1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 * minor incompatible change: under weak type checking policy integer
3 types are weakened less aggressively.
4 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
5 allowed to return, which causes SBCL to quit with exit status 0. Previously
6 if the function returned with a small integer return value, that value
7 was accidentally reused as the exit status.
8 * new feature: SB-EXT:DEFINE-HASH-TABLE-TEST allows defining new arguments
9 to MAKE-HASH-TABLE :TEST, and MAKE-HASH-TABLE has been extended with
10 :HASH-FUNCTION argument. Refer to user manual for details.
11 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
13 * new feature: SB-EXT:GET-TIME-OF-DAY provides access to seconds and
14 microseconds since the Unix epoch on all platforms.
15 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
16 allows the compiler to safely elide boundedness checks for special
18 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
19 the symbol, prohibits both lexical and dynamic binding. This is mainly an
20 efficiency measure for threaded platforms, but also valueable in
22 * new feature: UNC pathnames are now understood by the system on Windows.
23 * optimization: the compiler uses a specialized version of FILL when the
24 element type is know in more cases, making eg. (UNSIGNED-BYTE 8) case
26 * optimization: accesses to potentially non-simple arrays where element type
27 is known are 50% faster.
28 * optimization: compiler now generates faster array typechecking code.
29 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
31 * optimization: multidimensional array accesses in the absence of type
32 information regarding array rank are approximately 10% faster due to
33 open coding of ARRAY-RANK.
34 * optimization: result of (FILL (MAKE-ARRAY ...) ...) and (REPLACE
35 (MAKE-ARRAY ...) ...) can be stack allocated if the result of MAKE-ARRAY
37 * optimization: result of call to VECTOR can now be stack allocated.
38 * optimization: MAKE-ARRAY with :INITIAL-CONTENTS is now vastly faster
39 as long as the resulting array is one-dimensional and has a known
40 element type. In particular, :INITIAL-CONTENTS (LIST ...) where the
41 length of the list matches the known length of the vector does not
42 allocate the list as an intermediate step. Ditto for VECTOR and simple
44 * optimization: MAKE-ARRAY can now stack allocate in the presence of
45 :INITIAL-CONTENTS and :INITIAL-ELEMENT as long as the result has a
46 known element type, and is known to be simple and one dimensional.
47 * improvement: SBCL now emits a compiler note where stack allocation was
48 requested but could not be provided (not in all cases, unfortunately)
49 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
50 * improvement: pretty-printing loop has been implemented properly. (thanks
52 * documentation: CLOS slot typechecing policy has been documented.
53 * bug fix: SB-SPROF could be foiled by foreign code not have a frame
54 pointer, leading to memory faults. (thanks to Bart Botta)
55 * bug fix: better floating point exception handling on x86/OpenBSD.
56 (thanks to Josh Elsasser)
57 * bug fix: exit status from QUIT when called under --script was lost
58 (reported by Hubert Kauker)
59 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
60 same implementation of FILL to initialize the array, even if a more
61 efficient one was available (reported by Stas Boukarev, thanks to
63 * bug fix: potential miscompilation of array stack allocation on x86 and
64 x86-64. (reported by Time Tossavainen)
65 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
66 result in their subforms being treated as top level forms. (reported by
68 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
69 than a machine register could leave garbage in the high bits of the
70 result register (bug 316325).
71 * bug fix: disable address space randomization Linux/x86-64 as well,
72 not just x86-64. (reported by Ken Olum)
73 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
74 generate incorrect code.
75 * bug fix: #201; type inference for CONS and ARRAY types could derive
76 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
77 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
78 could not be assigned to or bound with PROGV. (reported by Lorenz
80 * bug fix: the value of CL:- in the inspector was the previous expression
81 evaluated rather than the expression being evaluated.
82 * bug fix: constants can no longer be locally declared special.
83 * bug fix: signals delivered to threads started from foreign land (read:
84 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
85 thread by blocking all signals and resignalling.
87 changes in sbcl-1.0.28 relative to 1.0.27:
88 * a number of bugs in cross-compilation have been fixed, with the ultimate
89 result that building under (at least) clisp should be much more reliable.
90 * minor incompatible changes: echo-streams now propagate unread-char to the
91 underlying input stream, and no longer permit unreading more than one
93 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
94 frames, allowing for instance more reliable backtraces.
95 * improvement: the debugger REPL can now reference lexical variables
96 by name directly for code compiled with (DEBUG 3).
