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 * improvement: the Unicode character database has been upgraded to the
7 Unicode 5.0.1 standard, giving names and properties to a number of new
8 characters, and providing a few extra characters with case
10 * new feature: experimental :EMIT-CFASL parameter to COMPILE-FILE can
11 be used to output toplevel compile-time effects into a separate .CFASL
13 * optimization: COERCE to VECTOR, STRING, SIMPLE-STRING and recognizable
14 one-dimenstional subtypes of ARRAY is upto 70% faster when the coercion is
16 * optimization: division of floating point numbers by constants uses
17 multiplication by reciprocal when an exact reciprocal exists.
18 * optimization: multiplication of single- and double-floats floats by
19 constant two has been optimized.
20 * optimization: ARRAY-IN-BOUNDS-P is resolved at compile-time when
21 sufficient type information is available. (thanks to Leslie Polzer)
22 * improvement: a STYLE-WARNING is signalled when a generic function
23 clobbers an earlier FTYPE proclamation.
24 * improvement: the compiler is able to track the effective type of
25 generic function across method addition and removal even in the
26 absence of an explicit DEFGENERIC.
27 * improvement: DESCRIBE now reports on symbols naming undefined
28 but assumed or declared function as well.
29 * improvement: recompilation of systems using SB-GROVEL now works
30 (thanks to Leslie Polzer)
31 * improvements to SB-CLTL2 (thanks to Larry D'Anna):
32 ** functions DECLARATION-INFORMATION, PARSE-MACRO, and ENCLOSE have been
34 ** AUGMENT-ENVIRONMENT and DEFINE-DECLARATION have been implemented.
35 ** DECLARATION-INFORMATION now supports declaration name DECLARATION as
36 well as user defined declaration names.
37 ** VARIABLE-INFORMATION is now aware of alien variables.
38 * improvement: improved address space layout on OpenBSD (thanks to Josh
40 * improvement: pretty-printing of various Lisp forms has been improved
41 (thanks to Tobias Rittweiler)
42 * bug fix: a failing AVER in CONVERT-MV-CALL has been fixed. (thanks to
44 * bug fix: SLEEP supports times over 100 million seconds on long on OpenBSD
45 as well. (reported by Josh Elsasser)
46 * bug fix: DELETE-FILE on streams no longer closes the stream with :ABORT T,
47 leading to possible attempts to delete the same file twice. See docstring
48 on DELETE-FILE for details. (reported by John Fremlin)
49 * bug fix: the low-level debugger had 32-bit assumptions and was missing
50 information about some array types. (thanks to Luis Oliveira)
51 * bug fix: moderately complex combinations of inline expansions could
52 be miscompiled if the result was declared to be dynamic extent.
53 * bug fix: in some cases no compiler note about failure to stack allocate
54 was emitted, even if the objects were in fact heap allocated.
55 * bug fix: minor violation of "otherwise inaccessible" rule for stack
56 allocation could cause objects users might reasonably expect to
57 be heap allocated to be stack allocated.
58 * bug fix: DESCRIBE signalled an error for generic functions under
59 certain circumstances. (thanks to Leslie Polzer)
60 * bug fix: Fixed spelling of an error message.
62 changes in sbcl-1.0.30 relative to sbcl-1.0.29:
63 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
64 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
65 of SB-THREAD:THREAD-ERROR-THREAD.
66 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
67 * new feature: docstrings for local and anonymous functions are no longer
68 discarded. (thanks to Leslie Polzer)
69 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
70 values in other threads.
71 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
72 about object allocation.
73 * optimization: division of a real float by a complex float is implemented
74 with a specialised code sequence.
75 * optimization: MAKE-INSTANCE with non-constant class-argument but constant
76 keywords is an order of magnitude faster.
77 * optimization: MAKE-INSTANCE with constant keyword arguments is x2-4 faster
78 in the presence of :AROUND or non-standard primary INITIALIZE-INSTANCE
79 methods, and similarly for non-standard metaclass classes as long as there
80 are no methods additional on MAKE-INSTANCE.
81 * optimization: more efficient type-checks for FIXNUMs when the value
82 is known to be a signed word on x86 and x86-64.
83 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
84 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
86 * optimization: compiler is smarter about delegating argument type checks to
88 * optimization: several character functions are now compiled somewhat more
89 efficiently. (reported by Lynn Quam)
90 * optimization: the compiler now derives simple types for LOAD-VALUE-FORMs.
91 * improvement: less unsafe constant folding in floating point arithmetic,
92 especially for mixed complex/real -float operations.
93 * optimization: constant double and single floats are stored in native
94 unboxed format on x86[-64].
95 * optimization: smarter code for arithmetic operations with constant floats,
96 complex floats, or integers on x86[-64].
97 * optimization: smarter code for conjugate/multiplication of float complexes
98 and abs/negate of floats on x86-64.
99 * optimization: more efficient complex float and real float operations on
101 * improvement: complex float division is slightly more stable.
102 * improvement: DESCRIBE output has been reworked to be easier to read and
103 contains more pertinent information.
104 * improvement: failure to provide requested stack allocation compiler notes
105 provided in all cases (requested stack allocation not happening without a
106 note being issued is now considered a bug.)
107 * bug fix: SB-POSIX exports the documented types and functions
108 FILE-DESCRIPTOR and FILENAME, and also the corresponding -DESCRIPTOR
109 types. (reported by "abhi")
110 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
111 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
112 (thanks to Paul Khuong)
113 * bug fix: looping around HANDLER-CASE could silently consume stack space
114 on each iteration. (reported by "foobar")
115 * bug fix: better error signalling when calls to functions seeking elements
116 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
117 (reported by Tobias Rittweiler)
118 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
119 directory components now works as it used to. (various prolems reported by
120 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
121 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
122 (reported by Elliot Slaughter, thanks to Stas Boukarev)
123 * bug fix: bogus undefined variable warnings from fopcompiled references to
124 global variables. (thanks to Lars Rune Nøstdal)
125 * bug fix: foreign function names should now appear in backtraces on
126 FC6 as well. (reported by Tomasz Skutnik and Tobias Rautenkranz)
127 * bug fix: SETF compiler macro documentation strings are not discarded
129 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
130 (thanks to Alex Plotnick)
131 * bug fix: SLEEP accepts large integer arguments, truncating them to
132 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
133 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
134 * bug fix: compiler notes for expensive slot type checks could be emitted
135 at runtime MAKE-INSTANCE calls. (reported by Samium Gromoff)
137 changes in sbcl-1.0.29 relative to 1.0.28:
138 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
139 https://bugs.launchpad.net/sbcl
140 Bugs can be reported directly there, or by sending email to
141 sbcl-bugs@lists.sourceforge.net
142 (no subscription required.)
143 * minor incompatible change: under weak type checking policy integer
144 types are weakened less aggressively.
145 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
146 allowed to return, which causes SBCL to quit with exit status 0. Previously
147 if the function returned with a small integer return value, that value
148 was accidentally reused as the exit status.
149 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
150 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
151 :HASH-FUNCTION argument. Refer to user manual for details.
152 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
154 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
155 microseconds since the Unix epoch on all platforms.
156 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
157 allows the compiler to safely elide boundedness checks for special
159 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
160 the symbol, prohibits both lexical and dynamic binding. This is mainly an
161 efficiency measure for threaded platforms, but also valueable in
163 * new feature: UNC pathnames are now understood by the system on Windows.
164 * optimization: the compiler uses a specialized version of FILL when the
165 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
167 * optimization: accesses to potentially non-simple arrays where element type
168 is known are 50% faster.
169 * optimization: compiler now generates faster array typechecking code.
170 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
172 * optimization: multidimensional array accesses in the absence of type
173 information regarding array rank are approximately 10% faster due to
174 open coding of ARRAY-RANK.
175 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
176 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
178 * optimization: result of call to VECTOR can now be stack allocated.
179 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
180 as long as the resulting array is one-dimensional and has a known
181 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
182 length of the list matches the known length of the vector does not
183 allocate the list as an intermediate step. Ditto for VECTOR and simple
185 * optimization: MAKE-ARRAY can now stack allocate in the presence of
186 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
187 known element type, and is known to be simple and one dimensional.
188 * improvement: SBCL now emits a compiler note where stack allocation was
189 requested but could not be provided (not in all cases, unfortunately)
190 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
191 * improvement: pretty-printing loop has been implemented properly. (thanks
192 to Tobias Rittweiler)
193 * documentation: CLOS slot typechecing policy has been documented.
194 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
195 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
196 pointer, leading to memory faults. (thanks to Bart Botta)
197 * bug fix: better floating point exception handling on x86/OpenBSD.
198 (thanks to Josh Elsasser)
199 * bug fix: exit status from QUIT when called under --script was lost
200 (reported by Hubert Kauker)
201 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
202 same implementation of FILL to initialize the array, even if a more
203 efficient one was available (reported by Stas Boukarev, thanks to
205 * bug fix: potential miscompilation of array stack allocation on x86 and
206 x86-64. (reported by Time Tossavainen)
207 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
208 result in their subforms being treated as top level forms. (reported by
210 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
211 than a machine register could leave garbage in the high bits of the
212 result register (bug 316325).
213 * bug fix: disable address space randomization Linux/x86-64 as well,
214 not just x86-64. (reported by Ken Olum)
215 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
216 generate incorrect code.
217 * bug fix: #201; type inference for CONS and ARRAY types could derive
218 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
219 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
220 could not be assigned to or bound with PROGV. (reported by Lorenz
222 * bug fix: the value of CL:- in the inspector was the previous expression
223 evaluated rather than the expression being evaluated.
224 * bug fix: constants can no longer be locally declared special.
225 * bug fix: signals delivered to threads started from foreign land (read:
226 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
227 thread by blocking all signals and resignalling.
228 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
229 from :INITFORM, if any.
231 changes in sbcl-1.0.28 relative to 1.0.27:
232 * a number of bugs in cross-compilation have been fixed, with the ultimate
233 result that building under (at least) clisp should be much more reliable.
234 * minor incompatible changes: echo-streams now propagate unread-char to the
235 underlying input stream, and no longer permit unreading more than one
237 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
238 frames, allowing for instance more reliable backtraces.
239 * improvement: the debugger REPL can now reference lexical variables
240 by name directly for code compiled with (DEBUG 3).
241 * improvement: errors from malformed declarations now have better source
242 paths associated with them. (thanks to Tobias Rittweiler)
243 * optimization: faster local calls on x86/x86-64
244 * bug fix: some error messages for out-of-bound array indexes confused the
245 index and the bound. (thanks to Stas Boukarev)
246 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
248 * bug fix: running regressions tests in shells without OSTYPE set now works.
249 (reported by Harald Hanche-Olsen)
250 * bug fix: more robust static space exhaustion signalling from
251 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
252 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
253 docstring away instead of storing it under names such as (LAMBDA (X)).
254 (reported by Leslie Polzer)
255 * bug fix: timers could go off in the wrong order, be delayed indefinitely
256 (thanks to Ole Arndt for the patch)
257 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
258 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
259 potentially-invalid effective methods in its cache.
260 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
261 instances as well (thanks to Paul Khuong)
262 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
263 SB-POSIX:WAITPID and their C equivalents.
264 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
266 changes in sbcl-1.0.27 relative to 1.0.26:
267 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
268 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
269 * improvement: the system either recovers from stack exhaustion or dies
270 properly as opposed to leaving the user uncertain of whether the handler
271 trampled on some random memory next to the stack or having to rely on
272 --lose-on-corruption (which is still a good idea to use in production
273 because stack exhaustion can happen in signal handlers which will likely
275 * bug fix: fix gc related interrupt handling bug on ppc (regression from
276 1.0.25.37, reported by Harald Hanche-Olsen)
277 * bug fix: work around signal delivery bug in darwin (regression from
278 1.0.25.44, reported by Sidney Markowitz)
279 * bug fix: fix ERROR leaking memory (reported by David Thompson)
281 changes in sbcl-1.0.26 relative to 1.0.25:
282 * incompatible change: an interruption (be it a function passed to
283 INTERRUPT-THREAD or a timer function) runs in an environment where
284 interrupts can be enabled. The interruption can use
285 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
286 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
287 potentially running out of stack. Keep in mind that in the absence
288 of WITHOUT-INTERRUPTS some potentially blocking operation such as
289 acquiring a lock can enable interrupts.
290 * incompatible change: GC-OFF and GC-ON are removed, as they were
291 always unsafe. Use WITHOUT-GCING instead.
292 * new feature: runtime option --disable-ldb
293 * new feature: runtime option --lose-on-corruption to die at the
294 slightest hint of possibly non-recoverable errors: running out of
295 memory, stack, alien stack, binding stack, encountering a memory
296 fault, etc. In the absence of --lose-on-corruption a warning is
298 * enhancement: detect binding stack exhaustion
299 * enhancement: detect alien stack exhaustion on x86/x86-64
300 * improvement: generally more stable and reliable interrupt handling
301 * improvement: there is a per thread interruption queue,
302 interruptions are executed in order of arrival
303 * improvement: a repeating timer reschedules itself when the it has
304 finished, but expiration times are spaced equally. If an
305 expiration time is in the past it will trigger after a short grace
306 period that may give a chance to other things to run.
307 * optimization: slightly faster gc on multithreaded builds
308 * optimization: faster WITHOUT-GCING
309 * bug fix: when JOIN-THREAD signals an error, do it when not holding
310 important locks so that the debugger/handler doesn't produce
311 recursive errors or deadlock.
312 * bug fix: real-time signals are not used anymore, so no more
313 hanging when the system wide real-time signal queue gets full.
314 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
316 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
317 * bug fix: fix random memory faults related to interrupts on alpha
318 * bug fix: fix random memory faults related to interrupts on sparc
319 * bug fix: fix deadlocks related to starting threads
320 * bug fix: fix deadlines on locks on futex platforms
321 * bug fix: restore errno in signal handlers
322 * bug fix: fix deadlocks related to hash tables
323 * bug fix: fix deadlocks in pcl
325 changes in sbcl-1.0.25 relative to 1.0.24:
326 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
327 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
328 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
329 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
330 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
331 better name and does not return values so stale on multiprocessor systems.
332 Also, HOLDING-MUTEX-P was added for about the only sane usage of
334 * improvement: unithread builds keep track of MUTEX-VALUE.
335 * improvement: reading from a TWO-WAY-STREAM does not touch the output
336 stream anymore making it thread safe to have a concurrent reader and
337 a writer, for instance, in a pipe.
338 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
339 arguments, making compiler-macros for SETF-functions able to inspect
340 their constant arguments.
341 * improvement: COMPILE-FILE reports times with millisecond accuracy
342 (thanks to Luis Oliveira)
343 * optimization: CHAR-CODE type derivation has been improved, making
344 TYPEP elimination on subtypes of CHARACTER work better. (reported
345 by Tobias Rittweiler, patch by Paul Khuong)
346 * bug fix: setting alien structure fields of type struct by value now
347 computes the right offset for the memory copy.
348 * bug fix: compilation problem involving inlined calls to aliens with
349 result type VOID. (reported by Ken Olum)
350 * bug fix: #235a; sequential inline expasion in different policies no
351 longer reuses the functional from the previous expansion site.
352 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
353 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
354 * bug fix: named ENUMs in multiply-referenced alien record types no longer
355 cause an implied type redefinition. Regression from 1.0.21.29.
356 * bug fix: sign-extension of small signed return values now works with C
357 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
358 * improvements to the Windows port:
359 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
360 Westfold, thanks to Rudi Schlatte)
362 changes in sbcl-1.0.24 relative to 1.0.23:
363 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
364 vector of a multidimensional SIMPLE-ARRAY.
365 * new feature: the system now signals a continuable error if standard
366 readtable modification is attempted. (thanks to Tobias Rittweiler)
367 * new feature: DIRECTORY has been extended with a non-standard keyword
368 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
369 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
370 has been robustified and documented better. (thanks to Robert Goldman)
371 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
372 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
373 * tradeoff: constant FORMAT control strings are now compiled unless
374 SPACE > SPEED (previously only when SPEED > SPACE.)
375 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
377 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
379 * bug fix: lexical type declarations are now correctly reported by
380 SB-CLTL2. (reported by Larry D'Anna)
381 * bug fix: STRING-TO-OCTETS did not handle :START properly when
382 using UTF-8 as external format. (thanks to Luis Oliveira)
383 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
384 are signalled correctly. (thanks to Stas Boukarev)
385 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
386 designator. (thanks to Tobias Rittweiler)
387 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
388 designator, and returns T instead of the function. (thanks to
390 * bug fix: direct superclasses of STANDARD-CLASS and
391 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
392 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
393 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
394 when no class with the named slot yet exists no longer causes a
395 compile-time style-warning.
396 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
397 in safe code. (reported by Didier Verna)
398 * bug fix: #430; nested structure constructors can stack allocate.
399 * bug fix: on some 64-bit platforms dynamic space size was truncated
400 to #xffff0000 bytes. (reported by Benjamin Lambert)
401 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
402 REPL unusable. (reported by Daniel Herring)
403 * bug fix: return values of READ-SEQUENCE did not take :START into
404 account on file streams, regressions since 1.0.12.22. (reported by
405 Thomas Russ, patch by Paul Khuong)
406 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
407 method specializer used to confuse permuation vector optimization.
408 * bug fix: system inserted bogus implicit type declarations for local
409 special variables in DEFMETHOD bodies.
410 * bug fix: #354; duplicated frames in backtraces due to
411 non-tail-call-optimized XEPs to functions with return type NIL
413 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
414 initializes structure object slots according to DEFSTRUCT initforms,
415 and DEFSTRUCT forms :INCLUDEind structure classes defined using
416 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
417 (reported by Bruno Haible and Stephen Wilson)
418 * bug fix: #395; fill-pointer output streams used now support
419 element-type BASE-CHAR as well.
420 * bug fix: compiler error when attempting to derive return value of
421 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
423 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
425 * bug fix: Mac OS X binaries should now be portable between Leopard
428 changes in sbcl-1.0.23 relative to 1.0.22:
429 * enhancement: when disassembling method functions, disassembly
430 for the associated fast function is also produced.
431 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
433 * optimization: printing with *PRINT-PRETTY* true is now more
434 efficient as long as the object being printed doesn't require
435 special handling by the pretty printer.
436 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
437 now interact correctly with type declarations.
438 * partial bug fix: PCL detects infinite recursion during wrapper
439 validation. (thanks to Attila Lendvai)
440 * bug fix: #426; nested function calls are inlined properly.
441 Previously if FOO was an inline function, in calls of the form
442 (FOO (FOO ...)) the outer call was not inlined.
443 * bug fix: long long arguments passed in registers to alien
444 functions on PowerPC Linux and NetBSD are now handled correctly
445 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
447 changes in sbcl-1.0.22 relative to 1.0.21:
448 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
449 for the shared object in the current directory, but passes the native
450 namestring of the designated pathname to the operation system's shared
451 object loading function as-it.
452 * minor incompatible change: --disable-debugger toplevel option now takes
453 effect before processing of initialization files and --eval or --load
455 * new feature: new commandline argument: --script, which supports
456 shebang lines. See documentation for details. (based on work by
458 * new feature: SAVE-LISP-AND-DIE can save current values of
459 --dynamic-space-size and --control-stack-size in the executable core,
460 causing it to skip normal runtime option processing. See documentation
461 for details. (thanks to Zach Beane)
462 * enhancement: inoccous calls to EVAL or generic functions dispatching
463 on subclasses of eg. STREAM no longer cause compiler notes to appear.
464 * enhancement: the system no longer resignals errors from --load and
465 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
466 associated with the original error to be lost. (thanks to Ariel
468 * enhancement: :DONT-SAVE keyword argument has been added to
469 LOAD-SHARED-OBJECT for controlling interaction with
470 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
471 the effects of an earlier LOAD-SHARED-OBJECT call.
472 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
473 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
474 provided. (reported by Cedric St-Jean)
475 * bug fix: circularity handling in the reader did not treat raw
476 structure slots correctly. (reported by Cedric St-Jean)
477 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
478 file descriptors when there were none.
479 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
480 was broken. (reporteed by Willem Broekema)
481 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
482 pathnames without a directory.
483 * bug fix: COMPILE-FILE did not warn about undefined variable
484 references at toplevel, and LOAD of the resulting fasl did
486 * bug fix: functions with non-required arguments used to end up with
487 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
488 * bug fix: redefining a function with non-required arguments didn't
489 update the system's knowledge about its call signature properly.
490 * bug fix: fixed #431; incompatible alien record type redefinitions
491 are detected and handled. (reported by Neil Haven)
492 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
493 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
495 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
496 after alien stack frames.
497 * bug fix: strings passed to mk{s,d}temp lacked null termination
499 changes in sbcl-1.0.21 relative to 1.0.20:
500 * new feature: the compiler is able to track the effective type of a
501 generic function across method addition and removal.
502 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
503 appropriately typed structure slots without locking.
504 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
505 information like those gathered by TIME using a programming-friendly
507 * new feature: TIME reports time taken even if the form performs a
508 non-local transfer of control.
509 * enhancement: reduced conservativism on GENCGC platforms: on
510 average 45% less pages pinned (measured from SBCL self build).
511 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
512 mutate constant symbols or violate declaimed type of the symbol.
513 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
514 sleeping on the semaphore in heavily contested semaphores.
515 * bug fix: semaphores and condition variables were not interrupt
517 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
518 owned by other threads anymore.
519 * bug fix: FIND on lists called KEY outside the specified
520 subsequence. (reported by budden)
521 * bug fix: LOG doesn't use single-float intermediate results when
522 given mixed integer and double-float arguments, leading to better
523 precision. (reported by Bob Felts)
524 * bug fix: LOG with base zero returned values of inconsistent type.
526 changes in sbcl-1.0.20 relative to 1.0.19:
527 * minor incompatible change: OPTIMIZE qualities
528 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
529 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
530 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
531 * documentation: some slot access efficiency guidelines have been
532 added to the user manual.
533 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
534 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
535 as ASSOC and MEMEBER.
536 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
537 to more efficient EQ-comparison versions more often.
538 * optimization: enhanced derivation of DOLIST iteration variable type
540 * optimization: constant folding of simple (LIST ...) forms as DOLIST
542 * optimization: runtime lookup of function definitions can be
543 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
544 * optimization: compiler is able to derive the return type of
545 (AREF (THE STRING X) Y) as being CHARACTER.
546 * optimization: CLRHASH on empty hash-tables no longer does pointless
547 work. (thanks to Alec Berryman)
548 * optimization: REPLACE deftransforms don't punt when :START1 or
550 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
551 no longer emits a code deletion note. (reported by Andrew
553 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
554 functions with an odd number of &OPTIONAL arguments, a &REST
555 argument, and one or more &KEY arguments at the call site.
556 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
557 arguments appeared at call sites as well.
558 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
559 READ-LINE &co can no longer read from them afterwards. (reported
561 * bug fix: fixed #427: unused local aliens no longer cause compiler
562 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
564 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
565 alien-stack leakage. (reported by Andy Hefner)
566 * bug fix: PROGV signals an error when an attempt to violate declared
567 type of a variable or bind a constant is made.
568 * bug fix: SET signals an error when an attempt to violate declared
569 type of a variable is made.
570 * bug fix: restart computation during the execution of a restart
571 test function no longer causes infinite recursion. (reported by
573 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
574 signals an error. (thanks to Pierre Mai)
575 * bug fix: EXPORT left symbol unexported in conflict situations.
576 (thanks to Michael Weber)
577 * bug fix: correctly handle name conflicts involving CL:NIL.
578 (thanks to Michael Weber)
579 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
580 conflicts arising from USEing package with conflicting symbols
581 correctly. (thanks to Michael Weber)
583 changes in sbcl-1.0.19 relative to 1.0.18:
584 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
585 warnings that go otherwise unhandled will be muffled if they are
586 of the type that's the value of this variable.
587 * optimization: stack allocation is slightly more efficient on x86
589 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
590 where a raw slot always is initialized using the initform whose
591 type is not know sufficiently well a compile-time are now compiled
592 correctly. (reported by John Morrison)
593 * bug fix: compiler no longer makes erronous assumptions in the
594 presence of non-foldable SATISFIES types.
595 * bug fix: stack analysis missed cleanups of dynamic-extent
596 arguments in non-let-converted calls to local functions.
597 * improvements to the Windows port:
598 ** adjusted address spaces for building on both Win32 and
599 Win64. (thanks for John Connors)
600 * fixed some bugs revealed by Paul Dietz' test suite:
601 ** interval arithmetic during type derivation used inexact integer
602 to single-float coercions.
603 ** arithmetic operations involving large integers and single
604 floats give the same results in compiled and interpreted code.
605 ** deriving the result type of COERCE no longer signals an error
606 if the derived type of the second argument is a MEMBER type
607 containing invalid type specifiers.
608 ** ADJOIN with constant NIL as second argument works correctly.
609 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
611 changes in sbcl-1.0.18 relative to 1.0.17:
612 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
613 profiles only the current thread.
614 * minor incompatible change: changes to SYMBOL-VALUE of constants
615 defined with DEFCONSTANT now signal an error.
616 * enhancement: SB-SPROF now has support for wallclock profiling,
617 and is also able to profile specific threads. REPORT output
618 has also additional sorting options.
619 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
621 * optimization: structure allocation has been improved
622 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
623 ** out of line constructors are ~10% faster.
624 ** inline constructors are ~15% faster.
625 ** inline constructors are capable of dynamic extent allocation
626 (generally on x86 and x86-64, in some cases on other platforms
628 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
630 * optimization: file compiler is now able to coalesce non-circular
631 lists, non-base strings, and bit-vectors. Additionally, constants
632 are never referenced through SYMBOL-VALUE at runtime.
633 * optimization: code defining methods on PRINT-OBJECT (and other
634 generic functions in the COMMON-LISP package) now loads faster.
635 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
637 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
638 fasl is now deleted. (reported by Attila Lendvai)
639 * bug fix: READ-LINE always returned NIL for the last line in files.
640 (reported by Yoshinori Tahara)
641 * bug fix: more accurate disassembly annotations of foreign function
642 calls. (thanks to Andy Hefner)
643 * bug fix: trimming non-simple strings and non-string string
644 designators when the there is nothing to trim works properly.
