot.
2170 ** the type-error required when a stream is not associated with
2171 a file has the stream as its datum.
2172 ** type-errors on single-floats on x86-64 no longer have
2173 :INVALID-OBJECT as the datum
2174 ** the type-errors signalled for invalid function names now have
2175 a correct expected type
2176 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2177 for structures defined with a :TYPE. Documentation strings for
2178 typed structures are no longer immediately discarded
2179 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2180 broadcast streams and synonym streams. FILE-LENGTH now also works
2181 on broadcast streams.
2183 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2184 * incompatible change: the --noprogrammer option, deprecated since
2185 version 0.7.5, has been removed. Please use the equivalent
2186 --disable-debugger option instead.
2187 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2189 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2190 inoperational for a while now) has been completely removed.
2191 * null lexical environments are now printed as #<NULL-LEXENV>,
2192 significantly reducing the amount of clutter in typical backtraces.
2193 * documentation on weak pointers, finalization, and after GC hooks
2194 has been added to the manual.
2195 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2196 as other specialized array types, is much faster. SUBSEQ and
2197 COPY-SEQ on such arrays have also been sped up.
2198 * optimization: EQL is now more efficient when at least other argument
2199 is known to be of type (OR FIXNUM (NOT NUMBER)).
2200 * fixed bug: the runtime is now less vulnerable to changes in the
2201 size of the SBCL object on OS X, and virtual memory is reserved for
2202 all spaces that need to be at a fixed address.
2203 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2204 * fixed bug: finalizers and after GC hooks that cause consing are now
2206 * fixed bug: compiler error messages and summaries are now printed to
2207 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2208 * fixed inference of the upper bound of an iteration variable.
2209 (reported by Rajat Datta).
2210 * fixed bug 211e: calling local functions with duplicated constant
2211 keyword argument no longer causes a bogus style warning about an
2213 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2214 ftype declaration to be disregarded. (reported by Dave Roberts)
2215 * fixed bug 373: caused by erronous compilation of references to alien
2216 variables in the runtime on ppc/darwin.
2217 * fixed bug 376: CONJUGATE type deriver.
2218 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2219 (reported by Baughn on #lisp)
2220 * a cross-compiler bug on non-x86 platforms has been identified and
2221 fixed. (thanks to Bruno Haible)
2222 * improvements to the MIPS runtime code for increased stability.
2223 (thanks to Thiemo Seufer)
2224 * increased the maximimum compact environment size to allow
2225 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2226 * partial workaround for bug 135: don't name the function we're
2227 calling for hairy cases of EVAL, so as not to accumulate one environment
2228 entry per call to EVAL. (reported by Kevin Reid)
2229 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2230 * optimization: type testing for non-vector arrays should be faster.
2231 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2233 * bug fix: setting 31st element of a bit vector to zero did not work
2235 * bug fix: redefining a class definition which failed due to a
2236 previous accessor / function clash now works (but see BUGS entry
2237 #380 for more problems in this area). (thanks to Zach Beane)
2238 * the long form of DEFINE-METHOD-COMBINATION disables method group
2239 checking when given a single method group with pattern *. (thanks
2241 * on x86 compiler supports stack allocation of results of simple
2242 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2243 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2244 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2245 * fixed some bugs related to Unicode integration:
2246 ** the restarts for recovering from input and output encoding
2247 errors only appear when there is in fact such an error to
2249 * fixed some bugs revealed by Paul Dietz' test suite:
2250 ** MISC.549 and similar: late transformation of unsafe type
2251 assertions into derived types caused unexpected code
2253 ** SCALE-FLOAT type deriver is less wrong.
2254 ** type derivers for EXP, LOG and similar functions compute result
2255 types for complex arguments better.
2256 ** (MISC.563) CONJUGATE type deriver works for very restricted
2258 ** out-of-line type testers for character strings are available.
2259 ** EQUAL compiler transform understands specialness of objects
2261 ** accessing double-floats stored on the stack now works on x86-64.
2262 ** debugger internals could sometimes create invalid lispobjs,
2263 resulting in GC crashes.
2264 ** MISC.548: type check weakening can convert required type into
2266 ** initialization forms for bindings are not in scope of free special
2269 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2270 * incompatible change: thread support for non-NPTL systems has
2271 been removed - locking is buggy and unreliable. A threaded
2272 SBCL build will now warn at startup and refuse to create
2273 new threads, unless futex support is detected in the kernel
2274 * incompatible change: the top level REPL now has only an ABORT
2275 restart associated with it, not TOPLEVEL and ABORT as it used to.
2276 TOP and TOPLEVEL are now available as debugger commands for
2277 returning to the top level.
2278 * incompatible change: forms evaluated in the REPL now use the
2279 global optimization policy.
2280 * incompatible change: user- and system-initialization files are
2281 no longer processed with LOAD, but by READ and EVAL; hence the
2282 global optimization policy, startup package, readtable, etc,
2284 * The .fasl file format number has been incremented because of
2285 various incompatible changes.
2286 * internal entry point details and argument counts no longer appear
2287 in backtraces unless explicitly requested by setting
2288 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2289 * built-in and standard functions no longer have names like "top
2290 level local call to FOO".
2291 * fixed bug 32: functions defined in non-null lexical environments
2292 now have more legible printed representation
2293 * fixed bug 33: functions defined in non-null lexical environemnts
2294 are now more amenable to inspection by INSPECT.
2295 * workaround for bug 354: XEPs no longer appear in backtraces unless
2296 explicitly requested.
2297 * fixed bug: receiving the signal which results from attempting to
2298 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2299 SIGBUS on other *BSDs) on architectures where the C stack is also
2300 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2301 to debugger confusion or wild instability. (reported by Cheuksan
2303 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2304 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2305 (reported by Lutz Euler)
2306 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2307 compiled code. (reported by Paul Dietz)
2308 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2309 specialized parameter to a method to something that is not TYPEP
2310 the specializer is now possible.
2311 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2312 face of package deletion.
2313 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2314 pathname. (reported by David Tolpin on c.l.l)
2315 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2316 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2317 STRING-STREAM and FILE-STREAM, respectively.
2318 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2319 than previously; in addition, its test suite is now run on build.
2320 (thanks to Luke Gorrie)
2321 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2322 * optimization: added a immediate representation for single-floats
2324 * fixed some bugs revealed by Paul Dietz' test suite:
2325 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2326 ** attempting to create a package with a colliding nickname causes
2327 correctable errors to be signalled.
2328 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2329 ** several x86-64 backend bugs related to sign-extension of immediate
2332 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2333 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2334 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2336 * fixed loading of multiply forward-referenced layouts.
2337 (thanks to Cheuksan Wang)
2338 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2339 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2340 (see COMPILE-FILE documentation for details of :PRINT option)
2341 * fixed bugs 19 and 317: fixed-format floating point printing is
2342 more accurate. This also fixes a bug reported by Adam Warner
2343 related to the ~@F format directive.
2344 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2346 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2347 dispatch table if the from-char is a dispatch macro character.
2348 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2349 by the wordsize no longer ignore the last word. (reported by Lutz
2351 * fixed bug in type checking of dynamic-extent variables. (reported
2353 * optimization: sequence traversal functions use their freedom to
2354 coerce function designators to functions.
2355 * optimization: code with many calls to CLOS methods specialized on
2356 CLOS classes has had redundant type checks removed.
2357 * optimization: added declarations to speed up operations that access
2358 the internal character database (for example STRING-UPCASE)
2359 * optimization: comparison operations between floats and sufficiently small
2360 fixnums no longer create extra rationals
2361 * fixed some bugs related to Unicode integration:
2362 ** portions of multibyte characters at the end of buffers for
2363 character-based file input are correctly transferred to the
2364 start of the buffer at the next read.
2365 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2366 passing it through to OPEN.
2367 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2368 argument given to internal calls to OPEN.
2369 ** the built-in comment readers (introduced by character sequences
2370 ";" and "#|") are more forgiving to encoding errors; they will
2371 STYLE-WARN and then attempt to resync the stream at a character
2372 boundary. (thanks to Teemu Kalvas)
2373 * fixed some bugs in the x86-64 port:
2374 ** Negative short int return values from c-calls are sign-extended
2376 ** The stack is aligned to 16-bytes for c-calls, as required by
2377 the ABI. (thanks to Cheuksan Wang)
2378 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2379 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2380 for enums. (thanks to Vincent Arkesteijn)
2381 ** Multiple small optimizations and bugfixes for floating point
2383 * fixed some bugs revealed by Paul Dietz' test suite:
2384 ** Space, Tab, Linefeed, Return and Page have the invalid
2385 secondary constituent character trait.
2386 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2388 ** WITH-INPUT-FROM-STRING should only update the index place on
2390 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2391 works more reliably.
2392 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2393 with one-dimensional arrays with fill pointers.
2394 ** TYPE-OF failed on a complex with an integer realpart and a
2396 ** compiler failure during type inference for the code of form
2397 (IF (EQL X (THE ...)) ...) (MISC.535).
2399 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2400 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2401 platform. The port supports 61-bit fixnums, large memory spaces
2402 and reloading shared object files.
2403 * enhancement: saving cores with foreign code loaded is now
2404 supported on ppc/Darwin in addition to the previously supported
2406 * enhancement: the statistical profiler now walks deeper into the
2407 call stack for more meaningful call-graphs and accrued time
2408 reports (x86/x86-64 only). It also now reports time spent in
2410 * enhancement: it is now possible to trace most individual methods
2411 of a generic function in addition to tracing the generic function
2413 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2415 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2416 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2417 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2418 produces an error. (thanks to Vincent Arkesteijn)
2419 * bug fix: NAMESTRING on pathnames with :WILD components in their
2420 directories works correctly. (thanks to Artem V. Andreev)
2421 * fixed bug 125: compiler preserves identity of closures. (reported
2423 * bug fixed: functions with &REST arguments sometimes failed with
2424 "Undefined function" when compiled with (DEBUG 3). (reported by
2426 * bug fix: overflow during compiling of setting element of a bit
2427 vector with constant index and value. (reported by Timmy Douglas)
2428 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2429 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2430 * on x86 compiler partially supports stack allocation of dynamic-extent
2432 * GO and RETURN-FROM do not check the extent of their exit points
2433 when compiled with SAFETY 0.
2434 * fixed some bugs related to Unicode integration:
2435 ** encoding and decoding errors are now much more robustly
2436 handled; it should now be possible to recover even from invalid
2437 input or output to the terminal. (thanks to Teemu Kalvas)
2438 ** provided a first cut at implementing STRING-TO-OCTETS and
2439 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2440 ** altered the SB-MD5 contributed module slightly, changing the
2441 interface just enough for it to be supportable for builds where
2442 lisp characters are not eight bits.
2443 * fixed some bugs revealed by Paul Dietz' test suite:
2444 ** the FORMATTER-generated functions for ~V[ conditionals require
2445 the correct number of arguments.
2446 ** READ-FROM-STRING returns the mandated second value when applied
2447 to displaced strings.
2448 ** the #\Rubout and #\Backspace characters are treated as invalid
2449 constituent characters by the tokenizer.
2451 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2452 * new feature: reloading changed shared object files with
2453 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2454 * new feature: references to foreign variables and functions
2455 can now be compiled and loaded before the corresponding shared
2456 object file is loaded, as long as the foreign definitions are
2457 available at runtime.
2458 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2459 supported. (thanks to Dan Debertin)
2460 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2461 just keywords. (thanks to Vincent Arkesteijn)
2462 * fixed bug #331: structure-class instances corresponding to
2463 DEFSTRUCT forms are now created eagerly.
2464 * fixed bug #345: backtraces from calls to undefined functions work
2465 on x86 as well. Related bug #61 is now also partially fixed on x86
2466 (backtraces from throws to unknown catch tags.)
2467 * bug fix: lambda-list parsing is now stricter vrt. order and number
2468 of lambda-list keywords.
2469 * bug fix: as specified by AMOP, an error is signalled if a
2470 class-option appears multiple times in a DEFCLASS form. (reported
2472 * bug fix: RANDOM can be compiled when the compiler derives the type
2473 of its numeric argument as a disjoint set of small integers.
2474 (reported by Paul Dietz)
2475 * bug fix: starting a core saved with shared objects loaded when
2476 those objects are not available no longer causes threaded SBCL to
2477 hang. (reported by Sean Ross)
2478 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2480 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2481 argument, which must always be NIL. (reported by Kalle Niemitalo)
2482 * bug fix: printing 1.0d+23 no longer results in an error.
2483 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2484 * bug fix: structure slot setters preserve evaluation order. (thanks
2486 * fixed some bugs related to Unicode integration:
2487 ** RUN-PROGRAM can allow its child to take input from a Lisp
2488 stream. (reported by Stefan Scholl)
2489 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2490 locales. (reported by Ken Causey)
2491 * fixed some bugs revealed by Paul Dietz' test suite:
2492 ** INCF, DECF and REMF evaluate their place form as specified in
2494 ** FORMATTER expands ~{ ~} iteration directives with V or #
2495 parameters correctly.
2496 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2497 consequent uses no arguments correctly.
2498 ** the system has a partial understanding of the (COMPLEX RATIO)
2500 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2503 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2504 * new feature: a build-time option (controlled by the :SB-UNICODE
2505 keyword feature, enabled by default) for building the system with
2506 support for the entire 21-bit character space defined by the
2508 * new feature: the system now has rudimentary external-format
2509 support; the primary user-visible change at this time is that
2510 characters with the high bit set (such as lower-case-e-acute) will
2511 print correctly to a terminal in a UTF-8 environment.
2512 * minor incompatible change: BASE-CHAR no longer names a class;
2513 however, CHARACTER continues to do so, as required by ANSI.
2514 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2515 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2516 has been moved to the SB-EXT package (temporarily re-exported from
2518 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2519 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2521 * fixed bug #351: better error-handlind and reporting for malformed
2523 * fixed bug #350: bignum-printing is now more memory-efficient,
2524 allowing printing of very large bignums, eg. (expt 2 10000000).
2525 (reported by Bruno Haible)
2526 * fixed bug #302: better primitive-type selection for intersection
2528 * fixed bug #308: non-graphic characters now all have names, as
2529 required. (reported by Bruno Haible)
2530 * bug fix: redefining a class with different superclasses now correctly
2531 removes it from the direct-subclasses of its previous superclasses.
2532 (reported by David Morse)
2533 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2534 new value now works. (reported by Bruno Haible)
2535 * bug fix: correct canonicalization of multiple non-standard slot
2536 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2537 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2538 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2539 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2540 now exists, an signals an error.
2541 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2542 during macroexpansion for non-iteration variables as well. (reported
2543 by Bruno Haible for CMUCL)
2544 * bug fix: Cyclic structures and unprintable objects in compiler
2545 messages no longer cause errors. (reported by Bruno Haible)
2546 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2547 and READ-FROM-STRING all now return a primary value of NIL if
2548 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2549 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2550 (reported by Bruno Haible for CMUCL)
2551 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2552 do not designate a condition. (reported by Bruno Haible for
2554 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2555 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2556 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2557 permitting resolution in favour of any of the conflicting symbols.
2558 (reported by Bruno Haible for CMUCL)
2559 * bug fix: EQUAL compiler optimizations is less aggressive on
2560 strings which can potentially compare true despite having distinct
2561 specialized array element types.
2562 * bug fix: unit enumerations can be defined without dividing by
2563 zero. (thanks to Vincent Arkesteijn)
2564 * FORMAT compile-time argument count checking has been enhanced.
