1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 * minor incompatible change: SB-THREAD:JOIN-THREAD-ERROR-THREAD and
3 SB-THREAD:INTERRUPT-THREAD-ERROR-THREAD have been deprecated in favor
4 of SB-THREAD:THREAD-ERROR-THREAD.
5 * new contrib module: SB-QUEUE provides thread-safe lockless FIFO queues.
6 * new feature: docstrings for local and anonymous functions are no longer
7 discarded. (thanks to Leslie Polzer)
8 * new feature: SB-THREAD:SYMBOL-VALUE-IN-THREAD provides access to symbol
9 values in other threads.
10 * new feature: SB-INTROSPECT:ALLOCATION-INFORMATION provides information
11 about object allocation.
12 * optimization: more efficient type-checks for FIXNUMs when the value
13 is known to be a signed word on x86 and x86-64.
14 * optimization: compiler now optimizes (EXPT -1 INTEGER), (EXPT -1.0 INTEGER),
15 and (EXPT -1.0d0 INTEGER) into an ODDP test. (thanks to Stas Boukarev and
17 * improvement: DESCRIBE output has been reworked to be easier to read and
18 contains more pertinent information.
19 * improvement: failure to provide requested stack allocation compiler notes
20 provided in all cases (requested stack allocation not happening without a
21 note being issued is now considered a bug.)
22 * optimization: compiler is smarter about delegating argument type checks to
24 * bug fix: on 64 bit platforms FILL worked incorrectly on arrays with
25 upgraded element type (COMPLEX SINGLE-FLOAT), regression from 1.0.28.55.
26 (thanks to Paul Khuong)
27 * bug fix: looping around HANDLER-CASE could silently consume stack space
28 on each iteration. (reported by "foobar")
29 * bug fix: better error signalling when calls to functions seeking elements
30 from lists (eg. ADJOIN) are compiled with both :TEST and :TEST-NOT.
31 (reported by Tobias Rittweiler)
32 * bug fix: regressions in DIRECTORY from 1.0.28.61: pattern matching of
33 directory components now works as it used to. (various prolems reported by
34 Michael Becker, Gabriel Dos Reis, Cyrus Harmon, and Harald Hanche-Olsen)
35 * bug fix: :PTY option in RUN-PROGRAM was broken with stream arguments.
36 (reported by Elliot Slaughter, thanks to Stas Boukarev)
37 * bug fix: bogus undefined variable warnings from fopcompiled references to
38 global variables. (thanks to Lars Rune Nøstdal)
39 * bug fix: foreign function names should now appear in backtraces on
40 FC6 as well. (reported by Tomasz Skutnik and obias Rautenkranz)
41 * bug fix: SETF compiler macro documentation strings are not discarded
43 * bug fix: GENTEMP is now unaffected by pretty printer dispatch table.
44 (thanks to Alex Plotnick)
45 * bug fix: SLEEP accepts large integer arguments, truncating them to
46 SIGNED-WORD on the assumption that sleeping for 68 years is sufficient
47 for anyone. (reported by Leslie Polzer, thanks to Stas Boukarev)
49 changes in sbcl-1.0.29 relative to 1.0.28:
50 * IMPORTANT: bug database has moved from the BUGS file to Launchpad
51 https://bugs.launchpad.net/sbcl
52 Bugs can be reported directly there, or by sending email to
53 sbcl-bugs@lists.sourceforge.net
54 (no subscription required.)
55 * minor incompatible change: under weak type checking policy integer
56 types are weakened less aggressively.
57 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
58 allowed to return, which causes SBCL to quit with exit status 0. Previously
59 if the function returned with a small integer return value, that value
60 was accidentally reused as the exit status.
61 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
62 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
63 :HASH-FUNCTION argument. Refer to user manual for details.
64 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
66 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
67 microseconds since the Unix epoch on all platforms.
68 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
69 allows the compiler to safely elide boundedness checks for special
71 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
72 the symbol, prohibits both lexical and dynamic binding. This is mainly an
73 efficiency measure for threaded platforms, but also valueable in
75 * new feature: UNC pathnames are now understood by the system on Windows.
76 * optimization: the compiler uses a specialized version of FILL when the
77 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
79 * optimization: accesses to potentially non-simple arrays where element type
80 is known are 50% faster.
81 * optimization: compiler now generates faster array typechecking code.
82 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
84 * optimization: multidimensional array accesses in the absence of type
85 information regarding array rank are approximately 10% faster due to
86 open coding of ARRAY-RANK.
87 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
88 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
90 * optimization: result of call to VECTOR can now be stack allocated.
91 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
92 as long as the resulting array is one-dimensional and has a known
93 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
94 length of the list matches the known length of the vector does not
95 allocate the list as an intermediate step. Ditto for VECTOR and simple
97 * optimization: MAKE-ARRAY can now stack allocate in the presence of
98 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
99 known element type, and is known to be simple and one dimensional.
100 * improvement: SBCL now emits a compiler note where stack allocation was
101 requested but could not be provided (not in all cases, unfortunately)
102 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
103 * improvement: pretty-printing loop has been implemented properly. (thanks
104 to Tobias Rittweiler)
105 * documentation: CLOS slot typechecing policy has been documented.
106 * bug fix: FILE-AUTHOR no longer signals an error on Windows.
107 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
108 pointer, leading to memory faults. (thanks to Bart Botta)
109 * bug fix: better floating point exception handling on x86/OpenBSD.
110 (thanks to Josh Elsasser)
111 * bug fix: exit status from QUIT when called under --script was lost
112 (reported by Hubert Kauker)
113 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
114 same implementation of FILL to initialize the array, even if a more
115 efficient one was available (reported by Stas Boukarev, thanks to
117 * bug fix: potential miscompilation of array stack allocation on x86 and
118 x86-64. (reported by Time Tossavainen)
119 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
120 result in their subforms being treated as top level forms. (reported by
122 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
123 than a machine register could leave garbage in the high bits of the
124 result register (bug 316325).
125 * bug fix: disable address space randomization Linux/x86-64 as well,
126 not just x86-64. (reported by Ken Olum)
127 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
128 generate incorrect code.
129 * bug fix: #201; type inference for CONS and ARRAY types could derive
130 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
131 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
132 could not be assigned to or bound with PROGV. (reported by Lorenz
134 * bug fix: the value of CL:- in the inspector was the previous expression
135 evaluated rather than the expression being evaluated.
136 * bug fix: constants can no longer be locally declared special.
137 * bug fix: signals delivered to threads started from foreign land (read:
138 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
139 thread by blocking all signals and resignalling.
140 * bug fix: SHARED-INITIALIZE initialized unbound :ALLOCATION :CLASS slots
141 from :INITFORM, if any.
143 changes in sbcl-1.0.28 relative to 1.0.27:
144 * a number of bugs in cross-compilation have been fixed, with the ultimate
145 result that building under (at least) clisp should be much more reliable.
146 * minor incompatible changes: echo-streams now propagate unread-char to the
147 underlying input stream, and no longer permit unreading more than one
149 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
150 frames, allowing for instance more reliable backtraces.
151 * improvement: the debugger REPL can now reference lexical variables
152 by name directly for code compiled with (DEBUG 3).
153 * improvement: errors from malformed declarations now have better source
154 paths associated with them. (thanks to Tobias Rittweiler)
155 * optimization: faster local calls on x86/x86-64
156 * bug fix: some error messages for out-of-bound array indexes confused the
157 index and the bound. (thanks to Stas Boukarev)
158 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
160 * bug fix: running regressions tests in shells without OSTYPE set now works.
161 (reported by Harald Hanche-Olsen)
162 * bug fix: more robust static space exhaustion signalling from
163 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
164 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
165 docstring away instead of storing it under names such as (LAMBDA (X)).
166 (reported by Leslie Polzer)
167 * bug fix: timers could go off in the wrong order, be delayed indefinitely
168 (thanks to Ole Arndt for the patch)
169 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
170 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
171 potentially-invalid effective methods in its cache.
172 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
173 instances as well (thanks to Paul Khuong)
174 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
175 SB-POSIX:WAITPID and their C equivalents.
176 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
178 changes in sbcl-1.0.27 relative to 1.0.26:
179 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
180 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
181 * improvement: the system either recovers from stack exhaustion or dies
182 properly as opposed to leaving the user uncertain of whether the handler
183 trampled on some random memory next to the stack or having to rely on
184 --lose-on-corruption (which is still a good idea to use in production
185 because stack exhaustion can happen in signal handlers which will likely
187 * bug fix: fix gc related interrupt handling bug on ppc (regression from
188 1.0.25.37, reported by Harald Hanche-Olsen)
189 * bug fix: work around signal delivery bug in darwin (regression from
190 1.0.25.44, reported by Sidney Markowitz)
191 * bug fix: fix ERROR leaking memory (reported by David Thompson)
193 changes in sbcl-1.0.26 relative to 1.0.25:
194 * incompatible change: an interruption (be it a function passed to
195 INTERRUPT-THREAD or a timer function) runs in an environment where
196 interrupts can be enabled. The interruption can use
197 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
198 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
199 potentially running out of stack. Keep in mind that in the absence
200 of WITHOUT-INTERRUPTS some potentially blocking operation such as
201 acquiring a lock can enable interrupts.
202 * incompatible change: GC-OFF and GC-ON are removed, as they were
203 always unsafe. Use WITHOUT-GCING instead.
204 * new feature: runtime option --disable-ldb
205 * new feature: runtime option --lose-on-corruption to die at the
206 slightest hint of possibly non-recoverable errors: running out of
207 memory, stack, alien stack, binding stack, encountering a memory
208 fault, etc. In the absence of --lose-on-corruption a warning is
210 * enhancement: detect binding stack exhaustion
211 * enhancement: detect alien stack exhaustion on x86/x86-64
212 * improvement: generally more stable and reliable interrupt handling
213 * improvement: there is a per thread interruption queue,
214 interruptions are executed in order of arrival
215 * improvement: a repeating timer reschedules itself when the it has
216 finished, but expiration times are spaced equally. If an
217 expiration time is in the past it will trigger after a short grace
218 period that may give a chance to other things to run.
219 * optimization: slightly faster gc on multithreaded builds
220 * optimization: faster WITHOUT-GCING
221 * bug fix: when JOIN-THREAD signals an error, do it when not holding
222 important locks so that the debugger/handler doesn't produce
223 recursive errors or deadlock.
224 * bug fix: real-time signals are not used anymore, so no more
225 hanging when the system wide real-time signal queue gets full.
226 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
228 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
229 * bug fix: fix random memory faults related to interrupts on alpha
230 * bug fix: fix random memory faults related to interrupts on sparc
231 * bug fix: fix deadlocks related to starting threads
232 * bug fix: fix deadlines on locks on futex platforms
233 * bug fix: restore errno in signal handlers
234 * bug fix: fix deadlocks related to hash tables
235 * bug fix: fix deadlocks in pcl
237 changes in sbcl-1.0.25 relative to 1.0.24:
238 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
239 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
240 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
241 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
242 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
243 better name and does not return values so stale on multiprocessor systems.
244 Also, HOLDING-MUTEX-P was added for about the only sane usage of
246 * improvement: unithread builds keep track of MUTEX-VALUE.
247 * improvement: reading from a TWO-WAY-STREAM does not touch the output
248 stream anymore making it thread safe to have a concurrent reader and
249 a writer, for instance, in a pipe.
250 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
251 arguments, making compiler-macros for SETF-functions able to inspect
252 their constant arguments.
253 * improvement: COMPILE-FILE reports times with millisecond accuracy
254 (thanks to Luis Oliveira)
255 * optimization: CHAR-CODE type derivation has been improved, making
256 TYPEP elimination on subtypes of CHARACTER work better. (reported
257 by Tobias Rittweiler, patch by Paul Khuong)
258 * bug fix: setting alien structure fields of type struct by value now
259 computes the right offset for the memory copy.
260 * bug fix: compilation problem involving inlined calls to aliens with
261 result type VOID. (reported by Ken Olum)
262 * bug fix: #235a; sequential inline expasion in different policies no
263 longer reuses the functional from the previous expansion site.
264 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
265 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
266 * bug fix: named ENUMs in multiply-referenced alien record types no longer
267 cause an implied type redefinition. Regression from 1.0.21.29.
268 * bug fix: sign-extension of small signed return values now works with C
269 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
270 * improvements to the Windows port:
271 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
272 Westfold, thanks to Rudi Schlatte)
274 changes in sbcl-1.0.24 relative to 1.0.23:
275 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
276 vector of a multidimensional SIMPLE-ARRAY.
277 * new feature: the system now signals a continuable error if standard
278 readtable modification is attempted. (thanks to Tobias Rittweiler)
279 * new feature: DIRECTORY has been extended with a non-standard keyword
280 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
281 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
282 has been robustified and documented better. (thanks to Robert Goldman)
283 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
284 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
285 * tradeoff: constant FORMAT control strings are now compiled unless
286 SPACE > SPEED (previously only when SPEED > SPACE.)
287 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
289 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
291 * bug fix: lexical type declarations are now correctly reported by
292 SB-CLTL2. (reported by Larry D'Anna)
293 * bug fix: STRING-TO-OCTETS did not handle :START properly when
294 using UTF-8 as external format. (thanks to Luis Oliveira)
295 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
296 are signalled correctly. (thanks to Stas Boukarev)
297 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
298 designator. (thanks to Tobias Rittweiler)
299 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
300 designator, and returns T instead of the function. (thanks to
302 * bug fix: direct superclasses of STANDARD-CLASS and
303 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
304 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
305 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
306 when no class with the named slot yet exists no longer causes a
307 compile-time style-warning.
308 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
309 in safe code. (reported by Didier Verna)
310 * bug fix: #430; nested structure constructors can stack allocate.
311 * bug fix: on some 64-bit platforms dynamic space size was truncated
312 to #xffff0000 bytes. (reported by Benjamin Lambert)
313 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
314 REPL unusable. (reported by Daniel Herring)
315 * bug fix: return values of READ-SEQUENCE did not take :START into
316 account on file streams, regressions since 1.0.12.22. (reported by
317 Thomas Russ, patch by Paul Khuong)
318 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
319 method specializer used to confuse permuation vector optimization.
320 * bug fix: system inserted bogus implicit type declarations for local
321 special variables in DEFMETHOD bodies.
322 * bug fix: #354; duplicated frames in backtraces due to
323 non-tail-call-optimized XEPs to functions with return type NIL
325 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
326 initializes structure object slots according to DEFSTRUCT initforms,
327 and DEFSTRUCT forms :INCLUDEind structure classes defined using
328 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
329 (reported by Bruno Haible and Stephen Wilson)
330 * bug fix: #395; fill-pointer output streams used now support
331 element-type BASE-CHAR as well.
332 * bug fix: compiler error when attempting to derive return value of
333 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
335 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
337 * bug fix: Mac OS X binaries should now be portable between Leopard
340 changes in sbcl-1.0.23 relative to 1.0.22:
341 * enhancement: when disassembling method functions, disassembly
342 for the associated fast function is also produced.
343 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
345 * optimization: printing with *PRINT-PRETTY* true is now more
346 efficient as long as the object being printed doesn't require
347 special handling by the pretty printer.
348 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
349 now interact correctly with type declarations.
350 * partial bug fix: PCL detects infinite recursion during wrapper
351 validation. (thanks to Attila Lendvai)
352 * bug fix: #426; nested function calls are inlined properly.
353 Previously if FOO was an inline function, in calls of the form
354 (FOO (FOO ...)) the outer call was not inlined.
355 * bug fix: long long arguments passed in registers to alien
356 functions on PowerPC Linux and NetBSD are now handled correctly
357 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
359 changes in sbcl-1.0.22 relative to 1.0.21:
360 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
361 for the shared object in the current directory, but passes the native
362 namestring of the designated pathname to the operation system's shared
363 object loading function as-it.
364 * minor incompatible change: --disable-debugger toplevel option now takes
365 effect before processing of initialization files and --eval or --load
367 * new feature: new commandline argument: --script, which supports
368 shebang lines. See documentation for details. (based on work by
370 * new feature: SAVE-LISP-AND-DIE can save current values of
371 --dynamic-space-size and --control-stack-size in the executable core,
372 causing it to skip normal runtime option processing. See documentation
373 for details. (thanks to Zach Beane)
374 * enhancement: inoccous calls to EVAL or generic functions dispatching
375 on subclasses of eg. STREAM no longer cause compiler notes to appear.
376 * enhancement: the system no longer resignals errors from --load and
377 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
378 associated with the original error to be lost. (thanks to Ariel
380 * enhancement: :DONT-SAVE keyword argument has been added to
381 LOAD-SHARED-OBJECT for controlling interaction with
382 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
383 the effects of an earlier LOAD-SHARED-OBJECT call.
384 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
385 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
386 provided. (reported by Cedric St-Jean)
387 * bug fix: circularity handling in the reader did not treat raw
388 structure slots correctly. (reported by Cedric St-Jean)
389 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
390 file descriptors when there were none.
391 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
392 was broken. (reporteed by Willem Broekema)
393 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
394 pathnames without a directory.
395 * bug fix: COMPILE-FILE did not warn about undefined variable
396 references at toplevel, and LOAD of the resulting fasl did
398 * bug fix: functions with non-required arguments used to end up with
399 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
400 * bug fix: redefining a function with non-required arguments didn't
401 update the system's knowledge about its call signature properly.
402 * bug fix: fixed #431; incompatible alien record type redefinitions
403 are detected and handled. (reported by Neil Haven)
404 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
405 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
407 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
408 after alien stack frames.
409 * bug fix: strings passed to mk{s,d}temp lacked null termination
411 changes in sbcl-1.0.21 relative to 1.0.20:
412 * new feature: the compiler is able to track the effective type of a
413 generic function across method addition and removal.
414 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
415 appropriately typed structure slots without locking.
416 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
417 information like those gathered by TIME using a programming-friendly
419 * new feature: TIME reports time taken even if the form performs a
420 non-local transfer of control.
421 * enhancement: reduced conservativism on GENCGC platforms: on
422 average 45% less pages pinned (measured from SBCL self build).
423 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
424 mutate constant symbols or violate declaimed type of the symbol.
425 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
426 sleeping on the semaphore in heavily contested semaphores.
427 * bug fix: semaphores and condition variables were not interrupt
429 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
430 owned by other threads anymore.
431 * bug fix: FIND on lists called KEY outside the specified
432 subsequence. (reported by budden)
433 * bug fix: LOG doesn't use single-float intermediate results when
434 given mixed integer and double-float arguments, leading to better
435 precision. (reported by Bob Felts)
436 * bug fix: LOG with base zero returned values of inconsistent type.
438 changes in sbcl-1.0.20 relative to 1.0.19:
439 * minor incompatible change: OPTIMIZE qualities
440 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
441 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
442 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
443 * documentation: some slot access efficiency guidelines have been
444 added to the user manual.
445 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
446 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
447 as ASSOC and MEMEBER.
448 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
449 to more efficient EQ-comparison versions more often.
450 * optimization: enhanced derivation of DOLIST iteration variable type
452 * optimization: constant folding of simple (LIST ...) forms as DOLIST
454 * optimization: runtime lookup of function definitions can be
455 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
456 * optimization: compiler is able to derive the return type of
457 (AREF (THE STRING X) Y) as being CHARACTER.
458 * optimization: CLRHASH on empty hash-tables no longer does pointless
459 work. (thanks to Alec Berryman)
460 * optimization: REPLACE deftransforms don't punt when :START1 or
462 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
463 no longer emits a code deletion note. (reported by Andrew
465 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
466 functions with an odd number of &OPTIONAL arguments, a &REST
467 argument, and one or more &KEY arguments at the call site.
468 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
469 arguments appeared at call sites as well.
470 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
471 READ-LINE &co can no longer read from them afterwards. (reported
473 * bug fix: fixed #427: unused local aliens no longer cause compiler
474 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
476 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
477 alien-stack leakage. (reported by Andy Hefner)
478 * bug fix: PROGV signals an error when an attempt to violate declared
479 type of a variable or bind a constant is made.
480 * bug fix: SET signals an error when an attempt to violate declared
481 type of a variable is made.
482 * bug fix: restart computation during the execution of a restart
483 test function no longer causes infinite recursion. (reported by
485 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
486 signals an error. (thanks to Pierre Mai)
487 * bug fix: EXPORT left symbol unexported in conflict situations.
488 (thanks to Michael Weber)
489 * bug fix: correctly handle name conflicts involving CL:NIL.
490 (thanks to Michael Weber)
491 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
492 conflicts arising from USEing package with conflicting symbols
493 correctly. (thanks to Michael Weber)
495 changes in sbcl-1.0.19 relative to 1.0.18:
496 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
497 warnings that go otherwise unhandled will be muffled if they are
498 of the type that's the value of this variable.
499 * optimization: stack allocation is slightly more efficient on x86
501 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
502 where a raw slot always is initialized using the initform whose
503 type is not know sufficiently well a compile-time are now compiled
504 correctly. (reported by John Morrison)
505 * bug fix: compiler no longer makes erronous assumptions in the
506 presence of non-foldable SATISFIES types.
507 * bug fix: stack analysis missed cleanups of dynamic-extent
508 arguments in non-let-converted calls to local functions.
509 * improvements to the Windows port:
510 ** adjusted address spaces for building on both Win32 and
511 Win64. (thanks for John Connors)
512 * fixed some bugs revealed by Paul Dietz' test suite:
513 ** interval arithmetic during type derivation used inexact integer
514 to single-float coercions.
515 ** arithmetic operations involving large integers and single
516 floats give the same results in compiled and interpreted code.
517 ** deriving the result type of COERCE no longer signals an error
518 if the derived type of the second argument is a MEMBER type
519 containing invalid type specifiers.
520 ** ADJOIN with constant NIL as second argument works correctly.
521 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
523 changes in sbcl-1.0.18 relative to 1.0.17:
524 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
525 profiles only the current thread.
526 * minor incompatible change: changes to SYMBOL-VALUE of constants
527 defined with DEFCONSTANT now signal an error.
528 * enhancement: SB-SPROF now has support for wallclock profiling,
529 and is also able to profile specific threads. REPORT output
530 has also additional sorting options.
531 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
533 * optimization: structure allocation has been improved
534 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
535 ** out of line constructors are ~10% faster.
536 ** inline constructors are ~15% faster.
537 ** inline constructors are capable of dynamic extent allocation
538 (generally on x86 and x86-64, in some cases on other platforms
540 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
542 * optimization: file compiler is now able to coalesce non-circular
543 lists, non-base strings, and bit-vectors. Additionally, constants
544 are never referenced through SYMBOL-VALUE at runtime.
545 * optimization: code defining methods on PRINT-OBJECT (and other
546 generic functions in the COMMON-LISP package) now loads faster.
547 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
549 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
550 fasl is now deleted. (reported by Attila Lendvai)
551 * bug fix: READ-LINE always returned NIL for the last line in files.
552 (reported by Yoshinori Tahara)
553 * bug fix: more accurate disassembly annotations of foreign function
554 calls. (thanks to Andy Hefner)
555 * bug fix: trimming non-simple strings and non-string string
556 designators when the there is nothing to trim works properly.
557 (thanks to James Knight)
558 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
559 (thanks to Travis Cross)
560 * fixed some bugs revealed by Paul Dietz' test suite:
561 ** NIL is a valid function name (regression at 1.0.13.38)
562 ** FILL on lists was missing its return value (regression at 1.0.12.27)
563 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
564 fill pointers properly (regression at 1.0.12.23)
565 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
566 displacement indices properly (regression at 1.0.12.23)
568 changes in sbcl-1.0.17 relative to 1.0.16:
569 * temporary regression: user code can no longer allocate closure
570 variable storage on stack, due to bug 419 without explicitly
571 requesting it. Please consult sbcl-devel for advice if you need to
572 use this feature in the meanwhile.
573 * new feature: runtime argument --control-stack-size can be used to
574 adjust thread default control stack size.
575 * enhancement: improved TIME output
576 ** all times are reported using the measured accuracy (milliseconds
577 for real and GC times, microseconds for everything else.)
578 ** processor cycle counts on x86 and x86-64.
579 ** interpreted forms are counted for both evaluator modes.
580 ** number of lambdas converted by the compiler is reported.
581 ** CPU percentage report (computed from real and total run time.)
582 ** more comprehensive run time reporting, using a condenced format
583 ** interperted form, lambda, and page fault counts are omitted
585 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
587 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
588 * optimization: two argument forms of LAST are upto ~10% faster
589 in normal SPEED policies.
590 * optimization: NCONC no longer needs to heap cons its &REST list
591 in normal SPEED policies.
592 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
594 * bug fix: UNION and NUNION work with :TEST-NOT once more,
595 regression since 1.0.9.1. (thanks to Eric Marsden)
596 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
597 since 1.0.15.36. (thanks to Paul Khuong)
598 * bug fix: LAST when always returned the whole list when given a bignum
599 as the second argument.
600 * bug fix: dynamic extent allocation of nested lists and vectors
601 could leak to otherwise accessible parts.
602 * bug fix: invalid optimization of heap-allocated alien variable
604 * bug fix: fasl header checking is less vulnerable to different
605 platform word lengths.
606 * bug fix: more correct assembler syntax for GNU binutils
607 2.18.50.0.4 support. (thanks to Marijn Schouten)
608 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
610 * bug fix: the fopcompiler can handle LOCALLY forms (with no
611 declarations) successfully. (reported by Attila Lendvai)
613 changes in sbcl-1.0.16 relative to 1.0.15:
614 * minor incompatible change: revert the changes to sb-posix's error
615 signaling added in 1.0.14.
616 * minor incompatible change: change PROBE-FILE back to returning
617 NIL whenever we can't get a truename, as was the case before 1.0.14.
618 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
619 inherits from ERROR instead of just CONDITION.