645 (thanks to James Knight)
646 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
647 (thanks to Travis Cross)
648 * fixed some bugs revealed by Paul Dietz' test suite:
649 ** NIL is a valid function name (regression at 1.0.13.38)
650 ** FILL on lists was missing its return value (regression at 1.0.12.27)
651 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
652 fill pointers properly (regression at 1.0.12.23)
653 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
654 displacement indices properly (regression at 1.0.12.23)
656 changes in sbcl-1.0.17 relative to 1.0.16:
657 * temporary regression: user code can no longer allocate closure
658 variable storage on stack, due to bug 419 without explicitly
659 requesting it. Please consult sbcl-devel for advice if you need to
660 use this feature in the meanwhile.
661 * new feature: runtime argument --control-stack-size can be used to
662 adjust thread default control stack size.
663 * enhancement: improved TIME output
664 ** all times are reported using the measured accuracy (milliseconds
665 for real and GC times, microseconds for everything else.)
666 ** processor cycle counts on x86 and x86-64.
667 ** interpreted forms are counted for both evaluator modes.
668 ** number of lambdas converted by the compiler is reported.
669 ** CPU percentage report (computed from real and total run time.)
670 ** more comprehensive run time reporting, using a condenced format
671 ** interperted form, lambda, and page fault counts are omitted
673 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
675 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
676 * optimization: two argument forms of LAST are upto ~10% faster
677 in normal SPEED policies.
678 * optimization: NCONC no longer needs to heap cons its &REST list
679 in normal SPEED policies.
680 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
682 * bug fix: UNION and NUNION work with :TEST-NOT once more,
683 regression since 1.0.9.1. (thanks to Eric Marsden)
684 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
685 since 1.0.15.36. (thanks to Paul Khuong)
686 * bug fix: LAST when always returned the whole list when given a bignum
687 as the second argument.
688 * bug fix: dynamic extent allocation of nested lists and vectors
689 could leak to otherwise accessible parts.
690 * bug fix: invalid optimization of heap-allocated alien variable
692 * bug fix: fasl header checking is less vulnerable to different
693 platform word lengths.
694 * bug fix: more correct assembler syntax for GNU binutils
695 2.18.50.0.4 support. (thanks to Marijn Schouten)
696 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
698 * bug fix: the fopcompiler can handle LOCALLY forms (with no
699 declarations) successfully. (reported by Attila Lendvai)
701 changes in sbcl-1.0.16 relative to 1.0.15:
702 * minor incompatible change: revert the changes to sb-posix's error
703 signaling added in 1.0.14.
704 * minor incompatible change: change PROBE-FILE back to returning
705 NIL whenever we can't get a truename, as was the case before 1.0.14.
706 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
707 inherits from ERROR instead of just CONDITION.
708 * new feature: SB-INTROSPECT can provide source locations for instances
709 as well. (thanks to Tobias Rittweiler)
710 * optimization: binding special variables now generates smaller code
711 on threaded platforms.
712 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
713 and cases where no :TEST is given but the compiler can infer that
714 the element to search is of type (OR FIXNUM (NOT NUMBER)).
715 * optimization: better LOGNOT on fixnums.
716 * optimization: modular arithmetic for a particular requested width
717 is implemented using a tagged representation unless a better
718 representation is available.
719 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
720 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
721 for wellformedness and returns a specialized vector. (reported by
722 Francois-Rene Rideau)
723 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
724 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
725 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
726 unscheduling timer at the same time as another timer fires.
727 * bug fix: don't reschedule timers for dead threads.
728 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
729 * bug fix: copying output from RUN-PROGRAM to a stream signalled
730 bogus errors if select() was interrupted.
731 * enhancement: add support for fcntl's struct flock to SB-POSIX.
733 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
734 * enhancement: cleaner backtraces for interactive interrupts, as
735 well as other cases where the interesting frames used to be
736 obscured by interrupt handling frames.
737 * enhancement: untracing a whole package using (UNTRACE "FOO") is
738 now supported, and tracing a whole package using (TRACE "FOO") now
739 traces SETF-functions as well.
740 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
741 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
742 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
743 * unreadably printed representation of hash-tables now includes
745 * bug fix: partially fixed #188: type propagation from assignments
746 is now more efficient.
747 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
748 'DOUBLE-FLOAT) are not flushable.
749 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
750 full address of the object, and none of the tag bits.
751 * bug fix: readably printing hash-tables now respects other printer
752 control variables. (reported by Cedric St-Jean)
753 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
754 keyword with MAKE-HASH-TABLE.
755 * bug fix: export SB-POSIX:MKSTEMP.
756 * bug fix: SORT was not interrupt safe.
757 * bug fix: XREF accounts for the last node of each basic-block as
759 * bug fix: MAKE-INSTANCE optimizations interacted badly with
760 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
761 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
762 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
763 create obsolete instances in the case of redefinition or
764 obsoletion of a superclass. (thanks to Andy Hefner)
765 * bug fix: Support for the Alpha architecture has been revived; it had
766 suffered somewhat from lack of maintenance since sbcl-1.0.
767 * improvements to the Windows port:
768 ** The system detects the codepage to use at startup, instead of
769 using the value from the saved core. (thanks to Kei Suzuki)
771 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
772 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
773 (see documentation for details.)
774 * revived support for OpenBSD (contributed by Josh Elsasser)
775 * partially fixed bug #108: ROOM no longer suffers from occasional
776 (AVER (SAP= CURRENT END)) failures.
777 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
778 bodies no longer cause a WARNING to be signalled. (reported by
780 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
781 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
782 the bug in TRUNCATE.) (reported by Michael Weber)
783 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
784 no samples. (reported by Andy Hefner)
785 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
786 no longer appear as (NIL ...) frames in backtraces.
787 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
788 is now actually exported from SB-EXT as documented. (reported by
790 * bug fix: sb-aclrepl now correctly understands how to inspect
791 single-floats on 64-bit platforms where single-floats are not boxed.
792 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
793 yet been finalized. (reported by Levente Meszaros)
794 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
795 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
796 SBCL-specific optimize qualities.
798 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
799 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
800 an executable in the search path, and does so in the child
801 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
802 been removed; it can be found in the CVS history, for somebody who
803 needs that search behavior (see the manual).
804 * minor incompatible change: compiler policy re. weakening type
805 checks has changed: now type checks are weakened only if SAFETY < 2
807 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
808 unparsing of directory pathnames as files. Analogously,
809 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
810 filename to parse into a directory pathname.
811 * enhancement: implicit generic function creation now signals a
812 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
813 users can bind handlers for and muffle around calls to LOAD.
814 * enhancement: RUN-PROGRAM allows unicode arguments and environments
815 to be used (using the default stream external format), and allows
816 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
817 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
818 strings and vectors whose element-type or simplicity is not fully
819 known at compile-time.
820 * optimization: STRING-TRIM and related functions no longer allocate
821 a new string when no trimming needs to be performed. These functions
822 are also faster than before when the input string has been declared
824 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
825 * optimization: READ-LINE is significantly faster for files containing
827 * optimization: non-open coded uses of character comparison operators
828 (e.g. char=) no longer cons when called with more than one parameter
829 on platforms supporting dynamic-extent allocation.
830 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
831 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
832 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
834 * bug fix: some sequence functions elided bounds checking when
836 * bug fix: too liberal weakening of union-type checks when SPEED >
838 * bug fix: more bogus fixnum declarations in ROOM implementation
841 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
842 * new feature: MAKE-HASH-TABLE now experimentally accepts a
843 :SYNCHRONIZED argument, which makes the hash-table safe for
844 concurrent accesses (but not iteration.) See also:
845 SB-EXT:WITH-LOCKED-HASH-TABLE, and
846 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
847 * optimization: CONCATENATE on strings is an order of magnitude faster
848 in code compiled with (> SPEED SPACE).
849 * optimization: SUBSEQ is ~50% faster on lists.
850 * bug fix: bug 417 fixed -- source location reporting is now more robust.
851 * bug fix: SUBSEQ on a list will now correctly signal an error if if
852 END is smaller then START.
853 * bug fix: SB-PROFILE will no longer report extra consing for nested
854 calls to profiled functions.
855 * bug fix: ROOM implementation had bogus fixnum declarations which
856 could cause type-errors when calling ROOM on large images.
857 * bug fix: if file compilation is aborted, the partial fasl is now
858 deleted, and COMPILE-FILE returns NIL as the primary value.
859 * bug fix: number of thread safety issues relating to SBCL's internal
860 hash-table usage have been fixed.
861 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
862 be returned from its body when the values were being returned
863 using unknown-values return convection and the W-P-O was wrapped
864 inside an UNWIND-PROTECT.
865 * bug fix: sb-posix should now compile again under Windows, enabling
868 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
869 * incompatible change: hash-table accessor functions are no longer
870 automatically protected by locks. Concurrent accesses on the same hash-table
871 from multiple threads can give inconsistent results or even corrupt the
872 hash-table completely. Multi-threaded applications should do their own
873 locking at the correct granularity. In the current implementation it is
874 still safe to have multiple readers access the same table, but it's not
875 guaranteed that this property will be maintained in future releases.
876 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
877 and will signal an error at runtime.
878 * enhancement: SB-THREAD package now exports a semaphore interface.
879 * enhancement: CONS can now stack-allocate on x86 and
880 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
882 * enhancement: nested lists can now be stack allocated on
883 platforms providing stack allocation support.
884 * enhancement: dynamic-extent support has been extended to support
885 cases where there are multiple possible sources for the stack
887 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
888 if the mutex is uncontested on Linux.
889 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
890 as specified by the CLHS. (thanks to Tobias Rittweiler)
891 * bug fix: NaN comparison now works on x86-64.
892 * bug fix: CLOSE :ABORT T on a stream with pending output now
894 * bug fix: instances of non-standard metaclasses using standard
895 instance structure protocol sometimes missed the slot type checks
897 * bug fix: known functions can be inlined in high-debug code.
899 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
900 * minor incompatible change: the MSI installer on Windows no longer
901 associates .lisp and .fasl files with the installed SBCL.
902 * minor incompatible change: :UNIX is no longer present in *FEATURES*
903 on Windows. (thanks to Luis Oliviera)
904 * new feature: SB-CLTL2 contrib module now implements
905 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
906 * optimization: scavenging weak pointers is now more efficient,
907 requiring O(1) instead of O(N) per weak pointer to identify
908 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
909 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
910 method bodies using constant slot-names when the first argument is
911 a specializer parameter for the method.
912 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
913 names, when the class of the instance is a direct instance
914 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
915 as slow as the constant slot-name case.)
916 * optimization: member type construction is now O(N) instead
918 * optimization: UNION and NUNION are now O(N+M) for large
919 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
921 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
922 now more readable in environments like Slime which display it.
923 (thanks to Tobias C. Rittweiler)
924 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
925 was able to infer the correct argument type for the object on which
926 the CAS operation was being performed.
927 * bug fix: copy propagation interfered with parallel assignment
928 semantics in local calls. (reported by Paul Khuong)
929 * bug fix: the signed modular fixnum shift compiled to wrong code on
930 x86 and x86-64. (spotted by a slight modification to some of
933 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
934 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
935 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
936 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
938 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
939 and generic function now print the object identity for anonymous
940 instances. (thanks to Attila Lendvai)
941 * enhancement: as an extension to MOP, SBCL now supports SETF of
942 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
943 (thanks to Attila Lendvai)
944 * enhancement: sb-cover has an alternate annotation mode which puts
945 the form annotations at the car of the form, instead of annotating
946 it completely. The benefit of this mode is that it shows better
947 which forms were instrumented by the compiler.
948 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
949 by printer control variables. (reported by Dan Corkill)
950 * bug fix: system leaked memory when delayed output was performed by
951 the OS in smaller chunks then expected. (thanks to David Smith)
952 * bug fix: system leaked memory when file streams were not closed
954 * bug fix: large objects written to slow streams that were modified
955 after the write could end up with the modified state written to
956 the underlying file descriptor.
957 * bug fix: multiple threads operating in parallel on the same stream
958 could cause buffer-overflows.
959 * bug fix: source location information is stored correctly
960 (broken since 1.0.6). This bug would generally show up as the
961 Slime debugger highlighting the wrong form when the sldb-show-source
963 * bug fix: Branch forms are again annotated as branches in the sb-cover
965 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
966 on x86-64. (reported by Gregory Vanuxem)
967 * bug fix: EQUALP could return wrong results for structures with raw
968 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
969 word). (reported by Vjacheslav Fyodorov)
970 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
973 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
974 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
975 atomic compare-and-swap operations on threaded platforms.
976 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
977 allows assigning a global minimum value to optimization qualities
978 (overriding proclamations and declarations).
979 * enhancement: closed over variables can be stack-allocated on x86
981 * performance bug fix: GETHASH and (SETF GETHASH) are once again
983 * optimization: slot definition lookup is now O(1). This speeds up
984 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
985 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
986 * optimization: ASSOC and MEMBER can now be open-coded for all
987 combinations of keyword arguments when second argument is constant
988 and SPEED >= SPACE. In other cases a specialized version is
990 * bug fix: using obsoleted structure instances with TYPEP and
991 generic functions now signals a sensible error.
992 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
993 (reported by Kristoffer Kvello)
994 * bug fix: backtrace construction is now more careful when making
995 lisp-objects from pointers on the stack, to avoid creating bogus
996 objects that can be seen by the GC.
997 * bug fix: defaulting of values in contexts expecting more than 7
998 variables now works on x86-64. (reported by Christopher Laux)
999 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
1001 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
1002 as the property-list of a symbol.
1003 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
1004 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
1005 situations requested, are once again file-compileable. (reported
1008 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
1009 * MOP improvement: support for user-defined subclasses of
1010 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
1011 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
1012 the expansion of DEFMETHOD, is responsible for generating a form
1013 which creates a list of specializers when evaluated. Additional
1014 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
1015 debugging and introspective support.
1016 * minor incompatible change: the (unsupported) spinlock interface
1017 has changed: free spinlock now has the value NIL, and a held spinlock
1018 has the owning thread as its value.
1019 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
1020 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
1022 * enhancement: name of a socket-stream is now "a socket" instead of
1023 "a constant string".
1024 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
1025 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
1026 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
1027 for generic functions. (Thanks to Tobias C. Rittweiler)
1028 * optimization: bignum printing speed has been improved by 20-40%
1029 (depending on the bignum size.)
1030 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
1032 * bug fix: the cache used by the CLOS to store precomputed effective
1033 methods, slot offsets, and constant return values is now thread and
1035 * bug fix: generic function dispatch function updating is now thread
1036 and interrupt safe (in the sense that the known issues have been
1038 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
1039 * bug fix: interning EQL-specializers is now thread and interrupt safe.
1040 * bug fix: asdf systems with dependencies to the SB-POSIX or
1041 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
1042 * bug fix: interrupt safety of applicable method computation has been
1045 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
1046 * new contrib: sb-cover, an experimental code coverage tool, is included
1047 as a contrib module.
1048 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
1049 significantly faster.
1050 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1051 produce more efficient code on x86-64 if the compiler can determine
1052 that (+ INDEX <constant>) does not require a bounds check and FOO
1053 has an element type at least 8 bits wide.
1054 * enhancement: a new, experimental synchronous timeout facility is
1055 provided. Refer to SB-SYS:WITH-DEADLINE for details.
1056 * enhancement: when a symbol name conflict error arises, the
1057 conflicting symbols are always printed with a package prefix.
1058 (thanks to Kevin Reid)
1059 * enhancement: stepping is now once again supported on the SPARC and
1060 MIPS platforms. (It is also now more likely to work on CheneyGC
1062 * enhancement: sb-sprof can now also track and report accurate call
1064 * bug fixes: the treatment of non-standard subclasses of
1065 SB-MOP:SPECIALIZER is more correct.
1066 * incompatible change: PURIFY no longer copies the data from the
1067 dynamic space into the static and read-only spaces on platforms
1068 that use the generational garbage collector
1069 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
1071 * bug fix: GC race condition occasionally resulting in crashes with
1072 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
1074 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
1076 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
1077 a STYLE-WARNING to be signalled (regression from 1.0.4.)
1078 * bug fix: an asynchronous interrupt could previously leave the
1079 system running with GC inhibited.
1080 * bug fix: a DECLARE form evaluated at top-level now causes an error
1081 rather than silently (or verbosely) returning NIL.
1082 * bug fix: trying to dissassemble functions compiled at high DEBUG could
1083 result in a error being signalled due to source form lookup errors.
1084 (reported by Peter Graves)
1086 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
1087 * incompatible change: removed writer methods for host-ent-name,
1088 host-ent-addresses -- changing the values did not update the DNS
1090 * minor incompatible change: changed experimental JOIN-THREAD interface
1091 * documentation: the manual now lists reader and writer methods
1092 in class slot documentation sections. (thanks to Richard M Kreuter)
1093 * documentation: unwinding from asyncronous events has been
1094 documented as unsafe.
1095 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1096 in multithreaded application code.
1097 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1098 platforms. (thanks to James Anderson for the optimization hint)
1099 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1101 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1102 variants no longer cons.
1103 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1104 their NOT- variants no longer cons.
1105 * optimization: EQUAL hash tables no longer use SXHASH for objects
1106 of all data types, but instead use an EQL hash for types for which
1107 EQUAL is the same as EQL.
1108 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1109 are significantly faster.
1110 * optimization: new STRING-OUTPUT-STREAM implementation is much
1111 faster for large outputs and conses 30% less on average.
1112 * enhancement: XREF information is now collected for references made
1113 to global variables using SYMBOL-VALUE and a constant argument.
1114 * enhancement: SIGINT now causes a specific condition
1115 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1116 * bug fix: FILE-POSITION used to signal an error in some cases where
1117 ANSI requires it to return NIL.
1118 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1119 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1120 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1122 * bug fix: bad type declaration in the CLOS implementation has
1123 been fixed. (reported by James Anderson)
1124 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1125 (thanks to Lutz Euler)
1126 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1127 required by the CLHS. (reported by Samium Gromoff)
1128 * bug fix: dead unbound variable references now signal an error.
1129 (reported by Marco Monteiro)
1130 * bug fix: / with an unused value was being deleted in safe code.
1131 (thanks to Marco Monteiro and Kevin Reid)
1132 * bug fix: number of characters that can be written onto a single
1133 line in a file is unlimited.
1134 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1135 been fixed by inhibiting interrupts for when GC is disabled.
1136 * bug fix: some interrupt safety issues with GETHASH, (SETF
1137 GETHASH), CLRHASH and REMHASH have been fixed.
1138 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1139 specifier no longer causes infinite recursion.
1140 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1141 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1142 * bug fix: an &environment argument with the correct variable information
1143 is passed to macros that are expanded during byte compilation.
1144 (reported by Samium Gromoff)
1145 * bug fix: a logic error was causing the hash function for bignums to
1146 have excessive amounts of collisions. (reported by Faré Rideau)
1147 * bug fix: modifying the contents of an array could change the return
1148 value of SXHASH on that array, which is only allowed for strings
1149 and bit vectors. (bug introduced in 0.9.16)
1150 * improvement: the x86-64/darwin port now passes all tests (except
1151 for the debugger tests) but should still be considered
1152 experimental until this is fixed.
1153 * improvement: a style-warning is signaled for CASE (etc) clauses with
1154 duplicate keys. (thanks to Kevin Reid)
1155 * improvement: macos/x86 and macos/x86-64 now use mach exception
1156 handlers for dealing with illegal instructions (for trapping and
1157 error handling) and memory protection violations (for GC).
1159 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1160 * new platform: experimental support for x86-64/darwin (MacOS).
1161 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1162 and getnameinfo sockets functions are used instead of gethostbyaddr
1163 and gethostbyname, on platforms where the newer functions are available.
1164 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1166 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1167 don't print the startup banner, but behave as if --noinform was passed
1168 as a command line argument. (thanks to Kevin Reid)
1169 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1170 * optimization: code using alien values with undeclared types is much faster.
1171 * optimization: the compiler is now able to open code SEARCH in more cases.
1172 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1173 * bug fix: using standardized COMMON-LISP special variables as loop
1174 variables no longer signals bogus package lock violations. (reported
1176 * bug fix: declaring local loop variables to be of a range-limited type
1177 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1178 (reported by Andras Simon)
1179 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1180 bugs remain on x86-64.)
1181 * bug fix: the #= and ## reader macros now interact reasonably with
1182 funcallable instances.
1183 * bug fix: type-checks for function arguments were compiled using the
1184 compiler policy settings of the wrong lexical environment (in 1.0.2
1186 * bug fix: SHADOW accepts characters as string designators, as required
1187 by the spec (thanks to Eric Marsden)
1188 * bug fix: fixed GC safety issues when foreign functions are called with
1189 non-base strings as arguments
1190 * bug fix: more consistent error output for fatal-compiler-warnings, like
1192 * bug fix: functions with proper names were showing up as NIL in
1193 backtraces (reported by Edi Weitz, regression in 1.0.2)
1195 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1196 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1197 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1198 produce more efficient code on the x86 if the compiler can determine
1199 that (+ INDEX <constant>) does not require a bounds check and FOO
1200 has an element type at least 8 bits wide.
1201 * bug fix: references to undefined variables are handled the same way
1202 in toplevel forms as in the normal compiler
1203 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1205 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1206 (thanks to Jon Buller)
1207 * improvement: faster compilation times for complex functions
1208 * improvement: added readlink support to SB-POSIX (thanks to Richard
1211 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1212 * improvement: experimental support for mach exception handling on
1213 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1215 * improvement: support for GBK external format.
1216 (thanks to Chun Tian (binghe))
1217 * improvement: the debugger now displays variables that have been closed
1218 over, in code compiled with (DEBUG 2) or higher
1219 * improvement: support for executable cores on NetBSD (thanks to
1221 * new feature: added a RESTART-FRAME debugger command
1222 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1223 be used to provide an implementation for FILE-POSITION on Gray streams
1224 (thanks to Eric Marsden)
1225 * new feature: add syslog, openlog and closelog support to SB-POSIX
1226 (thanks to Richard Kreuter)
1227 * optimization: the function call overhead in code compiled with
1228 a high DEBUG optimization setting is significantly
1229 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1230 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1231 (thanks to Tony Martinez)
1232 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1233 evaluated once (reported by Kevin Reid)
1234 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1235 works again (thanks to Kilian Sprotte)
1236 * bug fix: an error is signaled for tagbodies with duplicate tags
1237 (thanks to Stephen Wilson)
1238 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1239 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1240 stack frames from alien callbacks.
1241 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1242 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1243 65k pages (thanks to David Woodhouse)
1244 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1246 changes in sbcl-1.0.1 relative to sbcl-1.0:
1247 * new platform: FreeBSD/x86-64, including support for threading.
1248 * new feature: the compiler stores cross-referencing information
1249 about function calls (who-calls), macroexpansion (who-macroexpands)
1250 and special variables (who-binds, who-sets, who-references) for code
1251 compiled with (< SPACE 3). This information is available through the
1252 sb-introspect contrib.
1253 * new feature: users may subclass SEQUENCE, and have instances of
1254 these classes interoperate with standard Common Lisp functions if
1255 a number of methods are defined. (This feature is experimental
1256 and the interface subject to change based on feedback from SBCL
1257 users and the general community)
1258 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1259 x86/x86-64, rather than the previous fixed depth of 8
1260 * improvement: another pthread back-end of mutex "pthread-futex".
1261 * bug fix: non-ascii command-line arguments are processed correctly
1262 (thanks to Yaroslav Kavenchuk)
1263 * bug fix: non-required arguments were not passed correctly when a method
1264 defined using DEFMETHOD was called from a mop-generated method using
1265 CALL-NEXT-METHOD (reported by Pascal Costanza)
1266 * bug fix: recursion is now permitted in accessors through
1267 SLOT-UNBOUND. (reported by Pascal Costanza)
1268 * bug fix: an error was signaled at startup if the HOME environment
1269 variable was defined, but had an empty value (reported by Peter Van Eynde)
1270 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1271 signal a PROGRAM-ERROR, not a BUG.
1272 * bug fix: SB-POSIX:READDIR works when built with large file support.
1273 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1275 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1276 for better automatic indentation support. (thanks to Matt Pillsbury)
1277 * optimization: loading generic functions no longer takes O(n^2) time,
1278 proportional to the amount of methods in the generic function
1279 (reported by Todd Sabin and Jeremy Brown)
1280 * optimization: the FIND and POSITION family of sequence functions
1281 are significantly faster on arrays whose element types have been
1283 * improvements to the Windows port:
1284 ** Intermittent heap corruption problems have been fixed. (thanks
1285 to Alastair Bridgewater)
1286 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1288 ** Lisp is able to unwind foreign exception frames from alien
1289 callbacks. (thanks to Alastair Bridgewater)
1291 changes in sbcl-1.0 relative to sbcl-0.9.18:
1292 * improvement: experimental support for threading on FreeBSD/x86.
1293 (thanks to NIIMI Satoshi)
1294 * improvement: runtime option --dynamic-space-size can be used
1295 to set the size of the dynamic space reserved on startup.
1296 * improvement: floating point modes in effect are now saved in
1297 core, and restored on startup.
1298 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1299 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1300 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1301 threads are not running after *SAVE-HOOKS* have run.
1302 * improvement: writes to CLOS instance slots are type-checked in code
1303 compiled with (SAFETY 3)
1304 * improvement: floating-point exception handling on FreeBSD (thanks to
1306 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1307 (thanks to Zach Beane)
1308 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1310 * improvement: added support for the Shift-JIS external format.
1311 (contributed by NIIMI Satoshi)
1312 * improvement: callbacks are supported on Linux/PPC. (thanks to
1314 * bug fix: compiler bug triggered by a (non-standard) VALUES
1315 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1316 * bug fix: file compiler no longer confuses validated and already
1317 dumped structurres. (reported by Kaersten Poeck)
1318 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1319 fill-pointer signals a type-error as required. (thanks to
1321 * bug fix: disassemly of funcallable instances works.
1322 * bug fix: single stepping on PPC.