97 * improvement: errors from malformed declarations now have better source
98 paths associated with them. (thanks to Tobias Rittweiler)
99 * optimization: faster local calls on x86/x86-64
100 * bug fix: some error messages for out-of-bound array indexes confused the
101 index and the bound. (thanks to Stas Boukarev)
102 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
104 * bug fix: running regressions tests in shells without OSTYPE set now works.
105 (reported by Harald Hanche-Olsen)
106 * bug fix: more robust static space exhaustion signalling from
107 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
108 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
109 docstring away instead of storing it under names such as (LAMBDA (X)).
110 (reported by Leslie Polzer)
111 * bug fix: timers could go off in the wrong order, be delayed indefinitely
112 (thanks to Ole Arndt for the patch)
113 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
114 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
115 potentially-invalid effective methods in its cache.
116 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
117 instances as well (thanks to Paul Khuong)
118 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
119 SB-POSIX:WAITPID and their C equivalents.
120 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
122 changes in sbcl-1.0.27 relative to 1.0.26:
123 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
124 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
125 * improvement: the system either recovers from stack exhaustion or dies
126 properly as opposed to leaving the user uncertain of whether the handler
127 trampled on some random memory next to the stack or having to rely on
128 --lose-on-corruption (which is still a good idea to use in production
129 because stack exhaustion can happen in signal handlers which will likely
131 * bug fix: fix gc related interrupt handling bug on ppc (regression from
132 1.0.25.37, reported by Harald Hanche-Olsen)
133 * bug fix: work around signal delivery bug in darwin (regression from
134 1.0.25.44, reported by Sidney Markowitz)
135 * bug fix: fix ERROR leaking memory (reported by David Thompson)
137 changes in sbcl-1.0.26 relative to 1.0.25:
138 * incompatible change: an interruption (be it a function passed to
139 INTERRUPT-THREAD or a timer function) runs in an environment where
140 interrupts can be enabled. The interruption can use
141 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
142 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
143 potentially running out of stack. Keep in mind that in the absence
144 of WITHOUT-INTERRUPTS some potentially blocking operation such as
145 acquiring a lock can enable interrupts.
146 * incompatible change: GC-OFF and GC-ON are removed, as they were
147 always unsafe. Use WITHOUT-GCING instead.
148 * new feature: runtime option --disable-ldb
149 * new feature: runtime option --lose-on-corruption to die at the
150 slightest hint of possibly non-recoverable errors: running out of
151 memory, stack, alien stack, binding stack, encountering a memory
152 fault, etc. In the absence of --lose-on-corruption a warning is
154 * enhancement: detect binding stack exhaustion
155 * enhancement: detect alien stack exhaustion on x86/x86-64
156 * improvement: generally more stable and reliable interrupt handling
157 * improvement: there is a per thread interruption queue,
158 interruptions are executed in order of arrival
159 * improvement: a repeating timer reschedules itself when the it has
160 finished, but expiration times are spaced equally. If an
161 expiration time is in the past it will trigger after a short grace
162 period that may give a chance to other things to run.
163 * optimization: slightly faster gc on multithreaded builds
164 * optimization: faster WITHOUT-GCING
165 * bug fix: when JOIN-THREAD signals an error, do it when not holding
166 important locks so that the debugger/handler doesn't produce
167 recursive errors or deadlock.
168 * bug fix: real-time signals are not used anymore, so no more
169 hanging when the system wide real-time signal queue gets full.
170 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
172 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
173 * bug fix: fix random memory faults related to interrupts on alpha
174 * bug fix: fix random memory faults related to interrupts on sparc
175 * bug fix: fix deadlocks related to starting threads
176 * bug fix: fix deadlines on locks on futex platforms
177 * bug fix: restore errno in signal handlers
178 * bug fix: fix deadlocks related to hash tables
179 * bug fix: fix deadlocks in pcl
181 changes in sbcl-1.0.25 relative to 1.0.24:
182 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
183 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
184 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
185 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
186 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
187 better name and does not return values so stale on multiprocessor systems.
188 Also, HOLDING-MUTEX-P was added for about the only sane usage of
190 * improvement: unithread builds keep track of MUTEX-VALUE.
191 * improvement: reading from a TWO-WAY-STREAM does not touch the output
192 stream anymore making it thread safe to have a concurrent reader and
193 a writer, for instance, in a pipe.