2565 (report from Bruno Haible for CMUCL)
2566 * a partial workaround for the bug 262: the compiler does not try to
2567 inline-expand a local function doing RETURN-FROM from a deleted
2568 BLOCK. (thanks to Peter Denno for the bug report and to David
2569 Wragg for the simple test case)
2570 * fixed some bugs revealed by Paul Dietz' test suite:
2571 ** RENAME-PACKAGE allows all package designators as new package
2573 ** constraint propagation and lambda variable substitution are
2574 more cautious in dealing with partially deleted code.
2575 ** compiler sometimes lost reoptimization passes.
2576 ** CERROR, given a condition as condition designator, passes its
2577 remaining arguments to the continue format control without
2579 ** Case-altering FORMAT directives work correctly on non-ASCII
2581 ** The REAL type specifier handles bounds outside the floating
2582 point ranges without signalling FLOATING-POINT-OVERFLOW.
2583 ** Functions with IR1-transformations can create intercomponent
2584 references to global functions.
2585 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2587 ** FORMAT ~R treats a nil value for its first parameter correctly.
2589 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2590 * enhancement: saving cores with foreign code loaded is now
2591 supported on x86/NetBSD and sparc/Linux in addition to the previously
2592 supported platforms.
2593 * bug fix: on some platforms repeated installations caused multiple
2594 copies of HTML documentation to be installed -- should not happen
2595 any more. (reported by Stefan Scholl)
2596 * bug fix: parsing self-recursive alien record types multiple times
2597 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2598 original patch by Helmut Eller for CMUCL)
2599 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2600 (thanks to Richard Kreuter)
2601 * bug fix: defining classes whose accessors are methods on existing
2602 generic functions in other (locked) packages no longer signals
2603 bogus package lock violations. (reported by François-René Rideau)
2604 * bug fix: special variables as DEFMETHOD parameters no longer have
2605 associated bogus type declarations. (reported by David Wragg and
2607 * bug fix: read-write consistency on streams of element-type
2608 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2609 * bug fix: redefiniton of the only method of a generic function with
2610 no DEFGENERIC no longer emits a full WARNING. In addition,
2611 redefinition of generic functions with no DEFGENERIC to an
2612 incompatible lambda list now signals an error. (thanks to Zach
2614 * bug fix: DEFGENERIC now works even when there's a function of the
2615 same name in an enclosing lexical environment. (thanks to Zach
2617 * fixed compiler failure, caused by instrumenting code during
2618 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2619 * optimization: added loop analysis and improved register allocation
2620 * fixed some bugs revealed by Paul Dietz' test suite:
2621 ** POSITION on displaced vectors with non-zero displacement
2622 returns the right answer.
2623 ** (SIMPLE-STRING) is a valid type specifier for sequence
2625 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2627 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2628 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2630 ** The justification version of the FORMAT ~< directive treats
2631 non-zero minpad parameter correctly.
2633 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2634 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2635 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2636 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2637 the supported interface.
2638 * new feature: Single-stepping of code compiled with DEBUG 2 or
2639 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2640 * new feature: saving cores with foreign code loaded is now
2641 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2642 Timothy Moore's work for CMUCL)
2643 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2644 parameters to * instead of NIL if no initform is supplied.
2645 (reported by Johan Bockgaard)
2646 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2647 correctly when FUNCALL appears as the car of the form. Note:
2648 despite this FUNCALL forms are not currently subject to
2649 compiler-macro expansion. (port of Raymond Toy's fix for the
2650 same from CMUCL, reported by Johan Bockgaard)
2651 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2652 (thanks for Teemu Kalvas)
2653 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2654 case instead of terminating the iteration (thanks for Julian
2655 Squires, Sean Champ and Raymond Toy)
2656 * bug fix: incorrect expansion of defgeneric that caused a style
2657 warning. (thanks for Zach Beane)
2658 * bug fix: slot accessor effective method computation works properly
2659 for classes with multiple non-standard applicable methods on
2660 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2661 * on x86 compiler supports stack allocation of results of LIST and
2662 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2663 CMUCL implementation by Gerd Moellmann)
2664 * fixed some bugs revealed by Paul Dietz' test suite:
2665 ** FORMAT strings with both the ~<~:;~> form of the justification
2666 directive and pretty-printing directives cause an error.
2668 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2669 * incompatible change: the internal functions
2670 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2671 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2672 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2673 instead of the old functions.
2674 * new feature: on platforms where "dladdr" is available foreign
2675 function names now appear in backtraces. (based on Helmut Eller's
2677 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2678 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2680 * documentation: ASDF manual is now distributed with SBCL.
2681 * bug fix: SBCL can now load its contributed modules with REQUIRE
2682 even if the system-provided entries have been removed from
2683 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2685 * bug fix: backtraces involving undefined functions or assembly
2686 routines are more informative. (thanks to Brian Downing)
2687 * bug fix: mutually referent alien structures now work correctly.
2688 (reported by Rick Taube)
2689 * bug fix: structures defined by WITH-ALIEN can be referred to
2690 within other definitions in the same WITH-ALIEN.
2691 * bug fix: division operators (MOD, TRUNCATE and the like) with
2692 constant zero divisors and integer dividends no longer generate
2694 * bug fix: provide default methods for INPUT-STREAM-P and
2695 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2696 * bug fix: improve the detection and resolution of MOP metacycles
2697 (where effective method computation depends on the generic function
2698 having its effective method computed). (reported by Bruno Haible)
2699 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2700 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2701 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2702 * bug fix: pathnames with relative directory components can be
2703 represented relative to default pathnames.
2704 * optimization: in taking the GCD of bignums, reduce the two bignums
2705 to approximately the same size (using Euclid's algorithm) before
2706 applying the more sophisticated binary GCD. (thanks to Juho
2708 * optimization: COUNT on bitvectors now operates word-at-a-time.
2709 * optimization: ASH with a positive, but not necessarily constant,
2710 (leftwards) shift, when in a modular context, is compiled to a
2712 * fixed some bugs revealed by Paul Dietz' test suite:
2713 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2714 if the corresponding argument is NIL.
2715 ** FORMAT directives accepting a minpad parameter treat negative
2717 ** literal commas as character parameters to FORMAT directives are
2719 ** literal spaces directly after ~<Newline> directives within a
2720 format-logical-block (~:< ~@:>) do not induce :FILL-style
2721 conditional newlines.
2722 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2723 as specified: it no longer includes conditional newlines).
2724 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2726 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2727 :PER-LINE-PREFIX argument does not evaluate to a string.
2729 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2730 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2731 the manual for details; for now, package locks can be disabled by
2732 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2733 if no major problems are found then it is likely that they will be
2734 compiled in unconditionally.
2735 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2736 unsupported operators on all platforms. To load a shared library
2737 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2738 object file, link it into a shared library outside of SBCL and
2739 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2741 * bug workaround: ROOM T and the GC were not working together
2742 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2743 reporting facility aren't preserved. That reporting has been
2744 disabled, so now until and unless someone figures out how to make it
2745 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2746 * minor incompatible change: as threatened around sbcl-0.8.0, the
2747 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2748 The symbols are also exported from SB-PCL for backwards
2749 compatibility, but more so than before SB-PCL should be treated as
2750 an implementation-internal package.
2751 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2753 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2754 It is known as of this release not to work on the Alpha, however.
2755 * fixed bug #167: errors signalled due to illegal syntax in method
2756 bodies are now more legible.
2757 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2758 designators and can hence be used with TYPEP.
2759 * fixed bug #333: CHECK-TYPE now ensures that the type error
2760 signalled, if any, has the right object to be accessed by
2761 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2762 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2763 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2765 * fixed bug #334: programmatic addition of slots using specialized
2766 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2767 and :ALLOCATION :CLASS slots.
2768 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2769 float exponents. (rereported by Peter Seibel)
2770 * fixed a bug: #\Space (and other whitespace characters) are no
2771 longer considered to be macro characters in standard syntax by
2772 GET-MACRO-CHARACTER.
2773 * fixed bug: initialization of condition class metaobjects no longer
2774 causes an instance of the condition to be created. (reported by
2776 * fixed bug: it is now possible to have slots such that
2777 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2778 description is neither :INSTANCE nor :CLASS.
2779 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2780 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2781 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2782 non-standard SLOT-DEFINITION-ALLOCATION values.
2783 * fixed bugs: various race conditions handling exiting threads.
2784 CL-PPCRE's thread torture test now passes.
2785 * fixed arguably-a-bug: GC time proportional to number of threads in
2786 system even when most of them are idle
2787 * optimization: improved performance of BIT and SBIT on bit-vectors.
2788 * .fasl file incompatibility: The fasl file version number has
2789 been incremented because of changes associated with package locks.
2791 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2792 * minor incompatible change: the system no longer provides
2793 optimization hints (conditions which are TYPEP
2794 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2795 mode; these hints are still emitted when higher SPEED optimization
2797 * new contrib module: a sampling profiler (profiling by statistical
2798 sampling, rather than by instrumenting functions) is available as
2799 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2800 * the behaviour of the standard function ED is now customizeable by
2801 third parties through a hook variable: see ED's documentation
2802 string for information on the protocol.
2803 * the compiler no longer emits efficiency notes for (FUNCALL X)
2804 when the type of X is uncertain under default optimization
2806 * fixed bug 276: mutating a binding of a specialized parameter to a
2807 method to something that is not TYPEP the specializer is now
2809 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2810 exhibit the required behaviour on the x86 platform. (thanks to
2811 Peter van Eynde, Eric Marsden and Bruno Haible)
2812 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2813 even for difficult arguments. (reported by Peter Graves)
2814 * fixed bug 141a: the backquote printer now descends quoted
2816 * fixed another bug in backquote printing: no more destructive
2817 modification of the form's list structure. (reported by Brian
2819 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2820 move between its address being taken and the call to
2821 interrupt_thread, fixing a crashing race condition.
2822 * the SB-POSIX contrib implementation has been adjusted so that it
2823 no longer exhibits ridiculously poor performance when constructing
2824 instances corresponding to C structs.
2826 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2827 * minor incompatible change: the sb-grovel contrib now treats C
2828 structures as alien (in the sense of SB-ALIEN) objects rather than
2829 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2830 has implications for memory management of client code
2831 (sb-grovel-returned objects must now be manually managed) and for
2832 type safety (alien objects now have full types).
2833 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2834 used to control emission of compiler diagnostics, rather than the
2835 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2836 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2837 quality should be considered deprecated.
2838 * install.sh now installs the user manual as well
2839 * (not quite a new documentable feature, but worth considering in
2840 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2841 beginnings of a semantically meaningful condition hierarchy is
2842 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2844 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2845 designator as the defaults argument.
2846 * fixed bug: Displaced arrays whose displaced-to array has become
2847 too small now cause ARRAY-DIMENSION to signal an error, providing
2848 for safer bounds-checking. (reported by Bruno Haible)
2849 * fixed bug: DEFCLASS slot definitions with identical :READER and
2850 :WRITER names now signal a reasonable error. (reported by Thomas
2852 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2854 * fixed bug: Invalid slot specification errors now print correctly.
2855 (thanks to Zach Beane)
2856 * fixed bug 320: Shared to local slot value transfers in class
2857 redefinitions now happen corrently from superclasses as
2858 well. (reported by Bruno Haible)
2859 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2861 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2862 improper lists correctly. (reported by Bruno Haible)
2863 * fixed bug 313: source-transform for <fun-name> was erroneously
2864 applied to a call of a value of a variable with name <fun-name>.
2865 (reported by Antonio Menezes Leitao)
2866 * fixed bug 307: The obsolete instance protocol ensures that
2867 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2868 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2869 remove dead unknown-values globs from the middle of the stack.
2870 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2871 easier to resume long computations after using *BREAK-ON-SIGNALS*
2872 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2873 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2874 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2875 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2876 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2877 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2879 * fixed a (fairly theoretical) bug in string printing: if
2880 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2881 does not have array-element-type equal to the most general string
2883 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2884 function name. (thanks to Nikodemus Siivola)
2885 * fixed bug: compile-time format string checker failed on
2886 non-closed ~{. (reported by Thomas F Burdick)
2887 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2888 don't warn on legal loop code involving more than one aggregate
2889 boolean. (thanks to Nikodemus Siivola)
2890 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2891 in keyword position to MAKE-PACKAGE was not being reported as
2893 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2894 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2896 * fixed bugs 280 and 312: the checking for multiple definitions in a
2897 file is less likely to become confused by uses of inline
2899 * fixed bug: the #S reader macro performs the keyword coercion
2900 specified for slot names. (reported by Kalle Niemitalo)
2901 * fixed bug: lambda lists may contain symbols whose names start with
2902 & but are not lambda-list-keywords; their occurrence triggers a
2904 * fixed bug 321: define-method-combination argument lambda lists do
2905 not cause invalid code to be generated when &optional or &aux
2906 variables have default values. (reported by Bruno Haible)
2907 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2908 have CLOS classes; this appears also to have fixed reported
2909 instability in stack exhaustion detection.
2910 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2911 MUFFLE-WARNING when no associated restart is present is now
2913 * optimization: rearranged the expansion of various defining macros
2914 so that each expands into only one top-level form in a
2915 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2917 * optimization: used a previously little-used slot in symbols to
2918 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2920 * fixed some bugs revealed by Paul Dietz' test suite:
2921 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2923 ** Fixed an optimization bug related to inheritance of initforms
2924 from local to shared slots.
2925 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2926 ** CAST optimizer forgot to flush argument derived type.
2927 ** print/read consistency on floats is now orders of magnitude
2928 more likely. (thanks also to Bruno Haible for a similar report
2930 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2931 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2932 for objects of type REAL. Make it so.
2933 ** SXHASH is specified (infelicitously) to respect similarity,
2934 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2935 it so. (thanks to Markus Ziegler)
2936 ** on the Alpha, the compiler succeeds in compiling functions
2937 returning a known number of arguments greater than 63.
2938 ** fixed handling of invalid NIL arguments in keyword position
2940 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2942 ** provided a readably-printable representation for RANDOM-STATE
2944 ** ensured that pathnames, where they have a namestring, always
2945 print using #P"..." syntax.
2947 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2948 * Support for the forthcoming 2.0 version of the NetBSD kernel
2949 running on x86 hardware has been added. (thanks to Perry
2950 E. Metzger most immediately, and others for their past work)
2951 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2952 current release version; SBCL's OpenBSD support had been broken
2953 since about the time of OpenBSD's switch to ELF binary format.)
2954 * [placeholder for DX summary]
2955 ** user code with &REST lists declared dynamic-extent, under high
2956 speed or space and low safety and debug optimization policy.
2957 * The manual has been converted to Texinfo format and the debugger
2958 chapter from the cmucl manual has been added.
2959 * A facility has been added to extract documentation strings from
2960 sbcl and store them as Texinfo-formatted snippets for inclusion in
2961 the manual (via Texinfo's @include directive)
2962 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2963 the test case to Dave Roberts)
2964 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2965 pointers. (reported by Sean Ross)
2966 * bug fix: PROFILE output is printed nicely even for large numerical
2967 values. (thanks to Zach Beane)
2968 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2969 greater than 32 handle EOF correctly.
2970 * bug fix: on X86 an immediate argument of the IMUL instruction is
2971 correctly printed in disassembly. (thanks to Lutz Euler)
2972 * bug fix: class slots in redefined classes preserve their old
2973 values. (thanks to Bruno Haible and Nikodemus Siivola)
2974 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2975 succeeds. (reported by Marco Baringer)
2976 * fixed some bugs revealed by Paul Dietz' test suite:
2977 ** READ-SEQUENCE now works on ECHO-STREAMs.
2978 ** RATIONALIZE works more according to its specification. (thanks
2981 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2982 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2983 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2984 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2985 should matter to you only if you rebind the printer control
2986 variables and then find you want different bindings in the
2987 debugger than in the ordinary execution of your program.)