1323 * bug fix: fix thread-safety problems in the type system (generally
1324 manifesting as nonsensical errors like "STRING is a bad type specifier
1325 for sequences" or "The value 1 is not of type FIXNUM")
1326 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1327 (reported by Josip Gracin)
1328 * bug fix: an error is signaled for attempts to displace arrays with
1329 incompatible element types (thanks to Mario Mommer)
1330 * bug fix: more correct handling of wide characters in debug info
1331 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1332 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1333 and don't cause extra consing
1334 * optimization: MAP and MAP-INTO are significantly faster on vectors
1335 whose elements types have been declared.
1336 * Improvements to SB-SPROF:
1337 ** Support for allocation profiling
1338 ** Reduced profiling overhead, especially for long profiling runs
1339 * Improvements to the Windows port:
1340 ** floating point exceptions are now reported correctly.
1341 ** stack exhaustion detection works partially.
1342 ** more accurate GET-INTERNAL-REAL-TIME.
1343 ** better RUN-PROGRAM behaviour in case of errors in the child
1345 ** PROBE-FILE now simplifies pathnames correctly.
1346 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1347 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1348 1st 1970 (although time timezone data will be unreliable).
1349 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1351 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1352 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1353 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1355 * bug fix: two potential GC deadlocks affecting threaded builds.
1356 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1357 returns T (reported by Anton Kazennikov)
1358 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1359 with non-variable places
1360 * bug fix: remove a race condition in the setting of
1361 funcallable-instance functions, this should make threaded CLOS
1362 code more stable against memory faults.
1363 * bug fix: corruption of specials when unbinding is interrupted by an
1364 asynchronous unwind (reported by Hannu Koivisto)
1365 * improvement: the debugger will now also display local variables that
1366 are only used once, for code compiled with a DEBUG optimization quality
1369 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1370 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1371 * incompatible change: External-format support for FFI calls. The
1372 SB-ALIEN:C-STRING no longer implies an ASCII
1373 external-format. Instead the string is subject to external-format
1374 conversion. Additionally return values of type SB-ALIEN:C-STRING
1375 are by default of type (SIMPLE-ARRAY CHARACTER), not
1376 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1377 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1378 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1379 * incompatible change: SB-EXT package no longer contains the
1380 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1381 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1383 * incompatible change: the single-stepper is no longer available
1384 on Alpha, Mips and Sparc platforms.
1385 * minor incompatible change: the direct superclasses of
1386 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1387 not (STANDARD-OBJECT FUNCTION). This makes the
1388 class-precedence-lists of GENERIC-FUNCTION and
1389 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1391 * minor incompatible change: the default stream external format on
1392 non-unicode SBCL is no longer detected from the locale, but is always
1394 * new feature: Add a version of evaluator that uses an interpreter instead
1395 of the compiler. EVAL still uses the compiler by default, to switch it
1396 to use the interpreter, set the value of the variable
1397 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1398 * minor incompatible change: the single-stepper REPL has been merged
1399 with the normal debugger (see the "Stepping" heading of the debugger help
1400 for more details). The debugger command STEP will no longer switch
1401 to the single-stepper REPL.
1402 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1403 for a type now works.
1404 * bug fix: loading of default sysinit file works. (thanks to Leonid
1406 * bug fix: better detection of circularities in the file-compiler.
1407 (reported by Marco Monteiro)
1408 * bug fix: the CL pathname functions now work with files that have
1409 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1410 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1411 systems with Unix98 pty semantics.
1412 * bug fix: ASDF-INSTALL will now work with bsd tar.
1413 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1415 * bug fix: timers expiring in dead threads no longer cause a
1416 type-error (reported by Paul "Nonny Mouse").
1417 * bug fix: thanks to more lightweight single-stepper instrumentation,
1418 code compiled with (DEBUG 3) will compile and execute significantly faster,
1419 and will have more accurate type-inferencing than before
1420 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1421 whose bindings are modified
1422 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1423 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1424 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1425 CL:SOFTWARE-VERSION with proper version
1427 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1428 * feature: implemented the READER-METHOD-CLASS and
1429 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1430 as specified by AMOP.
1431 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1433 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1434 (thanks to Lutz Euler)
1435 * optimization: hashing of general arrays and vectors has been
1436 improved. (reported by Any Fingerhut)
1437 * enhancement: SB-INTROSPECT is now able to find definitions of
1438 profiled functions. (thanks to Troels Henriksen)
1439 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1440 as well, allowing compiler-macros for SETF-functions to expand.
1441 * enhancement: step-instrumentation no longer wraps known
1442 single-value functions with multiple-value context, allowing
1443 better type inference.
1444 * fixed bug #337: use of MAKE-METHOD in method combination now works
1445 even in the presence of user-defined method classes. (reported by
1446 Bruno Haible and Pascal Costanza)
1447 * fixed bug #339(c): if there are applicable methods not part of any
1448 long-form method-combination group, call INVALID-METHOD-ERROR.
1449 (reported by Bruno Haible)
1450 * fixed bug #361: the :FUNCTION initarg in the protocol for
1451 initialization of methods can now be used to override
1452 internally-produced optimized functions. (reported by Bruno
1454 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1455 system-provided lambda expression no longer cause warnings about
1456 unbound #:|pv-table| symbols.
1457 * bug fix: improved the handling of type declarations and the
1458 detection of violations for keyword arguments with non-constant
1460 * bug fix: potentially erronous calls to PATHNAME and
1461 MERGE-PATHNAMES were being flushed in some cases.
1462 (reported by Richard Kreuter)
1463 * bug fix: compiled calls to TYPEP were mishandling obsolete
1464 instances. (reported by James Bielman and Attila Lendvai)
1465 * bug fix: format strings with ~> without matching ~< no longer
1466 trigger an AVER, but signal an understandable error instead.
1467 (reported by Antonio Martinez)
1468 * bug fix: specifying an output-file without a pathname-type for
1469 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1470 of input-file instead of "fasl". (reported by Robert Dodier)
1471 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1472 (reported by James Y Knight).
1473 * bug fix: compiler-macros-function did not consider the environment
1474 argument for shadowing by local functions.
1475 * bug fix: compiler-macros expansion was inhibited by local INLINE
1477 * bug fix: inline expansions of known functions were subject to
1478 step-instrumentation in high debug policies, leading to problems
1479 with type-inference.
1480 * bug fix: compiler failed to differentiate between different CONS
1481 types in some cases.
1482 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1483 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1484 * bug fix: #368: incorrect use of expressed vs. upgraded array
1486 * bug fix: #306a: more precise unions of array types.
1487 * thread-safety improvements:
1488 ** CONDITION-WAIT could return early on Linux, if the thread was
1489 interrupted and subsequently continued with SIGCONT.
1490 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1492 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1493 * added support for the ucs-2 external format. (contributed by Ivan
1495 * minor incompatible change: pretty printing of objects of type
1496 (cons symbol) is, in the default pprint-dispatch-table, now
1497 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1499 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1500 called later in a class's lifetime, possibly as late as when the
1501 first instance of the class is created. Previously,
1502 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1503 class became finalizeable.
1504 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1505 for multibyte external-format streams. (thanks to "vbzoli")
1506 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1507 with :ALLOCATION :CLASS inherited from superclasses of the
1509 * fixed bug: anonymous classes can now be created using the :NAME
1510 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1511 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1512 * fixed bug: core-files saved with :EXECUTABLE T can again be
1513 executed when SBCL_HOME isn't set. (reported by James Knight)
1514 * fixed bug: toplevel LOCALLY forms with declarations could
1515 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1516 * fixed bug: printing from several different threads using different
1517 values of *print-case* could cause invalid output, due to
1518 some internal special variables of the printer not being bound
1519 thread-locally (reported by Max Mikhanosha)
1520 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1521 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1522 (reported by Pascal Costanza)
1523 * fixed bug: FINALIZE-INHERITANCE is called from
1524 REINITIALIZE-INSTANCE on classes when the class has previously
1525 been finalized, as required by AMOP.
1526 * minor code generation optimizations:
1527 ** better register allocation in CLOS dispatching functions
1528 ** overflow detection when coercing signed bytes to fixnums on x86-64
1529 ** is now implemented with one IMUL instruction instead of three shifts
1530 ** more efficient bit-vector access on x86 and x86-64
1531 ** more efficient access to raw structure slots on x86 and x86-64
1532 * fixed some bugs revealed by Paul Dietz' test suite:
1533 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1534 return its argument.
1536 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1537 * feature: thread support on Solaris/x86, and experimental thread support
1539 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1541 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1542 default method for SLOT-UNBOUND.
1543 * minor incompatible change: prevent the user from specializing the
1544 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1545 somewhat counter to the intent of the protocol, I (CSR) think, and
1546 additionally it just doesn't work in SBCL as currently
1547 implemented, thanks to optimizations (that are always valid for
1548 the other three SLOT-VALUEish functions, but not for the setter).
1549 * minor incompatible change: the :SB-LDB feature is now enabled by
1550 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1551 the low-level debugger.
1552 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1553 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1554 down to an SB-SYS:FD-STREAM, allowing direct communication in
1555 more cases. Temporary files and pipes are still used as a fallback
1557 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1558 on the non-reentrant C functions localtime(3) and gmtime(3).
1559 * bug fix: global optimization policy was not visible in LOCALLY and
1561 * bug fix: class objects can be used as specializers in methods.
1562 (reported by Pascal Costanza)
1563 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1564 with a type error. (reported by blitz_ on #lisp)
1565 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1566 failed with a type error. (reported by Pascal Bourguignon)
1567 * bug fix: merging pathnames against defaults with :DIRECTORY
1568 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1569 (reported by James Y Knight)
1570 * bug fix: saving large (>2GB) cores on x86-64 now works
1571 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1572 constant index and a (simple-array (signed-byte 32)) array
1573 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1574 error (patch by Robert J. Macomber)
1575 * feature: TIME now displays the amount of run-time spent in GC
1576 * bug fix: The debugger now does a better job of respecting
1577 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1578 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1579 workaround for bug 403.)
1580 * bug fix: floating point exception handling now works on FreeBSD
1581 (thanks to NIIMI Satoshi)
1582 * fixed some bugs revealed by Paul Dietz' test suite:
1583 ** MISC.641: LET-conversion were not supposed to work in late
1585 * improvements to the Win32/x86 port:
1586 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1587 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1589 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1590 * new feature: source path information is generated for macro-expansion
1591 errors for use in IDE's like Slime (thanks to Helmut Eller)
1592 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1593 * bug fix: compiler does not loop forever on an invalid type in
1595 * improvement: compilation of most CLOS applications is significantly
1597 * optimization: added a limited bytecode compiler for simple toplevel
1598 forms, speeding up compilation and FASL loading
1599 * bug fix: the statistical profiler now properly distinguishes anonymous
1602 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1603 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1604 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1605 into subclasses of the finalized class. Now user calls to
1606 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1607 system return before any subclasses are finalized.
1608 * minor incompatible change: The reader no longer ignores errors
1609 regarding non-existent packages in #+ and #- feature tests.
1610 * new feature: command line options --no-sysinit, --no-userinit to
1611 inhibit loading the corresponding init files
1612 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1613 for sending data through UDP sockets (thanks to François-René Rideau)
1614 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1615 error is signalled instead (thanks to François-René Rideau)
1616 * minor incompatible change: Error signalling behaviour of lexical
1617 operations violating package locks has changed slightly. Refer to
1618 documentation on package locks for details.
1619 * enhancement: EVAL can process IF-expressions without resorting to the
1621 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1622 constant-folded properly due to missing out-of-line functions.
1623 (reported by Mika Pihlajamäki)
1624 * bug fix: LISTEN sometimes returned T even in cases where no data was
1625 immediately available from the stream
1626 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1627 were reversed. (reported by Levente Mészáros)
1628 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1629 when given a vector displaced to an adjustable array. (reported
1631 * bug fix: derivation of float boundaries from numbers outside the
1632 appropriate float range (reported by John Wiseman)
1633 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1635 * bug fix: precomputing cache entries for generic functions with
1636 some subclasses of specializers as yet invalid does not attempt to
1637 fill a cache line with a negative offset. (reported by Levente
1639 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1640 allow condition class objects as arguments to DOCUMENTATION and
1641 (SETF DOCUMENTATION); only find and set documentation for
1642 structure names for the STRUCTURE doc-type. (suggested by Gary
1644 * improvements to the Win32/x86 port:
1645 ** (user-homedir-pathname) and default initialization file
1646 locations now know about the user's "Documents and Settings"
1647 directory (thanks to Yaroslav Kavenchuk)
1648 ** run-program is implemented (thanks to Mike Thomas)
1649 ** sockets support (thanks to Timothy Ritchey)
1650 ** better backtrace support (thanks to Alastair Bridgewater)
1651 ** sb-grovel supported
1652 ** asdf-install and sb-posix work somewhat
1653 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1654 * improvements to the Solaris/x86 port:
1655 ** works on Solaris 11/Solaris Express
1656 ** floating-point exception handling support
1657 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1658 * fixed some bugs revealed by Paul Dietz' test suite:
1659 ** REMOVE-METHOD returns its generic function argument even when
1660 the method is not one of the generic functions' methods.
1661 ** objects declared as MEMBER types can be admissible arguments to
1662 structure accessors.
1663 ** printing characters should simply be printed by the FORMAT ~:C
1665 ** compiler failure when compiling functions with hairy constant
1666 defaults for optional parameters.
1667 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1668 function, which is already optimized.
1670 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1671 * new platform: experimental support for SBCL x86/Darwin, including
1672 MacOS X 10.4.5 on Intel.
1673 * new feature: Unicode character names are now known to the system
1674 (through CHAR-NAME and NAME-CHAR).
1675 * new feature: the filesystem location of SBCL's core file is
1676 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1677 * minor incompatible change: the contrib modules SB-POSIX and
1678 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1679 this change is to make it easier to distribute
1680 location-independent binaries.
1681 * bug fix: as implied by AMOP, standardized classes no longer have
1682 slots named by external symbols of public packages. (reported by
1684 * optimization: calling functions via constant symbols -- as in
1685 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1686 via the function object as in (FUNCALL #'FOO).
1687 * optimization: on x86 and x86-64, the calling convention for the
1688 case of dealing with an unknown number of values has been altered
1689 to be friendlier to the prediction heuristics implemented,
1690 particularly with reference to CALL and RET pairing. (thanks to
1691 Alastair Bridgewater)
1692 * optimization: on x86, the code for access to thread-local symbol
1693 values and binding/unbinding of thread-local symbols is smaller.
1694 (thanks to Alastair Bridgewater)
1695 * enhancement: CONSTANTP is now able to determine constantness of
1696 more complex forms, including calls to constant-foldable standardized
1697 functions and some special forms beyond QUOTE.
1698 * fixed bug: occasional GC crashes on Solaris/x86
1699 * optimization: x86-64 supports stack allocation of results of simple
1700 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1701 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1702 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1703 (thanks to James Knight)
1704 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1705 by the standard (thanks to Stephen Compall)
1707 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1708 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1709 be used for bundling the runtime and the core file into one
1710 executable binary. This feature is not currently supported on all SBCL
1711 platforms. (thanks to James Bielman and NIIMI Satoshi)
1712 * new feature: a generational or ephemeral garbage collector is now
1713 the default on the PowerPC platform (both Linux and Darwin). The
1714 old Cheney (stop and copy) collector is a build-time option.
1715 (thanks to Cyrus Harmon, after Raymond Toy)
1716 * minor incompatible change: the method by which SBCL finds its
1717 contributed modules has changed; it no longer relies on symbolic
1718 links from an $SBCL_HOME/systems directory, but searches directly
1719 in the subdirectories of $SBCL_HOME.
1720 * enhancement: the dynamic heap size on the Linux/PPC platform is
1721 markedly larger, even using the older Cheney garbage collector.
1722 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1723 certain complicated string types. (reported by Gary King)
1724 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1725 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1726 * fixed bug: DOCUMENTATION on structures no longer signals an
1727 error if no structure type of the right name exists. (reported by
1729 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1730 applicable non-standard primary or :AROUND method on
1731 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1732 no longer cause unbound variable errors. (reported by Kevin Reid)
1733 * fixed bug: metacircle resolution in cases where methods have slots
1734 added before the slots from STANDARD-METHOD. (reported by Jean
1736 * fixed bug: the Power PC assembly code for calling into lisp
1737 sometimes computed its return address wrongly (depending on the
1738 15th bit of the address, fixed at link time). This used to
1739 manifest itself in a segmentation violation while building PCL.
1740 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1741 many others over the years)
1742 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1743 the stack pointer (treated as a signed integer) is negative.
1744 (thanks to Peter van Eynde)
1746 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1747 * new platform: experimental support for the Windows operating
1748 system has been added. (thanks to Alastair Bridgewater)
1749 * fixed several bugs in and robustified the PPC FFI (including
1750 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1751 * bug fix: allow non-simple string symbol names (reported by Paul
1753 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1755 * bug fix: saving a core corrupted callbacks on x86/x86-64
1756 * bug fix: closed a loophole in metacircularity detection and
1757 grounding in the PCL implementation of CLOS.
1758 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1759 (reported by tomppa on #lisp)
1760 * bug fix: clear the direction flag on Lisp -> C transitions, as
1761 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1762 (reported by Andrej Grozin and Hendrik Maryns)
1763 * optimization: major improvements to GC efficiency on GENCGC platforms
1764 * optimization: faster implementation of EQUAL
1765 * optimization: emit more efficient opcodes for some common
1766 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1767 * optimization: several other minor code-generation improvements on x86-64
1768 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1770 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1771 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1772 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1773 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1774 by ANSI as a generic function, and for consistency (SETF
1775 GENERIC-FUNCTION-NAME) is treated likewise.
1776 * fixed bug #233.b: make constraint propagation notice when a variable
1777 value is changed after it is referenced but before it is used
1778 * fixed bug #296: no more arbitrary behaviour from filenames with
1779 odd characters as --load arguments. (reported by Adam Warner)
1780 * fixed bug #390: :CHARACTER-SET pathname components now work as
1781 expected. (reported by Tim Daly Jr)
1782 * fixed bug #391: complicated :TYPE intersections in slot
1783 definitions no longer cause an error in PCL internals.
1784 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1785 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1786 returning the number of octets which would be written to the
1787 file-stream. (thanks to Robert J. Macomber)
1788 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1789 arguments to the internal stream functions. (thanks to David
1791 * bug fix: stack exhaustion is now handled on the x86/SunOS
1792 platform. (thanks to Daisuke Homma)
1793 * bug fix: more accurate ROOM results on GENCGC platforms
1794 * optimization: improved type inference for arithmetic-for
1795 index variables in LOOP
1796 * optimization: faster floating-point SQRT on x86-64
1797 * fixed some bugs revealed by Paul Dietz' test suite:
1798 ** DOCUMENTATION returns NIL instead of "" for method combinations
1799 that don't have a docstring
1801 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1802 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1803 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1804 therefore are not customizeable by user code (as seems to be at
1805 least permitted and maybe required by AMOP). As a consolation,
1806 however, the SBCL implementation of these functions now calls
1807 REINITIALIZE-INSTANCE as specified by AMOP.
1808 * bug fix: slot-definition documentation is propagated as per ANSI
1809 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1810 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1811 argument, as do eslotd initialization methods. (from Pascal
1812 Costanza's "Closer" project)
1813 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1814 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1816 * bug fix: it is now possible to have more than one subclass of
1817 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1818 (reported by Bruno Haible, Pascal Costanza and others)
1819 * bug fix: the dependent update protocol now works for generic
1820 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1821 and Pascal Costanza)
1822 * bug fix: condition-class instances corresponding to
1823 DEFINE-CONDITION forms are now created eagerly. (reported by
1824 Kalle Olavi Niemitalo on comp.lang.lisp)
1825 * bug fix: floating point printing is more accurate in some
1826 circumstances. (thanks to Simon Alexander)
1827 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1828 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1829 * bug fix: callbacks on OS X now preserve stack-alignment.
1830 (thanks to Cyrus Harmon)
1831 * enhancement: the x86-64 disassembler is much better at
1832 disassembling SSE instructions. (thanks to Lutz Euler)
1833 * enhancement: CHECK-TYPE on a variable now allows type-inference
1834 (thanks to Vincent Arkesteijn)
1835 * optimization: improved performance of EUC-JP external format.
1836 (thanks to NIIMI Satoshi)
1837 * optimization: performance improvements to IO on file streams of
1838 :ELEMENT-TYPE CHARACTER
1839 * optimization: much faster memory allocation on x86-64
1840 * optimization: faster garbage collections (latency and throughput)
1842 * optimization: faster bignum subtraction and division on x86 and x86-64
1843 (thanks to Lutz Euler)
1844 * optimization: more accurate type-inference for loops that use a
1845 floating point index variable or a negative step.
1847 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1848 * bug fix: add a workaround to SBCL looping infinitely at startup on
1849 Linux kernels with apparently buggy implementations of personality().
1850 (thanks to Svein Ove Aas)
1851 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1852 (thanks to David Lichteblau)
1853 * bug fix: local bindings shadow global symbol macros for
1854 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1855 * optimization: non-open coded uses of numeric comparison operators
1856 (e.g. >) no longer cons when called with more than one parameter
1857 on platforms supporting dynamic-extent allocation.
1858 * enhancement: saving cores with foreign code loaded is now supported
1859 on MIPS/Linux in addition to the previously supported platforms.
1860 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1861 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1862 * bug fix: division by zero in sb-sprof when no samples were collected
1863 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1864 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1865 (thanks to Alexander Kjeldaas)
1866 * new feature: ignoring the return values of destructive functions
1867 like NREVERSE signals a compile-time style-warning.
1868 (thanks to Kevin Reid)
1869 * new feature: the alignment of alien structure fields can be
1870 explicitly specified. (thanks to Cyrus Harmon)
1871 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1872 and dump core on SIGQUIT
1874 ** incompatible change: threads do not inherit values of specials
1875 from their parents (see manual)
1876 ** bug fix: threads stacks belonging to dead threads are freed by the
1877 next exiting thread, no need to gc to collect thread stacks anymore
1878 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1879 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1880 ** bug fix: dynamic variable and thread start related gc lossage
1881 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1883 * fixed some bugs revealed by Paul Dietz' test suite:
1884 ** SUBTYPEP is slightly more accurate on heinously complicated
1885 CONS types where some of the members have uncertain (in the
1886 NIL, NIL sense) type relationships to each other.
1888 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1889 no longer purified unless :PURIFY T is explicitly specified.
1890 ** Non-purified cores are significantly smaller than before
1892 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1893 * new feature: timers based on Zach Beane's excellent timer package
1894 * added support for the following external formats: koi8-u,
1895 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1896 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1897 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1898 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1899 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1900 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1901 * incompatible change: a threaded SBCL will no longer revert to
1902 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1903 * bug fix: interrupts are disabled until startup is complete; no
1904 more sigsegvs when receiving a signal to soon
1905 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1907 * bug fix: add a workaround for the memory randomization features in
1908 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1909 workaround will only be in effect on systems with the proc filesystem
1911 * bug fix: printing objects of type HASH-TABLE signals a
1912 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1914 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1915 have been running for over 50 days. (reported by Gilbert Baumann)
1916 * bug fix: the logic for getting names of functions gets less
1917 confused when confronded with alternate-metaclass
1918 funcallable-instances. (reported by Cyrus Harmon)
1919 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1920 consistent, even on internal alternate-metaclass objects.
1921 * bug fix: flush closure information collected by physical
1922 environment analysis prepass before the main pass. (bug reported
1924 * bug fix: compiler pack phase does not modify a hashtable, which is
1925 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1926 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1927 may be expanded. (reported by Alexander <malishev> on c.l.l)
1928 * bug fix: no more style-warnings when compiling code using LAST
1929 under high-SPEED low-SPACE settings. (reported by David Wallin
1931 * fixed some incompatibilities between SBCL's MOP and the MOP
1933 ** the METAOBJECT class is now implemented;
1934 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1935 STANDARD-OBJECT, as required;
1936 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1937 now compatible (as required by VALIDATE-SUPERCLASS); there
1938 remains a consistency requirement over the presence or absence
1939 of the FUNCTION class in the superclass list of finalized
1940 classes; see the manual for more details;
1941 ** the :AROUND method for COMPUTE-SLOTS on
1942 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1943 requested slot ordering.
1945 ** bug fix: parent thread now can be gc'ed even with a live
1947 ** bug fix: binding a special with PROGV to no value is not
1948 the same as that symbol not having been bound (thanks to
1950 ** bug fix: binding specials is thread safe (thanks to
1952 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1953 and SB-SPROF do not die with 'no handler for signal XX in
1954 interrupt_handle_now(..)' anymore
1955 ** bug fix: WITH-TIMEOUT works with multiple running threads
1956 * fixed some bugs revealed by Paul Dietz' test suite:
1957 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1958 the :method-class keyword argument.
1960 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1961 * new port: the Solaris operating system on x86 processors is now
1962 mostly supported, though some rough edges in the environment
1963 remain. (thanks to Daisuke Homma)
1964 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1965 space for its heap. (thanks to Thiemo Seufer)
1966 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1967 closures. (thanks to Thiemo Seufer)
1968 * minor incompatible change: eof selects abort in the debugger.
1969 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1970 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1972 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1973 deprecated, and will go away in a future revision of SBCL.
1974 * minor incompatible change: GC-ON and GC-OFF are no longer
1975 implemented with a counter, it does not matter how many times gc
1976 is switched on or off
1977 * bug fix: discriminating functions for generic function classes
1978 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1979 make invalid assumptions about method precedence order. (reported
1981 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1982 platforms. (thanks to Thiemo Seufer)
1983 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1984 * bug fix: correct pathname printing with printer escaping is on.
1985 (thanks to Kevin Reid)
1986 * bug fix: complex VOP definitions in "user-space" no longer trigger
1987 package locks. (reported by Zach Beane)
1988 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1989 for funcallable-instance functions, and is no different from
1991 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1992 characters in the current readtable. (reported by Nicholas Neuss)
1993 * bug fix: SUBTYPEP on various CONS types returns more a more
1994 accurate acknowledgment of its certainty.
1995 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1996 lists in some cases. This partially fixes bug 384.
1997 * flush all standard streams before prompting in the REPL and the
1999 * bug fix: signal handling and triggering gc do not conflict
2000 directly anymore, in particular a high frequency sb-sprof does
2001 not prevent gc from running
2002 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
2003 approximation for timezone and DST information between the
2004 universal time epoch and the smallest negative 32-bit time_t.