194 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
195 arguments, making compiler-macros for SETF-functions able to inspect
196 their constant arguments.
197 * improvement: COMPILE-FILE reports times with millisecond accuracy
198 (thanks to Luis Oliveira)
199 * optimization: CHAR-CODE type derivation has been improved, making
200 TYPEP elimination on subtypes of CHARACTER work better. (reported
201 by Tobias Rittweiler, patch by Paul Khuong)
202 * bug fix: setting alien structure fields of type struct by value now
203 computes the right offset for the memory copy.
204 * bug fix: compilation problem involving inlined calls to aliens with
205 result type VOID. (reported by Ken Olum)
206 * bug fix: #235a; sequential inline expasion in different policies no
207 longer reuses the functional from the previous expansion site.
208 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
209 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
210 * bug fix: named ENUMs in multiply-referenced alien record types no longer
211 cause an implied type redefinition. Regression from 1.0.21.29.
212 * bug fix: sign-extension of small signed return values now works with C
213 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
214 * improvements to the Windows port:
215 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
216 Westfold, thanks to Rudi Schlatte)
218 changes in sbcl-1.0.24 relative to 1.0.23:
219 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
220 vector of a multidimensional SIMPLE-ARRAY.
221 * new feature: the system now signals a continuable error if standard
222 readtable modification is attempted. (thanks to Tobias Rittweiler)
223 * new feature: DIRECTORY has been extended with a non-standard keyword
224 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
225 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
226 has been robustified and documented better. (thanks to Robert Goldman)
227 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
228 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
229 * tradeoff: constant FORMAT control strings are now compiled unless
230 SPACE > SPEED (previously only when SPEED > SPACE.)
231 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
233 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
235 * bug fix: lexical type declarations are now correctly reported by
236 SB-CLTL2. (reported by Larry D'Anna)
237 * bug fix: STRING-TO-OCTETS did not handle :START properly when
238 using UTF-8 as external format. (thanks to Luis Oliveira)
239 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
240 are signalled correctly. (thanks to Stas Boukarev)
241 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
242 designator. (thanks to Tobias Rittweiler)
243 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
244 designator, and returns T instead of the function. (thanks to
246 * bug fix: direct superclasses of STANDARD-CLASS and
247 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
248 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
249 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
250 when no class with the named slot yet exists no longer causes a
251 compile-time style-warning.
252 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
253 in safe code. (reported by Didier Verna)
254 * bug fix: #430; nested structure constructors can stack allocate.
255 * bug fix: on some 64-bit platforms dynamic space size was truncated
256 to #xffff0000 bytes. (reported by Benjamin Lambert)
257 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
258 REPL unusable. (reported by Daniel Herring)
259 * bug fix: return values of READ-SEQUENCE did not take :START into
260 account on file streams, regressions since 1.0.12.22. (reported by
261 Thomas Russ, patch by Paul Khuong)
262 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
263 method specializer used to confuse permuation vector optimization.
264 * bug fix: system inserted bogus implicit type declarations for local
265 special variables in DEFMETHOD bodies.
266 * bug fix: #354; duplicated frames in backtraces due to
267 non-tail-call-optimized XEPs to functions with return type NIL
269 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
270 initializes structure object slots according to DEFSTRUCT initforms,
271 and DEFSTRUCT forms :INCLUDEind structure classes defined using
272 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
273 (reported by Bruno Haible and Stephen Wilson)
274 * bug fix: #395; fill-pointer output streams used now support
275 element-type BASE-CHAR as well.
276 * bug fix: compiler error when attempting to derive return value of
277 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
279 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
281 * bug fix: Mac OS X binaries should now be portable between Leopard
284 changes in sbcl-1.0.23 relative to 1.0.22:
285 * enhancement: when disassembling method functions, disassembly
286 for the associated fast function is also produced.
287 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
289 * optimization: printing with *PRINT-PRETTY* true is now more
290 efficient as long as the object being printed doesn't require
291 special handling by the pretty printer.
292 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
293 now interact correctly with type declarations.
294 * partial bug fix: PCL detects infinite recursion during wrapper
295 validation. (thanks to Attila Lendvai)
296 * bug fix: #426; nested function calls are inlined properly.
297 Previously if FOO was an inline function, in calls of the form
298 (FOO (FOO ...)) the outer call was not inlined.