2988 * The runtime build system has been tweaked to support building
2989 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2990 assembler and linker. (thanks to Nikodemus Siivola)
2991 * Unbound, undefined, undeclared variables now trigger full
2992 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2993 more often programmer error than deliberate exploitation of undefined
2995 * optimization: the hash algorithm for strings has changed to one
2996 that is less vulnerable to spurious collisions. (thanks to Juho
2998 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2999 needless bounds checking. (thanks to Juho Snellman)
3000 * optimization: implemented multiplication as a modular
3001 (UNSIGNED-BYTE 32) operation on the PPC backend.
3002 * fixed some bugs revealed by Paul Dietz' test suite:
3003 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3005 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3006 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3007 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3009 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3011 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3013 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3014 their output stream on EOF from read.
3015 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3016 have been read to end-of-file.
3017 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3019 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3020 description of determination of which consecutive characters
3022 ** Printing the "Space" character with escaping on now yields "#\\ ",
3023 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3024 ** Reading floating-point numbers with *READ-BASE* set to a number
3025 less than 10 works correctly.
3026 ** Reading floating-point numbers with *READ-BASE* set to a number
3027 more than 10 works correctly.
3028 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3029 the readtable currently in effect.
3031 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3032 * minor incompatible change: parsing of namestrings on a physical
3033 (Unix) host has changed; numbers after the final #\. in a
3034 namestring are no longer interpreted as a version field. This is
3035 intented to be largely invisible to the user, except that the
3036 meaning of the namestring "*.*.*" has changed: it now refers to a
3037 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3038 should usually be replaced by
3039 (make-pathname :name :wild :type :wild :version :wild)
3040 with the added benefit that this is more likely to be portable.
3041 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3042 option to OPEN now signals an error if the file being opened
3043 exists; this may have an impact on existing code.
3044 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3045 signals received, on the PowerPC platforms (both Linux and
3046 Darwin). (thanks to Pierre Mai for pointing out the location of
3048 * several fixes on OS X: The system now builds and runs cleanly on
3049 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3050 (thanks to Brian Mastenbrook)
3051 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3052 second-resolution: integer multiples of 1/3600 between -24 and 24.
3053 (thanks to Vincent Arkesteijn)
3054 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3055 type when called with 1 argument; PEEK-CHAR checked type of
3056 PEEK-TYPE only after having read first character from a
3057 stream. (reported by Peter Graves)
3058 * bug fix: the garbage collector now has much better locality
3059 behaviour, and in particular no longer treats all memory as being
3060 exhausted when many small objects point to each other in a deeply
3062 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3063 recognized as being TYPEP their class.
3064 * bug fix: the PUSHNEW documentation string has been corrected.
3065 (thanks to Vincent Arkesteijn)
3066 * bug fix: defaulting of the value for the last of an atypically
3067 large number of multiple values being bound was not being
3068 performed correctly on the Alpha or PPC platforms
3069 * optimization: implemented multiplication as a modular
3070 (UNSIGNED-BYTE 32) operation on the x86 backend.
3071 * optimization: SEARCH on simple-base-strings can now be open-coded.
3072 (see also contrib/compiler-extras.lisp for inspiration for
3073 teaching the compiler about the Boyer-Moore algorithm).
3074 * value, returned by MAX (and MIN) called with several EQUALP, but
3075 not EQL, arguments now does not depend on compiler settings.
3076 * fixed some bugs revealed by Paul Dietz' test suite:
3077 ** in stack analysis liveness information is propagated from
3078 non-local entry points.
3079 ** pathwise CAST removing failed when the CAST node did not start
3081 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3082 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3084 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3085 host is already defined.
3086 ** RENAME-FILE works on streams instead of signalling an internal
3088 ** PEEK-CHAR uses the current readtable when determining whether
3089 or not a character is whitespace.
3090 ** MERGE-PATHNAMES handles the case when the pathname does not
3091 specify a name while the default-pathname specifies a version.
3092 ** Pathnames now stand a better chance of respecting print/read
3094 ** Attempting to use standardized file system operators with a
3095 pathname with invalid :DIRECTORY components signals a
3097 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3099 ** DIRECTORY on logical pathnames is more correct.
3100 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3101 signal a TYPE-ERROR if their argument is not a stream.
3102 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3103 designator argument does not designate a stream.
3104 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3105 examining the synonym.
3106 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3108 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3109 element-type larger than ([UN]SIGNED-BYTE 32).
3111 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3112 * When built with the :SB-FUTEX feature, threaded builds now take
3113 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3114 for faster/more reliable mutex and condition variable support.
3115 * Incompatible change (but one you probably shouldn't have been using
3116 anyway): the interface and code for arbitrating between multiple
3117 threads in the same user session has been redesigned.
3118 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3119 error when called without an explicit environment argument.
3120 (thanks to Nikodemus Siivola)
3121 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3122 confused about their position. (thanks to Adam Warner and Gerd
3124 * bug fix: Pretty printing backquoted forms with unquotations in the
3125 argument list position of various code constructs such as LAMBDA
3126 now works correctly. (reported by Paul Dietz)
3127 * bug fix: Pretty printing unquotations no longer loses all
3128 stream position information.
3129 * optimization: performance of string output streams is now less
3130 poor for multiple small sequence writes.
3131 * optimization: performance of CSUBTYPEP in the presence of complex
3132 expressions involving CONS and NOT many times has been improved.
3133 (reported by Paul Dietz)
3134 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3136 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3137 streams for the REPL, for the debugger, and so on. (thanks to
3139 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3140 * fixed some bugs revealed by Paul Dietz' test suite:
3141 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3142 is ignored if it is too small, rather than propagating through
3143 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3144 ** extremely complex negations of CONS types were not being
3145 sufficiently canonicalized, leading to inconsistencies in
3147 ** VALUES tranformer lost derived type.
3149 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3150 * fixed a bootstrapping bug: the build process no longer assumes
3151 that the various BOOLE-related constants have the same value in
3152 host and target lisps. (noted by Paul Dietz' test suite on an
3153 SBCL binary built from CLISP)
3154 * The system can now be dynamically linked on the MIPS platform,
3155 which enables dynamic loading of foreign code from Lisp. (thanks
3156 to Ralf Baechle for discussions on the MIPS ABI)
3157 * The system now records debugging information for its own source
3158 files in a filesystem-position-independent manner, relative to
3159 the "SYS" logical host.
3160 * fixed a compiler bug: MV-LET convertion did not check references
3161 to the "max args" entry point. (reported by Brian Downing)
3162 * tweaked disassembly notes to be less confident about proclaiming
3163 some instruction as an LRA. (thanks to Brian Downing)
3164 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3165 now each have their own history, command character, and other
3166 characteristics. (thanks to David Lichteblau)
3167 * fixed some bugs revealed by Paul Dietz' test suite:
3168 ** compiler failure in compiling LOGAND expressions including a
3170 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3171 shift greater than 32.
3172 ** FLUSH-DEST did not mark blocks for type check regeneration.
3173 ** HANDLER-CASE failed to accept declarations in handler clauses
3174 in some circumstances.
3176 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3177 * New code in contrib/sb-introspect (still probably not entirely
3178 stable yet) provides some support for smart Lisp development
3179 environments like SLIME.
3180 * The conditions signalled for errors occurring when loading .fasl
3181 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3182 in a way which should help ASDF recover gracefully.
3183 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3184 stuff has been cleaned up. If you code contrib/ stuff, this might
3185 affect you, and you can look at contrib/README, contrib/STANDARDS,
3186 and/or the 0.8.4.27 diff to check.
3187 * In full calls the compiler now does not generate checks for declared
3188 argument types for all arguments.
3189 * various threading fixes
3190 ** and some experimental patches which didn't make it into
3191 the main tree for this release, but which are shipped in
3192 contrib/experimental-thread.patch as a possible fix for some
3193 failures (deadlock, spinning...) in GC-intensive multithreaded
3195 * fixed PPC build problem (source code incompatibility of different
3196 library versions): added offsetof() hackery which attempts to divine
3197 where glibc maintainers put uc_mcontext today
3198 * fixed bug 282: compiler does not trust type assertions while passing
3199 arguments to a full call.
3200 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3201 &OPTIONAL VALUES type parameter.
3202 * fix bug 214: algorithm for noting rejected templates is now more
3203 similar to that of template seletion. (also reported by rydis on
3205 * fixed bug 141b: printing backquoted information readably and prettily
3206 inserts a space where necessary.
3207 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3208 (CHARACTER 10) are now reported as errors, rather than propagated
3209 as unknown types. (reported by piso on #lisp)
3210 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3211 with values NIL and :ERROR. (thanks to Milan Zamazal)
3212 * fixed bug 191c: CLOS now does proper keyword argument checking as
3213 described in CLHS 7.6.5 and 7.6.5.1.
3214 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3215 longer raise an error; further, using a list as a for-as-arithmetic
3216 counter now raises a meaningful error.
3217 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3218 now understood by sequence creation functions such as MAKE-SEQUENCE
3220 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3221 a STRING-INPUT-STREAM.
3222 * compiler enhancement: SIGNUM is now better able to derive the type
3224 * type declarations inside WITH-SLOTS are checked. (reported by
3226 * fixed some bugs revealed by Paul Dietz' test suite:
3227 ** incorrect optimization of TRUNCATE for a positive first
3228 argument and negative second.
3229 ** compiler failure in let-convertion during flushing dead code.
3230 ** compiler failure while deriving type of TRUNCATE on an
3231 interval, containing 0.
3232 ** ASH of a negative bignum by a negative bignum count now returns
3234 ** intersection of CONS types now canonicalizes properly, fixing
3235 inconsistencies in SUBTYPEP.
3237 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3238 * incompatible change: The --disable-debugger command line
3239 option now clobbers the debugger at a more fundamental
3240 level, by redefining #'INVOKE-DEBUGGER instead of by
3241 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3242 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3243 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3244 enter the debugger REPL and then suffer recursive errors
3245 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3246 while under the new system, BREAK in --disable-debugger mode
3247 terminates the system just as an unhandled error would.
3248 * fixed compiler performance when processing loops with a step >1;
3249 * bug fix: DOCUMENTATION now retrieves generic function
3250 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3251 support has been systematized, and now supports the methods
3252 specified by ANSI, along with a default method and a method for
3253 slot documentation. (reported by Nathan Froyd)
3254 * bug fix: effective methods associated with a generic function are
3255 no longer cached over a change of that generic function's method
3256 combination. (reported by Andreas Fuchs)
3257 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3258 names a non-existent directory. (thanks to Andreas Fuchs)
3259 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3260 the PPC platform, lead to stack corruption; this has been fixed.
3261 (reported by Rainer Joswig)
3262 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3263 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3264 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3265 arithmetic in general, is now much more reliable on the PPC
3267 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3268 platform now returns the right answer.
3269 * optimization: restored some effective method precomputation in
3270 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3271 precomputation is now tunable.
3272 * optimization: compiler-internal data structure use has been
3273 reviewed, and changes have been made that should improve the
3274 performance of the compiler by about 20%.
3275 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3276 simple-base-strings and simple-bit-vectors is improved.
3277 * optimization: the optimization of 32-bit logical and arithmetic
3278 functions introduced in version 0.8.3 on the x86 has been
3279 implemented on the mips, ppc and sparc platforms; an
3280 implementation of the same facility, but for 64-bit arithmetic,
3281 has been added for the alpha.
3282 * microoptimization: the compiler is better able to make use of the
3283 x86 LEA instruction for multiplication by constants.
3284 * bug fix: in some situations compiler did not report usage of
3285 generic arithmetic in (SPEED 3) policy.
3286 * bug 145b fix: compiler used wrong type specifier while converting
3287 MEMBER-types to numeric.
3288 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3290 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3291 index argument now works properly on the Alpha platform.
3292 * bug fix: floating point exception treatment on the Alpha platform
3294 * bug fix: FILE-POSITION works much better on string input and
3295 output streams. (thanks to Nikodemus Siivola)
3296 * bug fix: many threading/garbage collection symptoms sorted.
3297 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3298 might be pseudo-atomic.
3299 * internal change: Stopping for GC is now done with signals not
3300 ptrace. GC is now done in whichever thread wanted it, instead of
3302 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3304 * bug fix: result form in DO is not contained in the implicit
3306 * incompatible change: ICR structure is changed; the value part of
3307 CONTINUATION is now called LVAR; corresponding functions are
3308 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3309 * added type deriver for ISQRT (thanks to Robert E. Brown).
3310 * bug fix: better support for loading from the command line when an
3311 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3313 * fixed some bugs revealed by Paul Dietz' test suite:
3314 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3315 ** ROUND and FROUND now give the right answer when given very
3316 small float arguments.
3317 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3319 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3320 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3321 ** LCM should return a non-negative integer.
3322 ** PARSE-INTEGER returned the index of a terminator instead of the
3323 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3324 ** PARSE-INTEGER returned an incorrect index being applied to a
3326 ** LCM with two arguments of 0 returns 0 rather than signalling
3328 ** unsigned addition of a 32-bit constant with the high bit set no
3329 longer causes an internal compiler error on the x86.
3330 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3331 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3332 with negative last argument.
3333 ** byte specifiers with zero size and position no longer cause
3334 an error during type derivation.
3335 ** bignum multiplication on the Alpha platform now returns the
3337 * porting: The system now builds on SuSE AMD64, although it still
3338 generates a 32-bit binary.
3339 * .fasl file incompatibility: The fasl file version number has
3340 been incremented (because of the changes to internal compiler
3341 data structures referred to above).
3343 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3344 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3345 more accurately, on the Darwin kernel running on PowerPC hardware.
3346 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3347 * Compiler code deletion notes now signal a condition of type
3348 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3349 an associated MUFFLE-WARNING restart.
3350 * The compiler now performs limited argument count validation of
3351 constant format strings in FORMAT, and where appropriate in ERROR,
3352 CERROR and WARN. (thanks to Gerd Moellmann)
3353 * New ASDF-INSTALL contrib can be used for automatic download and
3354 installation of third-party Lisp code from CCLAN or other sites
3356 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3357 which forces another thread to execute a function supplied by the
3359 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3360 now accept and act upon their :ELEMENT-TYPE keyword argument.
3361 (reported by Martin Atzmueller, Edi Weitz)
3362 * bug fix: FILE-POSITION now accepts position designators up to
3363 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3364 is the greater. (thanks to Patrik Nordebo)
3365 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3366 Kalle Olavi Niemitalo)
3367 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3368 printable. (reported by Eric Marsden)
3369 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3370 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3371 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3372 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3373 (thanks to Stig Sandoe)
3374 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3375 UNREAD-CHAR now correctly includes the unread character in the
3376 target sequence. (thanks to Gerd Moellmann)
3377 * bug fix in threaded builds: the system can now be suspended and
3378 resumed by shell job control with minimal disruption.
3379 * bug fixes in times and timezones >2038AD
3380 * better handling of "where is GNU make?" problem in build scripts
3381 (thanks to Nikodemus Siivola)
3382 * new optimization: inside a named function any reference to a
3383 function with the same name is considered to be a self-reference;
3384 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3385 optimization quality.
3386 * new optimization on x86: logical functions and + now have
3387 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3388 used when the result is truncated to 32 bits.
3389 * VALUES declaration is partially enabled.
3390 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3391 * fixed some bugs revealed by Paul Dietz' test suite:
3392 ** The system now obeys the constraint imposed by
3393 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3394 types form a lattice under type intersection.
3395 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3396 ** ASSOC now ignores NIL elements in an alist.
3397 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3398 and (1+ MOST-POSITIVE-FIXNUM) answers.