620 * new feature: SB-INTROSPECT can provide source locations for instances
621 as well. (thanks to Tobias Rittweiler)
622 * optimization: binding special variables now generates smaller code
623 on threaded platforms.
624 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
625 and cases where no :TEST is given but the compiler can infer that
626 the element to search is of type (OR FIXNUM (NOT NUMBER)).
627 * optimization: better LOGNOT on fixnums.
628 * optimization: modular arithmetic for a particular requested width
629 is implemented using a tagged representation unless a better
630 representation is available.
631 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
632 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
633 for wellformedness and returns a specialized vector. (reported by
634 Francois-Rene Rideau)
635 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
636 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
637 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
638 unscheduling timer at the same time as another timer fires.
639 * bug fix: don't reschedule timers for dead threads.
640 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
641 * bug fix: copying output from RUN-PROGRAM to a stream signalled
642 bogus errors if select() was interrupted.
643 * enhancement: add support for fcntl's struct flock to SB-POSIX.
645 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
646 * enhancement: cleaner backtraces for interactive interrupts, as
647 well as other cases where the interesting frames used to be
648 obscured by interrupt handling frames.
649 * enhancement: untracing a whole package using (UNTRACE "FOO") is
650 now supported, and tracing a whole package using (TRACE "FOO") now
651 traces SETF-functions as well.
652 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
653 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
654 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
655 * unreadably printed representation of hash-tables now includes
657 * bug fix: partially fixed #188: type propagation from assignments
658 is now more efficient.
659 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
660 'DOUBLE-FLOAT) are not flushable.
661 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
662 full address of the object, and none of the tag bits.
663 * bug fix: readably printing hash-tables now respects other printer
664 control variables. (reported by Cedric St-Jean)
665 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
666 keyword with MAKE-HASH-TABLE.
667 * bug fix: export SB-POSIX:MKSTEMP.
668 * bug fix: SORT was not interrupt safe.
669 * bug fix: XREF accounts for the last node of each basic-block as
671 * bug fix: MAKE-INSTANCE optimizations interacted badly with
672 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
673 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
674 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
675 create obsolete instances in the case of redefinition or
676 obsoletion of a superclass. (thanks to Andy Hefner)
677 * bug fix: Support for the Alpha architecture has been revived; it had
678 suffered somewhat from lack of maintenance since sbcl-1.0.
679 * improvements to the Windows port:
680 ** The system detects the codepage to use at startup, instead of
681 using the value from the saved core. (thanks to Kei Suzuki)
683 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
684 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
685 (see documentation for details.)
686 * revived support for OpenBSD (contributed by Josh Elsasser)
687 * partially fixed bug #108: ROOM no longer suffers from occasional
688 (AVER (SAP= CURRENT END)) failures.
689 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
690 bodies no longer cause a WARNING to be signalled. (reported by
692 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
693 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
694 the bug in TRUNCATE.) (reported by Michael Weber)
695 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
696 no samples. (reported by Andy Hefner)
697 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
698 no longer appear as (NIL ...) frames in backtraces.
699 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
700 is now actually exported from SB-EXT as documented. (reported by
702 * bug fix: sb-aclrepl now correctly understands how to inspect
703 single-floats on 64-bit platforms where single-floats are not boxed.
704 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
705 yet been finalized. (reported by Levente Meszaros)
706 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
707 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
708 SBCL-specific optimize qualities.
710 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
711 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
712 an executable in the search path, and does so in the child
713 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
714 been removed; it can be found in the CVS history, for somebody who
715 needs that search behavior (see the manual).
716 * minor incompatible change: compiler policy re. weakening type
717 checks has changed: now type checks are weakened only if SAFETY < 2
719 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
720 unparsing of directory pathnames as files. Analogously,
721 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
722 filename to parse into a directory pathname.
723 * enhancement: implicit generic function creation now signals a
724 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
725 users can bind handlers for and muffle around calls to LOAD.
726 * enhancement: RUN-PROGRAM allows unicode arguments and environments
727 to be used (using the default stream external format), and allows
728 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
729 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
730 strings and vectors whose element-type or simplicity is not fully
731 known at compile-time.
732 * optimization: STRING-TRIM and related functions no longer allocate
733 a new string when no trimming needs to be performed. These functions
734 are also faster than before when the input string has been declared
736 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
737 * optimization: READ-LINE is significantly faster for files containing
739 * optimization: non-open coded uses of character comparison operators
740 (e.g. char=) no longer cons when called with more than one parameter
741 on platforms supporting dynamic-extent allocation.
742 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
743 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
744 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
746 * bug fix: some sequence functions elided bounds checking when
748 * bug fix: too liberal weakening of union-type checks when SPEED >
750 * bug fix: more bogus fixnum declarations in ROOM implementation
753 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
754 * new feature: MAKE-HASH-TABLE now experimentally accepts a
755 :SYNCHRONIZED argument, which makes the hash-table safe for
756 concurrent accesses (but not iteration.) See also:
757 SB-EXT:WITH-LOCKED-HASH-TABLE, and
758 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
759 * optimization: CONCATENATE on strings is an order of magnitude faster
760 in code compiled with (> SPEED SPACE).
761 * optimization: SUBSEQ is ~50% faster on lists.
762 * bug fix: bug 417 fixed -- source location reporting is now more robust.
763 * bug fix: SUBSEQ on a list will now correctly signal an error if if
764 END is smaller then START.
765 * bug fix: SB-PROFILE will no longer report extra consing for nested
766 calls to profiled functions.
767 * bug fix: ROOM implementation had bogus fixnum declarations which
768 could cause type-errors when calling ROOM on large images.
769 * bug fix: if file compilation is aborted, the partial fasl is now
770 deleted, and COMPILE-FILE returns NIL as the primary value.
771 * bug fix: number of thread safety issues relating to SBCL's internal
772 hash-table usage have been fixed.
773 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
774 be returned from its body when the values were being returned
775 using unknown-values return convection and the W-P-O was wrapped
776 inside an UNWIND-PROTECT.
777 * bug fix: sb-posix should now compile again under Windows, enabling
780 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
781 * incompatible change: hash-table accessor functions are no longer
782 automatically protected by locks. Concurrent accesses on the same hash-table
783 from multiple threads can give inconsistent results or even corrupt the
784 hash-table completely. Multi-threaded applications should do their own
785 locking at the correct granularity. In the current implementation it is
786 still safe to have multiple readers access the same table, but it's not
787 guaranteed that this property will be maintained in future releases.
788 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
789 and will signal an error at runtime.
790 * enhancement: SB-THREAD package now exports a semaphore interface.
791 * enhancement: CONS can now stack-allocate on x86 and
792 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
794 * enhancement: nested lists can now be stack allocated on
795 platforms providing stack allocation support.
796 * enhancement: dynamic-extent support has been extended to support
797 cases where there are multiple possible sources for the stack
799 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
800 if the mutex is uncontested on Linux.
801 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
802 as specified by the CLHS. (thanks to Tobias Rittweiler)
803 * bug fix: NaN comparison now works on x86-64.
804 * bug fix: CLOSE :ABORT T on a stream with pending output now
806 * bug fix: instances of non-standard metaclasses using standard
807 instance structure protocol sometimes missed the slot type checks
809 * bug fix: known functions can be inlined in high-debug code.
811 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
812 * minor incompatible change: the MSI installer on Windows no longer
813 associates .lisp and .fasl files with the installed SBCL.
814 * minor incompatible change: :UNIX is no longer present in *FEATURES*
815 on Windows. (thanks to Luis Oliviera)
816 * new feature: SB-CLTL2 contrib module now implements
817 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
818 * optimization: scavenging weak pointers is now more efficient,
819 requiring O(1) instead of O(N) per weak pointer to identify
820 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
821 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
822 method bodies using constant slot-names when the first argument is
823 a specializer parameter for the method.
824 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
825 names, when the class of the instance is a direct instance
826 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
827 as slow as the constant slot-name case.)
828 * optimization: member type construction is now O(N) instead
830 * optimization: UNION and NUNION are now O(N+M) for large
831 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
833 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
834 now more readable in environments like Slime which display it.
835 (thanks to Tobias C. Rittweiler)
836 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
837 was able to infer the correct argument type for the object on which
838 the CAS operation was being performed.
839 * bug fix: copy propagation interfered with parallel assignment
840 semantics in local calls. (reported by Paul Khuong)
841 * bug fix: the signed modular fixnum shift compiled to wrong code on
842 x86 and x86-64. (spotted by a slight modification to some of
845 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
846 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
847 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
848 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
850 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
851 and generic function now print the object identity for anonymous
852 instances. (thanks to Attila Lendvai)
853 * enhancement: as an extension to MOP, SBCL now supports SETF of
854 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
855 (thanks to Attila Lendvai)
856 * enhancement: sb-cover has an alternate annotation mode which puts
857 the form annotations at the car of the form, instead of annotating
858 it completely. The benefit of this mode is that it shows better
859 which forms were instrumented by the compiler.
860 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
861 by printer control variables. (reported by Dan Corkill)
862 * bug fix: system leaked memory when delayed output was performed by
863 the OS in smaller chunks then expected. (thanks to David Smith)
864 * bug fix: system leaked memory when file streams were not closed
866 * bug fix: large objects written to slow streams that were modified
867 after the write could end up with the modified state written to
868 the underlying file descriptor.
869 * bug fix: multiple threads operating in parallel on the same stream
870 could cause buffer-overflows.
871 * bug fix: source location information is stored correctly
872 (broken since 1.0.6). This bug would generally show up as the
873 Slime debugger highlighting the wrong form when the sldb-show-source
875 * bug fix: Branch forms are again annotated as branches in the sb-cover
877 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
878 on x86-64. (reported by Gregory Vanuxem)
879 * bug fix: EQUALP could return wrong results for structures with raw
880 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
881 word). (reported by Vjacheslav Fyodorov)
882 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
885 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
886 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
887 atomic compare-and-swap operations on threaded platforms.
888 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
889 allows assigning a global minimum value to optimization qualities
890 (overriding proclamations and declarations).
891 * enhancement: closed over variables can be stack-allocated on x86
893 * performance bug fix: GETHASH and (SETF GETHASH) are once again
895 * optimization: slot definition lookup is now O(1). This speeds up
896 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
897 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
898 * optimization: ASSOC and MEMBER can now be open-coded for all
899 combinations of keyword arguments when second argument is constant
900 and SPEED >= SPACE. In other cases a specialized version is
902 * bug fix: using obsoleted structure instances with TYPEP and
903 generic functions now signals a sensible error.
904 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
905 (reported by Kristoffer Kvello)
906 * bug fix: backtrace construction is now more careful when making
907 lisp-objects from pointers on the stack, to avoid creating bogus
908 objects that can be seen by the GC.
909 * bug fix: defaulting of values in contexts expecting more than 7
910 variables now works on x86-64. (reported by Christopher Laux)
911 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
913 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
914 as the property-list of a symbol.
915 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
916 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
917 situations requested, are once again file-compileable. (reported
920 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
921 * MOP improvement: support for user-defined subclasses of
922 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
923 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
924 the expansion of DEFMETHOD, is responsible for generating a form
925 which creates a list of specializers when evaluated. Additional
926 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
927 debugging and introspective support.
928 * minor incompatible change: the (unsupported) spinlock interface
929 has changed: free spinlock now has the value NIL, and a held spinlock
930 has the owning thread as its value.
931 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
932 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
934 * enhancement: name of a socket-stream is now "a socket" instead of
936 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
937 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
938 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
939 for generic functions. (Thanks to Tobias C. Rittweiler)
940 * optimization: bignum printing speed has been improved by 20-40%
941 (depending on the bignum size.)
942 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
944 * bug fix: the cache used by the CLOS to store precomputed effective
945 methods, slot offsets, and constant return values is now thread and
947 * bug fix: generic function dispatch function updating is now thread
948 and interrupt safe (in the sense that the known issues have been
950 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
951 * bug fix: interning EQL-specializers is now thread and interrupt safe.
952 * bug fix: asdf systems with dependencies to the SB-POSIX or
953 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
954 * bug fix: interrupt safety of applicable method computation has been
957 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
958 * new contrib: sb-cover, an experimental code coverage tool, is included
960 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
961 significantly faster.
962 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
963 produce more efficient code on x86-64 if the compiler can determine
964 that (+ INDEX <constant>) does not require a bounds check and FOO
965 has an element type at least 8 bits wide.
966 * enhancement: a new, experimental synchronous timeout facility is
967 provided. Refer to SB-SYS:WITH-DEADLINE for details.
968 * enhancement: when a symbol name conflict error arises, the
969 conflicting symbols are always printed with a package prefix.
970 (thanks to Kevin Reid)
971 * enhancement: stepping is now once again supported on the SPARC and
972 MIPS platforms. (It is also now more likely to work on CheneyGC
974 * enhancement: sb-sprof can now also track and report accurate call
976 * bug fixes: the treatment of non-standard subclasses of
977 SB-MOP:SPECIALIZER is more correct.
978 * incompatible change: PURIFY no longer copies the data from the
979 dynamic space into the static and read-only spaces on platforms
980 that use the generational garbage collector
981 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
983 * bug fix: GC race condition occasionally resulting in crashes with
984 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
986 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
988 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
989 a STYLE-WARNING to be signalled (regression from 1.0.4.)
990 * bug fix: an asynchronous interrupt could previously leave the
991 system running with GC inhibited.
992 * bug fix: a DECLARE form evaluated at top-level now causes an error
993 rather than silently (or verbosely) returning NIL.
994 * bug fix: trying to dissassemble functions compiled at high DEBUG could
995 result in a error being signalled due to source form lookup errors.
996 (reported by Peter Graves)
998 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
999 * incompatible change: removed writer methods for host-ent-name,
1000 host-ent-addresses -- changing the values did not update the DNS
1002 * minor incompatible change: changed experimental JOIN-THREAD interface
1003 * documentation: the manual now lists reader and writer methods
1004 in class slot documentation sections. (thanks to Richard M Kreuter)
1005 * documentation: unwinding from asyncronous events has been
1006 documented as unsafe.
1007 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
1008 in multithreaded application code.
1009 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
1010 platforms. (thanks to James Anderson for the optimization hint)
1011 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
1013 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
1014 variants no longer cons.
1015 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
1016 their NOT- variants no longer cons.
1017 * optimization: EQUAL hash tables no longer use SXHASH for objects
1018 of all data types, but instead use an EQL hash for types for which
1019 EQUAL is the same as EQL.
1020 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
1021 are significantly faster.
1022 * optimization: new STRING-OUTPUT-STREAM implementation is much
1023 faster for large outputs and conses 30% less on average.
1024 * enhancement: XREF information is now collected for references made
1025 to global variables using SYMBOL-VALUE and a constant argument.
1026 * enhancement: SIGINT now causes a specific condition
1027 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
1028 * bug fix: FILE-POSITION used to signal an error in some cases where
1029 ANSI requires it to return NIL.
1030 * bug fix: ADJUST-ARRAY is now interrupt-safe.
1031 * bug fix: adding and removing fd-handlers is now interrupt-safe.
1032 * bug fix: inlined calls to C now ensure 16byte stack alignment on
1034 * bug fix: bad type declaration in the CLOS implementation has
1035 been fixed. (reported by James Anderson)
1036 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
1037 (thanks to Lutz Euler)
1038 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
1039 required by the CLHS. (reported by Samium Gromoff)
1040 * bug fix: dead unbound variable references now signal an error.
1041 (reported by Marco Monteiro)
1042 * bug fix: / with an unused value was being deleted in safe code.
1043 (thanks to Marco Monteiro and Kevin Reid)
1044 * bug fix: number of characters that can be written onto a single
1045 line in a file is unlimited.
1046 * bug fix: some GC deadlocks caused by asynchronous interrupts have
1047 been fixed by inhibiting interrupts for when GC is disabled.
1048 * bug fix: some interrupt safety issues with GETHASH, (SETF
1049 GETHASH), CLRHASH and REMHASH have been fixed.
1050 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
1051 specifier no longer causes infinite recursion.
1052 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
1053 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
1054 * bug fix: an &environment argument with the correct variable information
1055 is passed to macros that are expanded during byte compilation.
1056 (reported by Samium Gromoff)
1057 * bug fix: a logic error was causing the hash function for bignums to
1058 have excessive amounts of collisions. (reported by Faré Rideau)
1059 * bug fix: modifying the contents of an array could change the return
1060 value of SXHASH on that array, which is only allowed for strings
1061 and bit vectors. (bug introduced in 0.9.16)
1062 * improvement: the x86-64/darwin port now passes all tests (except
1063 for the debugger tests) but should still be considered
1064 experimental until this is fixed.
1065 * improvement: a style-warning is signaled for CASE (etc) clauses with
1066 duplicate keys. (thanks to Kevin Reid)
1067 * improvement: macos/x86 and macos/x86-64 now use mach exception
1068 handlers for dealing with illegal instructions (for trapping and
1069 error handling) and memory protection violations (for GC).
1071 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1072 * new platform: experimental support for x86-64/darwin (MacOS).
1073 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1074 and getnameinfo sockets functions are used instead of gethostbyaddr
1075 and gethostbyname, on platforms where the newer functions are available.
1076 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1078 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1079 don't print the startup banner, but behave as if --noinform was passed
1080 as a command line argument. (thanks to Kevin Reid)
1081 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1082 * optimization: code using alien values with undeclared types is much faster.
1083 * optimization: the compiler is now able to open code SEARCH in more cases.
1084 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1085 * bug fix: using standardized COMMON-LISP special variables as loop
1086 variables no longer signals bogus package lock violations. (reported
1088 * bug fix: declaring local loop variables to be of a range-limited type
1089 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1090 (reported by Andras Simon)
1091 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1092 bugs remain on x86-64.)
1093 * bug fix: the #= and ## reader macros now interact reasonably with
1094 funcallable instances.
1095 * bug fix: type-checks for function arguments were compiled using the
1096 compiler policy settings of the wrong lexical environment (in 1.0.2
1098 * bug fix: SHADOW accepts characters as string designators, as required
1099 by the spec (thanks to Eric Marsden)
1100 * bug fix: fixed GC safety issues when foreign functions are called with
1101 non-base strings as arguments
1102 * bug fix: more consistent error output for fatal-compiler-warnings, like
1104 * bug fix: functions with proper names were showing up as NIL in
1105 backtraces (reported by Edi Weitz, regression in 1.0.2)
1107 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1108 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1109 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1110 produce more efficient code on the x86 if the compiler can determine
1111 that (+ INDEX <constant>) does not require a bounds check and FOO
1112 has an element type at least 8 bits wide.
1113 * bug fix: references to undefined variables are handled the same way
1114 in toplevel forms as in the normal compiler
1115 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1117 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1118 (thanks to Jon Buller)
1119 * improvement: faster compilation times for complex functions
1120 * improvement: added readlink support to SB-POSIX (thanks to Richard
1123 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1124 * improvement: experimental support for mach exception handling on
1125 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1127 * improvement: support for GBK external format.
1128 (thanks to Chun Tian (binghe))
1129 * improvement: the debugger now displays variables that have been closed
1130 over, in code compiled with (DEBUG 2) or higher
1131 * improvement: support for executable cores on NetBSD (thanks to
1133 * new feature: added a RESTART-FRAME debugger command
1134 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1135 be used to provide an implementation for FILE-POSITION on Gray streams
1136 (thanks to Eric Marsden)
1137 * new feature: add syslog, openlog and closelog support to SB-POSIX
1138 (thanks to Richard Kreuter)
1139 * optimization: the function call overhead in code compiled with
1140 a high DEBUG optimization setting is significantly
1141 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1142 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1143 (thanks to Tony Martinez)
1144 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1145 evaluated once (reported by Kevin Reid)
1146 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1147 works again (thanks to Kilian Sprotte)
1148 * bug fix: an error is signaled for tagbodies with duplicate tags
1149 (thanks to Stephen Wilson)
1150 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1151 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1152 stack frames from alien callbacks.
1153 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1154 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1155 65k pages (thanks to David Woodhouse)
1156 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1158 changes in sbcl-1.0.1 relative to sbcl-1.0:
1159 * new platform: FreeBSD/x86-64, including support for threading.
1160 * new feature: the compiler stores cross-referencing information
1161 about function calls (who-calls), macroexpansion (who-macroexpands)
1162 and special variables (who-binds, who-sets, who-references) for code
1163 compiled with (< SPACE 3). This information is available through the
1164 sb-introspect contrib.
1165 * new feature: users may subclass SEQUENCE, and have instances of
1166 these classes interoperate with standard Common Lisp functions if
1167 a number of methods are defined. (This feature is experimental
1168 and the interface subject to change based on feedback from SBCL
1169 users and the general community)
1170 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1171 x86/x86-64, rather than the previous fixed depth of 8
1172 * improvement: another pthread back-end of mutex "pthread-futex".
1173 * bug fix: non-ascii command-line arguments are processed correctly
1174 (thanks to Yaroslav Kavenchuk)
1175 * bug fix: non-required arguments were not passed correctly when a method
1176 defined using DEFMETHOD was called from a mop-generated method using
1177 CALL-NEXT-METHOD (reported by Pascal Costanza)
1178 * bug fix: recursion is now permitted in accessors through
1179 SLOT-UNBOUND. (reported by Pascal Costanza)
1180 * bug fix: an error was signaled at startup if the HOME environment
1181 variable was defined, but had an empty value (reported by Peter Van Eynde)
1182 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1183 signal a PROGRAM-ERROR, not a BUG.
1184 * bug fix: SB-POSIX:READDIR works when built with large file support.
1185 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1187 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1188 for better automatic indentation support. (thanks to Matt Pillsbury)
1189 * optimization: loading generic functions no longer takes O(n^2) time,
1190 proportional to the amount of methods in the generic function
1191 (reported by Todd Sabin and Jeremy Brown)
1192 * optimization: the FIND and POSITION family of sequence functions
1193 are significantly faster on arrays whose element types have been
1195 * improvements to the Windows port:
1196 ** Intermittent heap corruption problems have been fixed. (thanks
1197 to Alastair Bridgewater)
1198 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1200 ** Lisp is able to unwind foreign exception frames from alien
1201 callbacks. (thanks to Alastair Bridgewater)
1203 changes in sbcl-1.0 relative to sbcl-0.9.18:
1204 * improvement: experimental support for threading on FreeBSD/x86.
1205 (thanks to NIIMI Satoshi)
1206 * improvement: runtime option --dynamic-space-size can be used
1207 to set the size of the dynamic space reserved on startup.
1208 * improvement: floating point modes in effect are now saved in
1209 core, and restored on startup.
1210 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1211 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1212 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1213 threads are not running after *SAVE-HOOKS* have run.
1214 * improvement: writes to CLOS instance slots are type-checked in code
1215 compiled with (SAFETY 3)
1216 * improvement: floating-point exception handling on FreeBSD (thanks to
1218 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1219 (thanks to Zach Beane)
1220 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1222 * improvement: added support for the Shift-JIS external format.
1223 (contributed by NIIMI Satoshi)
1224 * improvement: callbacks are supported on Linux/PPC. (thanks to
1226 * bug fix: compiler bug triggered by a (non-standard) VALUES
1227 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1228 * bug fix: file compiler no longer confuses validated and already
1229 dumped structurres. (reported by Kaersten Poeck)
1230 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1231 fill-pointer signals a type-error as required. (thanks to
1233 * bug fix: disassemly of funcallable instances works.
1234 * bug fix: single stepping on PPC.
1235 * bug fix: fix thread-safety problems in the type system (generally
1236 manifesting as nonsensical errors like "STRING is a bad type specifier
1237 for sequences" or "The value 1 is not of type FIXNUM")
1238 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1239 (reported by Josip Gracin)
1240 * bug fix: an error is signaled for attempts to displace arrays with
1241 incompatible element types (thanks to Mario Mommer)
1242 * bug fix: more correct handling of wide characters in debug info
1243 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1244 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1245 and don't cause extra consing
1246 * optimization: MAP and MAP-INTO are significantly faster on vectors
1247 whose elements types have been declared.
1248 * Improvements to SB-SPROF:
1249 ** Support for allocation profiling
1250 ** Reduced profiling overhead, especially for long profiling runs
1251 * Improvements to the Windows port:
1252 ** floating point exceptions are now reported correctly.
1253 ** stack exhaustion detection works partially.
1254 ** more accurate GET-INTERNAL-REAL-TIME.
1255 ** better RUN-PROGRAM behaviour in case of errors in the child
1257 ** PROBE-FILE now simplifies pathnames correctly.
1258 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1259 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1260 1st 1970 (although time timezone data will be unreliable).
1261 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1263 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1264 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1265 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1267 * bug fix: two potential GC deadlocks affecting threaded builds.
1268 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1269 returns T (reported by Anton Kazennikov)
1270 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1271 with non-variable places
1272 * bug fix: remove a race condition in the setting of
1273 funcallable-instance functions, this should make threaded CLOS
1274 code more stable against memory faults.
1275 * bug fix: corruption of specials when unbinding is interrupted by an
1276 asynchronous unwind (reported by Hannu Koivisto)
1277 * improvement: the debugger will now also display local variables that
1278 are only used once, for code compiled with a DEBUG optimization quality
1281 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1282 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1283 * incompatible change: External-format support for FFI calls. The
1284 SB-ALIEN:C-STRING no longer implies an ASCII
1285 external-format. Instead the string is subject to external-format
1286 conversion. Additionally return values of type SB-ALIEN:C-STRING
1287 are by default of type (SIMPLE-ARRAY CHARACTER), not
1288 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1289 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1290 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1291 * incompatible change: SB-EXT package no longer contains the
1292 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1293 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1295 * incompatible change: the single-stepper is no longer available
1296 on Alpha, Mips and Sparc platforms.