299 * bug fix: long long arguments passed in registers to alien
300 functions on PowerPC Linux and NetBSD are now handled correctly
301 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
303 changes in sbcl-1.0.22 relative to 1.0.21:
304 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
305 for the shared object in the current directory, but passes the native
306 namestring of the designated pathname to the operation system's shared
307 object loading function as-it.
308 * minor incompatible change: --disable-debugger toplevel option now takes
309 effect before processing of initialization files and --eval or --load
311 * new feature: new commandline argument: --script, which supports
312 shebang lines. See documentation for details. (based on work by
314 * new feature: SAVE-LISP-AND-DIE can save current values of
315 --dynamic-space-size and --control-stack-size in the executable core,
316 causing it to skip normal runtime option processing. See documentation
317 for details. (thanks to Zach Beane)
318 * enhancement: inoccous calls to EVAL or generic functions dispatching
319 on subclasses of eg. STREAM no longer cause compiler notes to appear.
320 * enhancement: the system no longer resignals errors from --load and
321 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
322 associated with the original error to be lost. (thanks to Ariel
324 * enhancement: :DONT-SAVE keyword argument has been added to
325 LOAD-SHARED-OBJECT for controlling interaction with
326 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
327 the effects of an earlier LOAD-SHARED-OBJECT call.
328 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
329 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
330 provided. (reported by Cedric St-Jean)
331 * bug fix: circularity handling in the reader did not treat raw
332 structure slots correctly. (reported by Cedric St-Jean)
333 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
334 file descriptors when there were none.
335 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
336 was broken. (reporteed by Willem Broekema)
337 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
338 pathnames without a directory.
339 * bug fix: COMPILE-FILE did not warn about undefined variable
340 references at toplevel, and LOAD of the resulting fasl did
342 * bug fix: functions with non-required arguments used to end up with
343 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
344 * bug fix: redefining a function with non-required arguments didn't
345 update the system's knowledge about its call signature properly.
346 * bug fix: fixed #431; incompatible alien record type redefinitions
347 are detected and handled. (reported by Neil Haven)
348 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
349 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
351 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
352 after alien stack frames.
353 * bug fix: strings passed to mk{s,d}temp lacked null termination
355 changes in sbcl-1.0.21 relative to 1.0.20:
356 * new feature: the compiler is able to track the effective type of a
357 generic function across method addition and removal.
358 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
359 appropriately typed structure slots without locking.
360 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
361 information like those gathered by TIME using a programming-friendly
363 * new feature: TIME reports time taken even if the form performs a
364 non-local transfer of control.
365 * enhancement: reduced conservativism on GENCGC platforms: on
366 average 45% less pages pinned (measured from SBCL self build).
367 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
368 mutate constant symbols or violate declaimed type of the symbol.
369 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
370 sleeping on the semaphore in heavily contested semaphores.
371 * bug fix: semaphores and condition variables were not interrupt
373 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
374 owned by other threads anymore.
375 * bug fix: FIND on lists called KEY outside the specified
376 subsequence. (reported by budden)
377 * bug fix: LOG doesn't use single-float intermediate results when
378 given mixed integer and double-float arguments, leading to better
379 precision. (reported by Bob Felts)
380 * bug fix: LOG with base zero returned values of inconsistent type.
382 changes in sbcl-1.0.20 relative to 1.0.19:
383 * minor incompatible change: OPTIMIZE qualities
384 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
385 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
386 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
387 * documentation: some slot access efficiency guidelines have been
388 added to the user manual.
389 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
390 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
391 as ASSOC and MEMEBER.
392 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
393 to more efficient EQ-comparison versions more often.
394 * optimization: enhanced derivation of DOLIST iteration variable type
396 * optimization: constant folding of simple (LIST ...) forms as DOLIST
398 * optimization: runtime lookup of function definitions can be
399 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
400 * optimization: compiler is able to derive the return type of
401 (AREF (THE STRING X) Y) as being CHARACTER.
402 * optimization: CLRHASH on empty hash-tables no longer does pointless
403 work. (thanks to Alec Berryman)
404 * optimization: REPLACE deftransforms don't punt when :START1 or
406 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
407 no longer emits a code deletion note. (reported by Andrew
409 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
410 functions with an odd number of &OPTIONAL arguments, a &REST
411 argument, and one or more &KEY arguments at the call site.
412 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
413 arguments appeared at call sites as well.