3399 ** The addition of a method with invalid qualifiers to a generic
3400 function does not cause an error to be signalled immediately;
3401 a warning is signalled, and the error is generated only on
3402 calling the generic function.
3403 * changed .fasl file version number, in part to add type codes for
3404 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3405 obscure ANSI requirements
3407 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3408 * fixed bug 148: failure to inline-expand a local function left
3409 garbage, confusing the compiler.
3410 * fixed bugs 3cd: structure slot readers perform type check if the
3411 slot can have an invalid value (i.e. it is either not initialized
3412 or can be written with a less specific slot writer).
3413 * bug fix: the compiler now traps array references to elements off
3414 the end of an array; previously, the bounds checking in some
3415 circumstances could go off-by-one.
3416 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3418 * type declarations for array element types now obey the description
3419 on the CLHS page "Declaration TYPE", as per discussions on
3420 sbcl-help around 2003-05-08. This means that a declaration
3421 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3422 declaration, all references to BAR will be asserted or assumed
3423 (with THE, so dependent on compiler policy) to involve objects of
3424 type FOO. Note that no such declaration is implied in
3425 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3426 * declared types of functions from the "Conditions"
3427 chapter. (reported by Paul Dietz)
3428 * bug fix: CERROR accepts a function as its first argument.
3429 * bug fix: NTH an NTHCDR accept a bignum as index
3430 arguments. (reported by Adam Warner)
3431 * optimization: character compare routines now optimize comparing
3432 against a constant character. (reported by Gilbert Baumann)
3433 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3434 argument now works properly on the MIPS platform.
3435 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3436 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3437 declared to have a functional type, does not cause an error
3439 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3440 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3441 * fixed bug: if last continuation of a deleted block has a
3442 destination, this destination should be deleted too. (reported by
3444 * fixed a bug in the bootstrap process: the host compiler's values
3445 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3446 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3447 test case from Patrik Nordebo)
3448 * improved the ability of the disassembler on the PPC platform to
3449 provide helpful disassembly notes.
3450 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3451 the class in more cases than previously.
3452 * bug fix: FILE-POSITION now understands :START and :END for
3453 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3454 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3455 8) data. (thanks to David Lichteblau)
3456 * bug fix: it is possible to add a method to a generic function
3457 without lambda list.
3458 * bug fix: reader failed to signal END-OF-FILE inside an
3459 object representation. (reported by Nikodemus Siivola)
3460 * fixed some bugs revealed by Paul Dietz' test suite:
3461 ** LAST and [N]BUTLAST should accept a bignum.
3462 ** condition slot accessors are methods.
3463 ** (VECTOR NIL) is a subtype of STRING.
3465 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3466 * minor incompatible change: some nonsensical specialized lambda
3467 lists (used in DEFMETHOD) which were previously ignored now signal
3469 * minor incompatible change: the system is now aware of the types of
3470 variables in the COMMON-LISP package, and will signal errors for
3471 most violations of these type constraints (where previously they
3472 were silently accepted).
3473 * minor incompatible change: COMPILE-FILE now uses the freedom
3474 afforded (ANSI 3.2.2.3) to use derived function types for
3475 functions defined in the same file. This also permits the system
3476 to warn on static type mismatches and function
3477 redefinition. (Currently it does not work with high DEBUG level.)
3478 * minor incompatible change: VALUES declaration is disabled.
3479 * When issuing notes, the compiler now signals a condition of type
3480 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3481 restart for use in user handlers. It is expected that the
3482 COMPILER-NOTE condition will eventually become a condition
3483 supertype to a hierarchy of note types, which will then be
3484 handleable in a similar fashion. However, at the moment, no such
3485 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3486 but it's an implementation detail, not a classification for the
3488 * Changes in type checking closed the following bugs:
3489 ** type checking of unused values (192b, 194d, 203);
3490 ** template selection based on unsafe type assertions (192c, 236);
3491 ** type checking in branches (194bc).
3492 * A short form of VALUES type specifier has ANSI meaning (it has
3493 increased the number of situations when SBCL cannot perform type
3495 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3496 keywords or constants is permissible.
3497 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3498 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3499 classes. (thanks to Antonio Martinez)
3500 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3501 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3502 argument to a pretty-print stream. Instead, it leaves any such
3503 operations to DESCRIBE-OBJECT methods.
3504 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3506 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3507 lambda lists. (thanks to David Lichteblau)
3508 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3509 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3510 (thanks to Gerd Moellmann)
3511 * fixed bug 246: increased compilation speed of long
3512 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3514 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3516 * DEFCONSTANT now throws a condition of type
3517 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3518 constant to a non-EQL value; CONTINUE and ABORT restarts
3519 respectively change and preserve the value.
3520 * fixed bug 63: The code walker, part of the implementation of CLOS,
3521 is now better at handling symbol macros.
3522 * bug fix: There is no longer an internal implementation type named
3523 CL:LENGTH. (reported by Raymond Toy)
3524 * bug fix: In macro-like defining macros/special operators the
3525 implicit block does not enclose the lambda list.
3526 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3527 enclosed in parentheses) are not suitable as type specifiers, and
3528 their use properly signals an error now.
3529 * bug fix: An argument count mismatch for a type specifier in code
3530 being compiled no longer causes an unhandled error at compile
3531 time, but signals a compile-time warning.
3532 * fixed simple vector readable printing
3533 * bug fix: DESCRIBE takes more care over whether the class
3534 precedence list slot of a class is bound before accessing it.
3535 (reported by Markus Krummenacker)
3536 * bug fix: FORMATTER can successfully compile pretty-printer format
3537 strings which use variants of the ~* directive inside.
3538 * bug fix: SEARCH now applies its TEST predicate to the elements of
3539 the arguments in the correct order. (thanks to Wolfhard Buss)
3540 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3541 in type inference. (thanks to Robert E. Brown)
3542 * bug fix: Reading in symbols with an explicit package name of ""
3543 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3544 * fixed some bugs revealed by Paul Dietz' test suite:
3545 ** NIL is now allowed as a structure slot name.
3546 ** Arbitrary numbers, not just REALs, are allowed in certain
3547 circumstances in LOOP for-as-arithmetic clauses.
3548 ** Multiple class redefinitions before slot access no longer
3549 causes a type error.
3550 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3551 association between the name and a class.
3552 ** Generic functions with non-standard method-combination and over
3553 five methods all of which return constants no longer return NIL
3554 after the first few invocations. (thanks to Gerd Moellmann)
3555 ** CALL-NEXT-METHOD with no arguments now passes the original
3556 values of the arguments, even in the presence of assignment.
3557 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3559 ** DISASSEMBLE works with closures and funcallable instances.
3560 ** ADD-METHOD now returns the generic function, not the new method.
3561 ** FIND-METHOD signals an error if the lengths of the specializers
3562 is incompatible with the generic function, even if the ERRORP
3564 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3565 which its argument is a member.
3566 ** DEFCLASS only redefines the class named by its class-name
3567 argument if that name is the proper name of the class;
3568 otherwise, it creates a new class.
3569 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3570 of the UNBOUND-SLOT condition to the name of the slot.
3571 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3572 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3573 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3574 SLOT-MAKUNBOUND in the specified fashion.
3576 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3577 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3578 cross-compilation host. As a consequence, we can now bootstrap our
3579 way up to SBCL starting with a bare gcc toolchain and human-readable
3580 source code (first the source to CLISP, then the source to SBCL).
3581 * A contributed module containing a partial implementation of the
3582 simple-streams interface has been included. (thanks to Rudi
3584 * A contributed module implementing the RFC1321 Message Digest
3585 Algorithm, known as MD5, has been included.
3586 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3587 no longer has any effect, as the code controlled by this feature
3588 has been deleted. (As far as we know, no-one has ever built using
3589 this feature, and its semantics were confused in any case).
3590 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3591 work on conditions (as required by the ANSI specification),
3592 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3593 expected behaviour on conditions. Users should note, however,
3594 that such behaviour is not required by the ANSI specification,
3595 and so use of this behaviour may render their code unportable.
3596 * fixed some bugs revealed by Paul Dietz' test suite:
3597 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3599 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3600 ** SXHASH on condition objects no longer returns NIL.
3601 ** :ALLOCATION :CLASS slots are better treated; their values are
3602 updated on class redefinition, and initforms inherited from
3603 superclasses are applied.
3604 ** REMOVE-METHOD returns its generic function argument even when
3605 no method was removed.
3606 ** SHARED-INITIALIZE now initializes the values of the requested
3607 slots, including those with :ALLOCATION :CLASS.
3608 ** ALLOCATE-INSTANCE now works on structure classes defined via
3609 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3611 ** SLOT-EXISTS-P now works on conditions, as well as structures
3613 ** MAKE-LOAD-FORM now has the required methods on
3614 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3615 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3616 keyword, and now implements the SLOT-NAMES argument.
3617 ** methods with &OPTIONAL arguments no longer allow too many
3618 arguments to be passed in the call without error.
3619 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3620 option is consistent with the required arguments of the generic
3621 function lambda list.
3622 * bug fix: REQUIRE accepts a string designator. (Thanks to
3624 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3625 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3626 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3627 * bug fix: The compiler now checks for duplicated variables in macro
3629 * bug fix: SETQ on globals returns the correct value.
3630 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3631 not a COMPILER-ERROR (followed by some other strange error on
3632 choosing the CONTINUE restart).
3633 * bug fix: make.sh and friends are now more consistent in the way that
3634 they look for GNU "make".
3636 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3637 * experimental native threads support (on x86 Linux >=2.4 only).
3638 This is not compiled in by default: you need to add :SB-THREAD to
3639 the target features. See the "Beyond ANSI" chapter of the manual
3641 * fix for longstanding nonANSIism: The old distinction between
3642 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3643 The return value from CL:FIND-CLASS is now a CLOS class, and
3644 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3645 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3646 * An interface to the MetaObject Protocol, as described in Kiczales,
3647 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3648 MIT Press, 1991, is available from the SB-MOP package.
3649 * incompatible change: the SB-PCL package should now be considered
3650 a private implementation detail, and no longer a semi-private MOP
3652 * minor incompatible change: due to rearrangement for threads, the
3653 control stack and binding stack are now allocated at arbitrary
3654 addresses instead of being hardcoded per-port. Users affected by
3655 this probably have to be doing advanced things with shared
3656 libraries, and will know who they are.
3657 * minor incompatible change: Previously, all --eval forms used were
3658 processed with READ before any of them were processed with EVAL.
3659 Now each --eval form is processed with both READ and EVAL before
3660 the next --eval form is processed. (Thus package operations like
3661 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3662 work as the user might reasonably expect.)
3663 * minor incompatible change: *STANDARD-INPUT* is now only an
3664 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3666 * minor incompatible change: Y-OR-N-P is now character-oriented,
3667 not line oriented. Also, YES-OR-NO-P now works without errors.
3668 (thanks to Antonio Martinez)
3669 * sb-aclrepl module improvements: an integrated inspector, added
3670 repl features, and a bug fix to :trace command.
3671 * Known functions, which cannot be open coded by the backend, are
3672 considered to be able to check types of their arguments. (fixing
3673 a bug report by Nathan J. Froyd)
3674 * fixed a bug in computing method discriminating functions: It is
3675 now possible to define methods specialized on classes which have
3676 forward-referenced superclasses. (thanks to Gerd Moellmann)
3677 * fixed evaluation order in optional entries (reported by Gilbert
3679 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3680 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3681 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3682 required-by-AMOP NAME argument, as well as CLASS and
3683 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3684 * fixed bug 20: DEFMETHOD can define methods using names that are
3685 not the proper names of classes to designate class specializers.
3686 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3687 Unix file descriptors, instead of blowing up. (thanks to Antonio
3689 * Garbage collection refactoring: user-visible change is that a
3690 call to the GC function during WITHOUT-GCING will not do garbage
3691 collection until the end of the WITHOUT-GCING. If you were doing
3692 this you were probably losing anyway.
3693 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3694 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3695 TYPEP the latter but not the former.
3696 * The compiler issues a full WARNING for calls to undefined functions
3697 with names from the CL package.
3698 * MAP-INTO for a vector destination is open coded. (reported by
3699 Brian Downing on c.l.l)
3700 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3701 documentation string.
3702 * fixed some bugs revealed by Paul Dietz' test suite:
3703 ** COPY-ALIST now signals an error if its argument is a dotted
3705 ** Condition slots are now accessed more correctly in the presence
3706 of multiple initargs for a given slot.
3707 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3708 exclude restarts of the same name associated with a different
3710 ** DEFCLASS of forward-referenced classes with another
3711 forward-referenced class in the superclasses list no longer
3713 ** Condition slots are now initialized once each, not multiple
3714 times. (thanks to Gerd Moellmann)
3715 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3717 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3718 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3719 arguments contain duplicated elements.
3720 ** RESTART-CASE understands local macros.
3721 ** RESTART-CASE associates exactly its own restarts with a condition.
3722 ** ENDP in safe mode checks its argument to be of type LIST.
3723 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3724 in question is unbound.
3725 ** Optimized MAKE-INSTANCE functions no longer cause internal
3726 assertion failures in the presence of duplicate initargs.
3727 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3728 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3729 propagated even in the ctor optimized implementation of
3731 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3733 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3734 * a better implementation of SXHASH on (simple) bit vectors,
3735 measured both in execution speed and in distribution of results
3736 over the positive fixnums, has been installed. Likewise, a better
3737 implementation of EQUAL for simple bit vectors is now available.
3738 * fixed CEILING optimization for a divisor of form 2^k.
3739 * fixed bug 240 (emitting extra style warnings "using the lexical
3740 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3741 by Antonio Martinez)
3742 * fixed SXHASH, giving different results for NIL depending on type
3743 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3744 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3745 variable. (found by Rolf Wester)
3746 * SBCL does not ignore type declarations for special
3747 variables. (reported by rif on c.l.l 2003-03-05)
3748 * some bug fixes in contrib/sb-aclrepl/
3749 * fixed some bugs revealed by Paul Dietz' test suite:
3750 ** a bug in the CONS type specifier, whereby the CAR and CDR
3751 types got intertwined, has been fixed;
3752 ** the type system is now able to reason about the interaction
3753 between INTEGER and RATIO types more completely;
3754 ** APPEND, [N]REVERSE and NRECONC check that those their
3755 arguments, which must be proper lists, are really so;
3756 ** An array specialized to be unable to hold elements has been
3757 implemented, as required -- yes, really -- by ANSI;
3758 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3759 on malformed property lists;
3761 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3762 * incompatible packaging change: in line with Unix convention,
3763 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3764 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3765 * REQUIRE and PROVIDE are now optionally capable of doing something
3766 useful. See the documentation string for REQUIRE.
3767 * infrastructure for a managed SBCL contrib system: contributed
3768 modules in this release include:
3769 ** the ASDF system definition facility;
3770 ** an interface to the BSD Sockets API;
3771 ** an ACL-like convenience interface to the repl;
3772 (thanks to Kevin Rosenberg)
3773 ** an implementation of ROTATE-BYTE, with efficient implementation
3775 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3776 gives a closer approximation to the right answer than previously.
3777 (thanks to Raymond Toy)
3778 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3779 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3780 optional environment arguments, as required by ANSI.
3781 * fixed bugs in other functions taking environment objects, allowing
3782 calls with an explicit NIL environment argument to be compiled
3784 * fixed bug 228: primary return values from
3785 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3786 COMPILE or FUNCTION.
3787 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3788 :TYPE will no longer signal errors on innocuous objects.
3789 * fixed bug 231b: SETQ is better at respecting type declarations in
3790 the lexical environment.
3791 * fixed a bug in DEFCLASS: classes named by symbols with no or
3792 unprintable packages can now be defined.