1297 * minor incompatible change: the direct superclasses of
1298 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1299 not (STANDARD-OBJECT FUNCTION). This makes the
1300 class-precedence-lists of GENERIC-FUNCTION and
1301 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1303 * minor incompatible change: the default stream external format on
1304 non-unicode SBCL is no longer detected from the locale, but is always
1306 * new feature: Add a version of evaluator that uses an interpreter instead
1307 of the compiler. EVAL still uses the compiler by default, to switch it
1308 to use the interpreter, set the value of the variable
1309 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1310 * minor incompatible change: the single-stepper REPL has been merged
1311 with the normal debugger (see the "Stepping" heading of the debugger help
1312 for more details). The debugger command STEP will no longer switch
1313 to the single-stepper REPL.
1314 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1315 for a type now works.
1316 * bug fix: loading of default sysinit file works. (thanks to Leonid
1318 * bug fix: better detection of circularities in the file-compiler.
1319 (reported by Marco Monteiro)
1320 * bug fix: the CL pathname functions now work with files that have
1321 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1322 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1323 systems with Unix98 pty semantics.
1324 * bug fix: ASDF-INSTALL will now work with bsd tar.
1325 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1327 * bug fix: timers expiring in dead threads no longer cause a
1328 type-error (reported by Paul "Nonny Mouse").
1329 * bug fix: thanks to more lightweight single-stepper instrumentation,
1330 code compiled with (DEBUG 3) will compile and execute significantly faster,
1331 and will have more accurate type-inferencing than before
1332 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1333 whose bindings are modified
1334 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1335 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1336 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1337 CL:SOFTWARE-VERSION with proper version
1339 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1340 * feature: implemented the READER-METHOD-CLASS and
1341 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1342 as specified by AMOP.
1343 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1345 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1346 (thanks to Lutz Euler)
1347 * optimization: hashing of general arrays and vectors has been
1348 improved. (reported by Any Fingerhut)
1349 * enhancement: SB-INTROSPECT is now able to find definitions of
1350 profiled functions. (thanks to Troels Henriksen)
1351 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1352 as well, allowing compiler-macros for SETF-functions to expand.
1353 * enhancement: step-instrumentation no longer wraps known
1354 single-value functions with multiple-value context, allowing
1355 better type inference.
1356 * fixed bug #337: use of MAKE-METHOD in method combination now works
1357 even in the presence of user-defined method classes. (reported by
1358 Bruno Haible and Pascal Costanza)
1359 * fixed bug #339(c): if there are applicable methods not part of any
1360 long-form method-combination group, call INVALID-METHOD-ERROR.
1361 (reported by Bruno Haible)
1362 * fixed bug #361: the :FUNCTION initarg in the protocol for
1363 initialization of methods can now be used to override
1364 internally-produced optimized functions. (reported by Bruno
1366 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1367 system-provided lambda expression no longer cause warnings about
1368 unbound #:|pv-table| symbols.
1369 * bug fix: improved the handling of type declarations and the
1370 detection of violations for keyword arguments with non-constant
1372 * bug fix: potentially erronous calls to PATHNAME and
1373 MERGE-PATHNAMES were being flushed in some cases.
1374 (reported by Richard Kreuter)
1375 * bug fix: compiled calls to TYPEP were mishandling obsolete
1376 instances. (reported by James Bielman and Attila Lendvai)
1377 * bug fix: format strings with ~> without matching ~< no longer
1378 trigger an AVER, but signal an understandable error instead.
1379 (reported by Antonio Martinez)
1380 * bug fix: specifying an output-file without a pathname-type for
1381 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1382 of input-file instead of "fasl". (reported by Robert Dodier)
1383 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1384 (reported by James Y Knight).
1385 * bug fix: compiler-macros-function did not consider the environment
1386 argument for shadowing by local functions.
1387 * bug fix: compiler-macros expansion was inhibited by local INLINE
1389 * bug fix: inline expansions of known functions were subject to
1390 step-instrumentation in high debug policies, leading to problems
1391 with type-inference.
1392 * bug fix: compiler failed to differentiate between different CONS
1393 types in some cases.
1394 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1395 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1396 * bug fix: #368: incorrect use of expressed vs. upgraded array
1398 * bug fix: #306a: more precise unions of array types.
1399 * thread-safety improvements:
1400 ** CONDITION-WAIT could return early on Linux, if the thread was
1401 interrupted and subsequently continued with SIGCONT.
1402 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1404 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1405 * added support for the ucs-2 external format. (contributed by Ivan
1407 * minor incompatible change: pretty printing of objects of type
1408 (cons symbol) is, in the default pprint-dispatch-table, now
1409 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1411 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1412 called later in a class's lifetime, possibly as late as when the
1413 first instance of the class is created. Previously,
1414 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1415 class became finalizeable.
1416 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1417 for multibyte external-format streams. (thanks to "vbzoli")
1418 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1419 with :ALLOCATION :CLASS inherited from superclasses of the
1421 * fixed bug: anonymous classes can now be created using the :NAME
1422 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1423 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1424 * fixed bug: core-files saved with :EXECUTABLE T can again be
1425 executed when SBCL_HOME isn't set. (reported by James Knight)
1426 * fixed bug: toplevel LOCALLY forms with declarations could
1427 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1428 * fixed bug: printing from several different threads using different
1429 values of *print-case* could cause invalid output, due to
1430 some internal special variables of the printer not being bound
1431 thread-locally (reported by Max Mikhanosha)
1432 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1433 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1434 (reported by Pascal Costanza)
1435 * fixed bug: FINALIZE-INHERITANCE is called from
1436 REINITIALIZE-INSTANCE on classes when the class has previously
1437 been finalized, as required by AMOP.
1438 * minor code generation optimizations:
1439 ** better register allocation in CLOS dispatching functions
1440 ** overflow detection when coercing signed bytes to fixnums on x86-64
1441 ** is now implemented with one IMUL instruction instead of three shifts
1442 ** more efficient bit-vector access on x86 and x86-64
1443 ** more efficient access to raw structure slots on x86 and x86-64
1444 * fixed some bugs revealed by Paul Dietz' test suite:
1445 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1446 return its argument.
1448 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1449 * feature: thread support on Solaris/x86, and experimental thread support
1451 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1453 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1454 default method for SLOT-UNBOUND.
1455 * minor incompatible change: prevent the user from specializing the
1456 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1457 somewhat counter to the intent of the protocol, I (CSR) think, and
1458 additionally it just doesn't work in SBCL as currently
1459 implemented, thanks to optimizations (that are always valid for
1460 the other three SLOT-VALUEish functions, but not for the setter).
1461 * minor incompatible change: the :SB-LDB feature is now enabled by
1462 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1463 the low-level debugger.
1464 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1465 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1466 down to an SB-SYS:FD-STREAM, allowing direct communication in
1467 more cases. Temporary files and pipes are still used as a fallback
1469 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1470 on the non-reentrant C functions localtime(3) and gmtime(3).
1471 * bug fix: global optimization policy was not visible in LOCALLY and
1473 * bug fix: class objects can be used as specializers in methods.
1474 (reported by Pascal Costanza)
1475 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1476 with a type error. (reported by blitz_ on #lisp)
1477 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1478 failed with a type error. (reported by Pascal Bourguignon)
1479 * bug fix: merging pathnames against defaults with :DIRECTORY
1480 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1481 (reported by James Y Knight)
1482 * bug fix: saving large (>2GB) cores on x86-64 now works
1483 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1484 constant index and a (simple-array (signed-byte 32)) array
1485 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1486 error (patch by Robert J. Macomber)
1487 * feature: TIME now displays the amount of run-time spent in GC
1488 * bug fix: The debugger now does a better job of respecting
1489 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1490 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1491 workaround for bug 403.)
1492 * bug fix: floating point exception handling now works on FreeBSD
1493 (thanks to NIIMI Satoshi)
1494 * fixed some bugs revealed by Paul Dietz' test suite:
1495 ** MISC.641: LET-conversion were not supposed to work in late
1497 * improvements to the Win32/x86 port:
1498 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1499 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1501 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1502 * new feature: source path information is generated for macro-expansion
1503 errors for use in IDE's like Slime (thanks to Helmut Eller)
1504 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1505 * bug fix: compiler does not loop forever on an invalid type in
1507 * improvement: compilation of most CLOS applications is significantly
1509 * optimization: added a limited bytecode compiler for simple toplevel
1510 forms, speeding up compilation and FASL loading
1511 * bug fix: the statistical profiler now properly distinguishes anonymous
1514 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1515 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1516 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1517 into subclasses of the finalized class. Now user calls to
1518 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1519 system return before any subclasses are finalized.
1520 * minor incompatible change: The reader no longer ignores errors
1521 regarding non-existent packages in #+ and #- feature tests.
1522 * new feature: command line options --no-sysinit, --no-userinit to
1523 inhibit loading the corresponding init files
1524 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1525 for sending data through UDP sockets (thanks to François-René Rideau)
1526 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1527 error is signalled instead (thanks to François-René Rideau)
1528 * minor incompatible change: Error signalling behaviour of lexical
1529 operations violating package locks has changed slightly. Refer to
1530 documentation on package locks for details.
1531 * enhancement: EVAL can process IF-expressions without resorting to the
1533 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1534 constant-folded properly due to missing out-of-line functions.
1535 (reported by Mika Pihlajamäki)
1536 * bug fix: LISTEN sometimes returned T even in cases where no data was
1537 immediately available from the stream
1538 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1539 were reversed. (reported by Levente Mészáros)
1540 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1541 when given a vector displaced to an adjustable array. (reported
1543 * bug fix: derivation of float boundaries from numbers outside the
1544 appropriate float range (reported by John Wiseman)
1545 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1547 * bug fix: precomputing cache entries for generic functions with
1548 some subclasses of specializers as yet invalid does not attempt to
1549 fill a cache line with a negative offset. (reported by Levente
1551 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1552 allow condition class objects as arguments to DOCUMENTATION and
1553 (SETF DOCUMENTATION); only find and set documentation for
1554 structure names for the STRUCTURE doc-type. (suggested by Gary
1556 * improvements to the Win32/x86 port:
1557 ** (user-homedir-pathname) and default initialization file
1558 locations now know about the user's "Documents and Settings"
1559 directory (thanks to Yaroslav Kavenchuk)
1560 ** run-program is implemented (thanks to Mike Thomas)
1561 ** sockets support (thanks to Timothy Ritchey)
1562 ** better backtrace support (thanks to Alastair Bridgewater)
1563 ** sb-grovel supported
1564 ** asdf-install and sb-posix work somewhat
1565 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1566 * improvements to the Solaris/x86 port:
1567 ** works on Solaris 11/Solaris Express
1568 ** floating-point exception handling support
1569 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1570 * fixed some bugs revealed by Paul Dietz' test suite:
1571 ** REMOVE-METHOD returns its generic function argument even when
1572 the method is not one of the generic functions' methods.
1573 ** objects declared as MEMBER types can be admissible arguments to
1574 structure accessors.
1575 ** printing characters should simply be printed by the FORMAT ~:C
1577 ** compiler failure when compiling functions with hairy constant
1578 defaults for optional parameters.
1579 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1580 function, which is already optimized.
1582 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1583 * new platform: experimental support for SBCL x86/Darwin, including
1584 MacOS X 10.4.5 on Intel.
1585 * new feature: Unicode character names are now known to the system
1586 (through CHAR-NAME and NAME-CHAR).
1587 * new feature: the filesystem location of SBCL's core file is
1588 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1589 * minor incompatible change: the contrib modules SB-POSIX and
1590 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1591 this change is to make it easier to distribute
1592 location-independent binaries.
1593 * bug fix: as implied by AMOP, standardized classes no longer have
1594 slots named by external symbols of public packages. (reported by
1596 * optimization: calling functions via constant symbols -- as in
1597 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1598 via the function object as in (FUNCALL #'FOO).
1599 * optimization: on x86 and x86-64, the calling convention for the
1600 case of dealing with an unknown number of values has been altered
1601 to be friendlier to the prediction heuristics implemented,
1602 particularly with reference to CALL and RET pairing. (thanks to
1603 Alastair Bridgewater)
1604 * optimization: on x86, the code for access to thread-local symbol
1605 values and binding/unbinding of thread-local symbols is smaller.
1606 (thanks to Alastair Bridgewater)
1607 * enhancement: CONSTANTP is now able to determine constantness of
1608 more complex forms, including calls to constant-foldable standardized
1609 functions and some special forms beyond QUOTE.
1610 * fixed bug: occasional GC crashes on Solaris/x86
1611 * optimization: x86-64 supports stack allocation of results of simple
1612 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1613 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1614 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1615 (thanks to James Knight)
1616 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1617 by the standard (thanks to Stephen Compall)
1619 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1620 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1621 be used for bundling the runtime and the core file into one
1622 executable binary. This feature is not currently supported on all SBCL
1623 platforms. (thanks to James Bielman and NIIMI Satoshi)
1624 * new feature: a generational or ephemeral garbage collector is now
1625 the default on the PowerPC platform (both Linux and Darwin). The
1626 old Cheney (stop and copy) collector is a build-time option.
1627 (thanks to Cyrus Harmon, after Raymond Toy)
1628 * minor incompatible change: the method by which SBCL finds its
1629 contributed modules has changed; it no longer relies on symbolic
1630 links from an $SBCL_HOME/systems directory, but searches directly
1631 in the subdirectories of $SBCL_HOME.
1632 * enhancement: the dynamic heap size on the Linux/PPC platform is
1633 markedly larger, even using the older Cheney garbage collector.
1634 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1635 certain complicated string types. (reported by Gary King)
1636 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1637 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1638 * fixed bug: DOCUMENTATION on structures no longer signals an
1639 error if no structure type of the right name exists. (reported by
1641 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1642 applicable non-standard primary or :AROUND method on
1643 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1644 no longer cause unbound variable errors. (reported by Kevin Reid)
1645 * fixed bug: metacircle resolution in cases where methods have slots
1646 added before the slots from STANDARD-METHOD. (reported by Jean
1648 * fixed bug: the Power PC assembly code for calling into lisp
1649 sometimes computed its return address wrongly (depending on the
1650 15th bit of the address, fixed at link time). This used to
1651 manifest itself in a segmentation violation while building PCL.
1652 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1653 many others over the years)
1654 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1655 the stack pointer (treated as a signed integer) is negative.
1656 (thanks to Peter van Eynde)
1658 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1659 * new platform: experimental support for the Windows operating
1660 system has been added. (thanks to Alastair Bridgewater)
1661 * fixed several bugs in and robustified the PPC FFI (including
1662 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1663 * bug fix: allow non-simple string symbol names (reported by Paul
1665 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1667 * bug fix: saving a core corrupted callbacks on x86/x86-64
1668 * bug fix: closed a loophole in metacircularity detection and
1669 grounding in the PCL implementation of CLOS.
1670 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1671 (reported by tomppa on #lisp)
1672 * bug fix: clear the direction flag on Lisp -> C transitions, as
1673 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1674 (reported by Andrej Grozin and Hendrik Maryns)
1675 * optimization: major improvements to GC efficiency on GENCGC platforms
1676 * optimization: faster implementation of EQUAL
1677 * optimization: emit more efficient opcodes for some common
1678 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1679 * optimization: several other minor code-generation improvements on x86-64
1680 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1682 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1683 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1684 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1685 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1686 by ANSI as a generic function, and for consistency (SETF
1687 GENERIC-FUNCTION-NAME) is treated likewise.
1688 * fixed bug #233.b: make constraint propagation notice when a variable
1689 value is changed after it is referenced but before it is used
1690 * fixed bug #296: no more arbitrary behaviour from filenames with
1691 odd characters as --load arguments. (reported by Adam Warner)
1692 * fixed bug #390: :CHARACTER-SET pathname components now work as
1693 expected. (reported by Tim Daly Jr)
1694 * fixed bug #391: complicated :TYPE intersections in slot
1695 definitions no longer cause an error in PCL internals.
1696 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1697 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1698 returning the number of octets which would be written to the
1699 file-stream. (thanks to Robert J. Macomber)
1700 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1701 arguments to the internal stream functions. (thanks to David
1703 * bug fix: stack exhaustion is now handled on the x86/SunOS
1704 platform. (thanks to Daisuke Homma)
1705 * bug fix: more accurate ROOM results on GENCGC platforms
1706 * optimization: improved type inference for arithmetic-for
1707 index variables in LOOP
1708 * optimization: faster floating-point SQRT on x86-64
1709 * fixed some bugs revealed by Paul Dietz' test suite:
1710 ** DOCUMENTATION returns NIL instead of "" for method combinations
1711 that don't have a docstring
1713 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1714 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1715 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1716 therefore are not customizeable by user code (as seems to be at
1717 least permitted and maybe required by AMOP). As a consolation,
1718 however, the SBCL implementation of these functions now calls
1719 REINITIALIZE-INSTANCE as specified by AMOP.
1720 * bug fix: slot-definition documentation is propagated as per ANSI
1721 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1722 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1723 argument, as do eslotd initialization methods. (from Pascal
1724 Costanza's "Closer" project)
1725 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1726 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1728 * bug fix: it is now possible to have more than one subclass of
1729 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1730 (reported by Bruno Haible, Pascal Costanza and others)
1731 * bug fix: the dependent update protocol now works for generic
1732 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1733 and Pascal Costanza)
1734 * bug fix: condition-class instances corresponding to
1735 DEFINE-CONDITION forms are now created eagerly. (reported by
1736 Kalle Olavi Niemitalo on comp.lang.lisp)
1737 * bug fix: floating point printing is more accurate in some
1738 circumstances. (thanks to Simon Alexander)
1739 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1740 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1741 * bug fix: callbacks on OS X now preserve stack-alignment.
1742 (thanks to Cyrus Harmon)
1743 * enhancement: the x86-64 disassembler is much better at
1744 disassembling SSE instructions. (thanks to Lutz Euler)
1745 * enhancement: CHECK-TYPE on a variable now allows type-inference
1746 (thanks to Vincent Arkesteijn)
1747 * optimization: improved performance of EUC-JP external format.
1748 (thanks to NIIMI Satoshi)
1749 * optimization: performance improvements to IO on file streams of
1750 :ELEMENT-TYPE CHARACTER
1751 * optimization: much faster memory allocation on x86-64
1752 * optimization: faster garbage collections (latency and throughput)
1754 * optimization: faster bignum subtraction and division on x86 and x86-64
1755 (thanks to Lutz Euler)
1756 * optimization: more accurate type-inference for loops that use a
1757 floating point index variable or a negative step.
1759 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1760 * bug fix: add a workaround to SBCL looping infinitely at startup on
1761 Linux kernels with apparently buggy implementations of personality().
1762 (thanks to Svein Ove Aas)
1763 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1764 (thanks to David Lichteblau)
1765 * bug fix: local bindings shadow global symbol macros for
1766 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1767 * optimization: non-open coded uses of numeric comparison operators
1768 (e.g. >) no longer cons when called with more than one parameter
1769 on platforms supporting dynamic-extent allocation.
1770 * enhancement: saving cores with foreign code loaded is now supported
1771 on MIPS/Linux in addition to the previously supported platforms.
1772 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1773 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1774 * bug fix: division by zero in sb-sprof when no samples were collected
1775 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1776 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1777 (thanks to Alexander Kjeldaas)
1778 * new feature: ignoring the return values of destructive functions
1779 like NREVERSE signals a compile-time style-warning.
1780 (thanks to Kevin Reid)
1781 * new feature: the alignment of alien structure fields can be
1782 explicitly specified. (thanks to Cyrus Harmon)
1783 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1784 and dump core on SIGQUIT
1786 ** incompatible change: threads do not inherit values of specials
1787 from their parents (see manual)
1788 ** bug fix: threads stacks belonging to dead threads are freed by the
1789 next exiting thread, no need to gc to collect thread stacks anymore
1790 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1791 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1792 ** bug fix: dynamic variable and thread start related gc lossage
1793 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1795 * fixed some bugs revealed by Paul Dietz' test suite:
1796 ** SUBTYPEP is slightly more accurate on heinously complicated
1797 CONS types where some of the members have uncertain (in the
1798 NIL, NIL sense) type relationships to each other.
1800 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1801 no longer purified unless :PURIFY T is explicitly specified.
1802 ** Non-purified cores are significantly smaller than before
1804 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1805 * new feature: timers based on Zach Beane's excellent timer package
1806 * added support for the following external formats: koi8-u,
1807 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1808 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1809 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1810 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1811 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1812 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1813 * incompatible change: a threaded SBCL will no longer revert to
1814 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1815 * bug fix: interrupts are disabled until startup is complete; no
1816 more sigsegvs when receiving a signal to soon
1817 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1819 * bug fix: add a workaround for the memory randomization features in
1820 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1821 workaround will only be in effect on systems with the proc filesystem
1823 * bug fix: printing objects of type HASH-TABLE signals a
1824 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1826 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1827 have been running for over 50 days. (reported by Gilbert Baumann)
1828 * bug fix: the logic for getting names of functions gets less
1829 confused when confronded with alternate-metaclass
1830 funcallable-instances. (reported by Cyrus Harmon)
1831 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1832 consistent, even on internal alternate-metaclass objects.
1833 * bug fix: flush closure information collected by physical
1834 environment analysis prepass before the main pass. (bug reported
1836 * bug fix: compiler pack phase does not modify a hashtable, which is
1837 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1838 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1839 may be expanded. (reported by Alexander <malishev> on c.l.l)
1840 * bug fix: no more style-warnings when compiling code using LAST
1841 under high-SPEED low-SPACE settings. (reported by David Wallin
1843 * fixed some incompatibilities between SBCL's MOP and the MOP
1845 ** the METAOBJECT class is now implemented;
1846 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1847 STANDARD-OBJECT, as required;
1848 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1849 now compatible (as required by VALIDATE-SUPERCLASS); there
1850 remains a consistency requirement over the presence or absence
1851 of the FUNCTION class in the superclass list of finalized
1852 classes; see the manual for more details;
1853 ** the :AROUND method for COMPUTE-SLOTS on
1854 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1855 requested slot ordering.
1857 ** bug fix: parent thread now can be gc'ed even with a live
1859 ** bug fix: binding a special with PROGV to no value is not
1860 the same as that symbol not having been bound (thanks to
1862 ** bug fix: binding specials is thread safe (thanks to
1864 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1865 and SB-SPROF do not die with 'no handler for signal XX in
1866 interrupt_handle_now(..)' anymore
1867 ** bug fix: WITH-TIMEOUT works with multiple running threads
1868 * fixed some bugs revealed by Paul Dietz' test suite:
1869 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1870 the :method-class keyword argument.
1872 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1873 * new port: the Solaris operating system on x86 processors is now
1874 mostly supported, though some rough edges in the environment
1875 remain. (thanks to Daisuke Homma)
1876 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1877 space for its heap. (thanks to Thiemo Seufer)
1878 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1879 closures. (thanks to Thiemo Seufer)
1880 * minor incompatible change: eof selects abort in the debugger.
1881 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1882 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1884 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1885 deprecated, and will go away in a future revision of SBCL.
1886 * minor incompatible change: GC-ON and GC-OFF are no longer
1887 implemented with a counter, it does not matter how many times gc
1888 is switched on or off
1889 * bug fix: discriminating functions for generic function classes
1890 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1891 make invalid assumptions about method precedence order. (reported
1893 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1894 platforms. (thanks to Thiemo Seufer)
1895 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1896 * bug fix: correct pathname printing with printer escaping is on.
1897 (thanks to Kevin Reid)
1898 * bug fix: complex VOP definitions in "user-space" no longer trigger
1899 package locks. (reported by Zach Beane)
1900 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1901 for funcallable-instance functions, and is no different from
1903 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1904 characters in the current readtable. (reported by Nicholas Neuss)
1905 * bug fix: SUBTYPEP on various CONS types returns more a more
1906 accurate acknowledgment of its certainty.
1907 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1908 lists in some cases. This partially fixes bug 384.
1909 * flush all standard streams before prompting in the REPL and the
1911 * bug fix: signal handling and triggering gc do not conflict
1912 directly anymore, in particular a high frequency sb-sprof does
1913 not prevent gc from running
1914 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1915 approximation for timezone and DST information between the
1916 universal time epoch and the smallest negative 32-bit time_t.
1917 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1918 year 1899 so long as the encoded time is non-negative.
1919 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1920 the correct answer for non-constant rotate amounts when performing
1921 an inline 32-bit rotation.
1923 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1924 there is only one thread in the session
1925 ** bug fix: memory leak for streams created in one thread and
1926 written to in another
1927 ** bug fix: lockup when compiled with gcc4
1928 ** bug fix: race that allows the gc to be triggered when gc is
1930 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1931 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1933 ** bug fix: finished threads are gc'ed properly
1934 * fixed some bugs revealed by Paul Dietz' test suite:
1935 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1936 has arguments with a different set of applicable methods from
1937 the orignal arguments.
1938 ** The type error thrown by MAP now has a correct expected-type
1940 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1941 name a compiled function.
1942 ** Three MISC tests where a large bignum was improperly coerced to
1943 a float that couldn't represent that bignum during type
1944 derivation were fixed.
1945 ** SUBTYPEP can now handle types involving the negation of a
1946 list-form FUNCTION type.
1947 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1948 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1949 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1951 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1952 * New feature: Experimental support for bivalent streams: streams
1953 opened with :element-type :default now allow character and binary
1954 (unsigned-byte 8) I/O
1955 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1956 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1958 * bug fix: no more highly sporadic "couldn't check whether ~S is
1959 readable" when reading a stream and an interrupt hits in the middle
1960 of a select system call
1961 * compiler better recognizes complex arrays (reported by Eduardo
1963 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1964 for variables on SBCL built with linkage-tables. (reported by Luis
1966 * various error reporting improvements.
1967 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1968 (Thanks to Thiemo Seufer)
1969 * bug fix: nested reader invokations work correctly
1970 * bug fix: it is possible to have simultaneous references to foreign
1971 code and foreign data with the same name.