414 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
415 READ-LINE &co can no longer read from them afterwards. (reported
417 * bug fix: fixed #427: unused local aliens no longer cause compiler
418 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
420 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
421 alien-stack leakage. (reported by Andy Hefner)
422 * bug fix: PROGV signals an error when an attempt to violate declared
423 type of a variable or bind a constant is made.
424 * bug fix: SET signals an error when an attempt to violate declared
425 type of a variable is made.
426 * bug fix: restart computation during the execution of a restart
427 test function no longer causes infinite recursion. (reported by
429 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
430 signals an error. (thanks to Pierre Mai)
431 * bug fix: EXPORT left symbol unexported in conflict situations.
432 (thanks to Michael Weber)
433 * bug fix: correctly handle name conflicts involving CL:NIL.
434 (thanks to Michael Weber)
435 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
436 conflicts arising from USEing package with conflicting symbols
437 correctly. (thanks to Michael Weber)
439 changes in sbcl-1.0.19 relative to 1.0.18:
440 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
441 warnings that go otherwise unhandled will be muffled if they are
442 of the type that's the value of this variable.
443 * optimization: stack allocation is slightly more efficient on x86
445 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
446 where a raw slot always is initialized using the initform whose
447 type is not know sufficiently well a compile-time are now compiled
448 correctly. (reported by John Morrison)
449 * bug fix: compiler no longer makes erronous assumptions in the
450 presence of non-foldable SATISFIES types.
451 * bug fix: stack analysis missed cleanups of dynamic-extent
452 arguments in non-let-converted calls to local functions.
453 * improvements to the Windows port:
454 ** adjusted address spaces for building on both Win32 and
455 Win64. (thanks for John Connors)
456 * fixed some bugs revealed by Paul Dietz' test suite:
457 ** interval arithmetic during type derivation used inexact integer
458 to single-float coercions.
459 ** arithmetic operations involving large integers and single
460 floats give the same results in compiled and interpreted code.
461 ** deriving the result type of COERCE no longer signals an error
462 if the derived type of the second argument is a MEMBER type
463 containing invalid type specifiers.
464 ** ADJOIN with constant NIL as second argument works correctly.
465 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
467 changes in sbcl-1.0.18 relative to 1.0.17:
468 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
469 profiles only the current thread.
470 * minor incompatible change: changes to SYMBOL-VALUE of constants
471 defined with DEFCONSTANT now signal an error.
472 * enhancement: SB-SPROF now has support for wallclock profiling,
473 and is also able to profile specific threads. REPORT output
474 has also additional sorting options.
475 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
477 * optimization: structure allocation has been improved
478 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
479 ** out of line constructors are ~10% faster.
480 ** inline constructors are ~15% faster.
481 ** inline constructors are capable of dynamic extent allocation
482 (generally on x86 and x86-64, in some cases on other platforms
484 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
486 * optimization: file compiler is now able to coalesce non-circular
487 lists, non-base strings, and bit-vectors. Additionally, constants
488 are never referenced through SYMBOL-VALUE at runtime.
489 * optimization: code defining methods on PRINT-OBJECT (and other
490 generic functions in the COMMON-LISP package) now loads faster.
491 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
493 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
494 fasl is now deleted. (reported by Attila Lendvai)
495 * bug fix: READ-LINE always returned NIL for the last line in files.
496 (reported by Yoshinori Tahara)
497 * bug fix: more accurate disassembly annotations of foreign function
498 calls. (thanks to Andy Hefner)
499 * bug fix: trimming non-simple strings and non-string string
500 designators when the there is nothing to trim works properly.
501 (thanks to James Knight)
502 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
503 (thanks to Travis Cross)
504 * fixed some bugs revealed by Paul Dietz' test suite:
505 ** NIL is a valid function name (regression at 1.0.13.38)
506 ** FILL on lists was missing its return value (regression at 1.0.12.27)
507 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
508 fill pointers properly (regression at 1.0.12.23)
509 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
510 displacement indices properly (regression at 1.0.12.23)
512 changes in sbcl-1.0.17 relative to 1.0.16:
513 * temporary regression: user code can no longer allocate closure
514 variable storage on stack, due to bug 419 without explicitly
515 requesting it. Please consult sbcl-devel for advice if you need to
516 use this feature in the meanwhile.
517 * new feature: runtime argument --control-stack-size can be used to
518 adjust thread default control stack size.
519 * enhancement: improved TIME output
520 ** all times are reported using the measured accuracy (milliseconds
521 for real and GC times, microseconds for everything else.)