2005 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
2006 year 1899 so long as the encoded time is non-negative.
2007 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
2008 the correct answer for non-constant rotate amounts when performing
2009 an inline 32-bit rotation.
2011 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
2012 there is only one thread in the session
2013 ** bug fix: memory leak for streams created in one thread and
2014 written to in another
2015 ** bug fix: lockup when compiled with gcc4
2016 ** bug fix: race that allows the gc to be triggered when gc is
2018 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
2019 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
2021 ** bug fix: finished threads are gc'ed properly
2022 * fixed some bugs revealed by Paul Dietz' test suite:
2023 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
2024 has arguments with a different set of applicable methods from
2025 the orignal arguments.
2026 ** The type error thrown by MAP now has a correct expected-type
2028 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
2029 name a compiled function.
2030 ** Three MISC tests where a large bignum was improperly coerced to
2031 a float that couldn't represent that bignum during type
2032 derivation were fixed.
2033 ** SUBTYPEP can now handle types involving the negation of a
2034 list-form FUNCTION type.
2035 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
2036 as (COMPLEX (AND RATIO (NOT FIXNUM))).
2037 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
2039 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
2040 * New feature: Experimental support for bivalent streams: streams
2041 opened with :element-type :default now allow character and binary
2042 (unsigned-byte 8) I/O
2043 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
2044 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
2046 * bug fix: no more highly sporadic "couldn't check whether ~S is
2047 readable" when reading a stream and an interrupt hits in the middle
2048 of a select system call
2049 * compiler better recognizes complex arrays (reported by Eduardo
2051 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
2052 for variables on SBCL built with linkage-tables. (reported by Luis
2054 * various error reporting improvements.
2055 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
2056 (Thanks to Thiemo Seufer)
2057 * bug fix: nested reader invokations work correctly
2058 * bug fix: it is possible to have simultaneous references to foreign
2059 code and foreign data with the same name.
2061 ** added x86-64 support
2062 ** incompatible change: the threading api now works with thread
2063 objects instead of thread ids
2064 ** bug fix: threads are protected from signals and interruption when
2065 starting up or going down
2066 ** bug fix: a race where an exiting thread could lose its stack to gc
2067 ** bug fix: don't halt on infinite error in threads if possible
2068 ** fixed numerous gc deadlocks introduced in the pthread merge
2069 ** bug fix: fixed thread safety issues in read and print
2070 ** bug fix: debugger doesn't hang on session lock if interrupted at
2071 an inappropriate moment
2072 ** bug fix: run-program is now thread safe(r)
2073 ** bug fix: inner with-recursive-lock no longer releases the mutex
2074 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
2075 * fixed some bugs revealed by Paul Dietz' test suite:
2076 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
2077 EXPECTED-TYPE slots filled.
2078 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
2080 ** MAKE-CONDITION accepts classes as type-designators.
2081 ** COMPILE may never return NIL.
2082 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
2083 range before calling Unix time functions
2085 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
2086 * numerous signal handling fixes to increase stability
2087 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2088 * minor incompatible change: we now correctly canonize default
2089 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2090 as per the MOP, rather than the historical (INITARG INITFUNCTION
2091 INITFORM). (reported by Bruno Haible)
2092 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2093 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2094 for more information.
2095 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2096 pathname is a directory pathname.
2097 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2098 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2100 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2101 :METACLASS argument in addition to a class name. (reported by
2102 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2103 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2104 (reported by Bruno Haible; patch from Gerd Moellmann)
2105 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2107 * bug fix: more cleanups to the floating point exception handling on
2108 x86-64 (thanks to James Knight)
2109 * bug fix: the compiler does not try to constant fold calls of
2110 COERCE and BIT-* functions when they return freshly constructed
2111 objects. (reported by Nikodemus Siivola and Paul Dietz)
2112 * optimization: improved the MIPS versions of generic (in the
2113 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2114 * optimization: direct conversion of (unsigned-byte 32) to floats on
2115 the PowerPC platform.
2116 * optimization: structure instances with raw slots now use less
2117 memory, and probably show better memory locality. (thanks to
2119 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2120 are treated as requests for stack allocation on the x86-64,
2121 Alpha-32, PPC and SPARC platforms.
2122 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2123 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2124 (thanks to Tony Martinez)
2126 ** gcing a dead thread can no longer lead to lockups
2127 ** threads block signals until they are set up properly
2128 ** errno is no longer shared by threads
2129 ** interrupt-thread restores the eflags register on x86
2130 ** fixed some lockups due to gc/thread interaction
2131 * fixed some bugs revealed by Paul Dietz' test suite:
2132 ** invalid dotted lists no longer raise a read error when
2133 *READ-SUPPRESS* is T
2134 ** signal an error if a symbol that names a declaration is used
2135 as the name of a type, or vice versa
2136 ** allow using the (declare (typespec var*)) abbreviation for
2137 (declare (type typespec var*)) with all type specifiers
2138 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2139 ** always return NIL from PPRINT-POP when OBJECT is NIL
2140 ** don't signal errors when pretty-printing malformed LABELS,
2141 FLET or MACROLET forms
2142 ** declarations in a DOLIST body are also in scope for the
2144 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2146 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2149 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2150 * fixed cross-compiler leakages that prevented building a 32-bit
2151 target with a 64-bit host compiler.
2152 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2153 opened with :IF-EXISTS :OVERWRITE.
2154 * fixed bug 281: error for an invalid qualifier in a short-form method
2155 combination method is not signalled until the faulty method is called.
2156 * bug fix: iteration variable type inferrer failed to deal with open
2157 intervals. (reported by Alan Shields)
2158 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2159 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2160 or more runtime options were provided to the sbcl binary.
2161 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2162 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2163 * optimization: there's now a fast-path for fixnum arguments in the
2164 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2165 * optimization: the code generated on x86-64 is more compact thanks
2166 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2167 * bug fix: floating-point exception handling now partly works on
2168 x86-64. (Thanks to James Knight)
2169 * improvement to the MIPS backend from Thiemo Seufer: C-style
2170 64-bit long long arguments and return values to alien functions
2172 * fixed some bugs revealed by Paul Dietz' test suite:
2173 ** the type-error signalled from WARN has a filled-in DATUM slot.
2174 ** the type-error required when a stream is not associated with
2175 a file has the stream as its datum.
2176 ** type-errors on single-floats on x86-64 no longer have
2177 :INVALID-OBJECT as the datum
2178 ** the type-errors signalled for invalid function names now have
2179 a correct expected type
2180 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2181 for structures defined with a :TYPE. Documentation strings for
2182 typed structures are no longer immediately discarded
2183 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2184 broadcast streams and synonym streams. FILE-LENGTH now also works
2185 on broadcast streams.
2187 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2188 * incompatible change: the --noprogrammer option, deprecated since
2189 version 0.7.5, has been removed. Please use the equivalent
2190 --disable-debugger option instead.
2191 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2193 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2194 inoperational for a while now) has been completely removed.
2195 * null lexical environments are now printed as #<NULL-LEXENV>,
2196 significantly reducing the amount of clutter in typical backtraces.
2197 * documentation on weak pointers, finalization, and after GC hooks
2198 has been added to the manual.
2199 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2200 as other specialized array types, is much faster. SUBSEQ and
2201 COPY-SEQ on such arrays have also been sped up.
2202 * optimization: EQL is now more efficient when at least other argument
2203 is known to be of type (OR FIXNUM (NOT NUMBER)).
2204 * fixed bug: the runtime is now less vulnerable to changes in the
2205 size of the SBCL object on OS X, and virtual memory is reserved for
2206 all spaces that need to be at a fixed address.
2207 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2208 * fixed bug: finalizers and after GC hooks that cause consing are now
2210 * fixed bug: compiler error messages and summaries are now printed to
2211 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2212 * fixed inference of the upper bound of an iteration variable.
2213 (reported by Rajat Datta).
2214 * fixed bug 211e: calling local functions with duplicated constant
2215 keyword argument no longer causes a bogus style warning about an
2217 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2218 ftype declaration to be disregarded. (reported by Dave Roberts)
2219 * fixed bug 373: caused by erronous compilation of references to alien
2220 variables in the runtime on ppc/darwin.
2221 * fixed bug 376: CONJUGATE type deriver.
2222 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2223 (reported by Baughn on #lisp)
2224 * a cross-compiler bug on non-x86 platforms has been identified and
2225 fixed. (thanks to Bruno Haible)
2226 * improvements to the MIPS runtime code for increased stability.
2227 (thanks to Thiemo Seufer)
2228 * increased the maximimum compact environment size to allow
2229 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2230 * partial workaround for bug 135: don't name the function we're
2231 calling for hairy cases of EVAL, so as not to accumulate one environment
2232 entry per call to EVAL. (reported by Kevin Reid)
2233 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2234 * optimization: type testing for non-vector arrays should be faster.
2235 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2237 * bug fix: setting 31st element of a bit vector to zero did not work
2239 * bug fix: redefining a class definition which failed due to a
2240 previous accessor / function clash now works (but see BUGS entry
2241 #380 for more problems in this area). (thanks to Zach Beane)
2242 * the long form of DEFINE-METHOD-COMBINATION disables method group
2243 checking when given a single method group with pattern *. (thanks
2245 * on x86 compiler supports stack allocation of results of simple
2246 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2247 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2248 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2249 * fixed some bugs related to Unicode integration:
2250 ** the restarts for recovering from input and output encoding
2251 errors only appear when there is in fact such an error to
2253 * fixed some bugs revealed by Paul Dietz' test suite:
2254 ** MISC.549 and similar: late transformation of unsafe type
2255 assertions into derived types caused unexpected code
2257 ** SCALE-FLOAT type deriver is less wrong.
2258 ** type derivers for EXP, LOG and similar functions compute result
2259 types for complex arguments better.
2260 ** (MISC.563) CONJUGATE type deriver works for very restricted
2262 ** out-of-line type testers for character strings are available.
2263 ** EQUAL compiler transform understands specialness of objects
2265 ** accessing double-floats stored on the stack now works on x86-64.
2266 ** debugger internals could sometimes create invalid lispobjs,
2267 resulting in GC crashes.
2268 ** MISC.548: type check weakening can convert required type into
2270 ** initialization forms for bindings are not in scope of free special
2273 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2274 * incompatible change: thread support for non-NPTL systems has
2275 been removed - locking is buggy and unreliable. A threaded
2276 SBCL build will now warn at startup and refuse to create
2277 new threads, unless futex support is detected in the kernel
2278 * incompatible change: the top level REPL now has only an ABORT
2279 restart associated with it, not TOPLEVEL and ABORT as it used to.
2280 TOP and TOPLEVEL are now available as debugger commands for
2281 returning to the top level.
2282 * incompatible change: forms evaluated in the REPL now use the
2283 global optimization policy.
2284 * incompatible change: user- and system-initialization files are
2285 no longer processed with LOAD, but by READ and EVAL; hence the
2286 global optimization policy, startup package, readtable, etc,
2288 * The .fasl file format number has been incremented because of
2289 various incompatible changes.
2290 * internal entry point details and argument counts no longer appear
2291 in backtraces unless explicitly requested by setting
2292 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2293 * built-in and standard functions no longer have names like "top
2294 level local call to FOO".
2295 * fixed bug 32: functions defined in non-null lexical environments
2296 now have more legible printed representation
2297 * fixed bug 33: functions defined in non-null lexical environemnts
2298 are now more amenable to inspection by INSPECT.
2299 * workaround for bug 354: XEPs no longer appear in backtraces unless
2300 explicitly requested.
2301 * fixed bug: receiving the signal which results from attempting to
2302 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2303 SIGBUS on other *BSDs) on architectures where the C stack is also
2304 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2305 to debugger confusion or wild instability. (reported by Cheuksan
2307 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2308 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2309 (reported by Lutz Euler)
2310 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2311 compiled code. (reported by Paul Dietz)
2312 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2313 specialized parameter to a method to something that is not TYPEP
2314 the specializer is now possible.
2315 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2316 face of package deletion.
2317 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2318 pathname. (reported by David Tolpin on c.l.l)
2319 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2320 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2321 STRING-STREAM and FILE-STREAM, respectively.
2322 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2323 than previously; in addition, its test suite is now run on build.
2324 (thanks to Luke Gorrie)
2325 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2326 * optimization: added a immediate representation for single-floats
2328 * fixed some bugs revealed by Paul Dietz' test suite:
2329 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2330 ** attempting to create a package with a colliding nickname causes
2331 correctable errors to be signalled.
2332 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2333 ** several x86-64 backend bugs related to sign-extension of immediate
2336 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2337 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2338 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2340 * fixed loading of multiply forward-referenced layouts.
2341 (thanks to Cheuksan Wang)
2342 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2343 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2344 (see COMPILE-FILE documentation for details of :PRINT option)
2345 * fixed bugs 19 and 317: fixed-format floating point printing is
2346 more accurate. This also fixes a bug reported by Adam Warner
2347 related to the ~@F format directive.
2348 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2350 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2351 dispatch table if the from-char is a dispatch macro character.
2352 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2353 by the wordsize no longer ignore the last word. (reported by Lutz
2355 * fixed bug in type checking of dynamic-extent variables. (reported
2357 * optimization: sequence traversal functions use their freedom to
2358 coerce function designators to functions.
2359 * optimization: code with many calls to CLOS methods specialized on
2360 CLOS classes has had redundant type checks removed.
2361 * optimization: added declarations to speed up operations that access
2362 the internal character database (for example STRING-UPCASE)
2363 * optimization: comparison operations between floats and sufficiently small
2364 fixnums no longer create extra rationals
2365 * fixed some bugs related to Unicode integration:
2366 ** portions of multibyte characters at the end of buffers for
2367 character-based file input are correctly transferred to the
2368 start of the buffer at the next read.
2369 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2370 passing it through to OPEN.
2371 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2372 argument given to internal calls to OPEN.
2373 ** the built-in comment readers (introduced by character sequences
2374 ";" and "#|") are more forgiving to encoding errors; they will
2375 STYLE-WARN and then attempt to resync the stream at a character
2376 boundary. (thanks to Teemu Kalvas)
2377 * fixed some bugs in the x86-64 port:
2378 ** Negative short int return values from c-calls are sign-extended
2380 ** The stack is aligned to 16-bytes for c-calls, as required by
2381 the ABI. (thanks to Cheuksan Wang)
2382 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2383 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2384 for enums. (thanks to Vincent Arkesteijn)
2385 ** Multiple small optimizations and bugfixes for floating point
2387 * fixed some bugs revealed by Paul Dietz' test suite:
2388 ** Space, Tab, Linefeed, Return and Page have the invalid
2389 secondary constituent character trait.
2390 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2392 ** WITH-INPUT-FROM-STRING should only update the index place on
2394 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2395 works more reliably.
2396 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2397 with one-dimensional arrays with fill pointers.
2398 ** TYPE-OF failed on a complex with an integer realpart and a
2400 ** compiler failure during type inference for the code of form
2401 (IF (EQL X (THE ...)) ...) (MISC.535).
2403 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2404 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2405 platform. The port supports 61-bit fixnums, large memory spaces
2406 and reloading shared object files.
2407 * enhancement: saving cores with foreign code loaded is now
2408 supported on ppc/Darwin in addition to the previously supported
2410 * enhancement: the statistical profiler now walks deeper into the
2411 call stack for more meaningful call-graphs and accrued time
2412 reports (x86/x86-64 only). It also now reports time spent in
2414 * enhancement: it is now possible to trace most individual methods
2415 of a generic function in addition to tracing the generic function
2417 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2419 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2420 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2421 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2422 produces an error. (thanks to Vincent Arkesteijn)
2423 * bug fix: NAMESTRING on pathnames with :WILD components in their
2424 directories works correctly. (thanks to Artem V. Andreev)
2425 * fixed bug 125: compiler preserves identity of closures. (reported
2427 * bug fixed: functions with &REST arguments sometimes failed with
2428 "Undefined function" when compiled with (DEBUG 3). (reported by
2430 * bug fix: overflow during compiling of setting element of a bit
2431 vector with constant index and value. (reported by Timmy Douglas)
2432 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2433 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2434 * on x86 compiler partially supports stack allocation of dynamic-extent
2436 * GO and RETURN-FROM do not check the extent of their exit points
2437 when compiled with SAFETY 0.
2438 * fixed some bugs related to Unicode integration:
2439 ** encoding and decoding errors are now much more robustly
2440 handled; it should now be possible to recover even from invalid
2441 input or output to the terminal. (thanks to Teemu Kalvas)
2442 ** provided a first cut at implementing STRING-TO-OCTETS and
2443 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2444 ** altered the SB-MD5 contributed module slightly, changing the
2445 interface just enough for it to be supportable for builds where
2446 lisp characters are not eight bits.
2447 * fixed some bugs revealed by Paul Dietz' test suite:
2448 ** the FORMATTER-generated functions for ~V[ conditionals require
2449 the correct number of arguments.
2450 ** READ-FROM-STRING returns the mandated second value when applied
2451 to displaced strings.
2452 ** the #\Rubout and #\Backspace characters are treated as invalid
2453 constituent characters by the tokenizer.
2455 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2456 * new feature: reloading changed shared object files with
2457 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2458 * new feature: references to foreign variables and functions
2459 can now be compiled and loaded before the corresponding shared
2460 object file is loaded, as long as the foreign definitions are
2461 available at runtime.
2462 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2463 supported. (thanks to Dan Debertin)
2464 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2465 just keywords. (thanks to Vincent Arkesteijn)
2466 * fixed bug #331: structure-class instances corresponding to
2467 DEFSTRUCT forms are now created eagerly.
2468 * fixed bug #345: backtraces from calls to undefined functions work
2469 on x86 as well. Related bug #61 is now also partially fixed on x86
2470 (backtraces from throws to unknown catch tags.)
2471 * bug fix: lambda-list parsing is now stricter vrt. order and number
2472 of lambda-list keywords.
2473 * bug fix: as specified by AMOP, an error is signalled if a
2474 class-option appears multiple times in a DEFCLASS form. (reported
2476 * bug fix: RANDOM can be compiled when the compiler derives the type
2477 of its numeric argument as a disjoint set of small integers.
2478 (reported by Paul Dietz)
2479 * bug fix: starting a core saved with shared objects loaded when
2480 those objects are not available no longer causes threaded SBCL to
2481 hang. (reported by Sean Ross)
2482 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2484 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2485 argument, which must always be NIL. (reported by Kalle Niemitalo)
2486 * bug fix: printing 1.0d+23 no longer results in an error.
2487 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2488 * bug fix: structure slot setters preserve evaluation order. (thanks
2490 * fixed some bugs related to Unicode integration:
2491 ** RUN-PROGRAM can allow its child to take input from a Lisp
2492 stream. (reported by Stefan Scholl)
2493 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2494 locales. (reported by Ken Causey)
2495 * fixed some bugs revealed by Paul Dietz' test suite:
2496 ** INCF, DECF and REMF evaluate their place form as specified in
2498 ** FORMATTER expands ~{ ~} iteration directives with V or #
2499 parameters correctly.
2500 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2501 consequent uses no arguments correctly.
2502 ** the system has a partial understanding of the (COMPLEX RATIO)
2504 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2507 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2508 * new feature: a build-time option (controlled by the :SB-UNICODE
2509 keyword feature, enabled by default) for building the system with
2510 support for the entire 21-bit character space defined by the
2512 * new feature: the system now has rudimentary external-format
2513 support; the primary user-visible change at this time is that
2514 characters with the high bit set (such as lower-case-e-acute) will
2515 print correctly to a terminal in a UTF-8 environment.
2516 * minor incompatible change: BASE-CHAR no longer names a class;
2517 however, CHARACTER continues to do so, as required by ANSI.
2518 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2519 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2520 has been moved to the SB-EXT package (temporarily re-exported from
2522 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2523 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2525 * fixed bug #351: better error-handlind and reporting for malformed
2527 * fixed bug #350: bignum-printing is now more memory-efficient,
2528 allowing printing of very large bignums, eg. (expt 2 10000000).
2529 (reported by Bruno Haible)
2530 * fixed bug #302: better primitive-type selection for intersection
2532 * fixed bug #308: non-graphic characters now all have names, as
2533 required. (reported by Bruno Haible)
2534 * bug fix: redefining a class with different superclasses now correctly
2535 removes it from the direct-subclasses of its previous superclasses.
2536 (reported by David Morse)
2537 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2538 new value now works. (reported by Bruno Haible)
2539 * bug fix: correct canonicalization of multiple non-standard slot
2540 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2541 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2542 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2543 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2544 now exists, an signals an error.
2545 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2546 during macroexpansion for non-iteration variables as well. (reported
2547 by Bruno Haible for CMUCL)
2548 * bug fix: Cyclic structures and unprintable objects in compiler
2549 messages no longer cause errors. (reported by Bruno Haible)
2550 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2551 and READ-FROM-STRING all now return a primary value of NIL if
2552 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2553 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2554 (reported by Bruno Haible for CMUCL)
2555 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2556 do not designate a condition. (reported by Bruno Haible for
2558 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2559 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2560 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2561 permitting resolution in favour of any of the conflicting symbols.
2562 (reported by Bruno Haible for CMUCL)
2563 * bug fix: EQUAL compiler optimizations is less aggressive on
2564 strings which can potentially compare true despite having distinct
2565 specialized array element types.
2566 * bug fix: unit enumerations can be defined without dividing by
2567 zero. (thanks to Vincent Arkesteijn)
2568 * FORMAT compile-time argument count checking has been enhanced.
2569 (report from Bruno Haible for CMUCL)
2570 * a partial workaround for the bug 262: the compiler does not try to
2571 inline-expand a local function doing RETURN-FROM from a deleted
2572 BLOCK. (thanks to Peter Denno for the bug report and to David
2573 Wragg for the simple test case)
2574 * fixed some bugs revealed by Paul Dietz' test suite:
2575 ** RENAME-PACKAGE allows all package designators as new package
2577 ** constraint propagation and lambda variable substitution are
2578 more cautious in dealing with partially deleted code.
2579 ** compiler sometimes lost reoptimization passes.
2580 ** CERROR, given a condition as condition designator, passes its
2581 remaining arguments to the continue format control without
2583 ** Case-altering FORMAT directives work correctly on non-ASCII
2585 ** The REAL type specifier handles bounds outside the floating
2586 point ranges without signalling FLOATING-POINT-OVERFLOW.
2587 ** Functions with IR1-transformations can create intercomponent
2588 references to global functions.
2589 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2591 ** FORMAT ~R treats a nil value for its first parameter correctly.
2593 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2594 * enhancement: saving cores with foreign code loaded is now
2595 supported on x86/NetBSD and sparc/Linux in addition to the previously
2596 supported platforms.
2597 * bug fix: on some platforms repeated installations caused multiple
2598 copies of HTML documentation to be installed -- should not happen
2599 any more. (reported by Stefan Scholl)
2600 * bug fix: parsing self-recursive alien record types multiple times
2601 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2602 original patch by Helmut Eller for CMUCL)
2603 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2604 (thanks to Richard Kreuter)
2605 * bug fix: defining classes whose accessors are methods on existing
2606 generic functions in other (locked) packages no longer signals
2607 bogus package lock violations. (reported by François-René Rideau)
2608 * bug fix: special variables as DEFMETHOD parameters no longer have
2609 associated bogus type declarations. (reported by David Wragg and
2611 * bug fix: read-write consistency on streams of element-type
2612 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2613 * bug fix: redefiniton of the only method of a generic function with
2614 no DEFGENERIC no longer emits a full WARNING. In addition,
2615 redefinition of generic functions with no DEFGENERIC to an
2616 incompatible lambda list now signals an error. (thanks to Zach
2618 * bug fix: DEFGENERIC now works even when there's a function of the
2619 same name in an enclosing lexical environment. (thanks to Zach
2621 * fixed compiler failure, caused by instrumenting code during
2622 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2623 * optimization: added loop analysis and improved register allocation
2624 * fixed some bugs revealed by Paul Dietz' test suite:
2625 ** POSITION on displaced vectors with non-zero displacement
2626 returns the right answer.
2627 ** (SIMPLE-STRING) is a valid type specifier for sequence
2629 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2631 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2632 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2634 ** The justification version of the FORMAT ~< directive treats
2635 non-zero minpad parameter correctly.
2637 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2638 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2639 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2640 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2641 the supported interface.
2642 * new feature: Single-stepping of code compiled with DEBUG 2 or
2643 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2644 * new feature: saving cores with foreign code loaded is now
2645 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2646 Timothy Moore's work for CMUCL)
2647 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2648 parameters to * instead of NIL if no initform is supplied.
2649 (reported by Johan Bockgaard)
2650 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2651 correctly when FUNCALL appears as the car of the form. Note:
2652 despite this FUNCALL forms are not currently subject to
2653 compiler-macro expansion. (port of Raymond Toy's fix for the
2654 same from CMUCL, reported by Johan Bockgaard)
2655 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2656 (thanks for Teemu Kalvas)
2657 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2658 case instead of terminating the iteration (thanks for Julian
2659 Squires, Sean Champ and Raymond Toy)
2660 * bug fix: incorrect expansion of defgeneric that caused a style
2661 warning. (thanks for Zach Beane)
2662 * bug fix: slot accessor effective method computation works properly
2663 for classes with multiple non-standard applicable methods on
2664 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2665 * on x86 compiler supports stack allocation of results of LIST and
2666 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2667 CMUCL implementation by Gerd Moellmann)
2668 * fixed some bugs revealed by Paul Dietz' test suite:
2669 ** FORMAT strings with both the ~<~:;~> form of the justification
2670 directive and pretty-printing directives cause an error.
2672 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2673 * incompatible change: the internal functions
2674 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2675 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2676 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2677 instead of the old functions.
2678 * new feature: on platforms where "dladdr" is available foreign
2679 function names now appear in backtraces. (based on Helmut Eller's
2681 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2682 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2684 * documentation: ASDF manual is now distributed with SBCL.
2685 * bug fix: SBCL can now load its contributed modules with REQUIRE
2686 even if the system-provided entries have been removed from
2687 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2689 * bug fix: backtraces involving undefined functions or assembly
2690 routines are more informative. (thanks to Brian Downing)
2691 * bug fix: mutually referent alien structures now work correctly.