1973 ** added x86-64 support
1974 ** incompatible change: the threading api now works with thread
1975 objects instead of thread ids
1976 ** bug fix: threads are protected from signals and interruption when
1977 starting up or going down
1978 ** bug fix: a race where an exiting thread could lose its stack to gc
1979 ** bug fix: don't halt on infinite error in threads if possible
1980 ** fixed numerous gc deadlocks introduced in the pthread merge
1981 ** bug fix: fixed thread safety issues in read and print
1982 ** bug fix: debugger doesn't hang on session lock if interrupted at
1983 an inappropriate moment
1984 ** bug fix: run-program is now thread safe(r)
1985 ** bug fix: inner with-recursive-lock no longer releases the mutex
1986 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1987 * fixed some bugs revealed by Paul Dietz' test suite:
1988 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1989 EXPECTED-TYPE slots filled.
1990 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1992 ** MAKE-CONDITION accepts classes as type-designators.
1993 ** COMPILE may never return NIL.
1994 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1995 range before calling Unix time functions
1997 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1998 * numerous signal handling fixes to increase stability
1999 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
2000 * minor incompatible change: we now correctly canonize default
2001 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
2002 as per the MOP, rather than the historical (INITARG INITFUNCTION
2003 INITFORM). (reported by Bruno Haible)
2004 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
2005 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
2006 for more information.
2007 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
2008 pathname is a directory pathname.
2009 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
2010 * dynamic space size on PPC has been increased to 768Mb. (thanks to
2012 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
2013 :METACLASS argument in addition to a class name. (reported by
2014 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
2015 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
2016 (reported by Bruno Haible; patch from Gerd Moellmann)
2017 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
2019 * bug fix: more cleanups to the floating point exception handling on
2020 x86-64 (thanks to James Knight)
2021 * bug fix: the compiler does not try to constant fold calls of
2022 COERCE and BIT-* functions when they return freshly constructed
2023 objects. (reported by Nikodemus Siivola and Paul Dietz)
2024 * optimization: improved the MIPS versions of generic (in the
2025 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
2026 * optimization: direct conversion of (unsigned-byte 32) to floats on
2027 the PowerPC platform.
2028 * optimization: structure instances with raw slots now use less
2029 memory, and probably show better memory locality. (thanks to
2031 * optimization: DYNAMIC-EXTENT declarations for lists and closures
2032 are treated as requests for stack allocation on the x86-64,
2033 Alpha-32, PPC and SPARC platforms.
2034 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
2035 the wrong file descriptor; implementation of SOCKET-OPEN-P.
2036 (thanks to Tony Martinez)
2038 ** gcing a dead thread can no longer lead to lockups
2039 ** threads block signals until they are set up properly
2040 ** errno is no longer shared by threads
2041 ** interrupt-thread restores the eflags register on x86
2042 ** fixed some lockups due to gc/thread interaction
2043 * fixed some bugs revealed by Paul Dietz' test suite:
2044 ** invalid dotted lists no longer raise a read error when
2045 *READ-SUPPRESS* is T
2046 ** signal an error if a symbol that names a declaration is used
2047 as the name of a type, or vice versa
2048 ** allow using the (declare (typespec var*)) abbreviation for
2049 (declare (type typespec var*)) with all type specifiers
2050 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
2051 ** always return NIL from PPRINT-POP when OBJECT is NIL
2052 ** don't signal errors when pretty-printing malformed LABELS,
2053 FLET or MACROLET forms
2054 ** declarations in a DOLIST body are also in scope for the
2056 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2058 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2061 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2062 * fixed cross-compiler leakages that prevented building a 32-bit
2063 target with a 64-bit host compiler.
2064 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2065 opened with :IF-EXISTS :OVERWRITE.
2066 * fixed bug 281: error for an invalid qualifier in a short-form method
2067 combination method is not signalled until the faulty method is called.
2068 * bug fix: iteration variable type inferrer failed to deal with open
2069 intervals. (reported by Alan Shields)
2070 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2071 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2072 or more runtime options were provided to the sbcl binary.
2073 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2074 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2075 * optimization: there's now a fast-path for fixnum arguments in the
2076 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2077 * optimization: the code generated on x86-64 is more compact thanks
2078 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2079 * bug fix: floating-point exception handling now partly works on
2080 x86-64. (Thanks to James Knight)
2081 * improvement to the MIPS backend from Thiemo Seufer: C-style
2082 64-bit long long arguments and return values to alien functions
2084 * fixed some bugs revealed by Paul Dietz' test suite:
2085 ** the type-error signalled from WARN has a filled-in DATUM slot.
2086 ** the type-error required when a stream is not associated with
2087 a file has the stream as its datum.
2088 ** type-errors on single-floats on x86-64 no longer have
2089 :INVALID-OBJECT as the datum
2090 ** the type-errors signalled for invalid function names now have
2091 a correct expected type
2092 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2093 for structures defined with a :TYPE. Documentation strings for
2094 typed structures are no longer immediately discarded
2095 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2096 broadcast streams and synonym streams. FILE-LENGTH now also works
2097 on broadcast streams.
2099 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2100 * incompatible change: the --noprogrammer option, deprecated since
2101 version 0.7.5, has been removed. Please use the equivalent
2102 --disable-debugger option instead.
2103 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2105 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2106 inoperational for a while now) has been completely removed.
2107 * null lexical environments are now printed as #<NULL-LEXENV>,
2108 significantly reducing the amount of clutter in typical backtraces.
2109 * documentation on weak pointers, finalization, and after GC hooks
2110 has been added to the manual.
2111 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2112 as other specialized array types, is much faster. SUBSEQ and
2113 COPY-SEQ on such arrays have also been sped up.
2114 * optimization: EQL is now more efficient when at least other argument
2115 is known to be of type (OR FIXNUM (NOT NUMBER)).
2116 * fixed bug: the runtime is now less vulnerable to changes in the
2117 size of the SBCL object on OS X, and virtual memory is reserved for
2118 all spaces that need to be at a fixed address.
2119 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2120 * fixed bug: finalizers and after GC hooks that cause consing are now
2122 * fixed bug: compiler error messages and summaries are now printed to
2123 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2124 * fixed inference of the upper bound of an iteration variable.
2125 (reported by Rajat Datta).
2126 * fixed bug 211e: calling local functions with duplicated constant
2127 keyword argument no longer causes a bogus style warning about an
2129 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2130 ftype declaration to be disregarded. (reported by Dave Roberts)
2131 * fixed bug 373: caused by erronous compilation of references to alien
2132 variables in the runtime on ppc/darwin.
2133 * fixed bug 376: CONJUGATE type deriver.
2134 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2135 (reported by Baughn on #lisp)
2136 * a cross-compiler bug on non-x86 platforms has been identified and
2137 fixed. (thanks to Bruno Haible)
2138 * improvements to the MIPS runtime code for increased stability.
2139 (thanks to Thiemo Seufer)
2140 * increased the maximimum compact environment size to allow
2141 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2142 * partial workaround for bug 135: don't name the function we're
2143 calling for hairy cases of EVAL, so as not to accumulate one environment
2144 entry per call to EVAL. (reported by Kevin Reid)
2145 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2146 * optimization: type testing for non-vector arrays should be faster.
2147 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2149 * bug fix: setting 31st element of a bit vector to zero did not work
2151 * bug fix: redefining a class definition which failed due to a
2152 previous accessor / function clash now works (but see BUGS entry
2153 #380 for more problems in this area). (thanks to Zach Beane)
2154 * the long form of DEFINE-METHOD-COMBINATION disables method group
2155 checking when given a single method group with pattern *. (thanks
2157 * on x86 compiler supports stack allocation of results of simple
2158 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2159 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2160 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2161 * fixed some bugs related to Unicode integration:
2162 ** the restarts for recovering from input and output encoding
2163 errors only appear when there is in fact such an error to
2165 * fixed some bugs revealed by Paul Dietz' test suite:
2166 ** MISC.549 and similar: late transformation of unsafe type
2167 assertions into derived types caused unexpected code
2169 ** SCALE-FLOAT type deriver is less wrong.
2170 ** type derivers for EXP, LOG and similar functions compute result
2171 types for complex arguments better.
2172 ** (MISC.563) CONJUGATE type deriver works for very restricted
2174 ** out-of-line type testers for character strings are available.
2175 ** EQUAL compiler transform understands specialness of objects
2177 ** accessing double-floats stored on the stack now works on x86-64.
2178 ** debugger internals could sometimes create invalid lispobjs,
2179 resulting in GC crashes.
2180 ** MISC.548: type check weakening can convert required type into
2182 ** initialization forms for bindings are not in scope of free special
2185 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2186 * incompatible change: thread support for non-NPTL systems has
2187 been removed - locking is buggy and unreliable. A threaded
2188 SBCL build will now warn at startup and refuse to create
2189 new threads, unless futex support is detected in the kernel
2190 * incompatible change: the top level REPL now has only an ABORT
2191 restart associated with it, not TOPLEVEL and ABORT as it used to.
2192 TOP and TOPLEVEL are now available as debugger commands for
2193 returning to the top level.
2194 * incompatible change: forms evaluated in the REPL now use the
2195 global optimization policy.
2196 * incompatible change: user- and system-initialization files are
2197 no longer processed with LOAD, but by READ and EVAL; hence the
2198 global optimization policy, startup package, readtable, etc,
2200 * The .fasl file format number has been incremented because of
2201 various incompatible changes.
2202 * internal entry point details and argument counts no longer appear
2203 in backtraces unless explicitly requested by setting
2204 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2205 * built-in and standard functions no longer have names like "top
2206 level local call to FOO".
2207 * fixed bug 32: functions defined in non-null lexical environments
2208 now have more legible printed representation
2209 * fixed bug 33: functions defined in non-null lexical environemnts
2210 are now more amenable to inspection by INSPECT.
2211 * workaround for bug 354: XEPs no longer appear in backtraces unless
2212 explicitly requested.
2213 * fixed bug: receiving the signal which results from attempting to
2214 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2215 SIGBUS on other *BSDs) on architectures where the C stack is also
2216 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2217 to debugger confusion or wild instability. (reported by Cheuksan
2219 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2220 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2221 (reported by Lutz Euler)
2222 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2223 compiled code. (reported by Paul Dietz)
2224 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2225 specialized parameter to a method to something that is not TYPEP
2226 the specializer is now possible.
2227 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2228 face of package deletion.
2229 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2230 pathname. (reported by David Tolpin on c.l.l)
2231 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2232 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2233 STRING-STREAM and FILE-STREAM, respectively.
2234 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2235 than previously; in addition, its test suite is now run on build.
2236 (thanks to Luke Gorrie)
2237 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2238 * optimization: added a immediate representation for single-floats
2240 * fixed some bugs revealed by Paul Dietz' test suite:
2241 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2242 ** attempting to create a package with a colliding nickname causes
2243 correctable errors to be signalled.
2244 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2245 ** several x86-64 backend bugs related to sign-extension of immediate
2248 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2249 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2250 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2252 * fixed loading of multiply forward-referenced layouts.
2253 (thanks to Cheuksan Wang)
2254 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2255 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2256 (see COMPILE-FILE documentation for details of :PRINT option)
2257 * fixed bugs 19 and 317: fixed-format floating point printing is
2258 more accurate. This also fixes a bug reported by Adam Warner
2259 related to the ~@F format directive.
2260 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2262 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2263 dispatch table if the from-char is a dispatch macro character.
2264 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2265 by the wordsize no longer ignore the last word. (reported by Lutz
2267 * fixed bug in type checking of dynamic-extent variables. (reported
2269 * optimization: sequence traversal functions use their freedom to
2270 coerce function designators to functions.
2271 * optimization: code with many calls to CLOS methods specialized on
2272 CLOS classes has had redundant type checks removed.
2273 * optimization: added declarations to speed up operations that access
2274 the internal character database (for example STRING-UPCASE)
2275 * optimization: comparison operations between floats and sufficiently small
2276 fixnums no longer create extra rationals
2277 * fixed some bugs related to Unicode integration:
2278 ** portions of multibyte characters at the end of buffers for
2279 character-based file input are correctly transferred to the
2280 start of the buffer at the next read.
2281 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2282 passing it through to OPEN.
2283 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2284 argument given to internal calls to OPEN.
2285 ** the built-in comment readers (introduced by character sequences
2286 ";" and "#|") are more forgiving to encoding errors; they will
2287 STYLE-WARN and then attempt to resync the stream at a character
2288 boundary. (thanks to Teemu Kalvas)
2289 * fixed some bugs in the x86-64 port:
2290 ** Negative short int return values from c-calls are sign-extended
2292 ** The stack is aligned to 16-bytes for c-calls, as required by
2293 the ABI. (thanks to Cheuksan Wang)
2294 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2295 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2296 for enums. (thanks to Vincent Arkesteijn)
2297 ** Multiple small optimizations and bugfixes for floating point
2299 * fixed some bugs revealed by Paul Dietz' test suite:
2300 ** Space, Tab, Linefeed, Return and Page have the invalid
2301 secondary constituent character trait.
2302 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2304 ** WITH-INPUT-FROM-STRING should only update the index place on
2306 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2307 works more reliably.
2308 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2309 with one-dimensional arrays with fill pointers.
2310 ** TYPE-OF failed on a complex with an integer realpart and a
2312 ** compiler failure during type inference for the code of form
2313 (IF (EQL X (THE ...)) ...) (MISC.535).
2315 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2316 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2317 platform. The port supports 61-bit fixnums, large memory spaces
2318 and reloading shared object files.
2319 * enhancement: saving cores with foreign code loaded is now
2320 supported on ppc/Darwin in addition to the previously supported
2322 * enhancement: the statistical profiler now walks deeper into the
2323 call stack for more meaningful call-graphs and accrued time
2324 reports (x86/x86-64 only). It also now reports time spent in
2326 * enhancement: it is now possible to trace most individual methods
2327 of a generic function in addition to tracing the generic function
2329 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2331 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2332 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2333 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2334 produces an error. (thanks to Vincent Arkesteijn)
2335 * bug fix: NAMESTRING on pathnames with :WILD components in their
2336 directories works correctly. (thanks to Artem V. Andreev)
2337 * fixed bug 125: compiler preserves identity of closures. (reported
2339 * bug fixed: functions with &REST arguments sometimes failed with
2340 "Undefined function" when compiled with (DEBUG 3). (reported by
2342 * bug fix: overflow during compiling of setting element of a bit
2343 vector with constant index and value. (reported by Timmy Douglas)
2344 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2345 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2346 * on x86 compiler partially supports stack allocation of dynamic-extent
2348 * GO and RETURN-FROM do not check the extent of their exit points
2349 when compiled with SAFETY 0.
2350 * fixed some bugs related to Unicode integration:
2351 ** encoding and decoding errors are now much more robustly
2352 handled; it should now be possible to recover even from invalid
2353 input or output to the terminal. (thanks to Teemu Kalvas)
2354 ** provided a first cut at implementing STRING-TO-OCTETS and
2355 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2356 ** altered the SB-MD5 contributed module slightly, changing the
2357 interface just enough for it to be supportable for builds where
2358 lisp characters are not eight bits.
2359 * fixed some bugs revealed by Paul Dietz' test suite:
2360 ** the FORMATTER-generated functions for ~V[ conditionals require
2361 the correct number of arguments.
2362 ** READ-FROM-STRING returns the mandated second value when applied
2363 to displaced strings.
2364 ** the #\Rubout and #\Backspace characters are treated as invalid
2365 constituent characters by the tokenizer.
2367 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2368 * new feature: reloading changed shared object files with
2369 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2370 * new feature: references to foreign variables and functions
2371 can now be compiled and loaded before the corresponding shared
2372 object file is loaded, as long as the foreign definitions are
2373 available at runtime.
2374 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2375 supported. (thanks to Dan Debertin)
2376 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2377 just keywords. (thanks to Vincent Arkesteijn)
2378 * fixed bug #331: structure-class instances corresponding to
2379 DEFSTRUCT forms are now created eagerly.
2380 * fixed bug #345: backtraces from calls to undefined functions work
2381 on x86 as well. Related bug #61 is now also partially fixed on x86
2382 (backtraces from throws to unknown catch tags.)
2383 * bug fix: lambda-list parsing is now stricter vrt. order and number
2384 of lambda-list keywords.
2385 * bug fix: as specified by AMOP, an error is signalled if a
2386 class-option appears multiple times in a DEFCLASS form. (reported
2388 * bug fix: RANDOM can be compiled when the compiler derives the type
2389 of its numeric argument as a disjoint set of small integers.
2390 (reported by Paul Dietz)
2391 * bug fix: starting a core saved with shared objects loaded when
2392 those objects are not available no longer causes threaded SBCL to
2393 hang. (reported by Sean Ross)
2394 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2396 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2397 argument, which must always be NIL. (reported by Kalle Niemitalo)
2398 * bug fix: printing 1.0d+23 no longer results in an error.
2399 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2400 * bug fix: structure slot setters preserve evaluation order. (thanks
2402 * fixed some bugs related to Unicode integration:
2403 ** RUN-PROGRAM can allow its child to take input from a Lisp
2404 stream. (reported by Stefan Scholl)
2405 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2406 locales. (reported by Ken Causey)
2407 * fixed some bugs revealed by Paul Dietz' test suite:
2408 ** INCF, DECF and REMF evaluate their place form as specified in
2410 ** FORMATTER expands ~{ ~} iteration directives with V or #
2411 parameters correctly.
2412 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2413 consequent uses no arguments correctly.
2414 ** the system has a partial understanding of the (COMPLEX RATIO)
2416 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2419 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2420 * new feature: a build-time option (controlled by the :SB-UNICODE
2421 keyword feature, enabled by default) for building the system with
2422 support for the entire 21-bit character space defined by the
2424 * new feature: the system now has rudimentary external-format
2425 support; the primary user-visible change at this time is that
2426 characters with the high bit set (such as lower-case-e-acute) will
2427 print correctly to a terminal in a UTF-8 environment.
2428 * minor incompatible change: BASE-CHAR no longer names a class;
2429 however, CHARACTER continues to do so, as required by ANSI.
2430 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2431 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2432 has been moved to the SB-EXT package (temporarily re-exported from
2434 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2435 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2437 * fixed bug #351: better error-handlind and reporting for malformed
2439 * fixed bug #350: bignum-printing is now more memory-efficient,
2440 allowing printing of very large bignums, eg. (expt 2 10000000).
2441 (reported by Bruno Haible)
2442 * fixed bug #302: better primitive-type selection for intersection
2444 * fixed bug #308: non-graphic characters now all have names, as
2445 required. (reported by Bruno Haible)
2446 * bug fix: redefining a class with different superclasses now correctly
2447 removes it from the direct-subclasses of its previous superclasses.
2448 (reported by David Morse)
2449 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2450 new value now works. (reported by Bruno Haible)
2451 * bug fix: correct canonicalization of multiple non-standard slot
2452 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2453 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2454 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2455 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2456 now exists, an signals an error.
2457 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2458 during macroexpansion for non-iteration variables as well. (reported
2459 by Bruno Haible for CMUCL)
2460 * bug fix: Cyclic structures and unprintable objects in compiler
2461 messages no longer cause errors. (reported by Bruno Haible)
2462 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2463 and READ-FROM-STRING all now return a primary value of NIL if
2464 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2465 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2466 (reported by Bruno Haible for CMUCL)
2467 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2468 do not designate a condition. (reported by Bruno Haible for
2470 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2471 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2472 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2473 permitting resolution in favour of any of the conflicting symbols.
2474 (reported by Bruno Haible for CMUCL)
2475 * bug fix: EQUAL compiler optimizations is less aggressive on
2476 strings which can potentially compare true despite having distinct
2477 specialized array element types.
2478 * bug fix: unit enumerations can be defined without dividing by
2479 zero. (thanks to Vincent Arkesteijn)
2480 * FORMAT compile-time argument count checking has been enhanced.
2481 (report from Bruno Haible for CMUCL)
2482 * a partial workaround for the bug 262: the compiler does not try to
2483 inline-expand a local function doing RETURN-FROM from a deleted
2484 BLOCK. (thanks to Peter Denno for the bug report and to David
2485 Wragg for the simple test case)
2486 * fixed some bugs revealed by Paul Dietz' test suite:
2487 ** RENAME-PACKAGE allows all package designators as new package
2489 ** constraint propagation and lambda variable substitution are
2490 more cautious in dealing with partially deleted code.
2491 ** compiler sometimes lost reoptimization passes.
2492 ** CERROR, given a condition as condition designator, passes its
2493 remaining arguments to the continue format control without
2495 ** Case-altering FORMAT directives work correctly on non-ASCII
2497 ** The REAL type specifier handles bounds outside the floating
2498 point ranges without signalling FLOATING-POINT-OVERFLOW.
2499 ** Functions with IR1-transformations can create intercomponent
2500 references to global functions.
2501 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2503 ** FORMAT ~R treats a nil value for its first parameter correctly.
2505 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2506 * enhancement: saving cores with foreign code loaded is now
2507 supported on x86/NetBSD and sparc/Linux in addition to the previously
2508 supported platforms.
2509 * bug fix: on some platforms repeated installations caused multiple
2510 copies of HTML documentation to be installed -- should not happen
2511 any more. (reported by Stefan Scholl)
2512 * bug fix: parsing self-recursive alien record types multiple times
2513 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2514 original patch by Helmut Eller for CMUCL)
2515 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2516 (thanks to Richard Kreuter)
2517 * bug fix: defining classes whose accessors are methods on existing
2518 generic functions in other (locked) packages no longer signals
2519 bogus package lock violations. (reported by François-René Rideau)
2520 * bug fix: special variables as DEFMETHOD parameters no longer have
2521 associated bogus type declarations. (reported by David Wragg and
2523 * bug fix: read-write consistency on streams of element-type
2524 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2525 * bug fix: redefiniton of the only method of a generic function with
2526 no DEFGENERIC no longer emits a full WARNING. In addition,
2527 redefinition of generic functions with no DEFGENERIC to an
2528 incompatible lambda list now signals an error. (thanks to Zach
2530 * bug fix: DEFGENERIC now works even when there's a function of the
2531 same name in an enclosing lexical environment. (thanks to Zach
2533 * fixed compiler failure, caused by instrumenting code during
2534 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2535 * optimization: added loop analysis and improved register allocation
2536 * fixed some bugs revealed by Paul Dietz' test suite:
2537 ** POSITION on displaced vectors with non-zero displacement
2538 returns the right answer.
2539 ** (SIMPLE-STRING) is a valid type specifier for sequence
2541 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2543 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2544 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2546 ** The justification version of the FORMAT ~< directive treats
2547 non-zero minpad parameter correctly.
2549 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2550 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2551 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2552 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2553 the supported interface.
2554 * new feature: Single-stepping of code compiled with DEBUG 2 or
2555 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2556 * new feature: saving cores with foreign code loaded is now
2557 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2558 Timothy Moore's work for CMUCL)
2559 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2560 parameters to * instead of NIL if no initform is supplied.
2561 (reported by Johan Bockgaard)
2562 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2563 correctly when FUNCALL appears as the car of the form. Note:
2564 despite this FUNCALL forms are not currently subject to
2565 compiler-macro expansion. (port of Raymond Toy's fix for the
2566 same from CMUCL, reported by Johan Bockgaard)
2567 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2568 (thanks for Teemu Kalvas)
2569 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2570 case instead of terminating the iteration (thanks for Julian
2571 Squires, Sean Champ and Raymond Toy)
2572 * bug fix: incorrect expansion of defgeneric that caused a style
2573 warning. (thanks for Zach Beane)
2574 * bug fix: slot accessor effective method computation works properly
2575 for classes with multiple non-standard applicable methods on
2576 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2577 * on x86 compiler supports stack allocation of results of LIST and
2578 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2579 CMUCL implementation by Gerd Moellmann)
2580 * fixed some bugs revealed by Paul Dietz' test suite:
2581 ** FORMAT strings with both the ~<~:;~> form of the justification
2582 directive and pretty-printing directives cause an error.
2584 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2585 * incompatible change: the internal functions
2586 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2587 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2588 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2589 instead of the old functions.
2590 * new feature: on platforms where "dladdr" is available foreign
2591 function names now appear in backtraces. (based on Helmut Eller's
2593 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2594 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2596 * documentation: ASDF manual is now distributed with SBCL.
2597 * bug fix: SBCL can now load its contributed modules with REQUIRE
2598 even if the system-provided entries have been removed from
2599 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2601 * bug fix: backtraces involving undefined functions or assembly
2602 routines are more informative. (thanks to Brian Downing)
2603 * bug fix: mutually referent alien structures now work correctly.
2604 (reported by Rick Taube)
2605 * bug fix: structures defined by WITH-ALIEN can be referred to
2606 within other definitions in the same WITH-ALIEN.
2607 * bug fix: division operators (MOD, TRUNCATE and the like) with
2608 constant zero divisors and integer dividends no longer generate
2610 * bug fix: provide default methods for INPUT-STREAM-P and
2611 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2612 * bug fix: improve the detection and resolution of MOP metacycles
2613 (where effective method computation depends on the generic function
2614 having its effective method computed). (reported by Bruno Haible)
2615 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2616 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2617 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2618 * bug fix: pathnames with relative directory components can be
2619 represented relative to default pathnames.
2620 * optimization: in taking the GCD of bignums, reduce the two bignums
2621 to approximately the same size (using Euclid's algorithm) before
2622 applying the more sophisticated binary GCD. (thanks to Juho
2624 * optimization: COUNT on bitvectors now operates word-at-a-time.
2625 * optimization: ASH with a positive, but not necessarily constant,
2626 (leftwards) shift, when in a modular context, is compiled to a
2628 * fixed some bugs revealed by Paul Dietz' test suite:
2629 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2630 if the corresponding argument is NIL.
2631 ** FORMAT directives accepting a minpad parameter treat negative
2633 ** literal commas as character parameters to FORMAT directives are
2635 ** literal spaces directly after ~<Newline> directives within a
2636 format-logical-block (~:< ~@:>) do not induce :FILL-style
2637 conditional newlines.