3793 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3794 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3795 * fixed compiler failure related to checking types of functions.
3796 (reported by Robert E. Brown)
3797 * the compiler is now much more consistent in its error-checking
3798 treatment of bounding index arguments to sequence functions: in
3799 (SAFETY 3) code, errors will be signalled in almost all cases if
3800 invalid sequence bounding indices are passed to functions defined
3801 by ANSI to operate on sequences.
3802 * fixed a bug in the build procedure: documentation of SBCL-specific
3803 packages is now preserved and available in the final Lisp image.
3804 * lifted FDEFINITION lookup out of loops in the implementation of
3805 many list operations. (thanks to Robert E. Brown)
3806 * fixed a bug in the reader: the #n# reader macro now works for
3807 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3808 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3809 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3810 E. Brown and Tony Martinez respectively)
3811 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3812 SLOT-BOUNDP: in particular, it is now able to optimize them much
3813 better, and is now not vulnerable to having packages renamed.
3814 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3815 always called when a slot is not present in an instance. (thanks
3817 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3818 longer signal an error on unreading a character following EOF on
3819 the previous constituent stream. (thanks to Tony Martinez)
3820 * fixed some bugs revealed by Paul Dietz' test suite:
3821 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3822 not just nonnegative fixnums;
3823 ** the logical bit-array operators such as BIT-AND now accept an
3824 explicit NIL for their "opt-arg" argument (to indicate a
3825 freshly-consed result bit-array);
3826 ** ELT now signals an error on an invalid sequence index in safe
3828 ** the type system is now cleverer about negations of numeric
3829 types, and consequently understands the BIGNUM and RATIO types
3831 ** the type system is now cleverer about the interaction between
3832 INTEGER and RATIO types: while bugs still remain, many more
3833 cases are accurately computed;
3834 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3835 if it is in the last clause;
3836 ** CONSTANTLY now correctly returns a side-effect-free function in
3838 ** DECLARE is no longer treated as a special-operator; in
3839 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3840 * incremented fasl file version number due to the change in the
3841 DEFSTRUCT-SLOT-DESCRIPTION structure.
3843 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3844 * minor incompatible change: code processed by the "interpreter" or
3845 EVAL now has a compilation optimization policy of (DEBUG 2)
3846 (changed from (DEBUG 1)) to improve debuggability of interactive
3847 development, and to allow the use of the debug RETURN command in
3849 * an experimental implementation of the RETURN command for the
3850 debugger has been included. (thanks to Frederik Kuivinen)
3851 * fixed bug 62: constraints were not propagated into a loop.
3852 * fixed bug in embedded calls of SORT (reported and investigated by
3854 * fixed some bugs revealed by Paul Dietz' test suite:
3855 ** printing and reading of arrays with some dimensions having
3856 length 0 (thanks to Gerd Moellmann);
3857 ** BOA constructor with &AUX argument without a default value does
3858 not cause a type error;
3859 ** CONSTANTP now returns true for all self-evaluating objects.
3861 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3862 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3863 accessors that are related by inheritance, as specified in the
3864 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3865 Valtteri Vuorikoski)
3866 * The compiler is now able to inline functions that were defined in
3867 a complex lexical environment (e.g. inside a MACROLET).
3868 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3870 * Support for the upcoming FreeBSD-5.0 release has been included.
3871 (thanks to Dag-Erling Smorgrav)
3872 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3873 effect when it is not in a toplevel context.
3874 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3875 stand a better chance of being correct. (thanks to Gerd
3877 * fixed bug in COERCE, which now signals an error on coercing a
3878 rational to a bounded real type which excludes the expected
3880 * The compiler is now able to derive types more accurately from the
3881 COERCE and COMPILE functions.
3882 * fixed bug 223: functional binding is considered to be constant
3883 only for symbols in the CL package.
3884 * fixed bug 231: SETQ did not check the type of a variable being set
3885 (reported by Robert E. Brown)
3886 * A new optimization for MAKE-INSTANCE has been included, fixing
3887 various bugs (including relating to :ALLOCATION :CLASS slots and
3888 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3890 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3891 ** As required by ANSI, LOOP now disallows anonymous collection
3892 clauses such as COLLECT I in conjunction with aggregate boolean
3893 clauses such as THEREIS (= I 1);
3894 ** LOOP now signals an error when any variable is reused in the
3895 same loop (including the potentially useful construct analogous
3896 to WITH A = 1 WITH A = (1+ A);
3897 ** IT is only a special loop symbol within the first clause of a
3898 conditional loop clause;
3899 ** LOOP with a typed iteration variable over a hashtable now
3900 signals a type error iff it should.
3901 * fixed some other bugs revealed by Paul Dietz' test suite:
3902 ** FILE-STREAM now names the class previously known as FD-STREAM;
3903 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3904 argument) no longer signals an error;
3905 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3906 of the slot symbol, rather than using the current package
3907 ((:CONC-NAME "") continues to intern the slot's name in the
3909 * incremented fasl file version number, because of the incompatible
3910 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3911 of the new implementation of DEFINE-COMPILER-MACRO.
3913 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3914 * Support for building SBCL for MIPS platforms running in
3915 little-endian mode has now been checked in, and basic
3916 functionality on said platforms verified.
3917 * minor incompatible change: PCL now records the pathname of a file
3918 in which methods and the like are defined, rather than its
3920 * minor incompatible change: TRUENAME now considers the truename of
3921 a file naming a directory to be the pathname with :DIRECTORY
3922 component indicating that directory.
3923 * minor incompatible change: a NAMED clause in the extended form of
3924 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3925 reason for the previous behaviour is unclear.
3926 * more systematization and improvement of CLOS and MOP conformance
3927 in PCL (thanks to Gerd Moellman and Pierre Mai):
3928 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3930 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3931 primary methods with no specializers;
3932 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3934 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3935 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3936 FORWARD-REFERENCED-CLASSes; error reporting on
3937 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3939 ** SXHASH on CLOS instances now uses a slot internal to the
3940 instance to return different numbers on distinct instances,
3941 while preserving the same return value through invocations of
3943 ** DEFMETHOD signals errors when methods with longer incongruent
3944 lambda lists are added to generic functions;
3945 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3946 CLASS, as specified in AMOP;
3947 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3948 on the order returned by the primary method for classes of
3949 class STANDARD-CLASS;
3950 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3951 * fixed some bugs shown by Paul Dietz' test suite:
3952 ** DOLIST puts its body in TAGBODY;
3953 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3955 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3956 value producing form;
3957 ** if more variables are given to PROGV than values, extra
3958 variables are bound and made to have no value;
3959 ** NSUBSTITUTE on list arguments gets the right answer with
3961 ** ELT signals an error of type TYPE-ERROR when the index argument
3962 is not a valid sequence index;
3963 ** LOOP signals (at macroexpansion time) an error of type
3964 PROGRAM-ERROR when duplicate variable names are found;
3965 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3966 ** FUNCALL of special-operators now cause an error of type
3968 ** PSETQ now works as required in the presence of side-effecting
3969 symbol-macro places;
3970 ** NCONC accepts any object as its last argument;
3971 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3973 ** loop-for-as-package does not require a package to be explicitely
3975 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3977 * fixed bug 166: compiler preserves "there is a way to go"
3978 invariant when deleting code.
3979 * fixed bug 172: macro lambda lists with required arguments after
3980 &REST arguments now cause an error to be signalled. (thanks to
3982 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3983 bug. (thanks to Matthew Danish)
3984 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3986 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3987 arguments are lexically rebound. (thanks to Gerd Moellmann and
3989 * fixed bug 194: error messages are now more informative when there
3990 is no primary method applicable in a call to a generic
3991 function. (thanks to Gerd Moellmann)
3992 * fixed bug in command line argument checking (thanks to Julian
3994 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3995 (thanks to Matthew Danish)
3996 * incremented fasl file version number, because of the
3997 SXHASH-related changes in the layout of CLOS data structures
3999 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4000 * minor incompatible change: The runtime (the Unix executable named
4001 "sbcl") is now much pickier about the .core files it will load.
4002 Essentially it now requires .core files to descend from the same
4003 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4004 as the runtime does. (The intent is to prevent the crashes which
4005 can occur, and which can even be reported as mysterious failures,
4006 when people patch the sources or change the build parameters
4007 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4008 sbcl and .core files.)
4009 * fixed bug: VALUES-LIST is no longer optimized away.
4010 * fixed bug 142: The FFI conversion of C string values to Lisp
4011 string values no longer conses excessively. (thanks to Nathan
4012 Froyd porting Raymond Toy's fix to CMU CL)
4013 * began to systematize and improve MOP conformance in PCL (thanks to
4014 Nathan Froyd, Gerd Moellman and Pierre Mai):
4015 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4017 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4018 ** FINALIZE-INHERITANCE is now called on class finalization;
4019 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4020 argument precedence order.
4021 * fixed bug 202: The compiler no longer fails on functions whose
4022 derived types contradict their declared type.
4023 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4024 so it can be non-toplevel.
4025 * The fasl file version number has changed (because of the new
4026 implementation of DEFMACRO).
4027 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4028 safe code, that any length requirement by their type-specifier
4029 argument is valid. The exceptions to this are described in bug
4031 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4032 functions now check the types of their inputs as required by ANSI.
4033 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4034 introduced symbol is DECLAREd to be SPECIAL.
4035 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4036 * fixed bug: :COUNT argument to sequence functions may be negative
4037 * fixed bug: body of DO-SYMBOLS may contain declarations
4038 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4039 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4040 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4041 symbol macro only once
4042 * fixed printing of call frame when argument list is unavailable
4043 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4044 * compiler no longer signals WARNING on unknown keyword
4047 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4048 * A beta-quality port to the mips architecture running Linux,
4049 based on the old CMUCL backend, has been made. It has been tested
4050 on a big-endian kernel, and works sufficiently well to be able to
4051 rebuild itself; it has not been tested in little-endian mode.
4052 * fixed an inconsistency between gencgc.c and purify.c which made
4053 dumping/loading .core files unreliable
4054 * fixed bug 120a: The compiler now deals correctly with IFs where
4055 the consequent is the same as the alternative, instead of
4056 misderiving the return type. (thanks to Alexey Dejneka)
4057 * fixed bug 113: Logical pathnames are now dumpable (the logical
4058 host is resolved at load-time, throwing an error if it is not
4060 * fixed bug 174: FORMAT's error message is slightly clearer when a
4061 non-printing character is used in a format directive.
4062 * fixed several bugs in compiler checking of type declarations, i.e.
4063 violations of the Python "declarations are assertions" principle
4064 (thanks to Alexey Dejneka)
4065 * fixed several bugs in PCL's error checking (thanks to Gerd
4067 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4069 * fixed bug in compilation of functions as first class values
4070 (thanks to Antonio Martinez-Shotton)
4071 * The compiler's handling TYPE-ERRORs which it can prove will
4072 inevitably happen at runtime has been cleaned up and corrected
4073 in several ways. (thanks to Alexey Dejneka)
4074 * improved argument type checking for various basic arithmetic
4075 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4076 lavished on them in the past that they can be compiled in many
4077 ways in different special cases
4078 * fixed bug 181: compiler checks validity of user supplied type
4080 * cleaned up code flushing in optimization: Function calls which
4081 should signal errors for safety purposes (e.g. which ANSI says
4082 should signal errors when their arguments are of incorrect type)
4083 are no longer optimized away.
4084 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4085 * incremented fasl file version number, because changes in the
4086 implementation of sequence functions like COERCE caused
4087 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4088 in old inline expansions) to become undefined. (Actually these
4089 changes were later undone, so we might very well be binary
4090 compatible with 0.7.7 after all, but leaving the version number
4091 incremented seemed like the simplest and most conservative
4094 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4095 * An alpha-quality port to the parisc architecture running Linux,
4096 based on the old CMUCL backend, has been made. This, even more so
4097 than the other backends, should be considered still a work in
4098 progress; known problems include that the Linux kernel in 64-bit
4099 mode does not propagate the correct sigcontext structure to
4100 userspace, and consequently SBCL on a parisc64 kernel will not
4102 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4103 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4104 suppressed.) Also now that the compiler is looking at declarations
4105 in the environment, it checks optimization declarations as well,
4106 and suppresses inlining when (> DEBUG SPEED).
4107 * More fixes have been made to treatment of floating point exception
4108 treatment and other Unix signals. In particular, floating point
4109 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4110 * The detection and handling of control stack exhaustion (infinite
4111 or very deeply nested recursion) has changed. Stack exhaustion
4112 detection is now done by write-protecting pages at the OS level
4113 and applies at all optimization settings; when found, a
4114 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4115 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4116 be caught using IGNORE-ERRORS.
4117 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4118 correctly when transforming recursive LABELS functions to LETs.
4119 (thanks to Alexey Dejneka)
4120 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4121 that are names of constants or global variables.
4122 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4123 alien routines with docstrings.
4124 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4125 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4127 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4128 (thanks to Eric Marsden)
4129 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4130 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4131 * Bug fix: The SPARC backend can now compile functions involving
4132 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4133 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4134 to a functions expecting a PATHNAME-DESIGNATOR.
4135 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4136 lambda lists. (thanks to Alexey Dejneka)
4137 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4138 functions so that EXPORTing the name of the class doesn't cause
4139 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4140 to Antonio Martinez for reporting this)
4141 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4142 constant-folded without failing an assertion. (thanks to Einar
4143 Floystad Dorum for reporting this)
4144 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4145 types are now dealt with more correctly by the compiler.
4146 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4147 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4148 behaviour analogous to RENAME-FILE. This puts its behaviour more
4149 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4151 * The fasl file version number has changed again. (because of the
4152 bug fix involving the names of PCL MAKE-INSTANCE functions)
4154 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4155 * bug fix: Floating point exceptions are treated much more
4156 consistently on the x86/Linux and PPC/Linux platforms.
4157 * Array initialization with :INITIAL-ELEMENT is now much faster for
4158 cases when the compiler cannot open code the array creation, but
4159 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4160 array accesses have also seen a speed increase.
4161 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4162 specified. (This was at the root of some bad interactions between
4163 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4165 * bug fix: Internal error arguments for undefined functions are now
4166 computed correctly on the PPC/Linux platform.
4167 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4168 Raymond Toy's patch for CMU CL)
4169 * Support for the Solaris 9 operating environment has been included
4170 (thanks to Daniel Merritt)
4171 * A very ugly but hopefully complete draft of the missing FFI chapter
4172 of the manual has been created by reformatting the corresponding
4173 CMU CL manual chapter into (currently very ugly and incoherent)
4174 DocBook and bringing it up to date for SBCL behavior. Thus, the
4175 manual is now essentially complete, at least by my extreme
4176 once-and-only-once standards, whereby it's acceptable to refer to
4177 the doc strings of SB-EXT functions as the primary documentation.
4178 * The fasl file version number has changed again, due to cleanup of
4179 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4180 is no longer a static symbol.)
4182 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4183 * SBCL now builds with OpenMCL (version 0.12) as the
4184 cross-compilation host; also, more progress has been made toward
4185 bootstrapping under CLISP.
4186 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4188 * bug 158 fixed: The compiler can now deal with integer loop
4189 increments different from 1; fixing this turned out also to fix
4191 * bug 169 fixed: no more bogus warnings about using lexical bindings
4192 despite the presence of perfectly good SPECIAL declarations (thanks
4193 to David Lichteblau)
4194 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4195 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4196 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4198 * bug 180 fixed: Method combination specifications no longer ignore
4199 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4200 * bug fix: Structure type predicate functions now check their argument
4201 count as they should.