2692 (reported by Rick Taube)
2693 * bug fix: structures defined by WITH-ALIEN can be referred to
2694 within other definitions in the same WITH-ALIEN.
2695 * bug fix: division operators (MOD, TRUNCATE and the like) with
2696 constant zero divisors and integer dividends no longer generate
2698 * bug fix: provide default methods for INPUT-STREAM-P and
2699 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2700 * bug fix: improve the detection and resolution of MOP metacycles
2701 (where effective method computation depends on the generic function
2702 having its effective method computed). (reported by Bruno Haible)
2703 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2704 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2705 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2706 * bug fix: pathnames with relative directory components can be
2707 represented relative to default pathnames.
2708 * optimization: in taking the GCD of bignums, reduce the two bignums
2709 to approximately the same size (using Euclid's algorithm) before
2710 applying the more sophisticated binary GCD. (thanks to Juho
2712 * optimization: COUNT on bitvectors now operates word-at-a-time.
2713 * optimization: ASH with a positive, but not necessarily constant,
2714 (leftwards) shift, when in a modular context, is compiled to a
2716 * fixed some bugs revealed by Paul Dietz' test suite:
2717 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2718 if the corresponding argument is NIL.
2719 ** FORMAT directives accepting a minpad parameter treat negative
2721 ** literal commas as character parameters to FORMAT directives are
2723 ** literal spaces directly after ~<Newline> directives within a
2724 format-logical-block (~:< ~@:>) do not induce :FILL-style
2725 conditional newlines.
2726 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2727 as specified: it no longer includes conditional newlines).
2728 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2730 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2731 :PER-LINE-PREFIX argument does not evaluate to a string.
2733 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2734 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2735 the manual for details; for now, package locks can be disabled by
2736 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2737 if no major problems are found then it is likely that they will be
2738 compiled in unconditionally.
2739 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2740 unsupported operators on all platforms. To load a shared library
2741 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2742 object file, link it into a shared library outside of SBCL and
2743 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2745 * bug workaround: ROOM T and the GC were not working together
2746 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2747 reporting facility aren't preserved. That reporting has been
2748 disabled, so now until and unless someone figures out how to make it
2749 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2750 * minor incompatible change: as threatened around sbcl-0.8.0, the
2751 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2752 The symbols are also exported from SB-PCL for backwards
2753 compatibility, but more so than before SB-PCL should be treated as
2754 an implementation-internal package.
2755 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2757 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2758 It is known as of this release not to work on the Alpha, however.
2759 * fixed bug #167: errors signalled due to illegal syntax in method
2760 bodies are now more legible.
2761 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2762 designators and can hence be used with TYPEP.
2763 * fixed bug #333: CHECK-TYPE now ensures that the type error
2764 signalled, if any, has the right object to be accessed by
2765 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2766 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2767 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2769 * fixed bug #334: programmatic addition of slots using specialized
2770 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2771 and :ALLOCATION :CLASS slots.
2772 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2773 float exponents. (rereported by Peter Seibel)
2774 * fixed a bug: #\Space (and other whitespace characters) are no
2775 longer considered to be macro characters in standard syntax by
2776 GET-MACRO-CHARACTER.
2777 * fixed bug: initialization of condition class metaobjects no longer
2778 causes an instance of the condition to be created. (reported by
2780 * fixed bug: it is now possible to have slots such that
2781 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2782 description is neither :INSTANCE nor :CLASS.
2783 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2784 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2785 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2786 non-standard SLOT-DEFINITION-ALLOCATION values.
2787 * fixed bugs: various race conditions handling exiting threads.
2788 CL-PPCRE's thread torture test now passes.
2789 * fixed arguably-a-bug: GC time proportional to number of threads in
2790 system even when most of them are idle
2791 * optimization: improved performance of BIT and SBIT on bit-vectors.
2792 * .fasl file incompatibility: The fasl file version number has
2793 been incremented because of changes associated with package locks.
2795 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2796 * minor incompatible change: the system no longer provides
2797 optimization hints (conditions which are TYPEP
2798 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2799 mode; these hints are still emitted when higher SPEED optimization
2801 * new contrib module: a sampling profiler (profiling by statistical
2802 sampling, rather than by instrumenting functions) is available as
2803 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2804 * the behaviour of the standard function ED is now customizeable by
2805 third parties through a hook variable: see ED's documentation
2806 string for information on the protocol.
2807 * the compiler no longer emits efficiency notes for (FUNCALL X)
2808 when the type of X is uncertain under default optimization
2810 * fixed bug 276: mutating a binding of a specialized parameter to a
2811 method to something that is not TYPEP the specializer is now
2813 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2814 exhibit the required behaviour on the x86 platform. (thanks to
2815 Peter van Eynde, Eric Marsden and Bruno Haible)
2816 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2817 even for difficult arguments. (reported by Peter Graves)
2818 * fixed bug 141a: the backquote printer now descends quoted
2820 * fixed another bug in backquote printing: no more destructive
2821 modification of the form's list structure. (reported by Brian
2823 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2824 move between its address being taken and the call to
2825 interrupt_thread, fixing a crashing race condition.
2826 * the SB-POSIX contrib implementation has been adjusted so that it
2827 no longer exhibits ridiculously poor performance when constructing
2828 instances corresponding to C structs.
2830 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2831 * minor incompatible change: the sb-grovel contrib now treats C
2832 structures as alien (in the sense of SB-ALIEN) objects rather than
2833 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2834 has implications for memory management of client code
2835 (sb-grovel-returned objects must now be manually managed) and for
2836 type safety (alien objects now have full types).
2837 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2838 used to control emission of compiler diagnostics, rather than the
2839 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2840 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2841 quality should be considered deprecated.
2842 * install.sh now installs the user manual as well
2843 * (not quite a new documentable feature, but worth considering in
2844 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2845 beginnings of a semantically meaningful condition hierarchy is
2846 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2848 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2849 designator as the defaults argument.
2850 * fixed bug: Displaced arrays whose displaced-to array has become
2851 too small now cause ARRAY-DIMENSION to signal an error, providing
2852 for safer bounds-checking. (reported by Bruno Haible)
2853 * fixed bug: DEFCLASS slot definitions with identical :READER and
2854 :WRITER names now signal a reasonable error. (reported by Thomas
2856 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2858 * fixed bug: Invalid slot specification errors now print correctly.
2859 (thanks to Zach Beane)
2860 * fixed bug 320: Shared to local slot value transfers in class
2861 redefinitions now happen corrently from superclasses as
2862 well. (reported by Bruno Haible)
2863 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2865 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2866 improper lists correctly. (reported by Bruno Haible)
2867 * fixed bug 313: source-transform for <fun-name> was erroneously
2868 applied to a call of a value of a variable with name <fun-name>.
2869 (reported by Antonio Menezes Leitao)
2870 * fixed bug 307: The obsolete instance protocol ensures that
2871 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2872 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2873 remove dead unknown-values globs from the middle of the stack.
2874 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2875 easier to resume long computations after using *BREAK-ON-SIGNALS*
2876 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2877 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2878 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2879 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2880 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2881 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2883 * fixed a (fairly theoretical) bug in string printing: if
2884 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2885 does not have array-element-type equal to the most general string
2887 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2888 function name. (thanks to Nikodemus Siivola)
2889 * fixed bug: compile-time format string checker failed on
2890 non-closed ~{. (reported by Thomas F Burdick)
2891 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2892 don't warn on legal loop code involving more than one aggregate
2893 boolean. (thanks to Nikodemus Siivola)
2894 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2895 in keyword position to MAKE-PACKAGE was not being reported as
2897 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2898 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2900 * fixed bugs 280 and 312: the checking for multiple definitions in a
2901 file is less likely to become confused by uses of inline
2903 * fixed bug: the #S reader macro performs the keyword coercion
2904 specified for slot names. (reported by Kalle Niemitalo)
2905 * fixed bug: lambda lists may contain symbols whose names start with
2906 & but are not lambda-list-keywords; their occurrence triggers a
2908 * fixed bug 321: define-method-combination argument lambda lists do
2909 not cause invalid code to be generated when &optional or &aux
2910 variables have default values. (reported by Bruno Haible)
2911 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2912 have CLOS classes; this appears also to have fixed reported
2913 instability in stack exhaustion detection.
2914 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2915 MUFFLE-WARNING when no associated restart is present is now
2917 * optimization: rearranged the expansion of various defining macros
2918 so that each expands into only one top-level form in a
2919 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2921 * optimization: used a previously little-used slot in symbols to
2922 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2924 * fixed some bugs revealed by Paul Dietz' test suite:
2925 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2927 ** Fixed an optimization bug related to inheritance of initforms
2928 from local to shared slots.
2929 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2930 ** CAST optimizer forgot to flush argument derived type.
2931 ** print/read consistency on floats is now orders of magnitude
2932 more likely. (thanks also to Bruno Haible for a similar report
2934 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2935 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2936 for objects of type REAL. Make it so.
2937 ** SXHASH is specified (infelicitously) to respect similarity,
2938 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2939 it so. (thanks to Markus Ziegler)
2940 ** on the Alpha, the compiler succeeds in compiling functions
2941 returning a known number of arguments greater than 63.
2942 ** fixed handling of invalid NIL arguments in keyword position
2944 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2946 ** provided a readably-printable representation for RANDOM-STATE
2948 ** ensured that pathnames, where they have a namestring, always
2949 print using #P"..." syntax.
2951 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2952 * Support for the forthcoming 2.0 version of the NetBSD kernel
2953 running on x86 hardware has been added. (thanks to Perry
2954 E. Metzger most immediately, and others for their past work)
2955 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2956 current release version; SBCL's OpenBSD support had been broken
2957 since about the time of OpenBSD's switch to ELF binary format.)
2958 * [placeholder for DX summary]
2959 ** user code with &REST lists declared dynamic-extent, under high
2960 speed or space and low safety and debug optimization policy.
2961 * The manual has been converted to Texinfo format and the debugger
2962 chapter from the cmucl manual has been added.
2963 * A facility has been added to extract documentation strings from
2964 sbcl and store them as Texinfo-formatted snippets for inclusion in
2965 the manual (via Texinfo's @include directive)
2966 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2967 the test case to Dave Roberts)
2968 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2969 pointers. (reported by Sean Ross)
2970 * bug fix: PROFILE output is printed nicely even for large numerical
2971 values. (thanks to Zach Beane)
2972 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2973 greater than 32 handle EOF correctly.
2974 * bug fix: on X86 an immediate argument of the IMUL instruction is
2975 correctly printed in disassembly. (thanks to Lutz Euler)
2976 * bug fix: class slots in redefined classes preserve their old
2977 values. (thanks to Bruno Haible and Nikodemus Siivola)
2978 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2979 succeeds. (reported by Marco Baringer)
2980 * fixed some bugs revealed by Paul Dietz' test suite:
2981 ** READ-SEQUENCE now works on ECHO-STREAMs.
2982 ** RATIONALIZE works more according to its specification. (thanks
2985 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2986 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2987 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2988 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2989 should matter to you only if you rebind the printer control
2990 variables and then find you want different bindings in the
2991 debugger than in the ordinary execution of your program.)
2992 * The runtime build system has been tweaked to support building
2993 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2994 assembler and linker. (thanks to Nikodemus Siivola)
2995 * Unbound, undefined, undeclared variables now trigger full
2996 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2997 more often programmer error than deliberate exploitation of undefined
2999 * optimization: the hash algorithm for strings has changed to one
3000 that is less vulnerable to spurious collisions. (thanks to Juho
3002 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
3003 needless bounds checking. (thanks to Juho Snellman)
3004 * optimization: implemented multiplication as a modular
3005 (UNSIGNED-BYTE 32) operation on the PPC backend.
3006 * fixed some bugs revealed by Paul Dietz' test suite:
3007 ** ADJUST-ARRAY now copies the datum in a zero rank array if
3009 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
3010 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
3011 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
3013 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
3015 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
3017 ** ECHO-STREAMs no longer attempt to echo the end of file value to
3018 their output stream on EOF from read.
3019 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
3020 have been read to end-of-file.
3021 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
3023 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
3024 description of determination of which consecutive characters
3026 ** Printing the "Space" character with escaping on now yields "#\\ ",
3027 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
3028 ** Reading floating-point numbers with *READ-BASE* set to a number
3029 less than 10 works correctly.
3030 ** Reading floating-point numbers with *READ-BASE* set to a number
3031 more than 10 works correctly.
3032 ** Printing with *PRINT-READABLY* targets the standard readtable, not
3033 the readtable currently in effect.
3035 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
3036 * minor incompatible change: parsing of namestrings on a physical
3037 (Unix) host has changed; numbers after the final #\. in a
3038 namestring are no longer interpreted as a version field. This is
3039 intented to be largely invisible to the user, except that the
3040 meaning of the namestring "*.*.*" has changed: it now refers to a
3041 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
3042 should usually be replaced by
3043 (make-pathname :name :wild :type :wild :version :wild)
3044 with the added benefit that this is more likely to be portable.
3045 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
3046 option to OPEN now signals an error if the file being opened
3047 exists; this may have an impact on existing code.
3048 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
3049 signals received, on the PowerPC platforms (both Linux and
3050 Darwin). (thanks to Pierre Mai for pointing out the location of
3052 * several fixes on OS X: The system now builds and runs cleanly on
3053 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
3054 (thanks to Brian Mastenbrook)
3055 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
3056 second-resolution: integer multiples of 1/3600 between -24 and 24.
3057 (thanks to Vincent Arkesteijn)
3058 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
3059 type when called with 1 argument; PEEK-CHAR checked type of
3060 PEEK-TYPE only after having read first character from a
3061 stream. (reported by Peter Graves)
3062 * bug fix: the garbage collector now has much better locality
3063 behaviour, and in particular no longer treats all memory as being
3064 exhausted when many small objects point to each other in a deeply
3066 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
3067 recognized as being TYPEP their class.
3068 * bug fix: the PUSHNEW documentation string has been corrected.
3069 (thanks to Vincent Arkesteijn)
3070 * bug fix: defaulting of the value for the last of an atypically
3071 large number of multiple values being bound was not being
3072 performed correctly on the Alpha or PPC platforms
3073 * optimization: implemented multiplication as a modular
3074 (UNSIGNED-BYTE 32) operation on the x86 backend.
3075 * optimization: SEARCH on simple-base-strings can now be open-coded.
3076 (see also contrib/compiler-extras.lisp for inspiration for
3077 teaching the compiler about the Boyer-Moore algorithm).
3078 * value, returned by MAX (and MIN) called with several EQUALP, but
3079 not EQL, arguments now does not depend on compiler settings.
3080 * fixed some bugs revealed by Paul Dietz' test suite:
3081 ** in stack analysis liveness information is propagated from
3082 non-local entry points.
3083 ** pathwise CAST removing failed when the CAST node did not start
3085 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
3086 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3088 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3089 host is already defined.
3090 ** RENAME-FILE works on streams instead of signalling an internal
3092 ** PEEK-CHAR uses the current readtable when determining whether
3093 or not a character is whitespace.
3094 ** MERGE-PATHNAMES handles the case when the pathname does not
3095 specify a name while the default-pathname specifies a version.
3096 ** Pathnames now stand a better chance of respecting print/read
3098 ** Attempting to use standardized file system operators with a
3099 pathname with invalid :DIRECTORY components signals a
3101 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3103 ** DIRECTORY on logical pathnames is more correct.
3104 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3105 signal a TYPE-ERROR if their argument is not a stream.
3106 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3107 designator argument does not designate a stream.
3108 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3109 examining the synonym.
3110 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3112 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3113 element-type larger than ([UN]SIGNED-BYTE 32).
3115 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3116 * When built with the :SB-FUTEX feature, threaded builds now take
3117 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3118 for faster/more reliable mutex and condition variable support.
3119 * Incompatible change (but one you probably shouldn't have been using
3120 anyway): the interface and code for arbitrating between multiple
3121 threads in the same user session has been redesigned.
3122 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3123 error when called without an explicit environment argument.
3124 (thanks to Nikodemus Siivola)
3125 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3126 confused about their position. (thanks to Adam Warner and Gerd
3128 * bug fix: Pretty printing backquoted forms with unquotations in the
3129 argument list position of various code constructs such as LAMBDA
3130 now works correctly. (reported by Paul Dietz)
3131 * bug fix: Pretty printing unquotations no longer loses all
3132 stream position information.
3133 * optimization: performance of string output streams is now less
3134 poor for multiple small sequence writes.
3135 * optimization: performance of CSUBTYPEP in the presence of complex
3136 expressions involving CONS and NOT many times has been improved.
3137 (reported by Paul Dietz)
3138 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3140 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3141 streams for the REPL, for the debugger, and so on. (thanks to
3143 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3144 * fixed some bugs revealed by Paul Dietz' test suite:
3145 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3146 is ignored if it is too small, rather than propagating through
3147 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3148 ** extremely complex negations of CONS types were not being
3149 sufficiently canonicalized, leading to inconsistencies in
3151 ** VALUES tranformer lost derived type.
3153 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3154 * fixed a bootstrapping bug: the build process no longer assumes
3155 that the various BOOLE-related constants have the same value in
3156 host and target lisps. (noted by Paul Dietz' test suite on an
3157 SBCL binary built from CLISP)
3158 * The system can now be dynamically linked on the MIPS platform,
3159 which enables dynamic loading of foreign code from Lisp. (thanks
3160 to Ralf Baechle for discussions on the MIPS ABI)
3161 * The system now records debugging information for its own source
3162 files in a filesystem-position-independent manner, relative to
3163 the "SYS" logical host.
3164 * fixed a compiler bug: MV-LET convertion did not check references
3165 to the "max args" entry point. (reported by Brian Downing)
3166 * tweaked disassembly notes to be less confident about proclaiming
3167 some instruction as an LRA. (thanks to Brian Downing)
3168 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3169 now each have their own history, command character, and other
3170 characteristics. (thanks to David Lichteblau)
3171 * fixed some bugs revealed by Paul Dietz' test suite:
3172 ** compiler failure in compiling LOGAND expressions including a
3174 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3175 shift greater than 32.
3176 ** FLUSH-DEST did not mark blocks for type check regeneration.
3177 ** HANDLER-CASE failed to accept declarations in handler clauses
3178 in some circumstances.
3180 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3181 * New code in contrib/sb-introspect (still probably not entirely
3182 stable yet) provides some support for smart Lisp development
3183 environments like SLIME.
3184 * The conditions signalled for errors occurring when loading .fasl
3185 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3186 in a way which should help ASDF recover gracefully.
3187 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3188 stuff has been cleaned up. If you code contrib/ stuff, this might
3189 affect you, and you can look at contrib/README, contrib/STANDARDS,
3190 and/or the 0.8.4.27 diff to check.
3191 * In full calls the compiler now does not generate checks for declared
3192 argument types for all arguments.
3193 * various threading fixes
3194 ** and some experimental patches which didn't make it into
3195 the main tree for this release, but which are shipped in
3196 contrib/experimental-thread.patch as a possible fix for some
3197 failures (deadlock, spinning...) in GC-intensive multithreaded
3199 * fixed PPC build problem (source code incompatibility of different
3200 library versions): added offsetof() hackery which attempts to divine
3201 where glibc maintainers put uc_mcontext today
3202 * fixed bug 282: compiler does not trust type assertions while passing
3203 arguments to a full call.
3204 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3205 &OPTIONAL VALUES type parameter.
3206 * fix bug 214: algorithm for noting rejected templates is now more
3207 similar to that of template seletion. (also reported by rydis on
3209 * fixed bug 141b: printing backquoted information readably and prettily
3210 inserts a space where necessary.
3211 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3212 (CHARACTER 10) are now reported as errors, rather than propagated
3213 as unknown types. (reported by piso on #lisp)
3214 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3215 with values NIL and :ERROR. (thanks to Milan Zamazal)
3216 * fixed bug 191c: CLOS now does proper keyword argument checking as
3217 described in CLHS 7.6.5 and 7.6.5.1.
3218 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3219 longer raise an error; further, using a list as a for-as-arithmetic
3220 counter now raises a meaningful error.
3221 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3222 now understood by sequence creation functions such as MAKE-SEQUENCE
3224 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3225 a STRING-INPUT-STREAM.
3226 * compiler enhancement: SIGNUM is now better able to derive the type
3228 * type declarations inside WITH-SLOTS are checked. (reported by
3230 * fixed some bugs revealed by Paul Dietz' test suite:
3231 ** incorrect optimization of TRUNCATE for a positive first
3232 argument and negative second.
3233 ** compiler failure in let-convertion during flushing dead code.
3234 ** compiler failure while deriving type of TRUNCATE on an
3235 interval, containing 0.
3236 ** ASH of a negative bignum by a negative bignum count now returns
3238 ** intersection of CONS types now canonicalizes properly, fixing
3239 inconsistencies in SUBTYPEP.
3241 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3242 * incompatible change: The --disable-debugger command line
3243 option now clobbers the debugger at a more fundamental
3244 level, by redefining #'INVOKE-DEBUGGER instead of by
3245 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3246 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3247 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3248 enter the debugger REPL and then suffer recursive errors
3249 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3250 while under the new system, BREAK in --disable-debugger mode
3251 terminates the system just as an unhandled error would.
3252 * fixed compiler performance when processing loops with a step >1;
3253 * bug fix: DOCUMENTATION now retrieves generic function
3254 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3255 support has been systematized, and now supports the methods
3256 specified by ANSI, along with a default method and a method for
3257 slot documentation. (reported by Nathan Froyd)
3258 * bug fix: effective methods associated with a generic function are
3259 no longer cached over a change of that generic function's method
3260 combination. (reported by Andreas Fuchs)
3261 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3262 names a non-existent directory. (thanks to Andreas Fuchs)
3263 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3264 the PPC platform, lead to stack corruption; this has been fixed.
3265 (reported by Rainer Joswig)
3266 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3267 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3268 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3269 arithmetic in general, is now much more reliable on the PPC
3271 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3272 platform now returns the right answer.
3273 * optimization: restored some effective method precomputation in
3274 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3275 precomputation is now tunable.
3276 * optimization: compiler-internal data structure use has been
3277 reviewed, and changes have been made that should improve the
3278 performance of the compiler by about 20%.
3279 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3280 simple-base-strings and simple-bit-vectors is improved.
3281 * optimization: the optimization of 32-bit logical and arithmetic
3282 functions introduced in version 0.8.3 on the x86 has been
3283 implemented on the mips, ppc and sparc platforms; an
3284 implementation of the same facility, but for 64-bit arithmetic,
3285 has been added for the alpha.
3286 * microoptimization: the compiler is better able to make use of the
3287 x86 LEA instruction for multiplication by constants.
3288 * bug fix: in some situations compiler did not report usage of
3289 generic arithmetic in (SPEED 3) policy.
3290 * bug 145b fix: compiler used wrong type specifier while converting
3291 MEMBER-types to numeric.
3292 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3294 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3295 index argument now works properly on the Alpha platform.
3296 * bug fix: floating point exception treatment on the Alpha platform
3298 * bug fix: FILE-POSITION works much better on string input and
3299 output streams. (thanks to Nikodemus Siivola)
3300 * bug fix: many threading/garbage collection symptoms sorted.
3301 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3302 might be pseudo-atomic.
3303 * internal change: Stopping for GC is now done with signals not
3304 ptrace. GC is now done in whichever thread wanted it, instead of
3306 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3308 * bug fix: result form in DO is not contained in the implicit
3310 * incompatible change: ICR structure is changed; the value part of
3311 CONTINUATION is now called LVAR; corresponding functions are
3312 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3313 * added type deriver for ISQRT (thanks to Robert E. Brown).
3314 * bug fix: better support for loading from the command line when an
3315 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3317 * fixed some bugs revealed by Paul Dietz' test suite:
3318 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3319 ** ROUND and FROUND now give the right answer when given very
3320 small float arguments.
3321 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3323 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3324 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3325 ** LCM should return a non-negative integer.
3326 ** PARSE-INTEGER returned the index of a terminator instead of the
3327 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3328 ** PARSE-INTEGER returned an incorrect index being applied to a
3330 ** LCM with two arguments of 0 returns 0 rather than signalling
3332 ** unsigned addition of a 32-bit constant with the high bit set no
3333 longer causes an internal compiler error on the x86.
3334 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3335 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3336 with negative last argument.
3337 ** byte specifiers with zero size and position no longer cause
3338 an error during type derivation.
3339 ** bignum multiplication on the Alpha platform now returns the
3341 * porting: The system now builds on SuSE AMD64, although it still
3342 generates a 32-bit binary.
3343 * .fasl file incompatibility: The fasl file version number has
3344 been incremented (because of the changes to internal compiler
3345 data structures referred to above).
3347 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3348 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3349 more accurately, on the Darwin kernel running on PowerPC hardware.
3350 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3351 * Compiler code deletion notes now signal a condition of type
3352 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3353 an associated MUFFLE-WARNING restart.
3354 * The compiler now performs limited argument count validation of
3355 constant format strings in FORMAT, and where appropriate in ERROR,
3356 CERROR and WARN. (thanks to Gerd Moellmann)
3357 * New ASDF-INSTALL contrib can be used for automatic download and
3358 installation of third-party Lisp code from CCLAN or other sites
3360 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3361 which forces another thread to execute a function supplied by the
3363 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3364 now accept and act upon their :ELEMENT-TYPE keyword argument.
3365 (reported by Martin Atzmueller, Edi Weitz)
3366 * bug fix: FILE-POSITION now accepts position designators up to
3367 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3368 is the greater. (thanks to Patrik Nordebo)
3369 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3370 Kalle Olavi Niemitalo)
3371 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3372 printable. (reported by Eric Marsden)
3373 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3374 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3375 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3376 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3377 (thanks to Stig Sandoe)
3378 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3379 UNREAD-CHAR now correctly includes the unread character in the
3380 target sequence. (thanks to Gerd Moellmann)
3381 * bug fix in threaded builds: the system can now be suspended and
3382 resumed by shell job control with minimal disruption.
3383 * bug fixes in times and timezones >2038AD
3384 * better handling of "where is GNU make?" problem in build scripts
3385 (thanks to Nikodemus Siivola)
3386 * new optimization: inside a named function any reference to a
3387 function with the same name is considered to be a self-reference;
3388 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3389 optimization quality.