2638 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2639 as specified: it no longer includes conditional newlines).
2640 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2642 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2643 :PER-LINE-PREFIX argument does not evaluate to a string.
2645 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2646 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2647 the manual for details; for now, package locks can be disabled by
2648 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2649 if no major problems are found then it is likely that they will be
2650 compiled in unconditionally.
2651 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2652 unsupported operators on all platforms. To load a shared library
2653 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2654 object file, link it into a shared library outside of SBCL and
2655 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2657 * bug workaround: ROOM T and the GC were not working together
2658 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2659 reporting facility aren't preserved. That reporting has been
2660 disabled, so now until and unless someone figures out how to make it
2661 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2662 * minor incompatible change: as threatened around sbcl-0.8.0, the
2663 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2664 The symbols are also exported from SB-PCL for backwards
2665 compatibility, but more so than before SB-PCL should be treated as
2666 an implementation-internal package.
2667 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2669 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2670 It is known as of this release not to work on the Alpha, however.
2671 * fixed bug #167: errors signalled due to illegal syntax in method
2672 bodies are now more legible.
2673 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2674 designators and can hence be used with TYPEP.
2675 * fixed bug #333: CHECK-TYPE now ensures that the type error
2676 signalled, if any, has the right object to be accessed by
2677 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2678 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2679 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2681 * fixed bug #334: programmatic addition of slots using specialized
2682 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2683 and :ALLOCATION :CLASS slots.
2684 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2685 float exponents. (rereported by Peter Seibel)
2686 * fixed a bug: #\Space (and other whitespace characters) are no
2687 longer considered to be macro characters in standard syntax by
2688 GET-MACRO-CHARACTER.
2689 * fixed bug: initialization of condition class metaobjects no longer
2690 causes an instance of the condition to be created. (reported by
2692 * fixed bug: it is now possible to have slots such that
2693 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2694 description is neither :INSTANCE nor :CLASS.
2695 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2696 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2697 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2698 non-standard SLOT-DEFINITION-ALLOCATION values.
2699 * fixed bugs: various race conditions handling exiting threads.
2700 CL-PPCRE's thread torture test now passes.
2701 * fixed arguably-a-bug: GC time proportional to number of threads in
2702 system even when most of them are idle
2703 * optimization: improved performance of BIT and SBIT on bit-vectors.
2704 * .fasl file incompatibility: The fasl file version number has
2705 been incremented because of changes associated with package locks.
2707 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2708 * minor incompatible change: the system no longer provides
2709 optimization hints (conditions which are TYPEP
2710 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2711 mode; these hints are still emitted when higher SPEED optimization
2713 * new contrib module: a sampling profiler (profiling by statistical
2714 sampling, rather than by instrumenting functions) is available as
2715 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2716 * the behaviour of the standard function ED is now customizeable by
2717 third parties through a hook variable: see ED's documentation
2718 string for information on the protocol.
2719 * the compiler no longer emits efficiency notes for (FUNCALL X)
2720 when the type of X is uncertain under default optimization
2722 * fixed bug 276: mutating a binding of a specialized parameter to a
2723 method to something that is not TYPEP the specializer is now
2725 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2726 exhibit the required behaviour on the x86 platform. (thanks to
2727 Peter van Eynde, Eric Marsden and Bruno Haible)
2728 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2729 even for difficult arguments. (reported by Peter Graves)
2730 * fixed bug 141a: the backquote printer now descends quoted
2732 * fixed another bug in backquote printing: no more destructive
2733 modification of the form's list structure. (reported by Brian
2735 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2736 move between its address being taken and the call to
2737 interrupt_thread, fixing a crashing race condition.
2738 * the SB-POSIX contrib implementation has been adjusted so that it
2739 no longer exhibits ridiculously poor performance when constructing
2740 instances corresponding to C structs.
2742 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2743 * minor incompatible change: the sb-grovel contrib now treats C
2744 structures as alien (in the sense of SB-ALIEN) objects rather than
2745 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2746 has implications for memory management of client code
2747 (sb-grovel-returned objects must now be manually managed) and for
2748 type safety (alien objects now have full types).
2749 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2750 used to control emission of compiler diagnostics, rather than the
2751 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2752 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2753 quality should be considered deprecated.
2754 * install.sh now installs the user manual as well
2755 * (not quite a new documentable feature, but worth considering in
2756 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2757 beginnings of a semantically meaningful condition hierarchy is
2758 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2760 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2761 designator as the defaults argument.
2762 * fixed bug: Displaced arrays whose displaced-to array has become
2763 too small now cause ARRAY-DIMENSION to signal an error, providing
2764 for safer bounds-checking. (reported by Bruno Haible)
2765 * fixed bug: DEFCLASS slot definitions with identical :READER and
2766 :WRITER names now signal a reasonable error. (reported by Thomas
2768 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2770 * fixed bug: Invalid slot specification errors now print correctly.
2771 (thanks to Zach Beane)
2772 * fixed bug 320: Shared to local slot value transfers in class
2773 redefinitions now happen corrently from superclasses as
2774 well. (reported by Bruno Haible)
2775 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2777 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2778 improper lists correctly. (reported by Bruno Haible)
2779 * fixed bug 313: source-transform for <fun-name> was erroneously
2780 applied to a call of a value of a variable with name <fun-name>.
2781 (reported by Antonio Menezes Leitao)
2782 * fixed bug 307: The obsolete instance protocol ensures that
2783 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2784 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2785 remove dead unknown-values globs from the middle of the stack.
2786 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2787 easier to resume long computations after using *BREAK-ON-SIGNALS*
2788 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2789 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2790 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2791 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2792 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2793 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2795 * fixed a (fairly theoretical) bug in string printing: if
2796 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2797 does not have array-element-type equal to the most general string
2799 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2800 function name. (thanks to Nikodemus Siivola)
2801 * fixed bug: compile-time format string checker failed on
2802 non-closed ~{. (reported by Thomas F Burdick)
2803 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2804 don't warn on legal loop code involving more than one aggregate
2805 boolean. (thanks to Nikodemus Siivola)
2806 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2807 in keyword position to MAKE-PACKAGE was not being reported as
2809 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2810 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2812 * fixed bugs 280 and 312: the checking for multiple definitions in a
2813 file is less likely to become confused by uses of inline
2815 * fixed bug: the #S reader macro performs the keyword coercion
2816 specified for slot names. (reported by Kalle Niemitalo)
2817 * fixed bug: lambda lists may contain symbols whose names start with
2818 & but are not lambda-list-keywords; their occurrence triggers a
2820 * fixed bug 321: define-method-combination argument lambda lists do
2821 not cause invalid code to be generated when &optional or &aux
2822 variables have default values. (reported by Bruno Haible)
2823 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2824 have CLOS classes; this appears also to have fixed reported
2825 instability in stack exhaustion detection.
2826 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2827 MUFFLE-WARNING when no associated restart is present is now
2829 * optimization: rearranged the expansion of various defining macros
2830 so that each expands into only one top-level form in a
2831 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2833 * optimization: used a previously little-used slot in symbols to
2834 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2836 * fixed some bugs revealed by Paul Dietz' test suite:
2837 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2839 ** Fixed an optimization bug related to inheritance of initforms
2840 from local to shared slots.
2841 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2842 ** CAST optimizer forgot to flush argument derived type.
2843 ** print/read consistency on floats is now orders of magnitude
2844 more likely. (thanks also to Bruno Haible for a similar report
2846 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2847 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2848 for objects of type REAL. Make it so.
2849 ** SXHASH is specified (infelicitously) to respect similarity,
2850 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2851 it so. (thanks to Markus Ziegler)
2852 ** on the Alpha, the compiler succeeds in compiling functions
2853 returning a known number of arguments greater than 63.
2854 ** fixed handling of invalid NIL arguments in keyword position
2856 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2858 ** provided a readably-printable representation for RANDOM-STATE
2860 ** ensured that pathnames, where they have a namestring, always
2861 print using #P"..." syntax.
2863 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2864 * Support for the forthcoming 2.0 version of the NetBSD kernel
2865 running on x86 hardware has been added. (thanks to Perry
2866 E. Metzger most immediately, and others for their past work)
2867 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2868 current release version; SBCL's OpenBSD support had been broken
2869 since about the time of OpenBSD's switch to ELF binary format.)
2870 * [placeholder for DX summary]
2871 ** user code with &REST lists declared dynamic-extent, under high
2872 speed or space and low safety and debug optimization policy.
2873 * The manual has been converted to Texinfo format and the debugger
2874 chapter from the cmucl manual has been added.
2875 * A facility has been added to extract documentation strings from
2876 sbcl and store them as Texinfo-formatted snippets for inclusion in
2877 the manual (via Texinfo's @include directive)
2878 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2879 the test case to Dave Roberts)
2880 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2881 pointers. (reported by Sean Ross)
2882 * bug fix: PROFILE output is printed nicely even for large numerical
2883 values. (thanks to Zach Beane)
2884 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2885 greater than 32 handle EOF correctly.
2886 * bug fix: on X86 an immediate argument of the IMUL instruction is
2887 correctly printed in disassembly. (thanks to Lutz Euler)
2888 * bug fix: class slots in redefined classes preserve their old
2889 values. (thanks to Bruno Haible and Nikodemus Siivola)
2890 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2891 succeeds. (reported by Marco Baringer)
2892 * fixed some bugs revealed by Paul Dietz' test suite:
2893 ** READ-SEQUENCE now works on ECHO-STREAMs.
2894 ** RATIONALIZE works more according to its specification. (thanks
2897 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2898 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2899 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2900 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2901 should matter to you only if you rebind the printer control
2902 variables and then find you want different bindings in the
2903 debugger than in the ordinary execution of your program.)
2904 * The runtime build system has been tweaked to support building
2905 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2906 assembler and linker. (thanks to Nikodemus Siivola)
2907 * Unbound, undefined, undeclared variables now trigger full
2908 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2909 more often programmer error than deliberate exploitation of undefined
2911 * optimization: the hash algorithm for strings has changed to one
2912 that is less vulnerable to spurious collisions. (thanks to Juho
2914 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2915 needless bounds checking. (thanks to Juho Snellman)
2916 * optimization: implemented multiplication as a modular
2917 (UNSIGNED-BYTE 32) operation on the PPC backend.
2918 * fixed some bugs revealed by Paul Dietz' test suite:
2919 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2921 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2922 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2923 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2925 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2927 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2929 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2930 their output stream on EOF from read.
2931 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2932 have been read to end-of-file.
2933 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2935 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2936 description of determination of which consecutive characters
2938 ** Printing the "Space" character with escaping on now yields "#\\ ",
2939 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2940 ** Reading floating-point numbers with *READ-BASE* set to a number
2941 less than 10 works correctly.
2942 ** Reading floating-point numbers with *READ-BASE* set to a number
2943 more than 10 works correctly.
2944 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2945 the readtable currently in effect.
2947 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2948 * minor incompatible change: parsing of namestrings on a physical
2949 (Unix) host has changed; numbers after the final #\. in a
2950 namestring are no longer interpreted as a version field. This is
2951 intented to be largely invisible to the user, except that the
2952 meaning of the namestring "*.*.*" has changed: it now refers to a
2953 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2954 should usually be replaced by
2955 (make-pathname :name :wild :type :wild :version :wild)
2956 with the added benefit that this is more likely to be portable.
2957 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2958 option to OPEN now signals an error if the file being opened
2959 exists; this may have an impact on existing code.
2960 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2961 signals received, on the PowerPC platforms (both Linux and
2962 Darwin). (thanks to Pierre Mai for pointing out the location of
2964 * several fixes on OS X: The system now builds and runs cleanly on
2965 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2966 (thanks to Brian Mastenbrook)
2967 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2968 second-resolution: integer multiples of 1/3600 between -24 and 24.
2969 (thanks to Vincent Arkesteijn)
2970 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2971 type when called with 1 argument; PEEK-CHAR checked type of
2972 PEEK-TYPE only after having read first character from a
2973 stream. (reported by Peter Graves)
2974 * bug fix: the garbage collector now has much better locality
2975 behaviour, and in particular no longer treats all memory as being
2976 exhausted when many small objects point to each other in a deeply
2978 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2979 recognized as being TYPEP their class.
2980 * bug fix: the PUSHNEW documentation string has been corrected.
2981 (thanks to Vincent Arkesteijn)
2982 * bug fix: defaulting of the value for the last of an atypically
2983 large number of multiple values being bound was not being
2984 performed correctly on the Alpha or PPC platforms
2985 * optimization: implemented multiplication as a modular
2986 (UNSIGNED-BYTE 32) operation on the x86 backend.
2987 * optimization: SEARCH on simple-base-strings can now be open-coded.
2988 (see also contrib/compiler-extras.lisp for inspiration for
2989 teaching the compiler about the Boyer-Moore algorithm).
2990 * value, returned by MAX (and MIN) called with several EQUALP, but
2991 not EQL, arguments now does not depend on compiler settings.
2992 * fixed some bugs revealed by Paul Dietz' test suite:
2993 ** in stack analysis liveness information is propagated from
2994 non-local entry points.
2995 ** pathwise CAST removing failed when the CAST node did not start
2997 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2998 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
3000 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
3001 host is already defined.
3002 ** RENAME-FILE works on streams instead of signalling an internal
3004 ** PEEK-CHAR uses the current readtable when determining whether
3005 or not a character is whitespace.
3006 ** MERGE-PATHNAMES handles the case when the pathname does not
3007 specify a name while the default-pathname specifies a version.
3008 ** Pathnames now stand a better chance of respecting print/read
3010 ** Attempting to use standardized file system operators with a
3011 pathname with invalid :DIRECTORY components signals a
3013 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
3015 ** DIRECTORY on logical pathnames is more correct.
3016 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
3017 signal a TYPE-ERROR if their argument is not a stream.
3018 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
3019 designator argument does not designate a stream.
3020 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
3021 examining the synonym.
3022 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
3024 ** OPEN and WITH-OPEN-STREAM allow opening streams with
3025 element-type larger than ([UN]SIGNED-BYTE 32).
3027 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
3028 * When built with the :SB-FUTEX feature, threaded builds now take
3029 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
3030 for faster/more reliable mutex and condition variable support.
3031 * Incompatible change (but one you probably shouldn't have been using
3032 anyway): the interface and code for arbitrating between multiple
3033 threads in the same user session has been redesigned.
3034 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
3035 error when called without an explicit environment argument.
3036 (thanks to Nikodemus Siivola)
3037 * bug fix: buffered :DIRECTION :IO streams are less likely to become
3038 confused about their position. (thanks to Adam Warner and Gerd
3040 * bug fix: Pretty printing backquoted forms with unquotations in the
3041 argument list position of various code constructs such as LAMBDA
3042 now works correctly. (reported by Paul Dietz)
3043 * bug fix: Pretty printing unquotations no longer loses all
3044 stream position information.
3045 * optimization: performance of string output streams is now less
3046 poor for multiple small sequence writes.
3047 * optimization: performance of CSUBTYPEP in the presence of complex
3048 expressions involving CONS and NOT many times has been improved.
3049 (reported by Paul Dietz)
3050 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
3052 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
3053 streams for the REPL, for the debugger, and so on. (thanks to
3055 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3056 * fixed some bugs revealed by Paul Dietz' test suite:
3057 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3058 is ignored if it is too small, rather than propagating through
3059 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3060 ** extremely complex negations of CONS types were not being
3061 sufficiently canonicalized, leading to inconsistencies in
3063 ** VALUES tranformer lost derived type.
3065 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3066 * fixed a bootstrapping bug: the build process no longer assumes
3067 that the various BOOLE-related constants have the same value in
3068 host and target lisps. (noted by Paul Dietz' test suite on an
3069 SBCL binary built from CLISP)
3070 * The system can now be dynamically linked on the MIPS platform,
3071 which enables dynamic loading of foreign code from Lisp. (thanks
3072 to Ralf Baechle for discussions on the MIPS ABI)
3073 * The system now records debugging information for its own source
3074 files in a filesystem-position-independent manner, relative to
3075 the "SYS" logical host.
3076 * fixed a compiler bug: MV-LET convertion did not check references
3077 to the "max args" entry point. (reported by Brian Downing)
3078 * tweaked disassembly notes to be less confident about proclaiming
3079 some instruction as an LRA. (thanks to Brian Downing)
3080 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3081 now each have their own history, command character, and other
3082 characteristics. (thanks to David Lichteblau)
3083 * fixed some bugs revealed by Paul Dietz' test suite:
3084 ** compiler failure in compiling LOGAND expressions including a
3086 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3087 shift greater than 32.
3088 ** FLUSH-DEST did not mark blocks for type check regeneration.
3089 ** HANDLER-CASE failed to accept declarations in handler clauses
3090 in some circumstances.
3092 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3093 * New code in contrib/sb-introspect (still probably not entirely
3094 stable yet) provides some support for smart Lisp development
3095 environments like SLIME.
3096 * The conditions signalled for errors occurring when loading .fasl
3097 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3098 in a way which should help ASDF recover gracefully.
3099 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3100 stuff has been cleaned up. If you code contrib/ stuff, this might
3101 affect you, and you can look at contrib/README, contrib/STANDARDS,
3102 and/or the 0.8.4.27 diff to check.
3103 * In full calls the compiler now does not generate checks for declared
3104 argument types for all arguments.
3105 * various threading fixes
3106 ** and some experimental patches which didn't make it into
3107 the main tree for this release, but which are shipped in
3108 contrib/experimental-thread.patch as a possible fix for some
3109 failures (deadlock, spinning...) in GC-intensive multithreaded
3111 * fixed PPC build problem (source code incompatibility of different
3112 library versions): added offsetof() hackery which attempts to divine
3113 where glibc maintainers put uc_mcontext today
3114 * fixed bug 282: compiler does not trust type assertions while passing
3115 arguments to a full call.
3116 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3117 &OPTIONAL VALUES type parameter.
3118 * fix bug 214: algorithm for noting rejected templates is now more
3119 similar to that of template seletion. (also reported by rydis on
3121 * fixed bug 141b: printing backquoted information readably and prettily
3122 inserts a space where necessary.
3123 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3124 (CHARACTER 10) are now reported as errors, rather than propagated
3125 as unknown types. (reported by piso on #lisp)
3126 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3127 with values NIL and :ERROR. (thanks to Milan Zamazal)
3128 * fixed bug 191c: CLOS now does proper keyword argument checking as
3129 described in CLHS 7.6.5 and 7.6.5.1.
3130 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3131 longer raise an error; further, using a list as a for-as-arithmetic
3132 counter now raises a meaningful error.
3133 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3134 now understood by sequence creation functions such as MAKE-SEQUENCE
3136 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3137 a STRING-INPUT-STREAM.
3138 * compiler enhancement: SIGNUM is now better able to derive the type
3140 * type declarations inside WITH-SLOTS are checked. (reported by
3142 * fixed some bugs revealed by Paul Dietz' test suite:
3143 ** incorrect optimization of TRUNCATE for a positive first
3144 argument and negative second.
3145 ** compiler failure in let-convertion during flushing dead code.
3146 ** compiler failure while deriving type of TRUNCATE on an
3147 interval, containing 0.
3148 ** ASH of a negative bignum by a negative bignum count now returns
3150 ** intersection of CONS types now canonicalizes properly, fixing
3151 inconsistencies in SUBTYPEP.
3153 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3154 * incompatible change: The --disable-debugger command line
3155 option now clobbers the debugger at a more fundamental
3156 level, by redefining #'INVOKE-DEBUGGER instead of by
3157 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3158 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3159 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3160 enter the debugger REPL and then suffer recursive errors
3161 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3162 while under the new system, BREAK in --disable-debugger mode
3163 terminates the system just as an unhandled error would.
3164 * fixed compiler performance when processing loops with a step >1;
3165 * bug fix: DOCUMENTATION now retrieves generic function
3166 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3167 support has been systematized, and now supports the methods
3168 specified by ANSI, along with a default method and a method for
3169 slot documentation. (reported by Nathan Froyd)
3170 * bug fix: effective methods associated with a generic function are
3171 no longer cached over a change of that generic function's method
3172 combination. (reported by Andreas Fuchs)
3173 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3174 names a non-existent directory. (thanks to Andreas Fuchs)
3175 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3176 the PPC platform, lead to stack corruption; this has been fixed.
3177 (reported by Rainer Joswig)
3178 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3179 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3180 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3181 arithmetic in general, is now much more reliable on the PPC
3183 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3184 platform now returns the right answer.
3185 * optimization: restored some effective method precomputation in
3186 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3187 precomputation is now tunable.
3188 * optimization: compiler-internal data structure use has been
3189 reviewed, and changes have been made that should improve the
3190 performance of the compiler by about 20%.
3191 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3192 simple-base-strings and simple-bit-vectors is improved.
3193 * optimization: the optimization of 32-bit logical and arithmetic
3194 functions introduced in version 0.8.3 on the x86 has been
3195 implemented on the mips, ppc and sparc platforms; an
3196 implementation of the same facility, but for 64-bit arithmetic,
3197 has been added for the alpha.
3198 * microoptimization: the compiler is better able to make use of the
3199 x86 LEA instruction for multiplication by constants.
3200 * bug fix: in some situations compiler did not report usage of
3201 generic arithmetic in (SPEED 3) policy.
3202 * bug 145b fix: compiler used wrong type specifier while converting
3203 MEMBER-types to numeric.
3204 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3206 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3207 index argument now works properly on the Alpha platform.
3208 * bug fix: floating point exception treatment on the Alpha platform
3210 * bug fix: FILE-POSITION works much better on string input and
3211 output streams. (thanks to Nikodemus Siivola)
3212 * bug fix: many threading/garbage collection symptoms sorted.
3213 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3214 might be pseudo-atomic.
3215 * internal change: Stopping for GC is now done with signals not
3216 ptrace. GC is now done in whichever thread wanted it, instead of
3218 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3220 * bug fix: result form in DO is not contained in the implicit
3222 * incompatible change: ICR structure is changed; the value part of
3223 CONTINUATION is now called LVAR; corresponding functions are
3224 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3225 * added type deriver for ISQRT (thanks to Robert E. Brown).
3226 * bug fix: better support for loading from the command line when an
3227 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3229 * fixed some bugs revealed by Paul Dietz' test suite:
3230 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3231 ** ROUND and FROUND now give the right answer when given very
3232 small float arguments.
3233 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3235 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3236 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3237 ** LCM should return a non-negative integer.
3238 ** PARSE-INTEGER returned the index of a terminator instead of the
3239 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3240 ** PARSE-INTEGER returned an incorrect index being applied to a
3242 ** LCM with two arguments of 0 returns 0 rather than signalling
3244 ** unsigned addition of a 32-bit constant with the high bit set no
3245 longer causes an internal compiler error on the x86.
3246 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3247 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3248 with negative last argument.
3249 ** byte specifiers with zero size and position no longer cause
3250 an error during type derivation.
3251 ** bignum multiplication on the Alpha platform now returns the
3253 * porting: The system now builds on SuSE AMD64, although it still
3254 generates a 32-bit binary.
3255 * .fasl file incompatibility: The fasl file version number has
3256 been incremented (because of the changes to internal compiler
3257 data structures referred to above).
3259 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3260 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3261 more accurately, on the Darwin kernel running on PowerPC hardware.
3262 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3263 * Compiler code deletion notes now signal a condition of type
3264 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3265 an associated MUFFLE-WARNING restart.
3266 * The compiler now performs limited argument count validation of
3267 constant format strings in FORMAT, and where appropriate in ERROR,
3268 CERROR and WARN. (thanks to Gerd Moellmann)
3269 * New ASDF-INSTALL contrib can be used for automatic download and
3270 installation of third-party Lisp code from CCLAN or other sites
3272 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3273 which forces another thread to execute a function supplied by the
3275 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3276 now accept and act upon their :ELEMENT-TYPE keyword argument.
3277 (reported by Martin Atzmueller, Edi Weitz)
3278 * bug fix: FILE-POSITION now accepts position designators up to
3279 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3280 is the greater. (thanks to Patrik Nordebo)
3281 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3282 Kalle Olavi Niemitalo)
3283 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3284 printable. (reported by Eric Marsden)
3285 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3286 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3287 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3288 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3289 (thanks to Stig Sandoe)
3290 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3291 UNREAD-CHAR now correctly includes the unread character in the
3292 target sequence. (thanks to Gerd Moellmann)
3293 * bug fix in threaded builds: the system can now be suspended and
3294 resumed by shell job control with minimal disruption.
3295 * bug fixes in times and timezones >2038AD
3296 * better handling of "where is GNU make?" problem in build scripts
3297 (thanks to Nikodemus Siivola)
3298 * new optimization: inside a named function any reference to a
3299 function with the same name is considered to be a self-reference;
3300 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3301 optimization quality.
3302 * new optimization on x86: logical functions and + now have
3303 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3304 used when the result is truncated to 32 bits.
3305 * VALUES declaration is partially enabled.
3306 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3307 * fixed some bugs revealed by Paul Dietz' test suite:
3308 ** The system now obeys the constraint imposed by
3309 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3310 types form a lattice under type intersection.
3311 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3312 ** ASSOC now ignores NIL elements in an alist.
3313 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3314 and (1+ MOST-POSITIVE-FIXNUM) answers.
3315 ** The addition of a method with invalid qualifiers to a generic
3316 function does not cause an error to be signalled immediately;
3317 a warning is signalled, and the error is generated only on
3318 calling the generic function.
3319 * changed .fasl file version number, in part to add type codes for
3320 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3321 obscure ANSI requirements
3323 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3324 * fixed bug 148: failure to inline-expand a local function left
3325 garbage, confusing the compiler.
3326 * fixed bugs 3cd: structure slot readers perform type check if the
3327 slot can have an invalid value (i.e. it is either not initialized
3328 or can be written with a less specific slot writer).