4202 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4203 correctly. (thanks to Pierre Mai)
4204 * minor incompatible change: The --noprogrammer option is deprecated
4205 in favor of the new --disable-debugger option, which is very similar.
4206 (The major difference is that it takes effect at a slightly different
4207 time at startup, causing handling of errors in --sysinit and
4208 --userinit files will be affected differently.) The
4209 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4210 been added to allow this functionality to be controlled from ordinary
4211 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4212 maintainers, who might want to run non-interactive scripts to
4213 build SBCL cores which will later be used interactively.)
4214 * minor incompatible change: The LOAD function no longer, when given
4215 a wild pathname to load, loads all files matching that pathname.
4216 Instead, an error of type FILE-ERROR is signalled.
4218 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4219 * bug 147 fixed: The compiler preserves its block link/count
4220 invariants more correctly now so that it doesn't crash. (thanks
4222 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4224 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4225 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4226 no-value-for-this-character as NIL (as specified by ANSI).
4227 * HOST-NAMESTRING on physical pathnames now returns a string that is
4228 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4229 (thanks to Christophe Rhodes)
4230 * The Alpha port handles icache flushing more correctly. (thanks to
4232 * More progress has been made toward bootstrapping under CLISP. (thanks
4233 to Christophe Rhodes)
4234 * The fasl file format has changed again, because dynamic loading
4235 on OpenBSD (which has non-ELF object files) motivated some cleanups
4236 in the way that foreign symbols are transformed and passed around.
4237 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4238 is no longer treated as whitespace by the reader, but instead as
4239 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4240 |A
\7fB|, instead of A as it used to.
4242 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4243 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4244 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4245 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4246 this as of 0.7.1.45, but was left out of the previous news section
4247 (thanks to Dan Barlow)
4248 * SBCL now runs on the Solaris operating system on SPARC architectures
4249 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4250 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4251 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4252 * SPARC backend cleanups, allowing builds of cores optimized for V8
4253 and V9 SPARCS, and also emission of code targeted to a particular
4254 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4256 * SBCL is closer to bootstrapping under CLISP, thanks to various
4257 fixes by Christophe Rhodes.
4258 * The fasl file format has changed again, to allow the compiler's
4259 INFO database to support symbol macros.
4260 * The user manual (in doc/) is formatted into HTML more nicely.
4261 (thanks to coreythomas)
4262 * The system is smarter about SUBTYPEP relationships, especially
4263 those involving NOT types (including types such as ATOM which are
4264 represented internally using NOT types). Thus SUBTYPEP is less
4265 likely to return (VALUES NIL NIL) in general, and in particular
4266 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4268 * The fasl file format has changed again, because the internal
4269 representation of types now includes a new slot to support the new
4270 SUBTYPEP-of-NOT-types logic.
4271 * (not a change in the main branch of SBCL, but a related prototype
4272 which can hopefully be merged into the main branch of SBCL in the
4273 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4274 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4275 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4276 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4277 dispatch macro character combinations. (thanks to Alexey Dejneka)
4278 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4279 that superficially look like logical namestrings correctly.
4280 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4282 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4283 default supplied-p) for &optional and &key arguments. (thanks to
4286 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4287 * incompatible change: The compiler is now less aggressive about
4288 tail call optimization, doing it only when (> SPACE DEBUG) or
4289 (> SPEED DEBUG). (This is an incompatible change because there are
4290 programs which relied on the old CMU-CL-style behavior to optimize
4291 away their unbounded recursion which will now die of stack overflow.)
4292 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4293 for non-GENCGC systems has been increased to 20M (since that
4294 seems much closer to the likely performance optimum for modern
4295 systems than the old 4M value was)
4296 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4297 and *DEBUG-PRINT-LEVEL*
4298 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4299 of CMU CL's support for SPARC, and various endianness and other
4300 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4301 * new syntactic sugar for the Unix command line: --load foo.bar is now
4302 an alternate notation for --eval '(load "foo.bar")'.
4304 ** The system now detects stack overflow and handles it gracefully,
4305 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4306 optimization settings. (This is a good thing in general, and
4307 its introduction in this version should be particularly timely
4308 for anyone whose code fails because of suppression of tail
4310 ** The system now hunts for the C variable "environ" in a more
4311 devious way, to avoid segfaults when the C library version differs
4312 between compile time and run time. (thanks to Christophe Rhodes)
4313 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4314 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4315 ** The compiler no longer issues bogus style warnings for undefined
4316 classes in the same source file as the DEFCLASSes which defined
4317 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4319 ** fixes in CONDITION class precedence list for undefined function
4320 errors (thanks to Alexei Dejneka)
4321 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4322 correctly. (thanks to Dan Barlow)
4323 ** portability fixes aiming at bootstrapping under CLISP (thanks
4324 to Dave McDonald and Christophe Rhodes)
4325 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4326 ** fixes in type translation and and type inference (thanks to
4328 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4329 ** various fixes in the new ports (thanks to Dan Barlow)
4330 * several changes related to debugging:
4331 ** suppression of tail recursion, as noted above
4332 ** stack overflow detection, as noted above
4333 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4334 is now the default. (For some time encapsulation has been more
4335 reliable than the breakpoint-based :ENCAPSULATE NIL
4336 implementation, at least on X86 systems; and I just noticed that
4337 encapsulation also seems closer to the spirit of the ANSI
4340 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4342 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4343 up properly again. (There was a packaging bug in 0.7.0 which
4344 left their definitions in SB-SYS::LOAD-FOREIGN and
4345 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4346 vital for most things which interface to C-level interfaces,
4347 like extensions working with sockets or databases or
4348 Perl-compatible regexes or whatever, and the need to fix
4349 this bug is the main reason that 0.7.1 was released so
4351 ** DEFGENERIC is now choosier about the methods it redefines, so that
4352 reLOADing a previously-LOADed file containing DEFGENERICs does
4353 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4354 cycle now works as it should. (thanks to Alexey Dejneka)
4355 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4356 fixed by Christophe Rhodes. (He actually submitted this patch
4357 months ago, and I delayed until after 0.7.0.)
4358 ** Bug 111 (internal compiler confusion about runtime checks on
4359 FUNCTION types) was fixed by Alexey Dejneka.
4360 * Some internal cleanups (getting rid of variables which aren't
4361 needed now that the byte interpreter is gone) caused the fasl
4362 file format number to change again.
4364 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4365 * major incompatible change: The default fasl file extension, i.e. the
4366 default extension for files produced by COMPILE-FILE, has changed
4367 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4369 ** There are many changes in the implementation of the compiler.
4370 SBCL is now essentially a compiler-only implementation of ANSI
4371 Common Lisp. EVAL still "interprets" a few special cases, but
4372 almost all the interesting cases are handled by creating
4373 a LAMBDA expression, calling COMPILE on it, then calling
4374 FUNCALL on the result.
4375 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4376 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4377 Since the code is newer, there might still be some new bugs
4378 (though not as many as before Martin Atzmueller's fixes:-). But
4379 the new code is substantially simpler and clearer, and hopefully
4380 any remaining bugs will be simpler, less fundamental, and more
4381 fixable then the bugs in the old code.
4382 ** The revised compiler is still a little unsteady on its feet.
4384 *** The debugging information it produces (particularly the names
4385 of FUNCTION objects) is sometimes much less useful than what
4386 the old compiler produced.
4387 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4388 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4389 within a MACROLET) has been temporarily weakened.
4390 ** There are new compiler optimizations for various functions:
4391 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4392 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4393 *** the math functions TRUNCATE, FLOOR, and CEILING
4394 *** the function-of-all-trades COERCE
4395 Mostly these should be transparent, but there's one
4396 potentially-annoying problem (bug 117): when the compiler
4397 inline-expands a function and does type analysis on the result,
4398 it can create control paths which have type mismatches, and
4399 when it can't prove that those control paths aren't taken,
4400 it will issue WARNINGs about the type mismatches. This is
4401 a particular problem in practice for the new sequence functions.
4402 It's not clear how this should be fixed, and for now, a
4403 workaround is given in the entry for 117 in the BUGS file.
4404 ** (Because of the interaction between the two previous items --
4405 occasional inlining problems and new inline expansions -- some
4406 of the new sequence function optimizations won't really kick in
4407 completely until debugging information, and then inlining, are
4408 straightened out in some future version.)
4409 * minor incompatible changes:
4410 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4411 DIRECTORY no longer implicitly promotes NIL slots of its
4412 pathname argument to :WILD. In particular, when you ask for the
4413 contents of a directory (which you used to be able to do without
4414 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4415 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4416 ** changes in behavior that ANSI explicitly defines to be
4417 implementation dependent:
4418 *** The new compiler-only implementation still conforms with ANSI,
4419 but acts a little different than before. Besides the obvious
4420 changes in performance tradeoffs (that the cost per form passed
4421 to EVAL has gone up, and the cost per form executed by EVAL
4422 has gone down), the behavior of the system changes a little
4423 because there are no longer any interpreted function objects.
4424 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4425 e.g. doing COMPILE on the output of interactive DEFUN is
4427 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4429 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4430 has changed from (:CL) to NIL.
4431 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4432 e.g. #\Newline and #\Tab, don't have names specified in the
4433 ANSI Common Lisp standard, is now based on their ASCII symbolic
4434 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4435 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4436 are no longer used for output.
4437 ** changes in internal implementation constants:
4438 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4439 4 million. (If your application spends a lot of time GCing and
4440 you have a lot of RAM, you might want to experiment with
4441 increasing it even more.)
4442 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4443 However, almost all old code should still continue to work without
4444 immediate update, as SB-C-CALL is now a (deprecated) nickname
4446 ** Old operator names in the style DEF-FOO are now deprecated in
4447 favor of new corresponding names DEFINE-FOO, for consistency with
4448 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4449 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4450 internal symbols, but a few supported extensions like
4451 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4452 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4453 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4454 etc. as you get deeper into recursive calls to the debugger
4455 command loop, instead of the old "5]", "5]]", "5]]]"
4456 sequence. (I was motivated to do this when squabbles between
4457 ILISP and SBCL left me very deeply nested in the debugger. In the
4458 short term, this change will probably provoke more ILISP/SBCL
4459 squabbles, but hopefully it will be an improvement in the long run.)
4460 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4461 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4462 function, since declaiming return types involving aliens is
4463 (1) annoyingly messy to do by hand and (2) vital to efficient
4464 compilation of code which calls such functions.
4465 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4466 longer reexported by the SB-EXT package. They're solely useful
4467 for alien code, so it seems more logical that you should get
4468 them from the SB-ALIEN package, not in SB-EXT.
4469 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4470 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4471 features. Instead, the code that they used to control is always
4472 built into the system.
4473 * many other bug fixes
4474 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4475 advantage of the new EVAL-WHEN stuff and to clean them up in
4476 general, and they are now more ANSI-compliant in a number of
4477 ways. Martin Atzmueller is responsible for a lot of this.
4478 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4480 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4481 *** correct ERROR type for various file operations
4482 *** some fixes for Lisp streams
4483 *** DEFMETHOD syntax checking
4484 *** changing old weird representation of debug information as
4485 strings (which, among their other deficiencies, don't transform
4486 correctly when you rename packages, and don't change their
4487 print representation when you change things like *PACKAGE*
4488 and *PRINT-LENGTH*) to symbols and lists of symbols
4489 He also made several improvements and fixed several bugs in DESCRIBE.
4490 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4492 *** misbehavior of WRITE-STRING/WRITE-LINE
4493 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4494 and several other LOOP problems as well
4495 *** DIRECTORY when similar filenames are present
4496 *** DEFGENERIC with :METHOD options
4497 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4498 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4499 *** argument ordering in FIND with :TEST option
4500 *** mishandled package designator argument in APROPOS-LIST
4501 *** various problems in the backquote readmacro
4502 *** a bug in APROPOS
4503 *** probably some others that I'm not describing very well here,
4504 since the CVS log documents them by reference to sbcl-devel
4505 messages, and the SourceForge archives aren't working well.:-(
4506 ** Dan Barlow improved the Alpha port (and is making progress on the
4507 PPC port, for those of you who think different).
4508 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4509 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4510 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4511 continues to work on the SPARC port (for those of you in a position
4512 to look down upon our little PC-compatible boxes from a great height).
4513 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4514 and uses that copy, rather than the current dynamic value, when
4515 it's trying to decide whether to truncate output. Thus e.g.
4516 (let ((*print-lines* 50))
4517 (pprint-logical-block (stream nil)
4519 (let ((*print-lines* 8))
4520 (print (aref possiblybigthings i) stream)))))
4521 should now truncate the logical block only at 50 lines, instead of
4522 often truncating it at 8 lines, as it did before.
4523 * The doc/cmucl/ directory, containing old CMU CL documentation
4524 from the time of the fork, is no longer part of the base system.
4525 SourceForge has shut down its anonymous FTP service, and with it
4526 my original plan for distributing the old CMU CL documentation
4527 there. For now, if you need these files you can download an old
4528 SBCL source release and extract them from it.
4529 * The fasl file version number changed again, for dozens of reasons,
4530 some of which are apparent above.
4532 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4533 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4534 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4535 patch, so that DISASSEMBLE output is much nicer.
4536 * The code in the SB-PROFILE package now seems reasonably stable.
4537 I still haven't decided what the final interface should look like
4538 (I'd like PROFILE to interact cleanly with TRACE, since both
4539 facilities use function encapsulation) but if you have a need
4540 for profiling now, you can probably use it successfully with
4541 the current CMU-CL-style interface.
4542 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4543 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4544 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4545 link now returns the dangling link itself, and for similar
4546 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4547 link itself. (In these cases the old code signalled an error and
4548 looped endlessly, respectively.) Thus, DIRECTORY now works even
4549 in the presence of dangling and cyclic symbolic links.
4550 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4551 is now a supported extension again, since the consensus on
4552 sbcl-devel was that it can be useful for ordinary development
4553 work, not just for debugging SBCL itself.
4554 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4555 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4556 that currently-defined functions might be redefined later with
4557 different return types.
4558 * Hash tables can be printed readably, as inspired by CMU CL code
4559 of Eric Marsden and SBCL code of Martin Atzmueller.
4560 * better error handling in CLOS method combination, thanks to
4561 Martin Atzmueller porting Pierre Mai's CMU CL patches
4562 * more overflow fixes for >16Mbyte I/O buffers
4563 * A bug in READ has been fixed, so that now a single Ctrl-D
4564 character suffices to cause end-of-file on character streams.
4565 In particular, now you only need one Ctrl-D at the command
4566 line (not two) to exit SBCL.
4567 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4569 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4570 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4571 does the right thing.
4572 * The implementation of some type tests, especially for CONDITION
4573 types, is now tidier and maybe faster, due to CMU CL code
4574 originally by Douglas Crosher, ported by Martin Atzmueller.
4575 * Some math functions have been fixed, and there are new
4576 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4577 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4578 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4579 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4580 likely to be built into sbcl-0.7.0.)
4581 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4582 CL, ported to SBCL by Martin Atzmueller)
4583 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4584 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4585 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4586 * There's a new slam.sh hack to shorten the edit/compile/debug
4587 cycle for low-level changes to SBCL itself, and a new
4588 :SB-AFTER-XC-CORE target feature to control the generation of
4589 the after-xc.core file needed by slam.sh.