3390 * new optimization on x86: logical functions and + now have
3391 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3392 used when the result is truncated to 32 bits.
3393 * VALUES declaration is partially enabled.
3394 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3395 * fixed some bugs revealed by Paul Dietz' test suite:
3396 ** The system now obeys the constraint imposed by
3397 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3398 types form a lattice under type intersection.
3399 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3400 ** ASSOC now ignores NIL elements in an alist.
3401 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3402 and (1+ MOST-POSITIVE-FIXNUM) answers.
3403 ** The addition of a method with invalid qualifiers to a generic
3404 function does not cause an error to be signalled immediately;
3405 a warning is signalled, and the error is generated only on
3406 calling the generic function.
3407 * changed .fasl file version number, in part to add type codes for
3408 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3409 obscure ANSI requirements
3411 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3412 * fixed bug 148: failure to inline-expand a local function left
3413 garbage, confusing the compiler.
3414 * fixed bugs 3cd: structure slot readers perform type check if the
3415 slot can have an invalid value (i.e. it is either not initialized
3416 or can be written with a less specific slot writer).
3417 * bug fix: the compiler now traps array references to elements off
3418 the end of an array; previously, the bounds checking in some
3419 circumstances could go off-by-one.
3420 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3422 * type declarations for array element types now obey the description
3423 on the CLHS page "Declaration TYPE", as per discussions on
3424 sbcl-help around 2003-05-08. This means that a declaration
3425 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3426 declaration, all references to BAR will be asserted or assumed
3427 (with THE, so dependent on compiler policy) to involve objects of
3428 type FOO. Note that no such declaration is implied in
3429 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3430 * declared types of functions from the "Conditions"
3431 chapter. (reported by Paul Dietz)
3432 * bug fix: CERROR accepts a function as its first argument.
3433 * bug fix: NTH an NTHCDR accept a bignum as index
3434 arguments. (reported by Adam Warner)
3435 * optimization: character compare routines now optimize comparing
3436 against a constant character. (reported by Gilbert Baumann)
3437 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3438 argument now works properly on the MIPS platform.
3439 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3440 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3441 declared to have a functional type, does not cause an error
3443 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3444 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3445 * fixed bug: if last continuation of a deleted block has a
3446 destination, this destination should be deleted too. (reported by
3448 * fixed a bug in the bootstrap process: the host compiler's values
3449 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3450 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3451 test case from Patrik Nordebo)
3452 * improved the ability of the disassembler on the PPC platform to
3453 provide helpful disassembly notes.
3454 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3455 the class in more cases than previously.
3456 * bug fix: FILE-POSITION now understands :START and :END for
3457 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3458 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3459 8) data. (thanks to David Lichteblau)
3460 * bug fix: it is possible to add a method to a generic function
3461 without lambda list.
3462 * bug fix: reader failed to signal END-OF-FILE inside an
3463 object representation. (reported by Nikodemus Siivola)
3464 * fixed some bugs revealed by Paul Dietz' test suite:
3465 ** LAST and [N]BUTLAST should accept a bignum.
3466 ** condition slot accessors are methods.
3467 ** (VECTOR NIL) is a subtype of STRING.
3469 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3470 * minor incompatible change: some nonsensical specialized lambda
3471 lists (used in DEFMETHOD) which were previously ignored now signal
3473 * minor incompatible change: the system is now aware of the types of
3474 variables in the COMMON-LISP package, and will signal errors for
3475 most violations of these type constraints (where previously they
3476 were silently accepted).
3477 * minor incompatible change: COMPILE-FILE now uses the freedom
3478 afforded (ANSI 3.2.2.3) to use derived function types for
3479 functions defined in the same file. This also permits the system
3480 to warn on static type mismatches and function
3481 redefinition. (Currently it does not work with high DEBUG level.)
3482 * minor incompatible change: VALUES declaration is disabled.
3483 * When issuing notes, the compiler now signals a condition of type
3484 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3485 restart for use in user handlers. It is expected that the
3486 COMPILER-NOTE condition will eventually become a condition
3487 supertype to a hierarchy of note types, which will then be
3488 handleable in a similar fashion. However, at the moment, no such
3489 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3490 but it's an implementation detail, not a classification for the
3492 * Changes in type checking closed the following bugs:
3493 ** type checking of unused values (192b, 194d, 203);
3494 ** template selection based on unsafe type assertions (192c, 236);
3495 ** type checking in branches (194bc).
3496 * A short form of VALUES type specifier has ANSI meaning (it has
3497 increased the number of situations when SBCL cannot perform type
3499 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3500 keywords or constants is permissible.
3501 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3502 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3503 classes. (thanks to Antonio Martinez)
3504 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3505 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3506 argument to a pretty-print stream. Instead, it leaves any such
3507 operations to DESCRIBE-OBJECT methods.
3508 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3510 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3511 lambda lists. (thanks to David Lichteblau)
3512 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3513 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3514 (thanks to Gerd Moellmann)
3515 * fixed bug 246: increased compilation speed of long
3516 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3518 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3520 * DEFCONSTANT now throws a condition of type
3521 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3522 constant to a non-EQL value; CONTINUE and ABORT restarts
3523 respectively change and preserve the value.
3524 * fixed bug 63: The code walker, part of the implementation of CLOS,
3525 is now better at handling symbol macros.
3526 * bug fix: There is no longer an internal implementation type named
3527 CL:LENGTH. (reported by Raymond Toy)
3528 * bug fix: In macro-like defining macros/special operators the
3529 implicit block does not enclose the lambda list.
3530 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3531 enclosed in parentheses) are not suitable as type specifiers, and
3532 their use properly signals an error now.
3533 * bug fix: An argument count mismatch for a type specifier in code
3534 being compiled no longer causes an unhandled error at compile
3535 time, but signals a compile-time warning.
3536 * fixed simple vector readable printing
3537 * bug fix: DESCRIBE takes more care over whether the class
3538 precedence list slot of a class is bound before accessing it.
3539 (reported by Markus Krummenacker)
3540 * bug fix: FORMATTER can successfully compile pretty-printer format
3541 strings which use variants of the ~* directive inside.
3542 * bug fix: SEARCH now applies its TEST predicate to the elements of
3543 the arguments in the correct order. (thanks to Wolfhard Buss)
3544 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3545 in type inference. (thanks to Robert E. Brown)
3546 * bug fix: Reading in symbols with an explicit package name of ""
3547 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3548 * fixed some bugs revealed by Paul Dietz' test suite:
3549 ** NIL is now allowed as a structure slot name.
3550 ** Arbitrary numbers, not just REALs, are allowed in certain
3551 circumstances in LOOP for-as-arithmetic clauses.
3552 ** Multiple class redefinitions before slot access no longer
3553 causes a type error.
3554 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3555 association between the name and a class.
3556 ** Generic functions with non-standard method-combination and over
3557 five methods all of which return constants no longer return NIL
3558 after the first few invocations. (thanks to Gerd Moellmann)
3559 ** CALL-NEXT-METHOD with no arguments now passes the original
3560 values of the arguments, even in the presence of assignment.
3561 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3563 ** DISASSEMBLE works with closures and funcallable instances.
3564 ** ADD-METHOD now returns the generic function, not the new method.
3565 ** FIND-METHOD signals an error if the lengths of the specializers
3566 is incompatible with the generic function, even if the ERRORP
3568 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3569 which its argument is a member.
3570 ** DEFCLASS only redefines the class named by its class-name
3571 argument if that name is the proper name of the class;
3572 otherwise, it creates a new class.
3573 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3574 of the UNBOUND-SLOT condition to the name of the slot.
3575 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3576 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3577 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3578 SLOT-MAKUNBOUND in the specified fashion.
3580 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3581 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3582 cross-compilation host. As a consequence, we can now bootstrap our
3583 way up to SBCL starting with a bare gcc toolchain and human-readable
3584 source code (first the source to CLISP, then the source to SBCL).
3585 * A contributed module containing a partial implementation of the
3586 simple-streams interface has been included. (thanks to Rudi
3588 * A contributed module implementing the RFC1321 Message Digest
3589 Algorithm, known as MD5, has been included.
3590 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3591 no longer has any effect, as the code controlled by this feature
3592 has been deleted. (As far as we know, no-one has ever built using
3593 this feature, and its semantics were confused in any case).
3594 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3595 work on conditions (as required by the ANSI specification),
3596 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3597 expected behaviour on conditions. Users should note, however,
3598 that such behaviour is not required by the ANSI specification,
3599 and so use of this behaviour may render their code unportable.
3600 * fixed some bugs revealed by Paul Dietz' test suite:
3601 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3603 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3604 ** SXHASH on condition objects no longer returns NIL.
3605 ** :ALLOCATION :CLASS slots are better treated; their values are
3606 updated on class redefinition, and initforms inherited from
3607 superclasses are applied.
3608 ** REMOVE-METHOD returns its generic function argument even when
3609 no method was removed.
3610 ** SHARED-INITIALIZE now initializes the values of the requested
3611 slots, including those with :ALLOCATION :CLASS.
3612 ** ALLOCATE-INSTANCE now works on structure classes defined via
3613 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3615 ** SLOT-EXISTS-P now works on conditions, as well as structures
3617 ** MAKE-LOAD-FORM now has the required methods on
3618 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3619 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3620 keyword, and now implements the SLOT-NAMES argument.
3621 ** methods with &OPTIONAL arguments no longer allow too many
3622 arguments to be passed in the call without error.
3623 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3624 option is consistent with the required arguments of the generic
3625 function lambda list.
3626 * bug fix: REQUIRE accepts a string designator. (Thanks to
3628 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3629 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3630 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3631 * bug fix: The compiler now checks for duplicated variables in macro
3633 * bug fix: SETQ on globals returns the correct value.
3634 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3635 not a COMPILER-ERROR (followed by some other strange error on
3636 choosing the CONTINUE restart).
3637 * bug fix: make.sh and friends are now more consistent in the way that
3638 they look for GNU "make".
3640 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3641 * experimental native threads support (on x86 Linux >=2.4 only).
3642 This is not compiled in by default: you need to add :SB-THREAD to
3643 the target features. See the "Beyond ANSI" chapter of the manual
3645 * fix for longstanding nonANSIism: The old distinction between
3646 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3647 The return value from CL:FIND-CLASS is now a CLOS class, and
3648 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3649 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3650 * An interface to the MetaObject Protocol, as described in Kiczales,
3651 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3652 MIT Press, 1991, is available from the SB-MOP package.
3653 * incompatible change: the SB-PCL package should now be considered
3654 a private implementation detail, and no longer a semi-private MOP
3656 * minor incompatible change: due to rearrangement for threads, the
3657 control stack and binding stack are now allocated at arbitrary
3658 addresses instead of being hardcoded per-port. Users affected by
3659 this probably have to be doing advanced things with shared
3660 libraries, and will know who they are.
3661 * minor incompatible change: Previously, all --eval forms used were
3662 processed with READ before any of them were processed with EVAL.
3663 Now each --eval form is processed with both READ and EVAL before
3664 the next --eval form is processed. (Thus package operations like
3665 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3666 work as the user might reasonably expect.)
3667 * minor incompatible change: *STANDARD-INPUT* is now only an
3668 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3670 * minor incompatible change: Y-OR-N-P is now character-oriented,
3671 not line oriented. Also, YES-OR-NO-P now works without errors.
3672 (thanks to Antonio Martinez)
3673 * sb-aclrepl module improvements: an integrated inspector, added
3674 repl features, and a bug fix to :trace command.
3675 * Known functions, which cannot be open coded by the backend, are
3676 considered to be able to check types of their arguments. (fixing
3677 a bug report by Nathan J. Froyd)
3678 * fixed a bug in computing method discriminating functions: It is
3679 now possible to define methods specialized on classes which have
3680 forward-referenced superclasses. (thanks to Gerd Moellmann)
3681 * fixed evaluation order in optional entries (reported by Gilbert
3683 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3684 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3685 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3686 required-by-AMOP NAME argument, as well as CLASS and
3687 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3688 * fixed bug 20: DEFMETHOD can define methods using names that are
3689 not the proper names of classes to designate class specializers.
3690 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3691 Unix file descriptors, instead of blowing up. (thanks to Antonio
3693 * Garbage collection refactoring: user-visible change is that a
3694 call to the GC function during WITHOUT-GCING will not do garbage
3695 collection until the end of the WITHOUT-GCING. If you were doing
3696 this you were probably losing anyway.
3697 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3698 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3699 TYPEP the latter but not the former.
3700 * The compiler issues a full WARNING for calls to undefined functions
3701 with names from the CL package.
3702 * MAP-INTO for a vector destination is open coded. (reported by
3703 Brian Downing on c.l.l)
3704 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3705 documentation string.
3706 * fixed some bugs revealed by Paul Dietz' test suite:
3707 ** COPY-ALIST now signals an error if its argument is a dotted
3709 ** Condition slots are now accessed more correctly in the presence
3710 of multiple initargs for a given slot.
3711 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3712 exclude restarts of the same name associated with a different
3714 ** DEFCLASS of forward-referenced classes with another
3715 forward-referenced class in the superclasses list no longer
3717 ** Condition slots are now initialized once each, not multiple
3718 times. (thanks to Gerd Moellmann)
3719 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3721 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3722 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3723 arguments contain duplicated elements.
3724 ** RESTART-CASE understands local macros.
3725 ** RESTART-CASE associates exactly its own restarts with a condition.
3726 ** ENDP in safe mode checks its argument to be of type LIST.
3727 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3728 in question is unbound.
3729 ** Optimized MAKE-INSTANCE functions no longer cause internal
3730 assertion failures in the presence of duplicate initargs.
3731 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3732 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3733 propagated even in the ctor optimized implementation of
3735 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3737 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3738 * a better implementation of SXHASH on (simple) bit vectors,
3739 measured both in execution speed and in distribution of results
3740 over the positive fixnums, has been installed. Likewise, a better
3741 implementation of EQUAL for simple bit vectors is now available.
3742 * fixed CEILING optimization for a divisor of form 2^k.
3743 * fixed bug 240 (emitting extra style warnings "using the lexical
3744 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3745 by Antonio Martinez)
3746 * fixed SXHASH, giving different results for NIL depending on type
3747 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3748 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3749 variable. (found by Rolf Wester)
3750 * SBCL does not ignore type declarations for special
3751 variables. (reported by rif on c.l.l 2003-03-05)
3752 * some bug fixes in contrib/sb-aclrepl/
3753 * fixed some bugs revealed by Paul Dietz' test suite:
3754 ** a bug in the CONS type specifier, whereby the CAR and CDR
3755 types got intertwined, has been fixed;
3756 ** the type system is now able to reason about the interaction
3757 between INTEGER and RATIO types more completely;
3758 ** APPEND, [N]REVERSE and NRECONC check that those their
3759 arguments, which must be proper lists, are really so;
3760 ** An array specialized to be unable to hold elements has been
3761 implemented, as required -- yes, really -- by ANSI;
3762 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3763 on malformed property lists;
3765 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3766 * incompatible packaging change: in line with Unix convention,
3767 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3768 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3769 * REQUIRE and PROVIDE are now optionally capable of doing something
3770 useful. See the documentation string for REQUIRE.
3771 * infrastructure for a managed SBCL contrib system: contributed
3772 modules in this release include:
3773 ** the ASDF system definition facility;
3774 ** an interface to the BSD Sockets API;
3775 ** an ACL-like convenience interface to the repl;
3776 (thanks to Kevin Rosenberg)
3777 ** an implementation of ROTATE-BYTE, with efficient implementation
3779 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3780 gives a closer approximation to the right answer than previously.
3781 (thanks to Raymond Toy)
3782 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3783 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3784 optional environment arguments, as required by ANSI.
3785 * fixed bugs in other functions taking environment objects, allowing
3786 calls with an explicit NIL environment argument to be compiled
3788 * fixed bug 228: primary return values from
3789 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3790 COMPILE or FUNCTION.
3791 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3792 :TYPE will no longer signal errors on innocuous objects.
3793 * fixed bug 231b: SETQ is better at respecting type declarations in
3794 the lexical environment.
3795 * fixed a bug in DEFCLASS: classes named by symbols with no or
3796 unprintable packages can now be defined.
3797 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3798 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3799 * fixed compiler failure related to checking types of functions.
3800 (reported by Robert E. Brown)
3801 * the compiler is now much more consistent in its error-checking
3802 treatment of bounding index arguments to sequence functions: in
3803 (SAFETY 3) code, errors will be signalled in almost all cases if
3804 invalid sequence bounding indices are passed to functions defined
3805 by ANSI to operate on sequences.
3806 * fixed a bug in the build procedure: documentation of SBCL-specific
3807 packages is now preserved and available in the final Lisp image.
3808 * lifted FDEFINITION lookup out of loops in the implementation of
3809 many list operations. (thanks to Robert E. Brown)
3810 * fixed a bug in the reader: the #n# reader macro now works for
3811 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3812 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3813 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3814 E. Brown and Tony Martinez respectively)
3815 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3816 SLOT-BOUNDP: in particular, it is now able to optimize them much
3817 better, and is now not vulnerable to having packages renamed.
3818 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3819 always called when a slot is not present in an instance. (thanks
3821 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3822 longer signal an error on unreading a character following EOF on
3823 the previous constituent stream. (thanks to Tony Martinez)
3824 * fixed some bugs revealed by Paul Dietz' test suite:
3825 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3826 not just nonnegative fixnums;
3827 ** the logical bit-array operators such as BIT-AND now accept an
3828 explicit NIL for their "opt-arg" argument (to indicate a
3829 freshly-consed result bit-array);
3830 ** ELT now signals an error on an invalid sequence index in safe
3832 ** the type system is now cleverer about negations of numeric
3833 types, and consequently understands the BIGNUM and RATIO types
3835 ** the type system is now cleverer about the interaction between
3836 INTEGER and RATIO types: while bugs still remain, many more
3837 cases are accurately computed;
3838 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3839 if it is in the last clause;
3840 ** CONSTANTLY now correctly returns a side-effect-free function in
3842 ** DECLARE is no longer treated as a special-operator; in
3843 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3844 * incremented fasl file version number due to the change in the
3845 DEFSTRUCT-SLOT-DESCRIPTION structure.
3847 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3848 * minor incompatible change: code processed by the "interpreter" or
3849 EVAL now has a compilation optimization policy of (DEBUG 2)
3850 (changed from (DEBUG 1)) to improve debuggability of interactive
3851 development, and to allow the use of the debug RETURN command in
3853 * an experimental implementation of the RETURN command for the
3854 debugger has been included. (thanks to Frederik Kuivinen)
3855 * fixed bug 62: constraints were not propagated into a loop.
3856 * fixed bug in embedded calls of SORT (reported and investigated by
3858 * fixed some bugs revealed by Paul Dietz' test suite:
3859 ** printing and reading of arrays with some dimensions having
3860 length 0 (thanks to Gerd Moellmann);
3861 ** BOA constructor with &AUX argument without a default value does
3862 not cause a type error;
3863 ** CONSTANTP now returns true for all self-evaluating objects.
3865 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3866 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3867 accessors that are related by inheritance, as specified in the
3868 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3869 Valtteri Vuorikoski)
3870 * The compiler is now able to inline functions that were defined in
3871 a complex lexical environment (e.g. inside a MACROLET).
3872 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3874 * Support for the upcoming FreeBSD-5.0 release has been included.
3875 (thanks to Dag-Erling Smorgrav)
3876 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3877 effect when it is not in a toplevel context.
3878 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3879 stand a better chance of being correct. (thanks to Gerd
3881 * fixed bug in COERCE, which now signals an error on coercing a
3882 rational to a bounded real type which excludes the expected
3884 * The compiler is now able to derive types more accurately from the
3885 COERCE and COMPILE functions.
3886 * fixed bug 223: functional binding is considered to be constant
3887 only for symbols in the CL package.
3888 * fixed bug 231: SETQ did not check the type of a variable being set
3889 (reported by Robert E. Brown)
3890 * A new optimization for MAKE-INSTANCE has been included, fixing
3891 various bugs (including relating to :ALLOCATION :CLASS slots and
3892 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3894 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3895 ** As required by ANSI, LOOP now disallows anonymous collection
3896 clauses such as COLLECT I in conjunction with aggregate boolean
3897 clauses such as THEREIS (= I 1);
3898 ** LOOP now signals an error when any variable is reused in the
3899 same loop (including the potentially useful construct analogous
3900 to WITH A = 1 WITH A = (1+ A);
3901 ** IT is only a special loop symbol within the first clause of a
3902 conditional loop clause;
3903 ** LOOP with a typed iteration variable over a hashtable now
3904 signals a type error iff it should.
3905 * fixed some other bugs revealed by Paul Dietz' test suite:
3906 ** FILE-STREAM now names the class previously known as FD-STREAM;
3907 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3908 argument) no longer signals an error;
3909 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3910 of the slot symbol, rather than using the current package
3911 ((:CONC-NAME "") continues to intern the slot's name in the
3913 * incremented fasl file version number, because of the incompatible
3914 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3915 of the new implementation of DEFINE-COMPILER-MACRO.
3917 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3918 * Support for building SBCL for MIPS platforms running in
3919 little-endian mode has now been checked in, and basic
3920 functionality on said platforms verified.
3921 * minor incompatible change: PCL now records the pathname of a file
3922 in which methods and the like are defined, rather than its
3924 * minor incompatible change: TRUENAME now considers the truename of
3925 a file naming a directory to be the pathname with :DIRECTORY
3926 component indicating that directory.
3927 * minor incompatible change: a NAMED clause in the extended form of
3928 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3929 reason for the previous behaviour is unclear.
3930 * more systematization and improvement of CLOS and MOP conformance
3931 in PCL (thanks to Gerd Moellman and Pierre Mai):
3932 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3934 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3935 primary methods with no specializers;
3936 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3938 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3939 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3940 FORWARD-REFERENCED-CLASSes; error reporting on
3941 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3943 ** SXHASH on CLOS instances now uses a slot internal to the
3944 instance to return different numbers on distinct instances,
3945 while preserving the same return value through invocations of
3947 ** DEFMETHOD signals errors when methods with longer incongruent
3948 lambda lists are added to generic functions;
3949 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3950 CLASS, as specified in AMOP;
3951 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3952 on the order returned by the primary method for classes of
3953 class STANDARD-CLASS;
3954 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3955 * fixed some bugs shown by Paul Dietz' test suite:
3956 ** DOLIST puts its body in TAGBODY;
3957 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3959 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3960 value producing form;
3961 ** if more variables are given to PROGV than values, extra
3962 variables are bound and made to have no value;
3963 ** NSUBSTITUTE on list arguments gets the right answer with
3965 ** ELT signals an error of type TYPE-ERROR when the index argument
3966 is not a valid sequence index;
3967 ** LOOP signals (at macroexpansion time) an error of type
3968 PROGRAM-ERROR when duplicate variable names are found;
3969 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3970 ** FUNCALL of special-operators now cause an error of type
3972 ** PSETQ now works as required in the presence of side-effecting
3973 symbol-macro places;
3974 ** NCONC accepts any object as its last argument;
3975 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3977 ** loop-for-as-package does not require a package to be explicitely
3979 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3981 * fixed bug 166: compiler preserves "there is a way to go"
3982 invariant when deleting code.
3983 * fixed bug 172: macro lambda lists with required arguments after
3984 &REST arguments now cause an error to be signalled. (thanks to
3986 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3987 bug. (thanks to Matthew Danish)
3988 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3990 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3991 arguments are lexically rebound. (thanks to Gerd Moellmann and
3993 * fixed bug 194: error messages are now more informative when there
3994 is no primary method applicable in a call to a generic
3995 function. (thanks to Gerd Moellmann)
3996 * fixed bug in command line argument checking (thanks to Julian
3998 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3999 (thanks to Matthew Danish)
4000 * incremented fasl file version number, because of the
4001 SXHASH-related changes in the layout of CLOS data structures
4003 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
4004 * minor incompatible change: The runtime (the Unix executable named
4005 "sbcl") is now much pickier about the .core files it will load.
4006 Essentially it now requires .core files to descend from the same
4007 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
4008 as the runtime does. (The intent is to prevent the crashes which
4009 can occur, and which can even be reported as mysterious failures,
4010 when people patch the sources or change the build parameters
4011 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
4012 sbcl and .core files.)
4013 * fixed bug: VALUES-LIST is no longer optimized away.
4014 * fixed bug 142: The FFI conversion of C string values to Lisp
4015 string values no longer conses excessively. (thanks to Nathan
4016 Froyd porting Raymond Toy's fix to CMU CL)
4017 * began to systematize and improve MOP conformance in PCL (thanks to
4018 Nathan Froyd, Gerd Moellman and Pierre Mai):
4019 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
4021 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
4022 ** FINALIZE-INHERITANCE is now called on class finalization;
4023 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
4024 argument precedence order.
4025 * fixed bug 202: The compiler no longer fails on functions whose
4026 derived types contradict their declared type.
4027 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
4028 so it can be non-toplevel.
4029 * The fasl file version number has changed (because of the new
4030 implementation of DEFMACRO).
4031 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
4032 safe code, that any length requirement by their type-specifier
4033 argument is valid. The exceptions to this are described in bug
4035 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
4036 functions now check the types of their inputs as required by ANSI.
4037 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
4038 introduced symbol is DECLAREd to be SPECIAL.
4039 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
4040 * fixed bug: :COUNT argument to sequence functions may be negative
4041 * fixed bug: body of DO-SYMBOLS may contain declarations
4042 * fixed bug: PUSHNEW now evaluates its arguments from left to right
4043 (reported by Paul F. Dietz, fixed by Gerd Moellman)
4044 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
4045 symbol macro only once
4046 * fixed printing of call frame when argument list is unavailable
4047 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
4048 * compiler no longer signals WARNING on unknown keyword
4051 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
4052 * A beta-quality port to the mips architecture running Linux,
4053 based on the old CMUCL backend, has been made. It has been tested
4054 on a big-endian kernel, and works sufficiently well to be able to
4055 rebuild itself; it has not been tested in little-endian mode.