3329 * bug fix: the compiler now traps array references to elements off
3330 the end of an array; previously, the bounds checking in some
3331 circumstances could go off-by-one.
3332 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3334 * type declarations for array element types now obey the description
3335 on the CLHS page "Declaration TYPE", as per discussions on
3336 sbcl-help around 2003-05-08. This means that a declaration
3337 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3338 declaration, all references to BAR will be asserted or assumed
3339 (with THE, so dependent on compiler policy) to involve objects of
3340 type FOO. Note that no such declaration is implied in
3341 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3342 * declared types of functions from the "Conditions"
3343 chapter. (reported by Paul Dietz)
3344 * bug fix: CERROR accepts a function as its first argument.
3345 * bug fix: NTH an NTHCDR accept a bignum as index
3346 arguments. (reported by Adam Warner)
3347 * optimization: character compare routines now optimize comparing
3348 against a constant character. (reported by Gilbert Baumann)
3349 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3350 argument now works properly on the MIPS platform.
3351 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3352 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3353 declared to have a functional type, does not cause an error
3355 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3356 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3357 * fixed bug: if last continuation of a deleted block has a
3358 destination, this destination should be deleted too. (reported by
3360 * fixed a bug in the bootstrap process: the host compiler's values
3361 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3362 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3363 test case from Patrik Nordebo)
3364 * improved the ability of the disassembler on the PPC platform to
3365 provide helpful disassembly notes.
3366 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3367 the class in more cases than previously.
3368 * bug fix: FILE-POSITION now understands :START and :END for
3369 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3370 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3371 8) data. (thanks to David Lichteblau)
3372 * bug fix: it is possible to add a method to a generic function
3373 without lambda list.
3374 * bug fix: reader failed to signal END-OF-FILE inside an
3375 object representation. (reported by Nikodemus Siivola)
3376 * fixed some bugs revealed by Paul Dietz' test suite:
3377 ** LAST and [N]BUTLAST should accept a bignum.
3378 ** condition slot accessors are methods.
3379 ** (VECTOR NIL) is a subtype of STRING.
3381 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3382 * minor incompatible change: some nonsensical specialized lambda
3383 lists (used in DEFMETHOD) which were previously ignored now signal
3385 * minor incompatible change: the system is now aware of the types of
3386 variables in the COMMON-LISP package, and will signal errors for
3387 most violations of these type constraints (where previously they
3388 were silently accepted).
3389 * minor incompatible change: COMPILE-FILE now uses the freedom
3390 afforded (ANSI 3.2.2.3) to use derived function types for
3391 functions defined in the same file. This also permits the system
3392 to warn on static type mismatches and function
3393 redefinition. (Currently it does not work with high DEBUG level.)
3394 * minor incompatible change: VALUES declaration is disabled.
3395 * When issuing notes, the compiler now signals a condition of type
3396 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3397 restart for use in user handlers. It is expected that the
3398 COMPILER-NOTE condition will eventually become a condition
3399 supertype to a hierarchy of note types, which will then be
3400 handleable in a similar fashion. However, at the moment, no such
3401 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3402 but it's an implementation detail, not a classification for the
3404 * Changes in type checking closed the following bugs:
3405 ** type checking of unused values (192b, 194d, 203);
3406 ** template selection based on unsafe type assertions (192c, 236);
3407 ** type checking in branches (194bc).
3408 * A short form of VALUES type specifier has ANSI meaning (it has
3409 increased the number of situations when SBCL cannot perform type
3411 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3412 keywords or constants is permissible.
3413 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3414 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3415 classes. (thanks to Antonio Martinez)
3416 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3417 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3418 argument to a pretty-print stream. Instead, it leaves any such
3419 operations to DESCRIBE-OBJECT methods.
3420 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3422 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3423 lambda lists. (thanks to David Lichteblau)
3424 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3425 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3426 (thanks to Gerd Moellmann)
3427 * fixed bug 246: increased compilation speed of long
3428 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3430 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3432 * DEFCONSTANT now throws a condition of type
3433 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3434 constant to a non-EQL value; CONTINUE and ABORT restarts
3435 respectively change and preserve the value.
3436 * fixed bug 63: The code walker, part of the implementation of CLOS,
3437 is now better at handling symbol macros.
3438 * bug fix: There is no longer an internal implementation type named
3439 CL:LENGTH. (reported by Raymond Toy)
3440 * bug fix: In macro-like defining macros/special operators the
3441 implicit block does not enclose the lambda list.
3442 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3443 enclosed in parentheses) are not suitable as type specifiers, and
3444 their use properly signals an error now.
3445 * bug fix: An argument count mismatch for a type specifier in code
3446 being compiled no longer causes an unhandled error at compile
3447 time, but signals a compile-time warning.
3448 * fixed simple vector readable printing
3449 * bug fix: DESCRIBE takes more care over whether the class
3450 precedence list slot of a class is bound before accessing it.
3451 (reported by Markus Krummenacker)
3452 * bug fix: FORMATTER can successfully compile pretty-printer format
3453 strings which use variants of the ~* directive inside.
3454 * bug fix: SEARCH now applies its TEST predicate to the elements of
3455 the arguments in the correct order. (thanks to Wolfhard Buss)
3456 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3457 in type inference. (thanks to Robert E. Brown)
3458 * bug fix: Reading in symbols with an explicit package name of ""
3459 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3460 * fixed some bugs revealed by Paul Dietz' test suite:
3461 ** NIL is now allowed as a structure slot name.
3462 ** Arbitrary numbers, not just REALs, are allowed in certain
3463 circumstances in LOOP for-as-arithmetic clauses.
3464 ** Multiple class redefinitions before slot access no longer
3465 causes a type error.
3466 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3467 association between the name and a class.
3468 ** Generic functions with non-standard method-combination and over
3469 five methods all of which return constants no longer return NIL
3470 after the first few invocations. (thanks to Gerd Moellmann)
3471 ** CALL-NEXT-METHOD with no arguments now passes the original
3472 values of the arguments, even in the presence of assignment.
3473 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3475 ** DISASSEMBLE works with closures and funcallable instances.
3476 ** ADD-METHOD now returns the generic function, not the new method.
3477 ** FIND-METHOD signals an error if the lengths of the specializers
3478 is incompatible with the generic function, even if the ERRORP
3480 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3481 which its argument is a member.
3482 ** DEFCLASS only redefines the class named by its class-name
3483 argument if that name is the proper name of the class;
3484 otherwise, it creates a new class.
3485 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3486 of the UNBOUND-SLOT condition to the name of the slot.
3487 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3488 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3489 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3490 SLOT-MAKUNBOUND in the specified fashion.
3492 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3493 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3494 cross-compilation host. As a consequence, we can now bootstrap our
3495 way up to SBCL starting with a bare gcc toolchain and human-readable
3496 source code (first the source to CLISP, then the source to SBCL).
3497 * A contributed module containing a partial implementation of the
3498 simple-streams interface has been included. (thanks to Rudi
3500 * A contributed module implementing the RFC1321 Message Digest
3501 Algorithm, known as MD5, has been included.
3502 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3503 no longer has any effect, as the code controlled by this feature
3504 has been deleted. (As far as we know, no-one has ever built using
3505 this feature, and its semantics were confused in any case).
3506 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3507 work on conditions (as required by the ANSI specification),
3508 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3509 expected behaviour on conditions. Users should note, however,
3510 that such behaviour is not required by the ANSI specification,
3511 and so use of this behaviour may render their code unportable.
3512 * fixed some bugs revealed by Paul Dietz' test suite:
3513 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3515 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3516 ** SXHASH on condition objects no longer returns NIL.
3517 ** :ALLOCATION :CLASS slots are better treated; their values are
3518 updated on class redefinition, and initforms inherited from
3519 superclasses are applied.
3520 ** REMOVE-METHOD returns its generic function argument even when
3521 no method was removed.
3522 ** SHARED-INITIALIZE now initializes the values of the requested
3523 slots, including those with :ALLOCATION :CLASS.
3524 ** ALLOCATE-INSTANCE now works on structure classes defined via
3525 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3527 ** SLOT-EXISTS-P now works on conditions, as well as structures
3529 ** MAKE-LOAD-FORM now has the required methods on
3530 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3531 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3532 keyword, and now implements the SLOT-NAMES argument.
3533 ** methods with &OPTIONAL arguments no longer allow too many
3534 arguments to be passed in the call without error.
3535 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3536 option is consistent with the required arguments of the generic
3537 function lambda list.
3538 * bug fix: REQUIRE accepts a string designator. (Thanks to
3540 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3541 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3542 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3543 * bug fix: The compiler now checks for duplicated variables in macro
3545 * bug fix: SETQ on globals returns the correct value.
3546 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3547 not a COMPILER-ERROR (followed by some other strange error on
3548 choosing the CONTINUE restart).
3549 * bug fix: make.sh and friends are now more consistent in the way that
3550 they look for GNU "make".
3552 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3553 * experimental native threads support (on x86 Linux >=2.4 only).
3554 This is not compiled in by default: you need to add :SB-THREAD to
3555 the target features. See the "Beyond ANSI" chapter of the manual
3557 * fix for longstanding nonANSIism: The old distinction between
3558 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3559 The return value from CL:FIND-CLASS is now a CLOS class, and
3560 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3561 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3562 * An interface to the MetaObject Protocol, as described in Kiczales,
3563 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3564 MIT Press, 1991, is available from the SB-MOP package.
3565 * incompatible change: the SB-PCL package should now be considered
3566 a private implementation detail, and no longer a semi-private MOP
3568 * minor incompatible change: due to rearrangement for threads, the
3569 control stack and binding stack are now allocated at arbitrary
3570 addresses instead of being hardcoded per-port. Users affected by
3571 this probably have to be doing advanced things with shared
3572 libraries, and will know who they are.
3573 * minor incompatible change: Previously, all --eval forms used were
3574 processed with READ before any of them were processed with EVAL.
3575 Now each --eval form is processed with both READ and EVAL before
3576 the next --eval form is processed. (Thus package operations like
3577 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3578 work as the user might reasonably expect.)
3579 * minor incompatible change: *STANDARD-INPUT* is now only an
3580 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3582 * minor incompatible change: Y-OR-N-P is now character-oriented,
3583 not line oriented. Also, YES-OR-NO-P now works without errors.
3584 (thanks to Antonio Martinez)
3585 * sb-aclrepl module improvements: an integrated inspector, added
3586 repl features, and a bug fix to :trace command.
3587 * Known functions, which cannot be open coded by the backend, are
3588 considered to be able to check types of their arguments. (fixing
3589 a bug report by Nathan J. Froyd)
3590 * fixed a bug in computing method discriminating functions: It is
3591 now possible to define methods specialized on classes which have
3592 forward-referenced superclasses. (thanks to Gerd Moellmann)
3593 * fixed evaluation order in optional entries (reported by Gilbert
3595 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3596 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3597 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3598 required-by-AMOP NAME argument, as well as CLASS and
3599 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3600 * fixed bug 20: DEFMETHOD can define methods using names that are
3601 not the proper names of classes to designate class specializers.
3602 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3603 Unix file descriptors, instead of blowing up. (thanks to Antonio
3605 * Garbage collection refactoring: user-visible change is that a
3606 call to the GC function during WITHOUT-GCING will not do garbage
3607 collection until the end of the WITHOUT-GCING. If you were doing
3608 this you were probably losing anyway.
3609 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3610 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3611 TYPEP the latter but not the former.
3612 * The compiler issues a full WARNING for calls to undefined functions
3613 with names from the CL package.
3614 * MAP-INTO for a vector destination is open coded. (reported by
3615 Brian Downing on c.l.l)
3616 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3617 documentation string.
3618 * fixed some bugs revealed by Paul Dietz' test suite:
3619 ** COPY-ALIST now signals an error if its argument is a dotted
3621 ** Condition slots are now accessed more correctly in the presence
3622 of multiple initargs for a given slot.
3623 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3624 exclude restarts of the same name associated with a different
3626 ** DEFCLASS of forward-referenced classes with another
3627 forward-referenced class in the superclasses list no longer
3629 ** Condition slots are now initialized once each, not multiple
3630 times. (thanks to Gerd Moellmann)
3631 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3633 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3634 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3635 arguments contain duplicated elements.
3636 ** RESTART-CASE understands local macros.
3637 ** RESTART-CASE associates exactly its own restarts with a condition.
3638 ** ENDP in safe mode checks its argument to be of type LIST.
3639 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3640 in question is unbound.
3641 ** Optimized MAKE-INSTANCE functions no longer cause internal
3642 assertion failures in the presence of duplicate initargs.
3643 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3644 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3645 propagated even in the ctor optimized implementation of
3647 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3649 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3650 * a better implementation of SXHASH on (simple) bit vectors,
3651 measured both in execution speed and in distribution of results
3652 over the positive fixnums, has been installed. Likewise, a better
3653 implementation of EQUAL for simple bit vectors is now available.
3654 * fixed CEILING optimization for a divisor of form 2^k.
3655 * fixed bug 240 (emitting extra style warnings "using the lexical
3656 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3657 by Antonio Martinez)
3658 * fixed SXHASH, giving different results for NIL depending on type
3659 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3660 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3661 variable. (found by Rolf Wester)
3662 * SBCL does not ignore type declarations for special
3663 variables. (reported by rif on c.l.l 2003-03-05)
3664 * some bug fixes in contrib/sb-aclrepl/
3665 * fixed some bugs revealed by Paul Dietz' test suite:
3666 ** a bug in the CONS type specifier, whereby the CAR and CDR
3667 types got intertwined, has been fixed;
3668 ** the type system is now able to reason about the interaction
3669 between INTEGER and RATIO types more completely;
3670 ** APPEND, [N]REVERSE and NRECONC check that those their
3671 arguments, which must be proper lists, are really so;
3672 ** An array specialized to be unable to hold elements has been
3673 implemented, as required -- yes, really -- by ANSI;
3674 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3675 on malformed property lists;
3677 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3678 * incompatible packaging change: in line with Unix convention,
3679 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3680 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3681 * REQUIRE and PROVIDE are now optionally capable of doing something
3682 useful. See the documentation string for REQUIRE.
3683 * infrastructure for a managed SBCL contrib system: contributed
3684 modules in this release include:
3685 ** the ASDF system definition facility;
3686 ** an interface to the BSD Sockets API;
3687 ** an ACL-like convenience interface to the repl;
3688 (thanks to Kevin Rosenberg)
3689 ** an implementation of ROTATE-BYTE, with efficient implementation
3691 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3692 gives a closer approximation to the right answer than previously.
3693 (thanks to Raymond Toy)
3694 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3695 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3696 optional environment arguments, as required by ANSI.
3697 * fixed bugs in other functions taking environment objects, allowing
3698 calls with an explicit NIL environment argument to be compiled
3700 * fixed bug 228: primary return values from
3701 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3702 COMPILE or FUNCTION.
3703 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3704 :TYPE will no longer signal errors on innocuous objects.
3705 * fixed bug 231b: SETQ is better at respecting type declarations in
3706 the lexical environment.
3707 * fixed a bug in DEFCLASS: classes named by symbols with no or
3708 unprintable packages can now be defined.
3709 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3710 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3711 * fixed compiler failure related to checking types of functions.
3712 (reported by Robert E. Brown)
3713 * the compiler is now much more consistent in its error-checking
3714 treatment of bounding index arguments to sequence functions: in
3715 (SAFETY 3) code, errors will be signalled in almost all cases if
3716 invalid sequence bounding indices are passed to functions defined
3717 by ANSI to operate on sequences.
3718 * fixed a bug in the build procedure: documentation of SBCL-specific
3719 packages is now preserved and available in the final Lisp image.
3720 * lifted FDEFINITION lookup out of loops in the implementation of
3721 many list operations. (thanks to Robert E. Brown)
3722 * fixed a bug in the reader: the #n# reader macro now works for
3723 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3724 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3725 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3726 E. Brown and Tony Martinez respectively)
3727 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3728 SLOT-BOUNDP: in particular, it is now able to optimize them much
3729 better, and is now not vulnerable to having packages renamed.
3730 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3731 always called when a slot is not present in an instance. (thanks
3733 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3734 longer signal an error on unreading a character following EOF on
3735 the previous constituent stream. (thanks to Tony Martinez)
3736 * fixed some bugs revealed by Paul Dietz' test suite:
3737 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3738 not just nonnegative fixnums;
3739 ** the logical bit-array operators such as BIT-AND now accept an
3740 explicit NIL for their "opt-arg" argument (to indicate a
3741 freshly-consed result bit-array);
3742 ** ELT now signals an error on an invalid sequence index in safe
3744 ** the type system is now cleverer about negations of numeric
3745 types, and consequently understands the BIGNUM and RATIO types
3747 ** the type system is now cleverer about the interaction between
3748 INTEGER and RATIO types: while bugs still remain, many more
3749 cases are accurately computed;
3750 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3751 if it is in the last clause;
3752 ** CONSTANTLY now correctly returns a side-effect-free function in
3754 ** DECLARE is no longer treated as a special-operator; in
3755 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3756 * incremented fasl file version number due to the change in the
3757 DEFSTRUCT-SLOT-DESCRIPTION structure.
3759 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3760 * minor incompatible change: code processed by the "interpreter" or
3761 EVAL now has a compilation optimization policy of (DEBUG 2)
3762 (changed from (DEBUG 1)) to improve debuggability of interactive
3763 development, and to allow the use of the debug RETURN command in
3765 * an experimental implementation of the RETURN command for the
3766 debugger has been included. (thanks to Frederik Kuivinen)
3767 * fixed bug 62: constraints were not propagated into a loop.
3768 * fixed bug in embedded calls of SORT (reported and investigated by
3770 * fixed some bugs revealed by Paul Dietz' test suite:
3771 ** printing and reading of arrays with some dimensions having
3772 length 0 (thanks to Gerd Moellmann);
3773 ** BOA constructor with &AUX argument without a default value does
3774 not cause a type error;
3775 ** CONSTANTP now returns true for all self-evaluating objects.
3777 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3778 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3779 accessors that are related by inheritance, as specified in the
3780 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3781 Valtteri Vuorikoski)
3782 * The compiler is now able to inline functions that were defined in
3783 a complex lexical environment (e.g. inside a MACROLET).
3784 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3786 * Support for the upcoming FreeBSD-5.0 release has been included.
3787 (thanks to Dag-Erling Smorgrav)
3788 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3789 effect when it is not in a toplevel context.
3790 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3791 stand a better chance of being correct. (thanks to Gerd
3793 * fixed bug in COERCE, which now signals an error on coercing a
3794 rational to a bounded real type which excludes the expected
3796 * The compiler is now able to derive types more accurately from the
3797 COERCE and COMPILE functions.
3798 * fixed bug 223: functional binding is considered to be constant
3799 only for symbols in the CL package.
3800 * fixed bug 231: SETQ did not check the type of a variable being set
3801 (reported by Robert E. Brown)
3802 * A new optimization for MAKE-INSTANCE has been included, fixing
3803 various bugs (including relating to :ALLOCATION :CLASS slots and
3804 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3806 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3807 ** As required by ANSI, LOOP now disallows anonymous collection
3808 clauses such as COLLECT I in conjunction with aggregate boolean
3809 clauses such as THEREIS (= I 1);
3810 ** LOOP now signals an error when any variable is reused in the
3811 same loop (including the potentially useful construct analogous
3812 to WITH A = 1 WITH A = (1+ A);
3813 ** IT is only a special loop symbol within the first clause of a
3814 conditional loop clause;
3815 ** LOOP with a typed iteration variable over a hashtable now
3816 signals a type error iff it should.
3817 * fixed some other bugs revealed by Paul Dietz' test suite:
3818 ** FILE-STREAM now names the class previously known as FD-STREAM;
3819 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3820 argument) no longer signals an error;
3821 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3822 of the slot symbol, rather than using the current package
3823 ((:CONC-NAME "") continues to intern the slot's name in the
3825 * incremented fasl file version number, because of the incompatible
3826 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3827 of the new implementation of DEFINE-COMPILER-MACRO.
3829 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3830 * Support for building SBCL for MIPS platforms running in
3831 little-endian mode has now been checked in, and basic
3832 functionality on said platforms verified.
3833 * minor incompatible change: PCL now records the pathname of a file
3834 in which methods and the like are defined, rather than its
3836 * minor incompatible change: TRUENAME now considers the truename of
3837 a file naming a directory to be the pathname with :DIRECTORY
3838 component indicating that directory.
3839 * minor incompatible change: a NAMED clause in the extended form of
3840 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3841 reason for the previous behaviour is unclear.
3842 * more systematization and improvement of CLOS and MOP conformance
3843 in PCL (thanks to Gerd Moellman and Pierre Mai):
3844 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3846 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3847 primary methods with no specializers;
3848 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3850 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3851 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3852 FORWARD-REFERENCED-CLASSes; error reporting on
3853 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3855 ** SXHASH on CLOS instances now uses a slot internal to the
3856 instance to return different numbers on distinct instances,
3857 while preserving the same return value through invocations of
3859 ** DEFMETHOD signals errors when methods with longer incongruent
3860 lambda lists are added to generic functions;
3861 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3862 CLASS, as specified in AMOP;
3863 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3864 on the order returned by the primary method for classes of
3865 class STANDARD-CLASS;
3866 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3867 * fixed some bugs shown by Paul Dietz' test suite:
3868 ** DOLIST puts its body in TAGBODY;
3869 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3871 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3872 value producing form;
3873 ** if more variables are given to PROGV than values, extra
3874 variables are bound and made to have no value;
3875 ** NSUBSTITUTE on list arguments gets the right answer with
3877 ** ELT signals an error of type TYPE-ERROR when the index argument
3878 is not a valid sequence index;
3879 ** LOOP signals (at macroexpansion time) an error of type
3880 PROGRAM-ERROR when duplicate variable names are found;
3881 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3882 ** FUNCALL of special-operators now cause an error of type
3884 ** PSETQ now works as required in the presence of side-effecting
3885 symbol-macro places;
3886 ** NCONC accepts any object as its last argument;
3887 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3889 ** loop-for-as-package does not require a package to be explicitely
3891 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3893 * fixed bug 166: compiler preserves "there is a way to go"
3894 invariant when deleting code.
3895 * fixed bug 172: macro lambda lists with required arguments after
3896 &REST arguments now cause an error to be signalled. (thanks to
3898 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3899 bug. (thanks to Matthew Danish)
3900 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3902 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3903 arguments are lexically rebound. (thanks to Gerd Moellmann and
3905 * fixed bug 194: error messages are now more informative when there
3906 is no primary method applicable in a call to a generic
3907 function. (thanks to Gerd Moellmann)
3908 * fixed bug in command line argument checking (thanks to Julian
3910 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3911 (thanks to Matthew Danish)
3912 * incremented fasl file version number, because of the
3913 SXHASH-related changes in the layout of CLOS data structures
3915 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3916 * minor incompatible change: The runtime (the Unix executable named
3917 "sbcl") is now much pickier about the .core files it will load.
3918 Essentially it now requires .core files to descend from the same
3919 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3920 as the runtime does. (The intent is to prevent the crashes which
3921 can occur, and which can even be reported as mysterious failures,
3922 when people patch the sources or change the build parameters
3923 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3924 sbcl and .core files.)
3925 * fixed bug: VALUES-LIST is no longer optimized away.
3926 * fixed bug 142: The FFI conversion of C string values to Lisp
3927 string values no longer conses excessively. (thanks to Nathan
3928 Froyd porting Raymond Toy's fix to CMU CL)
3929 * began to systematize and improve MOP conformance in PCL (thanks to
3930 Nathan Froyd, Gerd Moellman and Pierre Mai):
3931 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3933 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3934 ** FINALIZE-INHERITANCE is now called on class finalization;
3935 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3936 argument precedence order.
3937 * fixed bug 202: The compiler no longer fails on functions whose
3938 derived types contradict their declared type.
3939 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3940 so it can be non-toplevel.
3941 * The fasl file version number has changed (because of the new
3942 implementation of DEFMACRO).
3943 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3944 safe code, that any length requirement by their type-specifier
3945 argument is valid. The exceptions to this are described in bug
3947 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3948 functions now check the types of their inputs as required by ANSI.
3949 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3950 introduced symbol is DECLAREd to be SPECIAL.
3951 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3952 * fixed bug: :COUNT argument to sequence functions may be negative
3953 * fixed bug: body of DO-SYMBOLS may contain declarations
3954 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3955 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3956 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3957 symbol macro only once
3958 * fixed printing of call frame when argument list is unavailable
3959 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3960 * compiler no longer signals WARNING on unknown keyword
3963 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3964 * A beta-quality port to the mips architecture running Linux,
3965 based on the old CMUCL backend, has been made. It has been tested
3966 on a big-endian kernel, and works sufficiently well to be able to
3967 rebuild itself; it has not been tested in little-endian mode.
3968 * fixed an inconsistency between gencgc.c and purify.c which made
3969 dumping/loading .core files unreliable
3970 * fixed bug 120a: The compiler now deals correctly with IFs where
3971 the consequent is the same as the alternative, instead of
3972 misderiving the return type. (thanks to Alexey Dejneka)
3973 * fixed bug 113: Logical pathnames are now dumpable (the logical
3974 host is resolved at load-time, throwing an error if it is not
3976 * fixed bug 174: FORMAT's error message is slightly clearer when a
3977 non-printing character is used in a format directive.
3978 * fixed several bugs in compiler checking of type declarations, i.e.