4590 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4591 COMPILE-FILE is no longer supported, so that now every function
4592 gets an entry point, so that block compilation looks a little
4593 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4594 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4595 returns the number of bytes consed since the system started,
4596 rather than the number consed since the first time the function
4597 was called. (The new definition parallels ANSI functions like
4598 CL:GET-INTERNAL-RUN-TIME.)
4599 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4600 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4601 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4602 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4603 * Fasl file version numbers are now independent of the target CPU,
4604 since historically most system changes which required version
4605 number changes have affected all CPUs equally. Similarly,
4606 the byte fasl file version is now equal to the ordinary
4609 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4610 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4611 is no longer recognized. I apologize for this, because it was
4612 listed in SB-EXT as a supported extension, but I found that
4613 its existing behavior was poorly specified, as well as incorrectly
4614 specified, and it looked like too much of a mess to straighten it
4615 out. I have enough on my hands trying to get ANSI stuff to work..
4616 * many patches ported from CMU CL by Martin Atzmueller, with
4617 half a dozen bug fixes in pretty-printing and the debugger, and
4618 half a dozen others elsewhere
4619 * fixed bug 13: Floating point infinities are now supported again.
4620 They might still be a little bit flaky, but thanks to bug reports
4621 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4622 as flaky as they were.
4623 * The --noprogrammer command line option is now supported. (Its
4624 behavior is slightly different in detail from what the old man
4625 page claimed it would do, but it's still appropriate under the
4626 same circumstances that the man page talks about.)
4627 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4628 are now supported, and enabled by default. Thus, the compiler can
4629 handle many floating point and complex operations much less
4630 inefficiently. (Thus e.g. you can implement a complex FFT
4632 * The compiler now detects type mismatches between DECLAIM FTYPE
4633 and DEFUN better, and implements CHECK-TYPE more correctly, and
4634 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4635 thanks to patches from Martin Atzmueller.
4636 * various fixes to make the cross-compiler more portable to
4637 ANSI-conforming-but-different cross-compilation hosts (notably
4638 Lispworks for Windows, following bug reports from Arthur Lemmens)
4639 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4640 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4641 fixed thanks to Pierre Mai's CMU CL patches.
4642 * fixes to keep the system from overflowing internal counters when
4643 it tries to use i/o buffers larger than 16M bytes
4644 * fixed bug 45a: Various internal functions required to support
4645 complex special functions have been merged from CMU CL sources.
4646 (When I was first setting up SBCL, I misunderstood a compile-time
4647 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4648 * improved support for type intersection and union, fixing bug 12
4649 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4650 more obscure bugs as well
4651 * some steps toward byte-compiling non-performance-critical
4652 parts of the system, courtesy of patches from Martin Atzmueller
4653 * Christophe Rhodes has made some debian packages of sbcl at
4654 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4655 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4656 stable, but are nonetheless usable. When he's ready, I'd be happy
4657 to add them to the SourceForge "File Releases" section. (And if
4658 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4659 * new fasl file format version number (because of changes in
4660 internal representation of (OR ..) types to accommodate the new
4661 support for (AND ..) types, among other things)
4663 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4664 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4666 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4668 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4669 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4670 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4671 * DESCRIBE now works on structure objects again.
4672 * Most function call argument type mismatches are now handled as
4673 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4674 know whether the function will be redefined before the call is
4675 executed. (The compiler could flag local calls with full WARNINGs,
4676 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4677 it doesn't keep track of enough information to know whether calls
4678 are local in this sense.)
4679 * Compiler output is now more verbose, with messages truncated
4680 later than before. (There should be some supported way for users
4681 to override the default verbosity, but I haven't decided how to
4682 provide it yet, so this behavior is still controlled by the internal
4683 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4684 src/compiler/ir1util.lisp.)
4685 * Fasl file format version numbers have increased again, because
4686 support for the Gray streams extension changes the layout of the
4687 system's STREAM objects.
4688 * The Gray subclassable streams extension now works, thanks to a
4689 patch from Martin Atzmueller.
4690 * The full LOAD-FOREIGN extension (not just the primitive
4691 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4692 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4693 but like most other programs, it defaults to copying the Unix
4694 environment from the original process instead of starting the
4695 new process in an empty environment.
4696 * Extensions which manipulate the Unix environment now support
4697 an :ENVIRONMENT keyword option which doesn't smash case or
4698 do other bad things. The CMU-CL-style :ENV option is retained
4699 for porting convenience.
4700 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4701 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4703 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4705 * A patch from Martin Atzmueller seems to have solved the SIGINT
4706 problem, and as far as we know, signal-handling now works cleanly.
4707 (If you find any new bugs, please report them!)
4708 * The system no longer defaults Lisp source file names to types
4709 ".l", ".cl", or ".lsp", but only to ".lisp".
4710 * The compiler no longer uses special default file extensions for
4711 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4712 seems to expect a single default extension for all compiled code,
4713 and there's no compelling reason to try to stretch the standard
4714 to allow two different extensions.) Instead, byte-compiled files
4715 default to the same extension as native-compiled files.
4716 * Fasl file format version numbers have increased again, because
4717 a rearrangement of internal implementation packages made some
4718 dumped symbols in old fasl files unreadable in new cores.
4719 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4720 many fewer weird special cases.
4721 * Bug #17 (differing COMPILE-FILE behavior between logical and
4722 physical pathnames) has been fixed, and some related misbehavior too,
4723 thanks to a patch from Martin Atzmueller.
4724 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4725 by Tim Moore, ported to SBCL by Martin Atzmueller.
4726 * Martin Atzmueller fixed several filesystem-related problems,
4727 including bug #36, in part by porting CMU CL patches, which were
4728 written in part by Paul Werkowski.
4729 * More compiler warnings in src/runtime/ are gone, thanks to
4730 more patches from Martin Atzmueller.
4731 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4734 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4736 * DESCRIBE now works on CONDITION objects.
4737 * The debugger now handles errors which arise when trying to print
4738 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4740 * The build system now uses an additional file, customize-target-features.lisp,
4741 to allow local modifications to the target *FEATURES* list. (The point of
4742 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4743 debugging features enabled, without having to worry about propagating it
4744 into everyone's system when I do a "cvs update".) When no
4745 customize-target-features.lisp file exists, the target *FEATURES* list
4746 should be constructed the same way as before.
4747 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4748 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4749 warn about behavior which is undefined under the ANSI spec. Note: This
4750 is specified by ANSI, but it's not very popular with programmers.
4751 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4752 macro in the SBCL sources for an example of a workaround which you
4753 might use to make portable ANSI-standard code which does what you want.
4754 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4755 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4756 when it's used as a non-toplevel form. (This is required in order
4757 to implement the DEFCONSTANT-EQX macro.)
4758 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4759 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4760 funny things, due to bugs in the implementation of EVAL-WHEN
4761 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4762 non-toplevel DEFCONSTANTs are uncommon.)
4763 * The core file version number and fasl file version number have been
4764 incremented, because the old noncompliant DEFCONSTANT behavior involved
4765 calling functions which no longer exist, and because I also took the
4766 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4767 * fixed bug 1 (error handling before read-eval-print loop starts), and
4768 redid debugger restarts and related debugger commands somewhat while
4770 ** The QUIT debugger command is gone, since it did something
4771 rather different than the SB-EXT:QUIT command, and since it never
4772 worked properly outside the main toplevel read/eval/print loop.
4773 Invoking the new TOPLEVEL restart provides the same functionality.
4774 ** The GO debugger command is also gone, since you can just invoke
4775 the CONTINUE restart directly instead.
4776 ** The TOP debugger command is also gone, since it's redundant with the
4777 FRAME 0 command, and since it interfered with abbreviations for the
4779 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4781 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4782 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4783 some porting work by Martin Atzmueller)
4784 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4785 with a boatload of patches which he ported to SBCL. Now that those
4787 ** The system tries to make sure that its low-priority messages
4788 are prefixed by semicolons, to help people who like to use
4789 syntax highlighting in their ILISP buffer. (This patch
4790 was originally due to Raymond Toy.)
4791 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4792 patches originally written by Raymond Toy.
4793 ** The compiler understands coercion between single-value and
4794 multiple-VALUES type expressions better, getting rid of some very
4795 weird behavior, thanks to patches originally by Robert MacLachlan
4796 and Douglas Crosher.
4797 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4798 lambda lists, thanks to a patch originally by Pierre Mai.
4799 ** The system no longer bogusly warns about "abbreviated type
4801 ** The compiler gets less confused by inlining and RETURN-FROM,
4802 thanks to some patches originally by Tim Moore.
4803 ** The system no longer hangs when dumping circular lists to fasl
4804 files, thanks to a patch originally from Douglas Crosher.
4805 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4806 undefined function error.
4807 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4808 in structure slot initforms) for now, documented workaround instead:-|
4809 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4810 * fixed bug 5: added stubs for various Gray stream functions called
4811 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4812 installed, at least appropriate type errors are generated
4813 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4814 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4815 consistently in DEFMETHOD forms.
4816 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4817 it doesn't seem to affect SBCL after all
4818 * The C runtime system now builds with better optimization and many
4819 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4821 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4823 * The system is now under CVS at SourceForge (instead of the
4824 CVS repository on my home machine).
4825 * The new signal handling code has been tweaked to treat register
4826 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4827 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4828 downstream, e.g. in debug-int.lisp, has implicit dependencies
4829 on the unsignedness of integer representation of machine words,
4830 and that caused the system to bomb out with infinite regress
4831 when trying to recover from type errors involving signed values,
4832 e.g. (BUTLAST '(1 2 3) -1).)
4833 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4834 (This was one of the bugs Peter Van Eynde reported back in July.)
4835 * The system now uses code inspired by Colin Walters' O(N)
4836 implementation of MAP (from the cmucl-imp@cons.org mailing
4837 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4838 inline the MAP operation, and there is more than one
4839 sequence argument to the MAP call (so that it can't just
4840 do ETYPECASE once and for all based on the type of the
4841 single sequence argument). (The old non-inline implementation
4842 of the general M-argument sequence-of-length-N case required
4843 O(M*N*N) time when any of the sequence arguments were LISTs.)
4844 * The QUIT :UNIX-CODE keyword argument has been renamed to
4845 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4846 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4848 * Raymond Wiker's patches to port dynamic loading from Linux to
4849 FreeBSD have been added.
4850 * The BUGS file is now more nearly up to date, thanks in large part
4851 to Martin Atzmueller's review of it.
4852 * The debugger now flushes standard output streams before it begins
4853 its output ("debugger invoked" and so forth).
4854 * The core version number and fasl file version number have both
4855 been incremented, because of incompatible changes in the layout
4857 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4858 used to not be called for a saved Lisp image.)
4859 * Martin Atzmueller's version of a patch to fix a compiler crash,
4860 as posted on sbcl-devel 13 September 2000, has been installed.
4861 * Instead of installing Martin Atzmueller's patch for the
4862 compiler transform for SUBSEQ, I deleted the compiler transform,
4863 and transforms for some similar consing operations.
4864 * A bug in signal handling which kept TRACE from working on OpenBSD
4866 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4867 away by constant folding
4868 * The system now defines its address space constants in one place
4869 (in the Lisp sources), and propagates them automatically elsewhere
4870 (through GENESIS and the sbcl.h file). Therefore, patching the
4871 address map is less unnecessarily tedious and error-prone. The
4872 Lisp names of address space constants have also been systematized.
4873 * CVS tags like dollar-Header-dollar have been removed from
4874 the sources, because they have never saved me trouble and
4875 they've been source of trouble working with patches and other
4876 diff-related operations.
4877 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4878 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4880 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4882 * The system has been ported to OpenBSD.
4883 * The system now compiles with a simple "sh make.sh" on the systems
4884 that it's supported on. I.e., now you no longer need to tweak
4885 text in the target-features.lisp-expr and symlinks in src/runtime/
4886 by hand, the make.sh takes care of it for you.
4887 * The system is no longer so grossly inefficient when compiling code
4888 involving vectors implemented as general (not simple) vectors (VECTOR T),
4889 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4890 which dares to use the various sequence functions on non-simple
4891 vectors, takes less of a performance hit.
4892 * There is now a primitive type predicate VECTOR-T-P
4893 to test for the (VECTOR T) type, so that e.g.
4894 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4895 can now be compiled with some semblance of efficiency. (The old code
4896 turned the type declaration into a full call to %TYPEP at runtime!)
4897 * AREF on (VECTOR T) is still not fast, since it's still compiled
4898 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4899 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4900 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4901 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4902 without making full calls to SUBTYPEP at runtime.
4903 (Some analogous efficiency issues for non-simple vectors specialized to
4904 element types other than T, or for non-simple multidimensional arrays,
4905 have not been addressed. They could almost certainly be handled the
4906 same way if anyone is motivated to do so.)
4907 * The changes in array handling break binary compatibility, so
4908 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4909 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4910 instead of (VALUES T T).
4911 * By following the instructions that Dan Barlow posted to sbcl-devel
4912 on 2 July 2000, I was able to enable primitive dynamic object
4913 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4914 functionality is not implemented (since it calls ld to resolve
4915 library references automatically, requiring RUN-PROGRAM for its
4916 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4917 try to resolve library references) is now supported.
4918 * The system now flushes the standard output streams when it terminates,
4919 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4920 them at several other probably-convenient times, e.g. in each pass of
4921 the toplevel read-eval-print loop, and after evaluating a form given
4922 as an "--eval" command-line option. (These changes were motivated by a
4923 discussion of stream flushing issues on cmucl-imp in August 2000.)
4924 * The source transform for TYPEP of array types no longer assumes
4925 that an array whose element type is a not-yet-defined type
4926 is implemented as an array of T, but instead punts, so that the
4927 type will be interpreted at runtime.
4928 * There is now some support for cross-compiling in make.sh: each of
4929 the phases of make.sh has its own script. (This should be transparent
4930 to people doing ordinary, non-cross-compile builds.)
4931 * Since my laptop doesn't have hundreds of megabytes of memory like
4932 my desktop machine, I became more motivated to do some items on
4933 my to-do list in order to reduce the size of the system a little:
4934 ** Arrange for various needed-only-at-cold-init things to be
4935 uninterned after cold init. To support this, those things have
4936 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4937 symbols with such names are now uninterned after cold init).
4938 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4939 fasl files for cold load.
4940 ** Remove the old compiler structure pooling code (which used to
4941 be conditional on the target feature :SB-ALLOC) completely.
4942 ** Redo the representation of some data in cold init to be more compact.
4943 (I also looked into supporting byte compiled code at bootstrap time,
4944 which would probably reduce the size of the system a lot, but that
4945 looked too complicated, so I punted for now.)
4946 * The maximum signal nesting depth in the src/runtime/ support code has
4947 been reduced from 4096 to 256. (I don't know any reason for the very
4948 large old value. If the new smaller value turns out to break something,
4949 I'll probably just bump it back up.)
4950 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4952 * Many, many bugs reported by Peter Van Eynde have been added to
4953 the BUGS list; some have even been fixed.
4954 * While enabling dynamic object file loading, I tried to make the
4955 code easier to understand, renaming various functions and variables
4956 with less ambiguous names, and changing some function calling
4957 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4958 * While trying to figure out how to do the OpenBSD port, I tried to
4959 clean up some of the code in src/runtime/. In particular, I dropped
4960 support for non-POSIX signal handling, added various comments,
4961 tweaked the code to reduce the number of compilation warnings, and
4962 renamed some files to increase consistency.
4963 * To support the new automatic configuration functionality in make.sh,
4964 the source file target-features.lisp-expr has been replaced with the
4965 source file base-target-features.lisp-expr and the machine-generated
4966 file local-target-features.lisp-expr.
4967 * fixed a stupid quoting error in make.sh so that using CMU CL
4968 "lisp -batch" as cross-compilation host works again
4970 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4972 * DESCRIBE no longer tries to call itself recursively to describe
4973 bound/fbound values, so that it no longer fails on symbols which are
4974 bound to themselves (like keywords, T, and NIL).
4975 * DESCRIBE now works on generic functions.