4056 * fixed an inconsistency between gencgc.c and purify.c which made
4057 dumping/loading .core files unreliable
4058 * fixed bug 120a: The compiler now deals correctly with IFs where
4059 the consequent is the same as the alternative, instead of
4060 misderiving the return type. (thanks to Alexey Dejneka)
4061 * fixed bug 113: Logical pathnames are now dumpable (the logical
4062 host is resolved at load-time, throwing an error if it is not
4064 * fixed bug 174: FORMAT's error message is slightly clearer when a
4065 non-printing character is used in a format directive.
4066 * fixed several bugs in compiler checking of type declarations, i.e.
4067 violations of the Python "declarations are assertions" principle
4068 (thanks to Alexey Dejneka)
4069 * fixed several bugs in PCL's error checking (thanks to Gerd
4071 * fixed bug: printing of FILE-ERROR (thanks to Antonio
4073 * fixed bug in compilation of functions as first class values
4074 (thanks to Antonio Martinez-Shotton)
4075 * The compiler's handling TYPE-ERRORs which it can prove will
4076 inevitably happen at runtime has been cleaned up and corrected
4077 in several ways. (thanks to Alexey Dejneka)
4078 * improved argument type checking for various basic arithmetic
4079 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
4080 lavished on them in the past that they can be compiled in many
4081 ways in different special cases
4082 * fixed bug 181: compiler checks validity of user supplied type
4084 * cleaned up code flushing in optimization: Function calls which
4085 should signal errors for safety purposes (e.g. which ANSI says
4086 should signal errors when their arguments are of incorrect type)
4087 are no longer optimized away.
4088 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4089 * incremented fasl file version number, because changes in the
4090 implementation of sequence functions like COERCE caused
4091 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4092 in old inline expansions) to become undefined. (Actually these
4093 changes were later undone, so we might very well be binary
4094 compatible with 0.7.7 after all, but leaving the version number
4095 incremented seemed like the simplest and most conservative
4098 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4099 * An alpha-quality port to the parisc architecture running Linux,
4100 based on the old CMUCL backend, has been made. This, even more so
4101 than the other backends, should be considered still a work in
4102 progress; known problems include that the Linux kernel in 64-bit
4103 mode does not propagate the correct sigcontext structure to
4104 userspace, and consequently SBCL on a parisc64 kernel will not
4106 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4107 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4108 suppressed.) Also now that the compiler is looking at declarations
4109 in the environment, it checks optimization declarations as well,
4110 and suppresses inlining when (> DEBUG SPEED).
4111 * More fixes have been made to treatment of floating point exception
4112 treatment and other Unix signals. In particular, floating point
4113 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4114 * The detection and handling of control stack exhaustion (infinite
4115 or very deeply nested recursion) has changed. Stack exhaustion
4116 detection is now done by write-protecting pages at the OS level
4117 and applies at all optimization settings; when found, a
4118 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4119 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4120 be caught using IGNORE-ERRORS.
4121 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4122 correctly when transforming recursive LABELS functions to LETs.
4123 (thanks to Alexey Dejneka)
4124 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4125 that are names of constants or global variables.
4126 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4127 alien routines with docstrings.
4128 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4129 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4131 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4132 (thanks to Eric Marsden)
4133 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4134 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4135 * Bug fix: The SPARC backend can now compile functions involving
4136 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4137 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4138 to a functions expecting a PATHNAME-DESIGNATOR.
4139 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4140 lambda lists. (thanks to Alexey Dejneka)
4141 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4142 functions so that EXPORTing the name of the class doesn't cause
4143 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4144 to Antonio Martinez for reporting this)
4145 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4146 constant-folded without failing an assertion. (thanks to Einar
4147 Floystad Dorum for reporting this)
4148 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4149 types are now dealt with more correctly by the compiler.
4150 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4151 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4152 behaviour analogous to RENAME-FILE. This puts its behaviour more
4153 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4155 * The fasl file version number has changed again. (because of the
4156 bug fix involving the names of PCL MAKE-INSTANCE functions)
4158 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4159 * bug fix: Floating point exceptions are treated much more
4160 consistently on the x86/Linux and PPC/Linux platforms.
4161 * Array initialization with :INITIAL-ELEMENT is now much faster for
4162 cases when the compiler cannot open code the array creation, but
4163 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4164 array accesses have also seen a speed increase.
4165 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4166 specified. (This was at the root of some bad interactions between
4167 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4169 * bug fix: Internal error arguments for undefined functions are now
4170 computed correctly on the PPC/Linux platform.
4171 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4172 Raymond Toy's patch for CMU CL)
4173 * Support for the Solaris 9 operating environment has been included
4174 (thanks to Daniel Merritt)
4175 * A very ugly but hopefully complete draft of the missing FFI chapter
4176 of the manual has been created by reformatting the corresponding
4177 CMU CL manual chapter into (currently very ugly and incoherent)
4178 DocBook and bringing it up to date for SBCL behavior. Thus, the
4179 manual is now essentially complete, at least by my extreme
4180 once-and-only-once standards, whereby it's acceptable to refer to
4181 the doc strings of SB-EXT functions as the primary documentation.
4182 * The fasl file version number has changed again, due to cleanup of
4183 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4184 is no longer a static symbol.)
4186 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4187 * SBCL now builds with OpenMCL (version 0.12) as the
4188 cross-compilation host; also, more progress has been made toward
4189 bootstrapping under CLISP.
4190 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4192 * bug 158 fixed: The compiler can now deal with integer loop
4193 increments different from 1; fixing this turned out also to fix
4195 * bug 169 fixed: no more bogus warnings about using lexical bindings
4196 despite the presence of perfectly good SPECIAL declarations (thanks
4197 to David Lichteblau)
4198 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4199 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4200 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4202 * bug 180 fixed: Method combination specifications no longer ignore
4203 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4204 * bug fix: Structure type predicate functions now check their argument
4205 count as they should.
4206 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4207 correctly. (thanks to Pierre Mai)
4208 * minor incompatible change: The --noprogrammer option is deprecated
4209 in favor of the new --disable-debugger option, which is very similar.
4210 (The major difference is that it takes effect at a slightly different
4211 time at startup, causing handling of errors in --sysinit and
4212 --userinit files will be affected differently.) The
4213 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4214 been added to allow this functionality to be controlled from ordinary
4215 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4216 maintainers, who might want to run non-interactive scripts to
4217 build SBCL cores which will later be used interactively.)
4218 * minor incompatible change: The LOAD function no longer, when given
4219 a wild pathname to load, loads all files matching that pathname.
4220 Instead, an error of type FILE-ERROR is signalled.
4222 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4223 * bug 147 fixed: The compiler preserves its block link/count
4224 invariants more correctly now so that it doesn't crash. (thanks
4226 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4228 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4229 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4230 no-value-for-this-character as NIL (as specified by ANSI).
4231 * HOST-NAMESTRING on physical pathnames now returns a string that is
4232 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4233 (thanks to Christophe Rhodes)
4234 * The Alpha port handles icache flushing more correctly. (thanks to
4236 * More progress has been made toward bootstrapping under CLISP. (thanks
4237 to Christophe Rhodes)
4238 * The fasl file format has changed again, because dynamic loading
4239 on OpenBSD (which has non-ELF object files) motivated some cleanups
4240 in the way that foreign symbols are transformed and passed around.
4241 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4242 is no longer treated as whitespace by the reader, but instead as
4243 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4244 |A
\7fB|, instead of A as it used to.
4246 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4247 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4248 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4249 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4250 this as of 0.7.1.45, but was left out of the previous news section
4251 (thanks to Dan Barlow)
4252 * SBCL now runs on the Solaris operating system on SPARC architectures
4253 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4254 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4255 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4256 * SPARC backend cleanups, allowing builds of cores optimized for V8
4257 and V9 SPARCS, and also emission of code targeted to a particular
4258 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4260 * SBCL is closer to bootstrapping under CLISP, thanks to various
4261 fixes by Christophe Rhodes.
4262 * The fasl file format has changed again, to allow the compiler's
4263 INFO database to support symbol macros.
4264 * The user manual (in doc/) is formatted into HTML more nicely.
4265 (thanks to coreythomas)
4266 * The system is smarter about SUBTYPEP relationships, especially
4267 those involving NOT types (including types such as ATOM which are
4268 represented internally using NOT types). Thus SUBTYPEP is less
4269 likely to return (VALUES NIL NIL) in general, and in particular
4270 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4272 * The fasl file format has changed again, because the internal
4273 representation of types now includes a new slot to support the new
4274 SUBTYPEP-of-NOT-types logic.
4275 * (not a change in the main branch of SBCL, but a related prototype
4276 which can hopefully be merged into the main branch of SBCL in the
4277 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4278 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4279 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4280 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4281 dispatch macro character combinations. (thanks to Alexey Dejneka)
4282 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4283 that superficially look like logical namestrings correctly.
4284 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4286 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4287 default supplied-p) for &optional and &key arguments. (thanks to
4290 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4291 * incompatible change: The compiler is now less aggressive about
4292 tail call optimization, doing it only when (> SPACE DEBUG) or
4293 (> SPEED DEBUG). (This is an incompatible change because there are
4294 programs which relied on the old CMU-CL-style behavior to optimize
4295 away their unbounded recursion which will now die of stack overflow.)
4296 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4297 for non-GENCGC systems has been increased to 20M (since that
4298 seems much closer to the likely performance optimum for modern
4299 systems than the old 4M value was)
4300 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4301 and *DEBUG-PRINT-LEVEL*
4302 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4303 of CMU CL's support for SPARC, and various endianness and other
4304 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4305 * new syntactic sugar for the Unix command line: --load foo.bar is now
4306 an alternate notation for --eval '(load "foo.bar")'.
4308 ** The system now detects stack overflow and handles it gracefully,
4309 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4310 optimization settings. (This is a good thing in general, and
4311 its introduction in this version should be particularly timely
4312 for anyone whose code fails because of suppression of tail
4314 ** The system now hunts for the C variable "environ" in a more
4315 devious way, to avoid segfaults when the C library version differs
4316 between compile time and run time. (thanks to Christophe Rhodes)
4317 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4318 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4319 ** The compiler no longer issues bogus style warnings for undefined
4320 classes in the same source file as the DEFCLASSes which defined
4321 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4323 ** fixes in CONDITION class precedence list for undefined function
4324 errors (thanks to Alexei Dejneka)
4325 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4326 correctly. (thanks to Dan Barlow)
4327 ** portability fixes aiming at bootstrapping under CLISP (thanks
4328 to Dave McDonald and Christophe Rhodes)
4329 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4330 ** fixes in type translation and and type inference (thanks to
4332 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4333 ** various fixes in the new ports (thanks to Dan Barlow)
4334 * several changes related to debugging:
4335 ** suppression of tail recursion, as noted above
4336 ** stack overflow detection, as noted above
4337 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4338 is now the default. (For some time encapsulation has been more
4339 reliable than the breakpoint-based :ENCAPSULATE NIL
4340 implementation, at least on X86 systems; and I just noticed that
4341 encapsulation also seems closer to the spirit of the ANSI
4344 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4346 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4347 up properly again. (There was a packaging bug in 0.7.0 which
4348 left their definitions in SB-SYS::LOAD-FOREIGN and
4349 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4350 vital for most things which interface to C-level interfaces,
4351 like extensions working with sockets or databases or
4352 Perl-compatible regexes or whatever, and the need to fix
4353 this bug is the main reason that 0.7.1 was released so
4355 ** DEFGENERIC is now choosier about the methods it redefines, so that
4356 reLOADing a previously-LOADed file containing DEFGENERICs does
4357 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4358 cycle now works as it should. (thanks to Alexey Dejneka)
4359 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4360 fixed by Christophe Rhodes. (He actually submitted this patch
4361 months ago, and I delayed until after 0.7.0.)
4362 ** Bug 111 (internal compiler confusion about runtime checks on
4363 FUNCTION types) was fixed by Alexey Dejneka.
4364 * Some internal cleanups (getting rid of variables which aren't
4365 needed now that the byte interpreter is gone) caused the fasl
4366 file format number to change again.
4368 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4369 * major incompatible change: The default fasl file extension, i.e. the
4370 default extension for files produced by COMPILE-FILE, has changed
4371 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4373 ** There are many changes in the implementation of the compiler.
4374 SBCL is now essentially a compiler-only implementation of ANSI
4375 Common Lisp. EVAL still "interprets" a few special cases, but
4376 almost all the interesting cases are handled by creating
4377 a LAMBDA expression, calling COMPILE on it, then calling
4378 FUNCALL on the result.
4379 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4380 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4381 Since the code is newer, there might still be some new bugs
4382 (though not as many as before Martin Atzmueller's fixes:-). But
4383 the new code is substantially simpler and clearer, and hopefully
4384 any remaining bugs will be simpler, less fundamental, and more
4385 fixable then the bugs in the old code.
4386 ** The revised compiler is still a little unsteady on its feet.
4388 *** The debugging information it produces (particularly the names
4389 of FUNCTION objects) is sometimes much less useful than what
4390 the old compiler produced.
4391 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4392 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4393 within a MACROLET) has been temporarily weakened.
4394 ** There are new compiler optimizations for various functions:
4395 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4396 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4397 *** the math functions TRUNCATE, FLOOR, and CEILING
4398 *** the function-of-all-trades COERCE
4399 Mostly these should be transparent, but there's one
4400 potentially-annoying problem (bug 117): when the compiler
4401 inline-expands a function and does type analysis on the result,
4402 it can create control paths which have type mismatches, and
4403 when it can't prove that those control paths aren't taken,
4404 it will issue WARNINGs about the type mismatches. This is
4405 a particular problem in practice for the new sequence functions.
4406 It's not clear how this should be fixed, and for now, a
4407 workaround is given in the entry for 117 in the BUGS file.
4408 ** (Because of the interaction between the two previous items --
4409 occasional inlining problems and new inline expansions -- some
4410 of the new sequence function optimizations won't really kick in
4411 completely until debugging information, and then inlining, are
4412 straightened out in some future version.)
4413 * minor incompatible changes:
4414 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4415 DIRECTORY no longer implicitly promotes NIL slots of its
4416 pathname argument to :WILD. In particular, when you ask for the
4417 contents of a directory (which you used to be able to do without
4418 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4419 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4420 ** changes in behavior that ANSI explicitly defines to be
4421 implementation dependent:
4422 *** The new compiler-only implementation still conforms with ANSI,
4423 but acts a little different than before. Besides the obvious
4424 changes in performance tradeoffs (that the cost per form passed
4425 to EVAL has gone up, and the cost per form executed by EVAL
4426 has gone down), the behavior of the system changes a little
4427 because there are no longer any interpreted function objects.
4428 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4429 e.g. doing COMPILE on the output of interactive DEFUN is
4431 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4433 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4434 has changed from (:CL) to NIL.
4435 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4436 e.g. #\Newline and #\Tab, don't have names specified in the
4437 ANSI Common Lisp standard, is now based on their ASCII symbolic
4438 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4439 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4440 are no longer used for output.
4441 ** changes in internal implementation constants:
4442 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4443 4 million. (If your application spends a lot of time GCing and
4444 you have a lot of RAM, you might want to experiment with
4445 increasing it even more.)
4446 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4447 However, almost all old code should still continue to work without
4448 immediate update, as SB-C-CALL is now a (deprecated) nickname
4450 ** Old operator names in the style DEF-FOO are now deprecated in
4451 favor of new corresponding names DEFINE-FOO, for consistency with
4452 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4453 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4454 internal symbols, but a few supported extensions like
4455 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4456 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4457 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4458 etc. as you get deeper into recursive calls to the debugger
4459 command loop, instead of the old "5]", "5]]", "5]]]"
4460 sequence. (I was motivated to do this when squabbles between
4461 ILISP and SBCL left me very deeply nested in the debugger. In the
4462 short term, this change will probably provoke more ILISP/SBCL
4463 squabbles, but hopefully it will be an improvement in the long run.)
4464 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4465 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4466 function, since declaiming return types involving aliens is
4467 (1) annoyingly messy to do by hand and (2) vital to efficient
4468 compilation of code which calls such functions.
4469 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4470 longer reexported by the SB-EXT package. They're solely useful
4471 for alien code, so it seems more logical that you should get
4472 them from the SB-ALIEN package, not in SB-EXT.
4473 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4474 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4475 features. Instead, the code that they used to control is always
4476 built into the system.
4477 * many other bug fixes
4478 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4479 advantage of the new EVAL-WHEN stuff and to clean them up in
4480 general, and they are now more ANSI-compliant in a number of
4481 ways. Martin Atzmueller is responsible for a lot of this.
4482 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4484 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4485 *** correct ERROR type for various file operations
4486 *** some fixes for Lisp streams
4487 *** DEFMETHOD syntax checking
4488 *** changing old weird representation of debug information as
4489 strings (which, among their other deficiencies, don't transform
4490 correctly when you rename packages, and don't change their
4491 print representation when you change things like *PACKAGE*
4492 and *PRINT-LENGTH*) to symbols and lists of symbols
4493 He also made several improvements and fixed several bugs in DESCRIBE.
4494 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4496 *** misbehavior of WRITE-STRING/WRITE-LINE
4497 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4498 and several other LOOP problems as well
4499 *** DIRECTORY when similar filenames are present
4500 *** DEFGENERIC with :METHOD options
4501 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4502 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4503 *** argument ordering in FIND with :TEST option
4504 *** mishandled package designator argument in APROPOS-LIST
4505 *** various problems in the backquote readmacro
4506 *** a bug in APROPOS
4507 *** probably some others that I'm not describing very well here,
4508 since the CVS log documents them by reference to sbcl-devel
4509 messages, and the SourceForge archives aren't working well.:-(
4510 ** Dan Barlow improved the Alpha port (and is making progress on the
4511 PPC port, for those of you who think different).
4512 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4513 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4514 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4515 continues to work on the SPARC port (for those of you in a position
4516 to look down upon our little PC-compatible boxes from a great height).
4517 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4518 and uses that copy, rather than the current dynamic value, when
4519 it's trying to decide whether to truncate output. Thus e.g.
4520 (let ((*print-lines* 50))
4521 (pprint-logical-block (stream nil)
4523 (let ((*print-lines* 8))
4524 (print (aref possiblybigthings i) stream)))))
4525 should now truncate the logical block only at 50 lines, instead of
4526 often truncating it at 8 lines, as it did before.
4527 * The doc/cmucl/ directory, containing old CMU CL documentation
4528 from the time of the fork, is no longer part of the base system.
4529 SourceForge has shut down its anonymous FTP service, and with it
4530 my original plan for distributing the old CMU CL documentation
4531 there. For now, if you need these files you can download an old
4532 SBCL source release and extract them from it.
4533 * The fasl file version number changed again, for dozens of reasons,
4534 some of which are apparent above.
4536 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4537 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4538 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4539 patch, so that DISASSEMBLE output is much nicer.
4540 * The code in the SB-PROFILE package now seems reasonably stable.
4541 I still haven't decided what the final interface should look like
4542 (I'd like PROFILE to interact cleanly with TRACE, since both
4543 facilities use function encapsulation) but if you have a need
4544 for profiling now, you can probably use it successfully with
4545 the current CMU-CL-style interface.
4546 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4547 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4548 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4549 link now returns the dangling link itself, and for similar
4550 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4551 link itself. (In these cases the old code signalled an error and
4552 looped endlessly, respectively.) Thus, DIRECTORY now works even
4553 in the presence of dangling and cyclic symbolic links.
4554 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4555 is now a supported extension again, since the consensus on
4556 sbcl-devel was that it can be useful for ordinary development
4557 work, not just for debugging SBCL itself.
4558 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4559 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4560 that currently-defined functions might be redefined later with
4561 different return types.
4562 * Hash tables can be printed readably, as inspired by CMU CL code
4563 of Eric Marsden and SBCL code of Martin Atzmueller.
4564 * better error handling in CLOS method combination, thanks to
4565 Martin Atzmueller porting Pierre Mai's CMU CL patches
4566 * more overflow fixes for >16Mbyte I/O buffers
4567 * A bug in READ has been fixed, so that now a single Ctrl-D
4568 character suffices to cause end-of-file on character streams.
4569 In particular, now you only need one Ctrl-D at the command
4570 line (not two) to exit SBCL.
4571 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4573 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4574 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4575 does the right thing.
4576 * The implementation of some type tests, especially for CONDITION
4577 types, is now tidier and maybe faster, due to CMU CL code
4578 originally by Douglas Crosher, ported by Martin Atzmueller.
4579 * Some math functions have been fixed, and there are new
4580 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4581 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4582 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4583 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4584 likely to be built into sbcl-0.7.0.)
4585 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4586 CL, ported to SBCL by Martin Atzmueller)
4587 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4588 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4589 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4590 * There's a new slam.sh hack to shorten the edit/compile/debug
4591 cycle for low-level changes to SBCL itself, and a new
4592 :SB-AFTER-XC-CORE target feature to control the generation of
4593 the after-xc.core file needed by slam.sh.
4594 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4595 COMPILE-FILE is no longer supported, so that now every function
4596 gets an entry point, so that block compilation looks a little
4597 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4598 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4599 returns the number of bytes consed since the system started,
4600 rather than the number consed since the first time the function
4601 was called. (The new definition parallels ANSI functions like
4602 CL:GET-INTERNAL-RUN-TIME.)
4603 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4604 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4605 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4606 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4607 * Fasl file version numbers are now independent of the target CPU,
4608 since historically most system changes which required version
4609 number changes have affected all CPUs equally. Similarly,
4610 the byte fasl file version is now equal to the ordinary
4613 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4614 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4615 is no longer recognized. I apologize for this, because it was
4616 listed in SB-EXT as a supported extension, but I found that
4617 its existing behavior was poorly specified, as well as incorrectly
4618 specified, and it looked like too much of a mess to straighten it
4619 out. I have enough on my hands trying to get ANSI stuff to work..
4620 * many patches ported from CMU CL by Martin Atzmueller, with
4621 half a dozen bug fixes in pretty-printing and the debugger, and
4622 half a dozen others elsewhere
4623 * fixed bug 13: Floating point infinities are now supported again.
4624 They might still be a little bit flaky, but thanks to bug reports
4625 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4626 as flaky as they were.
4627 * The --noprogrammer command line option is now supported. (Its
4628 behavior is slightly different in detail from what the old man
4629 page claimed it would do, but it's still appropriate under the
4630 same circumstances that the man page talks about.)
4631 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4632 are now supported, and enabled by default. Thus, the compiler can
4633 handle many floating point and complex operations much less
4634 inefficiently. (Thus e.g. you can implement a complex FFT
4636 * The compiler now detects type mismatches between DECLAIM FTYPE
4637 and DEFUN better, and implements CHECK-TYPE more correctly, and
4638 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4639 thanks to patches from Martin Atzmueller.
4640 * various fixes to make the cross-compiler more portable to
4641 ANSI-conforming-but-different cross-compilation hosts (notably
4642 Lispworks for Windows, following bug reports from Arthur Lemmens)
4643 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4644 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4645 fixed thanks to Pierre Mai's CMU CL patches.
4646 * fixes to keep the system from overflowing internal counters when
4647 it tries to use i/o buffers larger than 16M bytes
4648 * fixed bug 45a: Various internal functions required to support
4649 complex special functions have been merged from CMU CL sources.
4650 (When I was first setting up SBCL, I misunderstood a compile-time
4651 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4652 * improved support for type intersection and union, fixing bug 12
4653 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4654 more obscure bugs as well
4655 * some steps toward byte-compiling non-performance-critical
4656 parts of the system, courtesy of patches from Martin Atzmueller
4657 * Christophe Rhodes has made some debian packages of sbcl at
4658 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4659 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4660 stable, but are nonetheless usable. When he's ready, I'd be happy
4661 to add them to the SourceForge "File Releases" section. (And if
4662 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4663 * new fasl file format version number (because of changes in
4664 internal representation of (OR ..) types to accommodate the new
4665 support for (AND ..) types, among other things)
4667 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4668 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4670 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4672 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4673 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4674 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4675 * DESCRIBE now works on structure objects again.
4676 * Most function call argument type mismatches are now handled as
4677 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4678 know whether the function will be redefined before the call is
4679 executed. (The compiler could flag local calls with full WARNINGs,
4680 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4681 it doesn't keep track of enough information to know whether calls
4682 are local in this sense.)
4683 * Compiler output is now more verbose, with messages truncated
4684 later than before. (There should be some supported way for users
4685 to override the default verbosity, but I haven't decided how to
4686 provide it yet, so this behavior is still controlled by the internal
4687 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4688 src/compiler/ir1util.lisp.)
4689 * Fasl file format version numbers have increased again, because
4690 support for the Gray streams extension changes the layout of the
4691 system's STREAM objects.
4692 * The Gray subclassable streams extension now works, thanks to a
4693 patch from Martin Atzmueller.
4694 * The full LOAD-FOREIGN extension (not just the primitive
4695 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4696 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4697 but like most other programs, it defaults to copying the Unix
4698 environment from the original process instead of starting the
4699 new process in an empty environment.
4700 * Extensions which manipulate the Unix environment now support
4701 an :ENVIRONMENT keyword option which doesn't smash case or
4702 do other bad things. The CMU-CL-style :ENV option is retained
4703 for porting convenience.
4704 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4705 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4707 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4709 * A patch from Martin Atzmueller seems to have solved the SIGINT
4710 problem, and as far as we know, signal-handling now works cleanly.
4711 (If you find any new bugs, please report them!)
4712 * The system no longer defaults Lisp source file names to types
4713 ".l", ".cl", or ".lsp", but only to ".lisp".
4714 * The compiler no longer uses special default file extensions for
4715 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4716 seems to expect a single default extension for all compiled code,
4717 and there's no compelling reason to try to stretch the standard
4718 to allow two different extensions.) Instead, byte-compiled files
4719 default to the same extension as native-compiled files.
4720 * Fasl file format version numbers have increased again, because
4721 a rearrangement of internal implementation packages made some
4722 dumped symbols in old fasl files unreadable in new cores.
4723 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4724 many fewer weird special cases.
4725 * Bug #17 (differing COMPILE-FILE behavior between logical and
4726 physical pathnames) has been fixed, and some related misbehavior too,
4727 thanks to a patch from Martin Atzmueller.
4728 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4729 by Tim Moore, ported to SBCL by Martin Atzmueller.
4730 * Martin Atzmueller fixed several filesystem-related problems,
4731 including bug #36, in part by porting CMU CL patches, which were
4732 written in part by Paul Werkowski.