3979 violations of the Python "declarations are assertions" principle
3980 (thanks to Alexey Dejneka)
3981 * fixed several bugs in PCL's error checking (thanks to Gerd
3983 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3985 * fixed bug in compilation of functions as first class values
3986 (thanks to Antonio Martinez-Shotton)
3987 * The compiler's handling TYPE-ERRORs which it can prove will
3988 inevitably happen at runtime has been cleaned up and corrected
3989 in several ways. (thanks to Alexey Dejneka)
3990 * improved argument type checking for various basic arithmetic
3991 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3992 lavished on them in the past that they can be compiled in many
3993 ways in different special cases
3994 * fixed bug 181: compiler checks validity of user supplied type
3996 * cleaned up code flushing in optimization: Function calls which
3997 should signal errors for safety purposes (e.g. which ANSI says
3998 should signal errors when their arguments are of incorrect type)
3999 are no longer optimized away.
4000 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
4001 * incremented fasl file version number, because changes in the
4002 implementation of sequence functions like COERCE caused
4003 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
4004 in old inline expansions) to become undefined. (Actually these
4005 changes were later undone, so we might very well be binary
4006 compatible with 0.7.7 after all, but leaving the version number
4007 incremented seemed like the simplest and most conservative
4010 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
4011 * An alpha-quality port to the parisc architecture running Linux,
4012 based on the old CMUCL backend, has been made. This, even more so
4013 than the other backends, should be considered still a work in
4014 progress; known problems include that the Linux kernel in 64-bit
4015 mode does not propagate the correct sigcontext structure to
4016 userspace, and consequently SBCL on a parisc64 kernel will not
4018 * fixed bug 189: The compiler now respects NOTINLINE declarations for
4019 functions declared in FLET and LABELS. (I.e. "LET conversion" is
4020 suppressed.) Also now that the compiler is looking at declarations
4021 in the environment, it checks optimization declarations as well,
4022 and suppresses inlining when (> DEBUG SPEED).
4023 * More fixes have been made to treatment of floating point exception
4024 treatment and other Unix signals. In particular, floating point
4025 exceptions no longer cause Bus errors on the SPARC/Linux platform.
4026 * The detection and handling of control stack exhaustion (infinite
4027 or very deeply nested recursion) has changed. Stack exhaustion
4028 detection is now done by write-protecting pages at the OS level
4029 and applies at all optimization settings; when found, a
4030 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
4031 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
4032 be caught using IGNORE-ERRORS.
4033 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
4034 correctly when transforming recursive LABELS functions to LETs.
4035 (thanks to Alexey Dejneka)
4036 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
4037 that are names of constants or global variables.
4038 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
4039 alien routines with docstrings.
4040 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
4041 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
4043 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
4044 (thanks to Eric Marsden)
4045 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
4046 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
4047 * Bug fix: The SPARC backend can now compile functions involving
4048 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
4049 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
4050 to a functions expecting a PATHNAME-DESIGNATOR.
4051 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
4052 lambda lists. (thanks to Alexey Dejneka)
4053 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
4054 functions so that EXPORTing the name of the class doesn't cause
4055 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4056 to Antonio Martinez for reporting this)
4057 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4058 constant-folded without failing an assertion. (thanks to Einar
4059 Floystad Dorum for reporting this)
4060 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4061 types are now dealt with more correctly by the compiler.
4062 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4063 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4064 behaviour analogous to RENAME-FILE. This puts its behaviour more
4065 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4067 * The fasl file version number has changed again. (because of the
4068 bug fix involving the names of PCL MAKE-INSTANCE functions)
4070 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4071 * bug fix: Floating point exceptions are treated much more
4072 consistently on the x86/Linux and PPC/Linux platforms.
4073 * Array initialization with :INITIAL-ELEMENT is now much faster for
4074 cases when the compiler cannot open code the array creation, but
4075 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4076 array accesses have also seen a speed increase.
4077 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4078 specified. (This was at the root of some bad interactions between
4079 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4081 * bug fix: Internal error arguments for undefined functions are now
4082 computed correctly on the PPC/Linux platform.
4083 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4084 Raymond Toy's patch for CMU CL)
4085 * Support for the Solaris 9 operating environment has been included
4086 (thanks to Daniel Merritt)
4087 * A very ugly but hopefully complete draft of the missing FFI chapter
4088 of the manual has been created by reformatting the corresponding
4089 CMU CL manual chapter into (currently very ugly and incoherent)
4090 DocBook and bringing it up to date for SBCL behavior. Thus, the
4091 manual is now essentially complete, at least by my extreme
4092 once-and-only-once standards, whereby it's acceptable to refer to
4093 the doc strings of SB-EXT functions as the primary documentation.
4094 * The fasl file version number has changed again, due to cleanup of
4095 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4096 is no longer a static symbol.)
4098 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4099 * SBCL now builds with OpenMCL (version 0.12) as the
4100 cross-compilation host; also, more progress has been made toward
4101 bootstrapping under CLISP.
4102 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4104 * bug 158 fixed: The compiler can now deal with integer loop
4105 increments different from 1; fixing this turned out also to fix
4107 * bug 169 fixed: no more bogus warnings about using lexical bindings
4108 despite the presence of perfectly good SPECIAL declarations (thanks
4109 to David Lichteblau)
4110 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4111 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4112 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4114 * bug 180 fixed: Method combination specifications no longer ignore
4115 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4116 * bug fix: Structure type predicate functions now check their argument
4117 count as they should.
4118 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4119 correctly. (thanks to Pierre Mai)
4120 * minor incompatible change: The --noprogrammer option is deprecated
4121 in favor of the new --disable-debugger option, which is very similar.
4122 (The major difference is that it takes effect at a slightly different
4123 time at startup, causing handling of errors in --sysinit and
4124 --userinit files will be affected differently.) The
4125 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4126 been added to allow this functionality to be controlled from ordinary
4127 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4128 maintainers, who might want to run non-interactive scripts to
4129 build SBCL cores which will later be used interactively.)
4130 * minor incompatible change: The LOAD function no longer, when given
4131 a wild pathname to load, loads all files matching that pathname.
4132 Instead, an error of type FILE-ERROR is signalled.
4134 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4135 * bug 147 fixed: The compiler preserves its block link/count
4136 invariants more correctly now so that it doesn't crash. (thanks
4138 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4140 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4141 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4142 no-value-for-this-character as NIL (as specified by ANSI).
4143 * HOST-NAMESTRING on physical pathnames now returns a string that is
4144 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4145 (thanks to Christophe Rhodes)
4146 * The Alpha port handles icache flushing more correctly. (thanks to
4148 * More progress has been made toward bootstrapping under CLISP. (thanks
4149 to Christophe Rhodes)
4150 * The fasl file format has changed again, because dynamic loading
4151 on OpenBSD (which has non-ELF object files) motivated some cleanups
4152 in the way that foreign symbols are transformed and passed around.
4153 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4154 is no longer treated as whitespace by the reader, but instead as
4155 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4156 |A
\7fB|, instead of A as it used to.
4158 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4159 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4160 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4161 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4162 this as of 0.7.1.45, but was left out of the previous news section
4163 (thanks to Dan Barlow)
4164 * SBCL now runs on the Solaris operating system on SPARC architectures
4165 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4166 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4167 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4168 * SPARC backend cleanups, allowing builds of cores optimized for V8
4169 and V9 SPARCS, and also emission of code targeted to a particular
4170 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4172 * SBCL is closer to bootstrapping under CLISP, thanks to various
4173 fixes by Christophe Rhodes.
4174 * The fasl file format has changed again, to allow the compiler's
4175 INFO database to support symbol macros.
4176 * The user manual (in doc/) is formatted into HTML more nicely.
4177 (thanks to coreythomas)
4178 * The system is smarter about SUBTYPEP relationships, especially
4179 those involving NOT types (including types such as ATOM which are
4180 represented internally using NOT types). Thus SUBTYPEP is less
4181 likely to return (VALUES NIL NIL) in general, and in particular
4182 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4184 * The fasl file format has changed again, because the internal
4185 representation of types now includes a new slot to support the new
4186 SUBTYPEP-of-NOT-types logic.
4187 * (not a change in the main branch of SBCL, but a related prototype
4188 which can hopefully be merged into the main branch of SBCL in the
4189 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4190 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4191 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4192 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4193 dispatch macro character combinations. (thanks to Alexey Dejneka)
4194 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4195 that superficially look like logical namestrings correctly.
4196 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4198 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4199 default supplied-p) for &optional and &key arguments. (thanks to
4202 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4203 * incompatible change: The compiler is now less aggressive about
4204 tail call optimization, doing it only when (> SPACE DEBUG) or
4205 (> SPEED DEBUG). (This is an incompatible change because there are
4206 programs which relied on the old CMU-CL-style behavior to optimize
4207 away their unbounded recursion which will now die of stack overflow.)
4208 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4209 for non-GENCGC systems has been increased to 20M (since that
4210 seems much closer to the likely performance optimum for modern
4211 systems than the old 4M value was)
4212 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4213 and *DEBUG-PRINT-LEVEL*
4214 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4215 of CMU CL's support for SPARC, and various endianness and other
4216 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4217 * new syntactic sugar for the Unix command line: --load foo.bar is now
4218 an alternate notation for --eval '(load "foo.bar")'.
4220 ** The system now detects stack overflow and handles it gracefully,
4221 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4222 optimization settings. (This is a good thing in general, and
4223 its introduction in this version should be particularly timely
4224 for anyone whose code fails because of suppression of tail
4226 ** The system now hunts for the C variable "environ" in a more
4227 devious way, to avoid segfaults when the C library version differs
4228 between compile time and run time. (thanks to Christophe Rhodes)
4229 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4230 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4231 ** The compiler no longer issues bogus style warnings for undefined
4232 classes in the same source file as the DEFCLASSes which defined
4233 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4235 ** fixes in CONDITION class precedence list for undefined function
4236 errors (thanks to Alexei Dejneka)
4237 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4238 correctly. (thanks to Dan Barlow)
4239 ** portability fixes aiming at bootstrapping under CLISP (thanks
4240 to Dave McDonald and Christophe Rhodes)
4241 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4242 ** fixes in type translation and and type inference (thanks to
4244 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4245 ** various fixes in the new ports (thanks to Dan Barlow)
4246 * several changes related to debugging:
4247 ** suppression of tail recursion, as noted above
4248 ** stack overflow detection, as noted above
4249 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4250 is now the default. (For some time encapsulation has been more
4251 reliable than the breakpoint-based :ENCAPSULATE NIL
4252 implementation, at least on X86 systems; and I just noticed that
4253 encapsulation also seems closer to the spirit of the ANSI
4256 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4258 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4259 up properly again. (There was a packaging bug in 0.7.0 which
4260 left their definitions in SB-SYS::LOAD-FOREIGN and
4261 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4262 vital for most things which interface to C-level interfaces,
4263 like extensions working with sockets or databases or
4264 Perl-compatible regexes or whatever, and the need to fix
4265 this bug is the main reason that 0.7.1 was released so
4267 ** DEFGENERIC is now choosier about the methods it redefines, so that
4268 reLOADing a previously-LOADed file containing DEFGENERICs does
4269 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4270 cycle now works as it should. (thanks to Alexey Dejneka)
4271 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4272 fixed by Christophe Rhodes. (He actually submitted this patch
4273 months ago, and I delayed until after 0.7.0.)
4274 ** Bug 111 (internal compiler confusion about runtime checks on
4275 FUNCTION types) was fixed by Alexey Dejneka.
4276 * Some internal cleanups (getting rid of variables which aren't
4277 needed now that the byte interpreter is gone) caused the fasl
4278 file format number to change again.
4280 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4281 * major incompatible change: The default fasl file extension, i.e. the
4282 default extension for files produced by COMPILE-FILE, has changed
4283 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4285 ** There are many changes in the implementation of the compiler.
4286 SBCL is now essentially a compiler-only implementation of ANSI
4287 Common Lisp. EVAL still "interprets" a few special cases, but
4288 almost all the interesting cases are handled by creating
4289 a LAMBDA expression, calling COMPILE on it, then calling
4290 FUNCALL on the result.
4291 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4292 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4293 Since the code is newer, there might still be some new bugs
4294 (though not as many as before Martin Atzmueller's fixes:-). But
4295 the new code is substantially simpler and clearer, and hopefully
4296 any remaining bugs will be simpler, less fundamental, and more
4297 fixable then the bugs in the old code.
4298 ** The revised compiler is still a little unsteady on its feet.
4300 *** The debugging information it produces (particularly the names
4301 of FUNCTION objects) is sometimes much less useful than what
4302 the old compiler produced.
4303 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4304 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4305 within a MACROLET) has been temporarily weakened.
4306 ** There are new compiler optimizations for various functions:
4307 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4308 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4309 *** the math functions TRUNCATE, FLOOR, and CEILING
4310 *** the function-of-all-trades COERCE
4311 Mostly these should be transparent, but there's one
4312 potentially-annoying problem (bug 117): when the compiler
4313 inline-expands a function and does type analysis on the result,
4314 it can create control paths which have type mismatches, and
4315 when it can't prove that those control paths aren't taken,
4316 it will issue WARNINGs about the type mismatches. This is
4317 a particular problem in practice for the new sequence functions.
4318 It's not clear how this should be fixed, and for now, a
4319 workaround is given in the entry for 117 in the BUGS file.
4320 ** (Because of the interaction between the two previous items --
4321 occasional inlining problems and new inline expansions -- some
4322 of the new sequence function optimizations won't really kick in
4323 completely until debugging information, and then inlining, are
4324 straightened out in some future version.)
4325 * minor incompatible changes:
4326 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4327 DIRECTORY no longer implicitly promotes NIL slots of its
4328 pathname argument to :WILD. In particular, when you ask for the
4329 contents of a directory (which you used to be able to do without
4330 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4331 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4332 ** changes in behavior that ANSI explicitly defines to be
4333 implementation dependent:
4334 *** The new compiler-only implementation still conforms with ANSI,
4335 but acts a little different than before. Besides the obvious
4336 changes in performance tradeoffs (that the cost per form passed
4337 to EVAL has gone up, and the cost per form executed by EVAL
4338 has gone down), the behavior of the system changes a little
4339 because there are no longer any interpreted function objects.
4340 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4341 e.g. doing COMPILE on the output of interactive DEFUN is
4343 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4345 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4346 has changed from (:CL) to NIL.
4347 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4348 e.g. #\Newline and #\Tab, don't have names specified in the
4349 ANSI Common Lisp standard, is now based on their ASCII symbolic
4350 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4351 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4352 are no longer used for output.
4353 ** changes in internal implementation constants:
4354 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4355 4 million. (If your application spends a lot of time GCing and
4356 you have a lot of RAM, you might want to experiment with
4357 increasing it even more.)
4358 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4359 However, almost all old code should still continue to work without
4360 immediate update, as SB-C-CALL is now a (deprecated) nickname
4362 ** Old operator names in the style DEF-FOO are now deprecated in
4363 favor of new corresponding names DEFINE-FOO, for consistency with
4364 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4365 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4366 internal symbols, but a few supported extensions like
4367 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4368 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4369 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4370 etc. as you get deeper into recursive calls to the debugger
4371 command loop, instead of the old "5]", "5]]", "5]]]"
4372 sequence. (I was motivated to do this when squabbles between
4373 ILISP and SBCL left me very deeply nested in the debugger. In the
4374 short term, this change will probably provoke more ILISP/SBCL
4375 squabbles, but hopefully it will be an improvement in the long run.)
4376 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4377 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4378 function, since declaiming return types involving aliens is
4379 (1) annoyingly messy to do by hand and (2) vital to efficient
4380 compilation of code which calls such functions.
4381 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4382 longer reexported by the SB-EXT package. They're solely useful
4383 for alien code, so it seems more logical that you should get
4384 them from the SB-ALIEN package, not in SB-EXT.
4385 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4386 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4387 features. Instead, the code that they used to control is always
4388 built into the system.
4389 * many other bug fixes
4390 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4391 advantage of the new EVAL-WHEN stuff and to clean them up in
4392 general, and they are now more ANSI-compliant in a number of
4393 ways. Martin Atzmueller is responsible for a lot of this.
4394 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4396 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4397 *** correct ERROR type for various file operations
4398 *** some fixes for Lisp streams
4399 *** DEFMETHOD syntax checking
4400 *** changing old weird representation of debug information as
4401 strings (which, among their other deficiencies, don't transform
4402 correctly when you rename packages, and don't change their
4403 print representation when you change things like *PACKAGE*
4404 and *PRINT-LENGTH*) to symbols and lists of symbols
4405 He also made several improvements and fixed several bugs in DESCRIBE.
4406 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4408 *** misbehavior of WRITE-STRING/WRITE-LINE
4409 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4410 and several other LOOP problems as well
4411 *** DIRECTORY when similar filenames are present
4412 *** DEFGENERIC with :METHOD options
4413 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4414 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4415 *** argument ordering in FIND with :TEST option
4416 *** mishandled package designator argument in APROPOS-LIST
4417 *** various problems in the backquote readmacro
4418 *** a bug in APROPOS
4419 *** probably some others that I'm not describing very well here,
4420 since the CVS log documents them by reference to sbcl-devel
4421 messages, and the SourceForge archives aren't working well.:-(
4422 ** Dan Barlow improved the Alpha port (and is making progress on the
4423 PPC port, for those of you who think different).
4424 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4425 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4426 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4427 continues to work on the SPARC port (for those of you in a position
4428 to look down upon our little PC-compatible boxes from a great height).
4429 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4430 and uses that copy, rather than the current dynamic value, when
4431 it's trying to decide whether to truncate output. Thus e.g.
4432 (let ((*print-lines* 50))
4433 (pprint-logical-block (stream nil)
4435 (let ((*print-lines* 8))
4436 (print (aref possiblybigthings i) stream)))))
4437 should now truncate the logical block only at 50 lines, instead of
4438 often truncating it at 8 lines, as it did before.
4439 * The doc/cmucl/ directory, containing old CMU CL documentation
4440 from the time of the fork, is no longer part of the base system.
4441 SourceForge has shut down its anonymous FTP service, and with it
4442 my original plan for distributing the old CMU CL documentation
4443 there. For now, if you need these files you can download an old
4444 SBCL source release and extract them from it.
4445 * The fasl file version number changed again, for dozens of reasons,
4446 some of which are apparent above.
4448 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4449 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4450 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4451 patch, so that DISASSEMBLE output is much nicer.
4452 * The code in the SB-PROFILE package now seems reasonably stable.
4453 I still haven't decided what the final interface should look like
4454 (I'd like PROFILE to interact cleanly with TRACE, since both
4455 facilities use function encapsulation) but if you have a need
4456 for profiling now, you can probably use it successfully with
4457 the current CMU-CL-style interface.
4458 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4459 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4460 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4461 link now returns the dangling link itself, and for similar
4462 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4463 link itself. (In these cases the old code signalled an error and
4464 looped endlessly, respectively.) Thus, DIRECTORY now works even
4465 in the presence of dangling and cyclic symbolic links.
4466 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4467 is now a supported extension again, since the consensus on
4468 sbcl-devel was that it can be useful for ordinary development
4469 work, not just for debugging SBCL itself.
4470 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4471 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4472 that currently-defined functions might be redefined later with
4473 different return types.
4474 * Hash tables can be printed readably, as inspired by CMU CL code
4475 of Eric Marsden and SBCL code of Martin Atzmueller.
4476 * better error handling in CLOS method combination, thanks to
4477 Martin Atzmueller porting Pierre Mai's CMU CL patches
4478 * more overflow fixes for >16Mbyte I/O buffers
4479 * A bug in READ has been fixed, so that now a single Ctrl-D
4480 character suffices to cause end-of-file on character streams.
4481 In particular, now you only need one Ctrl-D at the command
4482 line (not two) to exit SBCL.
4483 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4485 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4486 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4487 does the right thing.
4488 * The implementation of some type tests, especially for CONDITION
4489 types, is now tidier and maybe faster, due to CMU CL code
4490 originally by Douglas Crosher, ported by Martin Atzmueller.
4491 * Some math functions have been fixed, and there are new
4492 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4493 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4494 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4495 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4496 likely to be built into sbcl-0.7.0.)
4497 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4498 CL, ported to SBCL by Martin Atzmueller)
4499 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4500 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4501 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4502 * There's a new slam.sh hack to shorten the edit/compile/debug
4503 cycle for low-level changes to SBCL itself, and a new
4504 :SB-AFTER-XC-CORE target feature to control the generation of
4505 the after-xc.core file needed by slam.sh.
4506 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4507 COMPILE-FILE is no longer supported, so that now every function
4508 gets an entry point, so that block compilation looks a little
4509 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4510 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4511 returns the number of bytes consed since the system started,
4512 rather than the number consed since the first time the function
4513 was called. (The new definition parallels ANSI functions like
4514 CL:GET-INTERNAL-RUN-TIME.)
4515 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4516 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4517 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4518 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4519 * Fasl file version numbers are now independent of the target CPU,
4520 since historically most system changes which required version
4521 number changes have affected all CPUs equally. Similarly,
4522 the byte fasl file version is now equal to the ordinary
4525 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4526 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4527 is no longer recognized. I apologize for this, because it was
4528 listed in SB-EXT as a supported extension, but I found that
4529 its existing behavior was poorly specified, as well as incorrectly
4530 specified, and it looked like too much of a mess to straighten it
4531 out. I have enough on my hands trying to get ANSI stuff to work..
4532 * many patches ported from CMU CL by Martin Atzmueller, with
4533 half a dozen bug fixes in pretty-printing and the debugger, and
4534 half a dozen others elsewhere
4535 * fixed bug 13: Floating point infinities are now supported again.
4536 They might still be a little bit flaky, but thanks to bug reports
4537 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4538 as flaky as they were.
4539 * The --noprogrammer command line option is now supported. (Its
4540 behavior is slightly different in detail from what the old man
4541 page claimed it would do, but it's still appropriate under the
4542 same circumstances that the man page talks about.)
4543 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4544 are now supported, and enabled by default. Thus, the compiler can
4545 handle many floating point and complex operations much less
4546 inefficiently. (Thus e.g. you can implement a complex FFT
4548 * The compiler now detects type mismatches between DECLAIM FTYPE
4549 and DEFUN better, and implements CHECK-TYPE more correctly, and
4550 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4551 thanks to patches from Martin Atzmueller.
4552 * various fixes to make the cross-compiler more portable to
4553 ANSI-conforming-but-different cross-compilation hosts (notably
4554 Lispworks for Windows, following bug reports from Arthur Lemmens)
4555 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4556 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4557 fixed thanks to Pierre Mai's CMU CL patches.
4558 * fixes to keep the system from overflowing internal counters when
4559 it tries to use i/o buffers larger than 16M bytes
4560 * fixed bug 45a: Various internal functions required to support
4561 complex special functions have been merged from CMU CL sources.
4562 (When I was first setting up SBCL, I misunderstood a compile-time
4563 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4564 * improved support for type intersection and union, fixing bug 12
4565 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4566 more obscure bugs as well
4567 * some steps toward byte-compiling non-performance-critical
4568 parts of the system, courtesy of patches from Martin Atzmueller
4569 * Christophe Rhodes has made some debian packages of sbcl at
4570 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4571 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4572 stable, but are nonetheless usable. When he's ready, I'd be happy
4573 to add them to the SourceForge "File Releases" section. (And if
4574 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4575 * new fasl file format version number (because of changes in
4576 internal representation of (OR ..) types to accommodate the new
4577 support for (AND ..) types, among other things)
4579 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4580 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4582 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4584 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4585 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4586 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4587 * DESCRIBE now works on structure objects again.
4588 * Most function call argument type mismatches are now handled as
4589 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4590 know whether the function will be redefined before the call is
4591 executed. (The compiler could flag local calls with full WARNINGs,
4592 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4593 it doesn't keep track of enough information to know whether calls
4594 are local in this sense.)
4595 * Compiler output is now more verbose, with messages truncated
4596 later than before. (There should be some supported way for users
4597 to override the default verbosity, but I haven't decided how to
4598 provide it yet, so this behavior is still controlled by the internal
4599 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4600 src/compiler/ir1util.lisp.)
4601 * Fasl file format version numbers have increased again, because
4602 support for the Gray streams extension changes the layout of the
4603 system's STREAM objects.
4604 * The Gray subclassable streams extension now works, thanks to a
4605 patch from Martin Atzmueller.
4606 * The full LOAD-FOREIGN extension (not just the primitive
4607 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4608 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4609 but like most other programs, it defaults to copying the Unix
4610 environment from the original process instead of starting the
4611 new process in an empty environment.
4612 * Extensions which manipulate the Unix environment now support
4613 an :ENVIRONMENT keyword option which doesn't smash case or
4614 do other bad things. The CMU-CL-style :ENV option is retained
4615 for porting convenience.
4616 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4617 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4619 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4621 * A patch from Martin Atzmueller seems to have solved the SIGINT
4622 problem, and as far as we know, signal-handling now works cleanly.
4623 (If you find any new bugs, please report them!)
4624 * The system no longer defaults Lisp source file names to types
4625 ".l", ".cl", or ".lsp", but only to ".lisp".
4626 * The compiler no longer uses special default file extensions for
4627 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4628 seems to expect a single default extension for all compiled code,
4629 and there's no compelling reason to try to stretch the standard
4630 to allow two different extensions.) Instead, byte-compiled files
4631 default to the same extension as native-compiled files.
4632 * Fasl file format version numbers have increased again, because
4633 a rearrangement of internal implementation packages made some
4634 dumped symbols in old fasl files unreadable in new cores.
4635 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4636 many fewer weird special cases.
4637 * Bug #17 (differing COMPILE-FILE behavior between logical and
4638 physical pathnames) has been fixed, and some related misbehavior too,
4639 thanks to a patch from Martin Atzmueller.
4640 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4641 by Tim Moore, ported to SBCL by Martin Atzmueller.
4642 * Martin Atzmueller fixed several filesystem-related problems,
4643 including bug #36, in part by porting CMU CL patches, which were
4644 written in part by Paul Werkowski.
4645 * More compiler warnings in src/runtime/ are gone, thanks to
4646 more patches from Martin Atzmueller.
4647 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4650 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4652 * DESCRIBE now works on CONDITION objects.