522 ** processor cycle counts on x86 and x86-64.
523 ** interpreted forms are counted for both evaluator modes.
524 ** number of lambdas converted by the compiler is reported.
525 ** CPU percentage report (computed from real and total run time.)
526 ** more comprehensive run time reporting, using a condenced format
527 ** interperted form, lambda, and page fault counts are omitted
529 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
531 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
532 * optimization: two argument forms of LAST are upto ~10% faster
533 in normal SPEED policies.
534 * optimization: NCONC no longer needs to heap cons its &REST list
535 in normal SPEED policies.
536 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
538 * bug fix: UNION and NUNION work with :TEST-NOT once more,
539 regression since 1.0.9.1. (thanks to Eric Marsden)
540 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
541 since 1.0.15.36. (thanks to Paul Khuong)
542 * bug fix: LAST when always returned the whole list when given a bignum
543 as the second argument.
544 * bug fix: dynamic extent allocation of nested lists and vectors
545 could leak to otherwise accessible parts.
546 * bug fix: invalid optimization of heap-allocated alien variable
548 * bug fix: fasl header checking is less vulnerable to different
549 platform word lengths.
550 * bug fix: more correct assembler syntax for GNU binutils
551 2.18.50.0.4 support. (thanks to Marijn Schouten)
552 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
554 * bug fix: the fopcompiler can handle LOCALLY forms (with no
555 declarations) successfully. (reported by Attila Lendvai)
557 changes in sbcl-1.0.16 relative to 1.0.15:
558 * minor incompatible change: revert the changes to sb-posix's error
559 signaling added in 1.0.14.
560 * minor incompatible change: change PROBE-FILE back to returning
561 NIL whenever we can't get a truename, as was the case before 1.0.14.
562 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
563 inherits from ERROR instead of just CONDITION.
564 * new feature: SB-INTROSPECT can provide source locations for instances
565 as well. (thanks to Tobias Rittweiler)
566 * optimization: binding special variables now generates smaller code
567 on threaded platforms.
568 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
569 and cases where no :TEST is given but the compiler can infer that
570 the element to search is of type (OR FIXNUM (NOT NUMBER)).
571 * optimization: better LOGNOT on fixnums.
572 * optimization: modular arithmetic for a particular requested width
573 is implemented using a tagged representation unless a better
574 representation is available.
575 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
576 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
577 for wellformedness and returns a specialized vector. (reported by
578 Francois-Rene Rideau)
579 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
580 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
581 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
582 unscheduling timer at the same time as another timer fires.
583 * bug fix: don't reschedule timers for dead threads.
584 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
585 * bug fix: copying output from RUN-PROGRAM to a stream signalled
586 bogus errors if select() was interrupted.
587 * enhancement: add support for fcntl's struct flock to SB-POSIX.
589 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
590 * enhancement: cleaner backtraces for interactive interrupts, as
591 well as other cases where the interesting frames used to be
592 obscured by interrupt handling frames.
593 * enhancement: untracing a whole package using (UNTRACE "FOO") is
594 now supported, and tracing a whole package using (TRACE "FOO") now
595 traces SETF-functions as well.
596 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
597 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
598 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
599 * unreadably printed representation of hash-tables now includes
601 * bug fix: partially fixed #188: type propagation from assignments
602 is now more efficient.
603 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
604 'DOUBLE-FLOAT) are not flushable.
605 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
606 full address of the object, and none of the tag bits.
607 * bug fix: readably printing hash-tables now respects other printer
608 control variables. (reported by Cedric St-Jean)
609 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
610 keyword with MAKE-HASH-TABLE.
611 * bug fix: export SB-POSIX:MKSTEMP.
612 * bug fix: SORT was not interrupt safe.
613 * bug fix: XREF accounts for the last node of each basic-block as
615 * bug fix: MAKE-INSTANCE optimizations interacted badly with
616 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
617 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
618 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
619 create obsolete instances in the case of redefinition or
620 obsoletion of a superclass. (thanks to Andy Hefner)
621 * bug fix: Support for the Alpha architecture has been revived; it had
622 suffered somewhat from lack of maintenance since sbcl-1.0.
623 * improvements to the Windows port:
624 ** The system detects the codepage to use at startup, instead of
625 using the value from the saved core. (thanks to Kei Suzuki)
627 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
628 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
629 (see documentation for details.)