4733 * More compiler warnings in src/runtime/ are gone, thanks to
4734 more patches from Martin Atzmueller.
4735 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4738 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4740 * DESCRIBE now works on CONDITION objects.
4741 * The debugger now handles errors which arise when trying to print
4742 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4744 * The build system now uses an additional file, customize-target-features.lisp,
4745 to allow local modifications to the target *FEATURES* list. (The point of
4746 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4747 debugging features enabled, without having to worry about propagating it
4748 into everyone's system when I do a "cvs update".) When no
4749 customize-target-features.lisp file exists, the target *FEATURES* list
4750 should be constructed the same way as before.
4751 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4752 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4753 warn about behavior which is undefined under the ANSI spec. Note: This
4754 is specified by ANSI, but it's not very popular with programmers.
4755 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4756 macro in the SBCL sources for an example of a workaround which you
4757 might use to make portable ANSI-standard code which does what you want.
4758 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4759 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4760 when it's used as a non-toplevel form. (This is required in order
4761 to implement the DEFCONSTANT-EQX macro.)
4762 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4763 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4764 funny things, due to bugs in the implementation of EVAL-WHEN
4765 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4766 non-toplevel DEFCONSTANTs are uncommon.)
4767 * The core file version number and fasl file version number have been
4768 incremented, because the old noncompliant DEFCONSTANT behavior involved
4769 calling functions which no longer exist, and because I also took the
4770 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4771 * fixed bug 1 (error handling before read-eval-print loop starts), and
4772 redid debugger restarts and related debugger commands somewhat while
4774 ** The QUIT debugger command is gone, since it did something
4775 rather different than the SB-EXT:QUIT command, and since it never
4776 worked properly outside the main toplevel read/eval/print loop.
4777 Invoking the new TOPLEVEL restart provides the same functionality.
4778 ** The GO debugger command is also gone, since you can just invoke
4779 the CONTINUE restart directly instead.
4780 ** The TOP debugger command is also gone, since it's redundant with the
4781 FRAME 0 command, and since it interfered with abbreviations for the
4783 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4785 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4786 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4787 some porting work by Martin Atzmueller)
4788 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4789 with a boatload of patches which he ported to SBCL. Now that those
4791 ** The system tries to make sure that its low-priority messages
4792 are prefixed by semicolons, to help people who like to use
4793 syntax highlighting in their ILISP buffer. (This patch
4794 was originally due to Raymond Toy.)
4795 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4796 patches originally written by Raymond Toy.
4797 ** The compiler understands coercion between single-value and
4798 multiple-VALUES type expressions better, getting rid of some very
4799 weird behavior, thanks to patches originally by Robert MacLachlan
4800 and Douglas Crosher.
4801 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4802 lambda lists, thanks to a patch originally by Pierre Mai.
4803 ** The system no longer bogusly warns about "abbreviated type
4805 ** The compiler gets less confused by inlining and RETURN-FROM,
4806 thanks to some patches originally by Tim Moore.
4807 ** The system no longer hangs when dumping circular lists to fasl
4808 files, thanks to a patch originally from Douglas Crosher.
4809 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4810 undefined function error.
4811 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4812 in structure slot initforms) for now, documented workaround instead:-|
4813 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4814 * fixed bug 5: added stubs for various Gray stream functions called
4815 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4816 installed, at least appropriate type errors are generated
4817 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4818 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4819 consistently in DEFMETHOD forms.
4820 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4821 it doesn't seem to affect SBCL after all
4822 * The C runtime system now builds with better optimization and many
4823 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4825 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4827 * The system is now under CVS at SourceForge (instead of the
4828 CVS repository on my home machine).
4829 * The new signal handling code has been tweaked to treat register
4830 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4831 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4832 downstream, e.g. in debug-int.lisp, has implicit dependencies
4833 on the unsignedness of integer representation of machine words,
4834 and that caused the system to bomb out with infinite regress
4835 when trying to recover from type errors involving signed values,
4836 e.g. (BUTLAST '(1 2 3) -1).)
4837 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4838 (This was one of the bugs Peter Van Eynde reported back in July.)
4839 * The system now uses code inspired by Colin Walters' O(N)
4840 implementation of MAP (from the cmucl-imp@cons.org mailing
4841 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4842 inline the MAP operation, and there is more than one
4843 sequence argument to the MAP call (so that it can't just
4844 do ETYPECASE once and for all based on the type of the
4845 single sequence argument). (The old non-inline implementation
4846 of the general M-argument sequence-of-length-N case required
4847 O(M*N*N) time when any of the sequence arguments were LISTs.)
4848 * The QUIT :UNIX-CODE keyword argument has been renamed to
4849 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4850 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4852 * Raymond Wiker's patches to port dynamic loading from Linux to
4853 FreeBSD have been added.
4854 * The BUGS file is now more nearly up to date, thanks in large part
4855 to Martin Atzmueller's review of it.
4856 * The debugger now flushes standard output streams before it begins
4857 its output ("debugger invoked" and so forth).
4858 * The core version number and fasl file version number have both
4859 been incremented, because of incompatible changes in the layout
4861 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4862 used to not be called for a saved Lisp image.)
4863 * Martin Atzmueller's version of a patch to fix a compiler crash,
4864 as posted on sbcl-devel 13 September 2000, has been installed.
4865 * Instead of installing Martin Atzmueller's patch for the
4866 compiler transform for SUBSEQ, I deleted the compiler transform,
4867 and transforms for some similar consing operations.
4868 * A bug in signal handling which kept TRACE from working on OpenBSD
4870 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4871 away by constant folding
4872 * The system now defines its address space constants in one place
4873 (in the Lisp sources), and propagates them automatically elsewhere
4874 (through GENESIS and the sbcl.h file). Therefore, patching the
4875 address map is less unnecessarily tedious and error-prone. The
4876 Lisp names of address space constants have also been systematized.
4877 * CVS tags like dollar-Header-dollar have been removed from
4878 the sources, because they have never saved me trouble and
4879 they've been source of trouble working with patches and other
4880 diff-related operations.
4881 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4882 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4884 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4886 * The system has been ported to OpenBSD.
4887 * The system now compiles with a simple "sh make.sh" on the systems
4888 that it's supported on. I.e., now you no longer need to tweak
4889 text in the target-features.lisp-expr and symlinks in src/runtime/
4890 by hand, the make.sh takes care of it for you.
4891 * The system is no longer so grossly inefficient when compiling code
4892 involving vectors implemented as general (not simple) vectors (VECTOR T),
4893 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4894 which dares to use the various sequence functions on non-simple
4895 vectors, takes less of a performance hit.
4896 * There is now a primitive type predicate VECTOR-T-P
4897 to test for the (VECTOR T) type, so that e.g.
4898 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4899 can now be compiled with some semblance of efficiency. (The old code
4900 turned the type declaration into a full call to %TYPEP at runtime!)
4901 * AREF on (VECTOR T) is still not fast, since it's still compiled
4902 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4903 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4904 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4905 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4906 without making full calls to SUBTYPEP at runtime.
4907 (Some analogous efficiency issues for non-simple vectors specialized to
4908 element types other than T, or for non-simple multidimensional arrays,
4909 have not been addressed. They could almost certainly be handled the
4910 same way if anyone is motivated to do so.)
4911 * The changes in array handling break binary compatibility, so
4912 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4913 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4914 instead of (VALUES T T).
4915 * By following the instructions that Dan Barlow posted to sbcl-devel
4916 on 2 July 2000, I was able to enable primitive dynamic object
4917 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4918 functionality is not implemented (since it calls ld to resolve
4919 library references automatically, requiring RUN-PROGRAM for its
4920 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4921 try to resolve library references) is now supported.
4922 * The system now flushes the standard output streams when it terminates,
4923 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4924 them at several other probably-convenient times, e.g. in each pass of
4925 the toplevel read-eval-print loop, and after evaluating a form given
4926 as an "--eval" command-line option. (These changes were motivated by a
4927 discussion of stream flushing issues on cmucl-imp in August 2000.)
4928 * The source transform for TYPEP of array types no longer assumes
4929 that an array whose element type is a not-yet-defined type
4930 is implemented as an array of T, but instead punts, so that the
4931 type will be interpreted at runtime.
4932 * There is now some support for cross-compiling in make.sh: each of
4933 the phases of make.sh has its own script. (This should be transparent
4934 to people doing ordinary, non-cross-compile builds.)
4935 * Since my laptop doesn't have hundreds of megabytes of memory like
4936 my desktop machine, I became more motivated to do some items on
4937 my to-do list in order to reduce the size of the system a little:
4938 ** Arrange for various needed-only-at-cold-init things to be
4939 uninterned after cold init. To support this, those things have
4940 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4941 symbols with such names are now uninterned after cold init).
4942 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4943 fasl files for cold load.
4944 ** Remove the old compiler structure pooling code (which used to
4945 be conditional on the target feature :SB-ALLOC) completely.
4946 ** Redo the representation of some data in cold init to be more compact.
4947 (I also looked into supporting byte compiled code at bootstrap time,
4948 which would probably reduce the size of the system a lot, but that
4949 looked too complicated, so I punted for now.)
4950 * The maximum signal nesting depth in the src/runtime/ support code has
4951 been reduced from 4096 to 256. (I don't know any reason for the very
4952 large old value. If the new smaller value turns out to break something,
4953 I'll probably just bump it back up.)
4954 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4956 * Many, many bugs reported by Peter Van Eynde have been added to
4957 the BUGS list; some have even been fixed.
4958 * While enabling dynamic object file loading, I tried to make the
4959 code easier to understand, renaming various functions and variables
4960 with less ambiguous names, and changing some function calling
4961 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4962 * While trying to figure out how to do the OpenBSD port, I tried to
4963 clean up some of the code in src/runtime/. In particular, I dropped
4964 support for non-POSIX signal handling, added various comments,
4965 tweaked the code to reduce the number of compilation warnings, and
4966 renamed some files to increase consistency.
4967 * To support the new automatic configuration functionality in make.sh,
4968 the source file target-features.lisp-expr has been replaced with the
4969 source file base-target-features.lisp-expr and the machine-generated
4970 file local-target-features.lisp-expr.
4971 * fixed a stupid quoting error in make.sh so that using CMU CL
4972 "lisp -batch" as cross-compilation host works again
4974 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4976 * DESCRIBE no longer tries to call itself recursively to describe
4977 bound/fbound values, so that it no longer fails on symbols which are
4978 bound to themselves (like keywords, T, and NIL).
4979 * DESCRIBE now works on generic functions.
4980 * The printer now prints less-screwed-up representations of closures
4981 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4982 * A private symbol is used instead of the :EMPTY keyword previously
4983 used to mark empty slots in hash tables. Thus
4984 (DEFVAR *HT* (MAKE-HASH-TABLE))
4985 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4986 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4987 now does what ANSI says that it should. (You can still get
4988 similar noncompliant behavior if bang on the hash table
4989 implementation with all the symbols you get back from
4990 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4991 This breaks binary compatibility, since tests for equality to
4992 :EMPTY are wired into things like the macroexpansion of
4993 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4995 * There's now a minimal placeholder implementation for CL:STEP,
4996 as required by ANSI.
4997 * An obscure bug in the interaction of the normal compiler, the byte
4998 compiler, inlining, and structure predicates has been patched
4999 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
5000 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
5002 * Missing ordinary arguments in a macro call are now detected even
5003 when the macro lambda list contains &KEY or &REST.
5004 * The debugger no longer complains about encountering the top of the
5005 stack when you type "FRAME 0" to explicitly instruct it to go to
5006 the top of the stack. And it now prints the frame you request even
5007 if it's the current frame (instead of saying "You are here.").
5008 * As specified by ANSI, the system now always prints keywords
5009 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
5011 * The default initial SIZE of HASH-TABLEs is now smaller.
5012 * Type information from CLOS class dispatch is now propagated
5013 into DEFMETHOD bodies, so that e.g.
5014 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5016 is now basically equivalent to
5017 (DEFMETHOD FOO ((X SINGLE-FLOAT))
5018 (DECLARE (TYPE SINGLE-FLOAT X))
5020 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
5021 operation, without having to do run-time type dispatch.
5022 * The macroexpansion of DEFMETHOD has been tweaked so that it has
5023 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
5024 * Since I don't seem to be making big file reorganizations very often
5025 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
5026 is overflowing my ability to conveniently back them up), I've finally
5027 checked the system into CVS. (The CVS repository is on my home system,
5028 not at SourceForge -- putting it on SourceForge might come later.)
5029 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
5030 high-level GC-NOTIFY-FOO functions send their output. (There's
5031 still very little control of where low-level verbose GC functions
5032 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
5033 less than it used to -- the GC-NOTIFY-FOO functions are now under
5034 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
5035 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
5036 in only one place in the source code, and propagates it automatically
5037 everywhere that it's needed. Thus e.g. when I bump the version from
5038 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
5039 * The C source files now include boilerplate legalese and documentation
5040 at the head of each file (just as the Lisp source files already did).
5041 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
5042 to his page will be replaced with a link to his new CLiki service.
5044 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
5046 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
5047 * The build process now looks for GNU make under the default name "gmake",
5048 instead of "make" as it used to. If GNU make is not available as "gmake"
5049 on your system, you can change this default behavior by setting the
5050 GNUMAKE environment variable.
5051 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
5052 can build without error under CMU CL.
5054 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
5056 * There is now a partial SBCL user manual (with some new text and some
5057 text cribbed from the CMU CL manual).
5058 * The beginnings of a profiler have been added (starting with the
5059 CMU CL profiler and simplifying and cleaning up). Eventually the
5060 main interface should be through the TRACE macro, but for now,
5061 it's still accessed through vaguely CMU-CL-style functions and macros
5062 exported from the package SB-PROFILE.
5063 * Some problems left over from porting CMU CL to the new
5064 cross-compilation bootstrap process have been cleaned up:
5065 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
5066 instead of SB!XC:DEFMACRO, compounded by an oversight on my
5067 part when getting rid of the compiler *BACKEND* stuff.)
5068 ** The value of *NULL-TYPE* was screwed up, because it was
5069 being initialized before the type system knew the final
5070 definition of the 'NULL type. This screwed up several key
5071 optimizations in the compiler, causing inefficiency in all sorts
5072 of places. (I found it because I wanted to understand why
5073 GET-INTERNAL-RUN-TIME was consing.)
5074 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
5075 PROCLAIM FTYPE information. Unfortunately this broke binary
5076 compatibility again, since now the forms output by DEFGENERIC
5077 to refer to functions which didn't exist in 0.6.3.
5078 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
5079 can use the new (as of 0.6.3) transform for SOME into MAP into
5081 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
5082 values to (INTEGER 0 1000000), so that the time code will no longer
5083 occasionally get blown up by Linux returning 1000000 microseconds
5084 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
5085 its output conform to the ANSI spec. (Alas, this makes its output
5086 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
5087 * A full call to MAP NIL with a single sequence argument no longer conses.
5088 * fixes to problems pointed out by Martin Atzmueller:
5089 * The manual page no longer talks about multiprocessing as though
5090 it were currently supported.
5091 * The ILISP support patches have been removed from the distribution,
5092 because as of version 5.10.1, ILISP now supports SBCL without us
5093 having to maintain patches.
5094 * added a modified version of Raymond Toy's recent CMU CL patch for
5095 EQUALP comparison of HASH-TABLE
5097 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5099 * The system still can't cross-compile itself with
5100 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5101 entails), but at least it can compile more of itself that way
5102 than it used to be able to, and various buglets which were uncovered
5103 by trying to cross-compile itself that way have now been fixed.
5104 * This release breaks binary compatibility again. This time
5105 at least I've incremented the FASL file format version to 2, so that the
5106 problem can be detected reliably instead of just causing weird errors.
5107 * various new style warnings:
5108 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5109 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5110 ** using the lexical binding of a variable named in the *FOO* style
5111 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5112 as specified by ANSI.
5113 * *RANDOM-STATE* is no longer automatically initialized from
5114 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5115 default behavior of the system is to repeat its behavior every time
5116 it's run. If you'd like to change this behavior, you can always
5117 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5118 old convention there was no comparably easy way to get the system to
5119 repeat its behavior every time it was run.
5120 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5121 FTYPE declarations has been removed, in favor of their ANSI
5122 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5123 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5124 the types of their sequence arguments can be determined at compile time.
5125 This is done through a new open code expansion for MAP which eliminates
5126 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5127 argument types can be determined at compile time.
5128 * The optimizer now transforms COERCE into an identity operation when it
5129 can prove that the coerced object is already of the correct type. (This
5130 can be a win for machine generated code, including the output of other
5131 optimization transforms, such as the MAP transform above.)
5132 * Credit information has been moved from source file headers into CREDITS.
5133 * Source file headers have been made more standard.
5134 * The CASE macro now compiles without complaining even when it has
5137 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5139 * (Note that the way that the PCL macroexpansions were rewritten
5140 to accommodate the change in DEFGENERIC below breaks binary
5141 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5142 not run under sbcl-0.6.2. Once we get out of alpha releases,
5143 i.e. hit release 1.0.0, we'll probably try to maintain binary
5144 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5145 and sbcl-1.4.4. Until then, however, it might be fairly common
5146 for maintenance releases to break binary compatibility.)
5147 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5149 * The reporting of unhandled signals has been changed to print some
5150 explanatory text as well as the report form. (Previously only
5151 the report form was printed.)
5152 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5153 it defines, so that the compiler no longer issues undefined function
5154 warnings for compiled-but-not-yet-loaded generic functions.
5155 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5156 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5157 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5158 section of the ANSI spec are supported.
5159 * The "" nickname for the "KEYWORD" package has been removed.
5160 The reader still handles symbol tokens which begin with a package marker
5161 as keywords, but it doesn't expose its mechanism for doing so in the
5162 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5163 * The system now issues STYLE-WARNINGs for contradictory TYPE
5164 proclamations. (Warnings for contradictory FTYPE proclamations would
5165 be nice too, but those can't be done usefully unless the type system
5166 is made smarter about FUNCTION types.)
5167 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5168 systematized, so that "*target-*.lisp is supposed to exist only on the
5169 target and imply that there's a related file which exists on the
5170 host, and *host-*.lisp is supposed to exist only on the host and imply
5171 that there's a related file which exists on the target. This involves a
5172 lot of renaming. Hopefully the acute confusion caused by the renaming
5173 will be justified by the reduction in chronic confusion..
5174 ** runtime-type.lisp -> early-target-type.lisp
5175 ** target-type.lisp -> late-target-type.lisp
5176 ** early-host-format.lisp -> early-format.lisp
5177 ** late-host-format.lisp -> late-format.lisp
5178 ** host-error.lisp -> misc-error.lisp
5179 ** early-error.lisp -> early-target-error.lisp
5180 ** late-error.lisp -> late-target-error.lisp
5181 ** host-defboot.lisp -> early-defboot.lisp
5182 ** code/misc.lisp -> code/target-misc.lisp
5183 ** code/host-misc.lisp -> code/misc.lisp
5184 ** code/numbers.lisp -> code/target-numbers.lisp
5185 ** code/early-numbers.lisp -> numbers.lisp
5186 ** early-host-type.lisp -> early-type.lisp
5187 ** late-host-type.lisp -> late-type.lisp
5188 ** host-typep.lisp -> typep.lisp
5189 ** load.lisp -> target-load.lisp
5190 ** host-load.lisp -> load.lisp
5191 ** host-disassem.lisp -> disassem.lisp
5192 ** host-insts.lisp -> insts.lisp
5193 ** byte-comp.lisp -> target-byte-comp.lisp
5194 ** host-byte-comp.lisp -> byte-comp.lisp
5195 ** host-signal.lisp -> signal.lisp
5196 ** host-defstruct.lisp -> defstruct.lisp
5197 ** late-target-type.lisp -> deftypes-for-target.lisp
5198 Furthermore, several other previously target-only files foo.lisp (e.g.
5199 hash-table.lisp and random.lisp) have been split into a target-and-host
5200 foo.lisp file and a target-only target-foo.lisp file, with their key type
5201 definitions in the target-and-host part, so that the cross-compiler will
5202 know more about target types.
5203 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5204 gone away. In their place are various *BACKEND-FOO* variables
5205 corresponding to the slots of the old structure.
5206 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5207 into the target SBCL has been fixed.
5208 * The chill.lisp system for loading cold code into a running SBCL
5210 * Support for the CMU CL "scavenger hook" extension has been removed.
5211 (It was undocumented and unused in the CMU CL sources that SBCL was
5212 derived from, and stale in sbcl-0.6.1.)
5213 * Various errors in the cross-compiler type system were detected
5214 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5215 (enabling various consistency checks). Many of them were fixed,
5216 but some hard problems remain, so the compiler is back to
5217 running without *TYPE-SYSTEM-INITIALIZED* for now.
5218 * As part of the cross-compiler type system cleanup, I implemented
5219 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5220 * I have started adding UNCROSS calls throughout the type system
5221 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5222 on cross-compiler input files will be able to go away, and various
5224 * CONSTANTP now returns true for quoted forms (as explicitly required
5227 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5229 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5230 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5231 (MAKE-LIST -1). In the longer term, it ought to become true
5232 that declarations are assertions even at SAFETY 1. For now, it's not
5233 quite true even at SAFETY 3, but it's at least more nearly true..
5234 (Note that this change seems to increases the size of the system by
5235 O(5%) and to decrease the speed of the compiler by 20% or more.)
5236 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5237 for the problem of printing dozens of lines of distracting information
5238 about low-level system machinery as part of the top stack frame
5239 on entry to the debugger when an undefined function was called.
5240 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5241 is not reset to COMMON-LISP-USER.
5242 * Compilation of stuff related to dyncount.lisp has been made conditional
5243 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5244 smaller. The various dyncount-related symbols have been moved into
5245 a new "SB-DYNCOUNT" package.
5246 * tty-inspect.lisp has been renamed to inspect.lisp.
5247 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5248 feature has gone away. (When we eventually port to other flavors of
5249 libc and/or Unix, we'll try to make the differences between flavors
5250 invisible at the user level.)
5251 * Various other *FEATURES* tags, and/or their associated conditionals,
5252 have been removed if obsolescent, or given better documentation, or
5253 sometimes given more-mnemonic names.
5255 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5257 * tidied up "make.sh" script
5258 * tidied up system directory structure
5259 * better "clean.sh" behavior
5260 * added doc/FOR-CMUCL-DEVELOPERS
5261 * many many small tweaks to output format, e.g. removing possibly-confusing
5262 trailing #\. character in DESCRIBE-INSTANCE
5263 * (EQUALP #\A 'A) no longer signals an error.
5264 * new hashing code, including EQUALP hashing
5265 * tidied up Lisp initialization and toplevel
5266 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5267 * command line argument processing
5268 * added POSIX-GETENV function to deal with Unix-ish environment variables
5269 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5270 terminating SBCL on EOF
5271 * non-verbose GC by default
5272 * There is no more "sbcl" shell script; the sbcl file is now the C
5273 runtime executable (just like CMU CL).
5274 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5276 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5277 symbol and package information as Lisp native symbol and package objects
5278 instead of strings naming symbols and strings naming packages. This way,
5279 whenever packages are renamed (as in warm init), debug information is
5280 transformed along with everything else.
5281 * tweaked the optimization policy declarations which control the building
5282 of SBCL itself. Now, among other things, the system no longer saves
5283 source location debugging information. (This helps two problems at once
5284 by reducing SBCL size and by keeping SBCL from trying to look for its
5285 sources -- which may not exist -- when reporting errors.)
5286 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5287 debugging and testing purposes
5288 * cleaned up printing, making the printer call PRINT-OBJECT for
5289 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5290 methods, giving nearly-ANSI behavior
5291 * converted almost all special variables to use *FOO* naming convention
5292 * deleted PARSE-TIME functionality, since it can be done portably
5293 * moved some files out of cold init into warm init
5294 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5296 * regularized formatting of source files
5297 * added an install.sh script
5298 * fixed ridiculous memory usage of cross-compiler by making
5299 compiler/alloc.lisp not try to do pooling unless it can hook
5300 itself into the GC of the cross-compilation host. Now the system
5301 builds nicely on my old laptop.
5302 * added :SB-ALLOC in target-features.lisp-expr
5303 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5304 was not implemented)
5305 * re-did condition handling and note reporting in the compiler. Notes
5306 are no longer handled by signalling conditions. Style warnings
5307 and warnings are handled more correctly and reported in such a way
5308 that it's easy to find one or the other in your output (so that you
5309 can e.g. figure out which of many problems caused COMPILE-FILE to
5311 * changed the severity of several compiler warnings from full WARNING
5312 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5313 compiler note reporting so that it doesn't use the condition system
5314 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5315 COMPILE-FILE command)
5316 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5317 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5318 no longer supported, since their implementation was deeply intertwingled
5319 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5320 * removed bogus "support" for compiler macros named (SETF FOO), and
5321 removed the compiler macro for SETF INFO (but only after making a fool
5322 of myself on the cmucl-imp mailing list by posting a bogus patch for
5323 DEFINE-COMPILER-MACRO..)
5324 * Compiled files containing forms which have side effects on the Lisp
5325 reader (such as DEFPACKAGE forms) are now handled more correctly.
5326 (Compiler queuing of top level lambdas has been suppressed by setting
5327 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5328 * deleted various currently-unused source files, e.g. gengc.lisp. They
5329 may be added back at some point e.g. when porting to other architectures,
5330 but until they are it's distracting to distribute them and to try to
5332 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5333 there were so many of them they're just distractions, and UNCROSS is
5334 known to be able to handle the current sources
5335 * moved PROFILE functionality into TRACE, so that it will be clear
5336 how the wrapping and unwrapping of functions when you profile them
5337 interacts with the wrapping and unwrapping of functions when you
5338 trace them. (Actually, the functionality isn't there yet, but at least
5339 the interface specification is there. Hopefully, the functionality will
5340 arrive with some future maintenance release.)
5341 * removed host-oops.lisp
5342 * changed signature of QUIT function to allow UNIX-CODE argument
5343 * fixed READ-SEQUENCE bug
5344 * tweaked verbose GC output so that it looks more like the progress
5345 output that ANSI specifies for functions like LOAD
5346 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5347 * added <http://sbcl.sourceforge.com> to the banner information printed by