4653 * The debugger now handles errors which arise when trying to print
4654 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4656 * The build system now uses an additional file, customize-target-features.lisp,
4657 to allow local modifications to the target *FEATURES* list. (The point of
4658 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4659 debugging features enabled, without having to worry about propagating it
4660 into everyone's system when I do a "cvs update".) When no
4661 customize-target-features.lisp file exists, the target *FEATURES* list
4662 should be constructed the same way as before.
4663 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4664 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4665 warn about behavior which is undefined under the ANSI spec. Note: This
4666 is specified by ANSI, but it's not very popular with programmers.
4667 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4668 macro in the SBCL sources for an example of a workaround which you
4669 might use to make portable ANSI-standard code which does what you want.
4670 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4671 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4672 when it's used as a non-toplevel form. (This is required in order
4673 to implement the DEFCONSTANT-EQX macro.)
4674 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4675 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4676 funny things, due to bugs in the implementation of EVAL-WHEN
4677 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4678 non-toplevel DEFCONSTANTs are uncommon.)
4679 * The core file version number and fasl file version number have been
4680 incremented, because the old noncompliant DEFCONSTANT behavior involved
4681 calling functions which no longer exist, and because I also took the
4682 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4683 * fixed bug 1 (error handling before read-eval-print loop starts), and
4684 redid debugger restarts and related debugger commands somewhat while
4686 ** The QUIT debugger command is gone, since it did something
4687 rather different than the SB-EXT:QUIT command, and since it never
4688 worked properly outside the main toplevel read/eval/print loop.
4689 Invoking the new TOPLEVEL restart provides the same functionality.
4690 ** The GO debugger command is also gone, since you can just invoke
4691 the CONTINUE restart directly instead.
4692 ** The TOP debugger command is also gone, since it's redundant with the
4693 FRAME 0 command, and since it interfered with abbreviations for the
4695 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4697 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4698 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4699 some porting work by Martin Atzmueller)
4700 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4701 with a boatload of patches which he ported to SBCL. Now that those
4703 ** The system tries to make sure that its low-priority messages
4704 are prefixed by semicolons, to help people who like to use
4705 syntax highlighting in their ILISP buffer. (This patch
4706 was originally due to Raymond Toy.)
4707 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4708 patches originally written by Raymond Toy.
4709 ** The compiler understands coercion between single-value and
4710 multiple-VALUES type expressions better, getting rid of some very
4711 weird behavior, thanks to patches originally by Robert MacLachlan
4712 and Douglas Crosher.
4713 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4714 lambda lists, thanks to a patch originally by Pierre Mai.
4715 ** The system no longer bogusly warns about "abbreviated type
4717 ** The compiler gets less confused by inlining and RETURN-FROM,
4718 thanks to some patches originally by Tim Moore.
4719 ** The system no longer hangs when dumping circular lists to fasl
4720 files, thanks to a patch originally from Douglas Crosher.
4721 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4722 undefined function error.
4723 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4724 in structure slot initforms) for now, documented workaround instead:-|
4725 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4726 * fixed bug 5: added stubs for various Gray stream functions called
4727 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4728 installed, at least appropriate type errors are generated
4729 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4730 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4731 consistently in DEFMETHOD forms.
4732 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4733 it doesn't seem to affect SBCL after all
4734 * The C runtime system now builds with better optimization and many
4735 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4737 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4739 * The system is now under CVS at SourceForge (instead of the
4740 CVS repository on my home machine).
4741 * The new signal handling code has been tweaked to treat register
4742 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4743 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4744 downstream, e.g. in debug-int.lisp, has implicit dependencies
4745 on the unsignedness of integer representation of machine words,
4746 and that caused the system to bomb out with infinite regress
4747 when trying to recover from type errors involving signed values,
4748 e.g. (BUTLAST '(1 2 3) -1).)
4749 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4750 (This was one of the bugs Peter Van Eynde reported back in July.)
4751 * The system now uses code inspired by Colin Walters' O(N)
4752 implementation of MAP (from the cmucl-imp@cons.org mailing
4753 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4754 inline the MAP operation, and there is more than one
4755 sequence argument to the MAP call (so that it can't just
4756 do ETYPECASE once and for all based on the type of the
4757 single sequence argument). (The old non-inline implementation
4758 of the general M-argument sequence-of-length-N case required
4759 O(M*N*N) time when any of the sequence arguments were LISTs.)
4760 * The QUIT :UNIX-CODE keyword argument has been renamed to
4761 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4762 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4764 * Raymond Wiker's patches to port dynamic loading from Linux to
4765 FreeBSD have been added.
4766 * The BUGS file is now more nearly up to date, thanks in large part
4767 to Martin Atzmueller's review of it.
4768 * The debugger now flushes standard output streams before it begins
4769 its output ("debugger invoked" and so forth).
4770 * The core version number and fasl file version number have both
4771 been incremented, because of incompatible changes in the layout
4773 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4774 used to not be called for a saved Lisp image.)
4775 * Martin Atzmueller's version of a patch to fix a compiler crash,
4776 as posted on sbcl-devel 13 September 2000, has been installed.
4777 * Instead of installing Martin Atzmueller's patch for the
4778 compiler transform for SUBSEQ, I deleted the compiler transform,
4779 and transforms for some similar consing operations.
4780 * A bug in signal handling which kept TRACE from working on OpenBSD
4782 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4783 away by constant folding
4784 * The system now defines its address space constants in one place
4785 (in the Lisp sources), and propagates them automatically elsewhere
4786 (through GENESIS and the sbcl.h file). Therefore, patching the
4787 address map is less unnecessarily tedious and error-prone. The
4788 Lisp names of address space constants have also been systematized.
4789 * CVS tags like dollar-Header-dollar have been removed from
4790 the sources, because they have never saved me trouble and
4791 they've been source of trouble working with patches and other
4792 diff-related operations.
4793 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4794 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4796 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4798 * The system has been ported to OpenBSD.
4799 * The system now compiles with a simple "sh make.sh" on the systems
4800 that it's supported on. I.e., now you no longer need to tweak
4801 text in the target-features.lisp-expr and symlinks in src/runtime/
4802 by hand, the make.sh takes care of it for you.
4803 * The system is no longer so grossly inefficient when compiling code
4804 involving vectors implemented as general (not simple) vectors (VECTOR T),
4805 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4806 which dares to use the various sequence functions on non-simple
4807 vectors, takes less of a performance hit.
4808 * There is now a primitive type predicate VECTOR-T-P
4809 to test for the (VECTOR T) type, so that e.g.
4810 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4811 can now be compiled with some semblance of efficiency. (The old code
4812 turned the type declaration into a full call to %TYPEP at runtime!)
4813 * AREF on (VECTOR T) is still not fast, since it's still compiled
4814 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4815 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4816 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4817 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4818 without making full calls to SUBTYPEP at runtime.
4819 (Some analogous efficiency issues for non-simple vectors specialized to
4820 element types other than T, or for non-simple multidimensional arrays,
4821 have not been addressed. They could almost certainly be handled the
4822 same way if anyone is motivated to do so.)
4823 * The changes in array handling break binary compatibility, so
4824 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4825 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4826 instead of (VALUES T T).
4827 * By following the instructions that Dan Barlow posted to sbcl-devel
4828 on 2 July 2000, I was able to enable primitive dynamic object
4829 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4830 functionality is not implemented (since it calls ld to resolve
4831 library references automatically, requiring RUN-PROGRAM for its
4832 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4833 try to resolve library references) is now supported.
4834 * The system now flushes the standard output streams when it terminates,
4835 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4836 them at several other probably-convenient times, e.g. in each pass of
4837 the toplevel read-eval-print loop, and after evaluating a form given
4838 as an "--eval" command-line option. (These changes were motivated by a
4839 discussion of stream flushing issues on cmucl-imp in August 2000.)
4840 * The source transform for TYPEP of array types no longer assumes
4841 that an array whose element type is a not-yet-defined type
4842 is implemented as an array of T, but instead punts, so that the
4843 type will be interpreted at runtime.
4844 * There is now some support for cross-compiling in make.sh: each of
4845 the phases of make.sh has its own script. (This should be transparent
4846 to people doing ordinary, non-cross-compile builds.)
4847 * Since my laptop doesn't have hundreds of megabytes of memory like
4848 my desktop machine, I became more motivated to do some items on
4849 my to-do list in order to reduce the size of the system a little:
4850 ** Arrange for various needed-only-at-cold-init things to be
4851 uninterned after cold init. To support this, those things have
4852 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4853 symbols with such names are now uninterned after cold init).
4854 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4855 fasl files for cold load.
4856 ** Remove the old compiler structure pooling code (which used to
4857 be conditional on the target feature :SB-ALLOC) completely.
4858 ** Redo the representation of some data in cold init to be more compact.
4859 (I also looked into supporting byte compiled code at bootstrap time,
4860 which would probably reduce the size of the system a lot, but that
4861 looked too complicated, so I punted for now.)
4862 * The maximum signal nesting depth in the src/runtime/ support code has
4863 been reduced from 4096 to 256. (I don't know any reason for the very
4864 large old value. If the new smaller value turns out to break something,
4865 I'll probably just bump it back up.)
4866 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4868 * Many, many bugs reported by Peter Van Eynde have been added to
4869 the BUGS list; some have even been fixed.
4870 * While enabling dynamic object file loading, I tried to make the
4871 code easier to understand, renaming various functions and variables
4872 with less ambiguous names, and changing some function calling
4873 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4874 * While trying to figure out how to do the OpenBSD port, I tried to
4875 clean up some of the code in src/runtime/. In particular, I dropped
4876 support for non-POSIX signal handling, added various comments,
4877 tweaked the code to reduce the number of compilation warnings, and
4878 renamed some files to increase consistency.
4879 * To support the new automatic configuration functionality in make.sh,
4880 the source file target-features.lisp-expr has been replaced with the
4881 source file base-target-features.lisp-expr and the machine-generated
4882 file local-target-features.lisp-expr.
4883 * fixed a stupid quoting error in make.sh so that using CMU CL
4884 "lisp -batch" as cross-compilation host works again
4886 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4888 * DESCRIBE no longer tries to call itself recursively to describe
4889 bound/fbound values, so that it no longer fails on symbols which are
4890 bound to themselves (like keywords, T, and NIL).
4891 * DESCRIBE now works on generic functions.
4892 * The printer now prints less-screwed-up representations of closures
4893 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4894 * A private symbol is used instead of the :EMPTY keyword previously
4895 used to mark empty slots in hash tables. Thus
4896 (DEFVAR *HT* (MAKE-HASH-TABLE))
4897 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4898 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4899 now does what ANSI says that it should. (You can still get
4900 similar noncompliant behavior if bang on the hash table
4901 implementation with all the symbols you get back from
4902 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4903 This breaks binary compatibility, since tests for equality to
4904 :EMPTY are wired into things like the macroexpansion of
4905 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4907 * There's now a minimal placeholder implementation for CL:STEP,
4908 as required by ANSI.
4909 * An obscure bug in the interaction of the normal compiler, the byte
4910 compiler, inlining, and structure predicates has been patched
4911 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4912 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4914 * Missing ordinary arguments in a macro call are now detected even
4915 when the macro lambda list contains &KEY or &REST.
4916 * The debugger no longer complains about encountering the top of the
4917 stack when you type "FRAME 0" to explicitly instruct it to go to
4918 the top of the stack. And it now prints the frame you request even
4919 if it's the current frame (instead of saying "You are here.").
4920 * As specified by ANSI, the system now always prints keywords
4921 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4923 * The default initial SIZE of HASH-TABLEs is now smaller.
4924 * Type information from CLOS class dispatch is now propagated
4925 into DEFMETHOD bodies, so that e.g.
4926 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4928 is now basically equivalent to
4929 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4930 (DECLARE (TYPE SINGLE-FLOAT X))
4932 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4933 operation, without having to do run-time type dispatch.
4934 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4935 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4936 * Since I don't seem to be making big file reorganizations very often
4937 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4938 is overflowing my ability to conveniently back them up), I've finally
4939 checked the system into CVS. (The CVS repository is on my home system,
4940 not at SourceForge -- putting it on SourceForge might come later.)
4941 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4942 high-level GC-NOTIFY-FOO functions send their output. (There's
4943 still very little control of where low-level verbose GC functions
4944 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4945 less than it used to -- the GC-NOTIFY-FOO functions are now under
4946 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4947 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4948 in only one place in the source code, and propagates it automatically
4949 everywhere that it's needed. Thus e.g. when I bump the version from
4950 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4951 * The C source files now include boilerplate legalese and documentation
4952 at the head of each file (just as the Lisp source files already did).
4953 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4954 to his page will be replaced with a link to his new CLiki service.
4956 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4958 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4959 * The build process now looks for GNU make under the default name "gmake",
4960 instead of "make" as it used to. If GNU make is not available as "gmake"
4961 on your system, you can change this default behavior by setting the
4962 GNUMAKE environment variable.
4963 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4964 can build without error under CMU CL.
4966 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4968 * There is now a partial SBCL user manual (with some new text and some
4969 text cribbed from the CMU CL manual).
4970 * The beginnings of a profiler have been added (starting with the
4971 CMU CL profiler and simplifying and cleaning up). Eventually the
4972 main interface should be through the TRACE macro, but for now,
4973 it's still accessed through vaguely CMU-CL-style functions and macros
4974 exported from the package SB-PROFILE.
4975 * Some problems left over from porting CMU CL to the new
4976 cross-compilation bootstrap process have been cleaned up:
4977 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4978 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4979 part when getting rid of the compiler *BACKEND* stuff.)
4980 ** The value of *NULL-TYPE* was screwed up, because it was
4981 being initialized before the type system knew the final
4982 definition of the 'NULL type. This screwed up several key
4983 optimizations in the compiler, causing inefficiency in all sorts
4984 of places. (I found it because I wanted to understand why
4985 GET-INTERNAL-RUN-TIME was consing.)
4986 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4987 PROCLAIM FTYPE information. Unfortunately this broke binary
4988 compatibility again, since now the forms output by DEFGENERIC
4989 to refer to functions which didn't exist in 0.6.3.
4990 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4991 can use the new (as of 0.6.3) transform for SOME into MAP into
4993 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4994 values to (INTEGER 0 1000000), so that the time code will no longer
4995 occasionally get blown up by Linux returning 1000000 microseconds
4996 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4997 its output conform to the ANSI spec. (Alas, this makes its output
4998 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4999 * A full call to MAP NIL with a single sequence argument no longer conses.
5000 * fixes to problems pointed out by Martin Atzmueller:
5001 * The manual page no longer talks about multiprocessing as though
5002 it were currently supported.
5003 * The ILISP support patches have been removed from the distribution,
5004 because as of version 5.10.1, ILISP now supports SBCL without us
5005 having to maintain patches.
5006 * added a modified version of Raymond Toy's recent CMU CL patch for
5007 EQUALP comparison of HASH-TABLE
5009 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
5011 * The system still can't cross-compile itself with
5012 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
5013 entails), but at least it can compile more of itself that way
5014 than it used to be able to, and various buglets which were uncovered
5015 by trying to cross-compile itself that way have now been fixed.
5016 * This release breaks binary compatibility again. This time
5017 at least I've incremented the FASL file format version to 2, so that the
5018 problem can be detected reliably instead of just causing weird errors.
5019 * various new style warnings:
5020 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
5021 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
5022 ** using the lexical binding of a variable named in the *FOO* style
5023 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
5024 as specified by ANSI.
5025 * *RANDOM-STATE* is no longer automatically initialized from
5026 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
5027 default behavior of the system is to repeat its behavior every time
5028 it's run. If you'd like to change this behavior, you can always
5029 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
5030 old convention there was no comparably easy way to get the system to
5031 repeat its behavior every time it was run.
5032 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
5033 FTYPE declarations has been removed, in favor of their ANSI
5034 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
5035 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
5036 the types of their sequence arguments can be determined at compile time.
5037 This is done through a new open code expansion for MAP which eliminates
5038 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
5039 argument types can be determined at compile time.
5040 * The optimizer now transforms COERCE into an identity operation when it
5041 can prove that the coerced object is already of the correct type. (This
5042 can be a win for machine generated code, including the output of other
5043 optimization transforms, such as the MAP transform above.)
5044 * Credit information has been moved from source file headers into CREDITS.
5045 * Source file headers have been made more standard.
5046 * The CASE macro now compiles without complaining even when it has
5049 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
5051 * (Note that the way that the PCL macroexpansions were rewritten
5052 to accommodate the change in DEFGENERIC below breaks binary
5053 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
5054 not run under sbcl-0.6.2. Once we get out of alpha releases,
5055 i.e. hit release 1.0.0, we'll probably try to maintain binary
5056 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5057 and sbcl-1.4.4. Until then, however, it might be fairly common
5058 for maintenance releases to break binary compatibility.)
5059 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5061 * The reporting of unhandled signals has been changed to print some
5062 explanatory text as well as the report form. (Previously only
5063 the report form was printed.)
5064 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5065 it defines, so that the compiler no longer issues undefined function
5066 warnings for compiled-but-not-yet-loaded generic functions.
5067 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5068 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5069 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5070 section of the ANSI spec are supported.
5071 * The "" nickname for the "KEYWORD" package has been removed.
5072 The reader still handles symbol tokens which begin with a package marker
5073 as keywords, but it doesn't expose its mechanism for doing so in the
5074 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5075 * The system now issues STYLE-WARNINGs for contradictory TYPE
5076 proclamations. (Warnings for contradictory FTYPE proclamations would
5077 be nice too, but those can't be done usefully unless the type system
5078 is made smarter about FUNCTION types.)
5079 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5080 systematized, so that "*target-*.lisp is supposed to exist only on the
5081 target and imply that there's a related file which exists on the
5082 host, and *host-*.lisp is supposed to exist only on the host and imply
5083 that there's a related file which exists on the target. This involves a
5084 lot of renaming. Hopefully the acute confusion caused by the renaming
5085 will be justified by the reduction in chronic confusion..
5086 ** runtime-type.lisp -> early-target-type.lisp
5087 ** target-type.lisp -> late-target-type.lisp
5088 ** early-host-format.lisp -> early-format.lisp
5089 ** late-host-format.lisp -> late-format.lisp
5090 ** host-error.lisp -> misc-error.lisp
5091 ** early-error.lisp -> early-target-error.lisp
5092 ** late-error.lisp -> late-target-error.lisp
5093 ** host-defboot.lisp -> early-defboot.lisp
5094 ** code/misc.lisp -> code/target-misc.lisp
5095 ** code/host-misc.lisp -> code/misc.lisp
5096 ** code/numbers.lisp -> code/target-numbers.lisp
5097 ** code/early-numbers.lisp -> numbers.lisp
5098 ** early-host-type.lisp -> early-type.lisp
5099 ** late-host-type.lisp -> late-type.lisp
5100 ** host-typep.lisp -> typep.lisp
5101 ** load.lisp -> target-load.lisp
5102 ** host-load.lisp -> load.lisp
5103 ** host-disassem.lisp -> disassem.lisp
5104 ** host-insts.lisp -> insts.lisp
5105 ** byte-comp.lisp -> target-byte-comp.lisp
5106 ** host-byte-comp.lisp -> byte-comp.lisp
5107 ** host-signal.lisp -> signal.lisp
5108 ** host-defstruct.lisp -> defstruct.lisp
5109 ** late-target-type.lisp -> deftypes-for-target.lisp
5110 Furthermore, several other previously target-only files foo.lisp (e.g.
5111 hash-table.lisp and random.lisp) have been split into a target-and-host
5112 foo.lisp file and a target-only target-foo.lisp file, with their key type
5113 definitions in the target-and-host part, so that the cross-compiler will
5114 know more about target types.
5115 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5116 gone away. In their place are various *BACKEND-FOO* variables
5117 corresponding to the slots of the old structure.
5118 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5119 into the target SBCL has been fixed.
5120 * The chill.lisp system for loading cold code into a running SBCL
5122 * Support for the CMU CL "scavenger hook" extension has been removed.
5123 (It was undocumented and unused in the CMU CL sources that SBCL was
5124 derived from, and stale in sbcl-0.6.1.)
5125 * Various errors in the cross-compiler type system were detected
5126 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5127 (enabling various consistency checks). Many of them were fixed,
5128 but some hard problems remain, so the compiler is back to
5129 running without *TYPE-SYSTEM-INITIALIZED* for now.
5130 * As part of the cross-compiler type system cleanup, I implemented
5131 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5132 * I have started adding UNCROSS calls throughout the type system
5133 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5134 on cross-compiler input files will be able to go away, and various
5136 * CONSTANTP now returns true for quoted forms (as explicitly required
5139 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5141 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5142 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5143 (MAKE-LIST -1). In the longer term, it ought to become true
5144 that declarations are assertions even at SAFETY 1. For now, it's not
5145 quite true even at SAFETY 3, but it's at least more nearly true..
5146 (Note that this change seems to increases the size of the system by
5147 O(5%) and to decrease the speed of the compiler by 20% or more.)
5148 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5149 for the problem of printing dozens of lines of distracting information
5150 about low-level system machinery as part of the top stack frame
5151 on entry to the debugger when an undefined function was called.
5152 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5153 is not reset to COMMON-LISP-USER.
5154 * Compilation of stuff related to dyncount.lisp has been made conditional
5155 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5156 smaller. The various dyncount-related symbols have been moved into
5157 a new "SB-DYNCOUNT" package.
5158 * tty-inspect.lisp has been renamed to inspect.lisp.
5159 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5160 feature has gone away. (When we eventually port to other flavors of
5161 libc and/or Unix, we'll try to make the differences between flavors
5162 invisible at the user level.)
5163 * Various other *FEATURES* tags, and/or their associated conditionals,
5164 have been removed if obsolescent, or given better documentation, or
5165 sometimes given more-mnemonic names.
5167 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5169 * tidied up "make.sh" script
5170 * tidied up system directory structure
5171 * better "clean.sh" behavior
5172 * added doc/FOR-CMUCL-DEVELOPERS
5173 * many many small tweaks to output format, e.g. removing possibly-confusing
5174 trailing #\. character in DESCRIBE-INSTANCE
5175 * (EQUALP #\A 'A) no longer signals an error.
5176 * new hashing code, including EQUALP hashing
5177 * tidied up Lisp initialization and toplevel
5178 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5179 * command line argument processing
5180 * added POSIX-GETENV function to deal with Unix-ish environment variables
5181 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5182 terminating SBCL on EOF
5183 * non-verbose GC by default
5184 * There is no more "sbcl" shell script; the sbcl file is now the C
5185 runtime executable (just like CMU CL).
5186 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5188 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5189 symbol and package information as Lisp native symbol and package objects
5190 instead of strings naming symbols and strings naming packages. This way,
5191 whenever packages are renamed (as in warm init), debug information is
5192 transformed along with everything else.
5193 * tweaked the optimization policy declarations which control the building
5194 of SBCL itself. Now, among other things, the system no longer saves
5195 source location debugging information. (This helps two problems at once
5196 by reducing SBCL size and by keeping SBCL from trying to look for its
5197 sources -- which may not exist -- when reporting errors.)
5198 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5199 debugging and testing purposes
5200 * cleaned up printing, making the printer call PRINT-OBJECT for
5201 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5202 methods, giving nearly-ANSI behavior
5203 * converted almost all special variables to use *FOO* naming convention
5204 * deleted PARSE-TIME functionality, since it can be done portably
5205 * moved some files out of cold init into warm init
5206 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5208 * regularized formatting of source files
5209 * added an install.sh script
5210 * fixed ridiculous memory usage of cross-compiler by making
5211 compiler/alloc.lisp not try to do pooling unless it can hook
5212 itself into the GC of the cross-compilation host. Now the system
5213 builds nicely on my old laptop.
5214 * added :SB-ALLOC in target-features.lisp-expr
5215 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5216 was not implemented)
5217 * re-did condition handling and note reporting in the compiler. Notes
5218 are no longer handled by signalling conditions. Style warnings
5219 and warnings are handled more correctly and reported in such a way
5220 that it's easy to find one or the other in your output (so that you
5221 can e.g. figure out which of many problems caused COMPILE-FILE to
5223 * changed the severity of several compiler warnings from full WARNING
5224 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5225 compiler note reporting so that it doesn't use the condition system
5226 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5227 COMPILE-FILE command)
5228 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5229 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5230 no longer supported, since their implementation was deeply intertwingled
5231 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5232 * removed bogus "support" for compiler macros named (SETF FOO), and
5233 removed the compiler macro for SETF INFO (but only after making a fool
5234 of myself on the cmucl-imp mailing list by posting a bogus patch for
5235 DEFINE-COMPILER-MACRO..)
5236 * Compiled files containing forms which have side effects on the Lisp
5237 reader (such as DEFPACKAGE forms) are now handled more correctly.
5238 (Compiler queuing of top level lambdas has been suppressed by setting
5239 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5240 * deleted various currently-unused source files, e.g. gengc.lisp. They
5241 may be added back at some point e.g. when porting to other architectures,
5242 but until they are it's distracting to distribute them and to try to
5244 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5245 there were so many of them they're just distractions, and UNCROSS is
5246 known to be able to handle the current sources
5247 * moved PROFILE functionality into TRACE, so that it will be clear
5248 how the wrapping and unwrapping of functions when you profile them
5249 interacts with the wrapping and unwrapping of functions when you
5250 trace them. (Actually, the functionality isn't there yet, but at least
5251 the interface specification is there. Hopefully, the functionality will
5252 arrive with some future maintenance release.)
5253 * removed host-oops.lisp
5254 * changed signature of QUIT function to allow UNIX-CODE argument
5255 * fixed READ-SEQUENCE bug
5256 * tweaked verbose GC output so that it looks more like the progress
5257 output that ANSI specifies for functions like LOAD
5258 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5259 * added <http://sbcl.sourceforge.com> to the banner information printed by