4976 * The printer now prints less-screwed-up representations of closures
4977 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4978 * A private symbol is used instead of the :EMPTY keyword previously
4979 used to mark empty slots in hash tables. Thus
4980 (DEFVAR *HT* (MAKE-HASH-TABLE))
4981 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4982 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4983 now does what ANSI says that it should. (You can still get
4984 similar noncompliant behavior if bang on the hash table
4985 implementation with all the symbols you get back from
4986 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4987 This breaks binary compatibility, since tests for equality to
4988 :EMPTY are wired into things like the macroexpansion of
4989 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4991 * There's now a minimal placeholder implementation for CL:STEP,
4992 as required by ANSI.
4993 * An obscure bug in the interaction of the normal compiler, the byte
4994 compiler, inlining, and structure predicates has been patched
4995 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4996 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4998 * Missing ordinary arguments in a macro call are now detected even
4999 when the macro lambda list contains &KEY or &REST.
5000 * The debugger no longer complains about encountering the top of the
5001 stack when you type "FRAME 0" to explicitly instruct it to go to
5002 the top of the stack. And it now prints the frame you request even
5003 if it's the current frame (instead of saying "You are here.").
5004 * As specified by ANSI, the system now always prints keywords
5005 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5007 * The default initial SIZE of HASH-TABLEs is now smaller.
5008 * Type information from CLOS class dispatch is now propagated
5009 into DEFMETHOD bodies, so that e.g.
5010 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5012 is now basically equivalent to
5013 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5014 (DECLARE (TYPE SINGLE-FLOAT X))
5016 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5017 operation, without having to do run-time type dispatch.
5018 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5019 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5020 * Since I don't seem to be making big file reorganizations very often
5021 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5022 is overflowing my ability to conveniently back them up), I've finally
5023 checked the system into CVS. (The CVS repository is on my home system,
5024 not at SourceForge -- putting it on SourceForge might come later.)
5025 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5026 high-level GC-NOTIFY-FOO functions send their output. (There's
5027 still very little control of where low-level verbose GC functions
5028 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5029 less than it used to -- the GC-NOTIFY-FOO functions are now under
5030 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5031 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5032 in only one place in the source code, and propagates it automatically
5033 everywhere that it's needed. Thus e.g. when I bump the version from
5034 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5035 * The C source files now include boilerplate legalese and documentation
5036 at the head of each file (just as the Lisp source files already did).
5037 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5038 to his page will be replaced with a link to his new CLiki service.
5040 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5042 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5043 * The build process now looks for GNU make under the default name "gmake",
5044 instead of "make" as it used to. If GNU make is not available as "gmake"
5045 on your system, you can change this default behavior by setting the
5046 GNUMAKE environment variable.
5047 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5048 can build without error under CMU CL.
5050 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5052 * There is now a partial SBCL user manual (with some new text and some
5053 text cribbed from the CMU CL manual).
5054 * The beginnings of a profiler have been added (starting with the
5055 CMU CL profiler and simplifying and cleaning up). Eventually the
5056 main interface should be through the TRACE macro, but for now,
5057 it's still accessed through vaguely CMU-CL-style functions and macros
5058 exported from the package SB-PROFILE.
5059 * Some problems left over from porting CMU CL to the new
5060 cross-compilation bootstrap process have been cleaned up:
5061 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5062 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5063 part when getting rid of the compiler *BACKEND* stuff.)
5064 ** The value of *NULL-TYPE* was screwed up, because it was
5065 being initialized before the type system knew the final
5066 definition of the 'NULL type. This screwed up several key
5067 optimizations in the compiler, causing inefficiency in all sorts
5068 of places. (I found it because I wanted to understand why
5069 GET-INTERNAL-RUN-TIME was consing.)
5070 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5071 PROCLAIM FTYPE information. Unfortunately this broke binary
5072 compatibility again, since now the forms output by DEFGENERIC
5073 to refer to functions which didn't exist in 0.6.3.
5074 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5075 can use the new (as of 0.6.3) transform for SOME into MAP into
5077 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5078 values to (INTEGER 0 1000000), so that the time code will no longer
5079 occasionally get blown up by Linux returning 1000000 microseconds
5080 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5081 its output conform to the ANSI spec. (Alas, this makes its output
5082 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5083 * A full call to MAP NIL with a single sequence argument no longer conses.
5084 * fixes to problems pointed out by Martin Atzmueller:
5085 * The manual page no longer talks about multiprocessing as though
5086 it were currently supported.
5087 * The ILISP support patches have been removed from the distribution,
5088 because as of version 5.10.1, ILISP now supports SBCL without us
5089 having to maintain patches.
5090 * added a modified version of Raymond Toy's recent CMU CL patch for
5091 EQUALP comparison of HASH-TABLE
5093 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5095 * The system still can't cross-compile itself with
5096 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5097 entails), but at least it can compile more of itself that way
5098 than it used to be able to, and various buglets which were uncovered
5099 by trying to cross-compile itself that way have now been fixed.
5100 * This release breaks binary compatibility again. This time
5101 at least I've incremented the FASL file format version to 2, so that the
5102 problem can be detected reliably instead of just causing weird errors.
5103 * various new style warnings:
5104 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5105 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5106 ** using the lexical binding of a variable named in the *FOO* style
5107 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5108 as specified by ANSI.
5109 * *RANDOM-STATE* is no longer automatically initialized from
5110 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5111 default behavior of the system is to repeat its behavior every time
5112 it's run. If you'd like to change this behavior, you can always
5113 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5114 old convention there was no comparably easy way to get the system to
5115 repeat its behavior every time it was run.
5116 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5117 FTYPE declarations has been removed, in favor of their ANSI
5118 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5119 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5120 the types of their sequence arguments can be determined at compile time.
5121 This is done through a new open code expansion for MAP which eliminates
5122 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5123 argument types can be determined at compile time.
5124 * The optimizer now transforms COERCE into an identity operation when it
5125 can prove that the coerced object is already of the correct type. (This
5126 can be a win for machine generated code, including the output of other
5127 optimization transforms, such as the MAP transform above.)
5128 * Credit information has been moved from source file headers into CREDITS.
5129 * Source file headers have been made more standard.
5130 * The CASE macro now compiles without complaining even when it has
5133 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5135 * (Note that the way that the PCL macroexpansions were rewritten
5136 to accommodate the change in DEFGENERIC below breaks binary
5137 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5138 not run under sbcl-0.6.2. Once we get out of alpha releases,
5139 i.e. hit release 1.0.0, we'll probably try to maintain binary
5140 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5141 and sbcl-1.4.4. Until then, however, it might be fairly common
5142 for maintenance releases to break binary compatibility.)
5143 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5145 * The reporting of unhandled signals has been changed to print some
5146 explanatory text as well as the report form. (Previously only
5147 the report form was printed.)
5148 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5149 it defines, so that the compiler no longer issues undefined function
5150 warnings for compiled-but-not-yet-loaded generic functions.
5151 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5152 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5153 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5154 section of the ANSI spec are supported.
5155 * The "" nickname for the "KEYWORD" package has been removed.
5156 The reader still handles symbol tokens which begin with a package marker
5157 as keywords, but it doesn't expose its mechanism for doing so in the
5158 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5159 * The system now issues STYLE-WARNINGs for contradictory TYPE
5160 proclamations. (Warnings for contradictory FTYPE proclamations would
5161 be nice too, but those can't be done usefully unless the type system
5162 is made smarter about FUNCTION types.)
5163 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5164 systematized, so that "*target-*.lisp is supposed to exist only on the
5165 target and imply that there's a related file which exists on the
5166 host, and *host-*.lisp is supposed to exist only on the host and imply
5167 that there's a related file which exists on the target. This involves a
5168 lot of renaming. Hopefully the acute confusion caused by the renaming
5169 will be justified by the reduction in chronic confusion..
5170 ** runtime-type.lisp -> early-target-type.lisp
5171 ** target-type.lisp -> late-target-type.lisp
5172 ** early-host-format.lisp -> early-format.lisp
5173 ** late-host-format.lisp -> late-format.lisp
5174 ** host-error.lisp -> misc-error.lisp
5175 ** early-error.lisp -> early-target-error.lisp
5176 ** late-error.lisp -> late-target-error.lisp
5177 ** host-defboot.lisp -> early-defboot.lisp
5178 ** code/misc.lisp -> code/target-misc.lisp
5179 ** code/host-misc.lisp -> code/misc.lisp
5180 ** code/numbers.lisp -> code/target-numbers.lisp
5181 ** code/early-numbers.lisp -> numbers.lisp
5182 ** early-host-type.lisp -> early-type.lisp
5183 ** late-host-type.lisp -> late-type.lisp
5184 ** host-typep.lisp -> typep.lisp
5185 ** load.lisp -> target-load.lisp
5186 ** host-load.lisp -> load.lisp
5187 ** host-disassem.lisp -> disassem.lisp
5188 ** host-insts.lisp -> insts.lisp
5189 ** byte-comp.lisp -> target-byte-comp.lisp
5190 ** host-byte-comp.lisp -> byte-comp.lisp
5191 ** host-signal.lisp -> signal.lisp
5192 ** host-defstruct.lisp -> defstruct.lisp
5193 ** late-target-type.lisp -> deftypes-for-target.lisp
5194 Furthermore, several other previously target-only files foo.lisp (e.g.
5195 hash-table.lisp and random.lisp) have been split into a target-and-host
5196 foo.lisp file and a target-only target-foo.lisp file, with their key type
5197 definitions in the target-and-host part, so that the cross-compiler will
5198 know more about target types.
5199 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5200 gone away. In their place are various *BACKEND-FOO* variables
5201 corresponding to the slots of the old structure.
5202 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5203 into the target SBCL has been fixed.
5204 * The chill.lisp system for loading cold code into a running SBCL
5206 * Support for the CMU CL "scavenger hook" extension has been removed.
5207 (It was undocumented and unused in the CMU CL sources that SBCL was
5208 derived from, and stale in sbcl-0.6.1.)
5209 * Various errors in the cross-compiler type system were detected
5210 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5211 (enabling various consistency checks). Many of them were fixed,
5212 but some hard problems remain, so the compiler is back to
5213 running without *TYPE-SYSTEM-INITIALIZED* for now.
5214 * As part of the cross-compiler type system cleanup, I implemented
5215 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5216 * I have started adding UNCROSS calls throughout the type system
5217 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5218 on cross-compiler input files will be able to go away, and various
5220 * CONSTANTP now returns true for quoted forms (as explicitly required
5223 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5225 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5226 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5227 (MAKE-LIST -1). In the longer term, it ought to become true
5228 that declarations are assertions even at SAFETY 1. For now, it's not
5229 quite true even at SAFETY 3, but it's at least more nearly true..
5230 (Note that this change seems to increases the size of the system by
5231 O(5%) and to decrease the speed of the compiler by 20% or more.)
5232 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5233 for the problem of printing dozens of lines of distracting information
5234 about low-level system machinery as part of the top stack frame
5235 on entry to the debugger when an undefined function was called.
5236 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5237 is not reset to COMMON-LISP-USER.
5238 * Compilation of stuff related to dyncount.lisp has been made conditional
5239 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5240 smaller. The various dyncount-related symbols have been moved into
5241 a new "SB-DYNCOUNT" package.
5242 * tty-inspect.lisp has been renamed to inspect.lisp.
5243 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5244 feature has gone away. (When we eventually port to other flavors of
5245 libc and/or Unix, we'll try to make the differences between flavors
5246 invisible at the user level.)
5247 * Various other *FEATURES* tags, and/or their associated conditionals,
5248 have been removed if obsolescent, or given better documentation, or
5249 sometimes given more-mnemonic names.
5251 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5253 * tidied up "make.sh" script
5254 * tidied up system directory structure
5255 * better "clean.sh" behavior
5256 * added doc/FOR-CMUCL-DEVELOPERS
5257 * many many small tweaks to output format, e.g. removing possibly-confusing
5258 trailing #\. character in DESCRIBE-INSTANCE
5259 * (EQUALP #\A 'A) no longer signals an error.
5260 * new hashing code, including EQUALP hashing
5261 * tidied up Lisp initialization and toplevel
5262 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5263 * command line argument processing
5264 * added POSIX-GETENV function to deal with Unix-ish environment variables
5265 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5266 terminating SBCL on EOF
5267 * non-verbose GC by default
5268 * There is no more "sbcl" shell script; the sbcl file is now the C
5269 runtime executable (just like CMU CL).
5270 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5272 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5273 symbol and package information as Lisp native symbol and package objects
5274 instead of strings naming symbols and strings naming packages. This way,
5275 whenever packages are renamed (as in warm init), debug information is
5276 transformed along with everything else.
5277 * tweaked the optimization policy declarations which control the building
5278 of SBCL itself. Now, among other things, the system no longer saves
5279 source location debugging information. (This helps two problems at once
5280 by reducing SBCL size and by keeping SBCL from trying to look for its
5281 sources -- which may not exist -- when reporting errors.)
5282 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5283 debugging and testing purposes
5284 * cleaned up printing, making the printer call PRINT-OBJECT for
5285 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5286 methods, giving nearly-ANSI behavior
5287 * converted almost all special variables to use *FOO* naming convention
5288 * deleted PARSE-TIME functionality, since it can be done portably
5289 * moved some files out of cold init into warm init
5290 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5292 * regularized formatting of source files
5293 * added an install.sh script
5294 * fixed ridiculous memory usage of cross-compiler by making
5295 compiler/alloc.lisp not try to do pooling unless it can hook
5296 itself into the GC of the cross-compilation host. Now the system
5297 builds nicely on my old laptop.
5298 * added :SB-ALLOC in target-features.lisp-expr
5299 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5300 was not implemented)
5301 * re-did condition handling and note reporting in the compiler. Notes
5302 are no longer handled by signalling conditions. Style warnings
5303 and warnings are handled more correctly and reported in such a way
5304 that it's easy to find one or the other in your output (so that you
5305 can e.g. figure out which of many problems caused COMPILE-FILE to
5307 * changed the severity of several compiler warnings from full WARNING
5308 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5309 compiler note reporting so that it doesn't use the condition system
5310 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5311 COMPILE-FILE command)
5312 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5313 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5314 no longer supported, since their implementation was deeply intertwingled
5315 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5316 * removed bogus "support" for compiler macros named (SETF FOO), and
5317 removed the compiler macro for SETF INFO (but only after making a fool
5318 of myself on the cmucl-imp mailing list by posting a bogus patch for
5319 DEFINE-COMPILER-MACRO..)
5320 * Compiled files containing forms which have side effects on the Lisp
5321 reader (such as DEFPACKAGE forms) are now handled more correctly.
5322 (Compiler queuing of top level lambdas has been suppressed by setting
5323 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5324 * deleted various currently-unused source files, e.g. gengc.lisp. They
5325 may be added back at some point e.g. when porting to other architectures,
5326 but until they are it's distracting to distribute them and to try to
5328 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5329 there were so many of them they're just distractions, and UNCROSS is
5330 known to be able to handle the current sources
5331 * moved PROFILE functionality into TRACE, so that it will be clear
5332 how the wrapping and unwrapping of functions when you profile them
5333 interacts with the wrapping and unwrapping of functions when you
5334 trace them. (Actually, the functionality isn't there yet, but at least
5335 the interface specification is there. Hopefully, the functionality will
5336 arrive with some future maintenance release.)
5337 * removed host-oops.lisp
5338 * changed signature of QUIT function to allow UNIX-CODE argument
5339 * fixed READ-SEQUENCE bug
5340 * tweaked verbose GC output so that it looks more like the progress
5341 output that ANSI specifies for functions like LOAD
5342 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5343 * added <http://sbcl.sourceforge.com> to the banner information printed by