630 * revived support for OpenBSD (contributed by Josh Elsasser)
631 * partially fixed bug #108: ROOM no longer suffers from occasional
632 (AVER (SAP= CURRENT END)) failures.
633 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
634 bodies no longer cause a WARNING to be signalled. (reported by
636 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
637 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
638 the bug in TRUNCATE.) (reported by Michael Weber)
639 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
640 no samples. (reported by Andy Hefner)
641 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
642 no longer appear as (NIL ...) frames in backtraces.
643 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
644 is now actually exported from SB-EXT as documented. (reported by
646 * bug fix: sb-aclrepl now correctly understands how to inspect
647 single-floats on 64-bit platforms where single-floats are not boxed.
648 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
649 yet been finalized. (reported by Levente Meszaros)
650 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
651 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
652 SBCL-specific optimize qualities.
654 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
655 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
656 an executable in the search path, and does so in the child
657 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
658 been removed; it can be found in the CVS history, for somebody who
659 needs that search behavior (see the manual).
660 * minor incompatible change: compiler policy re. weakening type
661 checks has changed: now type checks are weakened only if SAFETY < 2
663 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
664 unparsing of directory pathnames as files. Analogously,
665 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
666 filename to parse into a directory pathname.
667 * enhancement: implicit generic function creation now signals a
668 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
669 users can bind handlers for and muffle around calls to LOAD.
670 * enhancement: RUN-PROGRAM allows unicode arguments and environments
671 to be used (using the default stream external format), and allows
672 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
673 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
674 strings and vectors whose element-type or simplicity is not fully
675 known at compile-time.
676 * optimization: STRING-TRIM and related functions no longer allocate
677 a new string when no trimming needs to be performed. These functions
678 are also faster than before when the input string has been declared
680 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
681 * optimization: READ-LINE is significantly faster for files containing
683 * optimization: non-open coded uses of character comparison operators
684 (e.g. char=) no longer cons when called with more than one parameter
685 on platforms supporting dynamic-extent allocation.
686 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
687 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
688 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
690 * bug fix: some sequence functions elided bounds checking when
692 * bug fix: too liberal weakening of union-type checks when SPEED >
694 * bug fix: more bogus fixnum declarations in ROOM implementation
697 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
698 * new feature: MAKE-HASH-TABLE now experimentally accepts a
699 :SYNCHRONIZED argument, which makes the hash-table safe for
700 concurrent accesses (but not iteration.) See also:
701 SB-EXT:WITH-LOCKED-HASH-TABLE, and
702 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
703 * optimization: CONCATENATE on strings is an order of magnitude faster
704 in code compiled with (> SPEED SPACE).
705 * optimization: SUBSEQ is ~50% faster on lists.
706 * bug fix: bug 417 fixed -- source location reporting is now more robust.
707 * bug fix: SUBSEQ on a list will now correctly signal an error if if
708 END is smaller then START.
709 * bug fix: SB-PROFILE will no longer report extra consing for nested
710 calls to profiled functions.
711 * bug fix: ROOM implementation had bogus fixnum declarations which
712 could cause type-errors when calling ROOM on large images.
713 * bug fix: if file compilation is aborted, the partial fasl is now
714 deleted, and COMPILE-FILE returns NIL as the primary value.
715 * bug fix: number of thread safety issues relating to SBCL's internal
716 hash-table usage have been fixed.
717 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
718 be returned from its body when the values were being returned
719 using unknown-values return convection and the W-P-O was wrapped
720 inside an UNWIND-PROTECT.
721 * bug fix: sb-posix should now compile again under Windows, enabling
724 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
725 * incompatible change: hash-table accessor functions are no longer
726 automatically protected by locks. Concurrent accesses on the same hash-table
727 from multiple threads can give inconsistent results or even corrupt the
728 hash-table completely. Multi-threaded applications should do their own
729 locking at the correct granularity. In the current implementation it is
730 still safe to have multiple readers access the same table, but it's not
731 guaranteed that this property will be maintained in future releases.
732 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
733 and will signal an error at runtime.
734 * enhancement: SB-THREAD package now exports a semaphore interface.
735 * enhancement: CONS can now stack-allocate on x86 and
736 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
738 * enhancement: nested lists can now be stack allocated on
739 platforms providing stack allocation support.
740 * enhancement: dynamic-extent support has been extended to support
741 cases where there are multiple possible sources for the stack
743 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
744 if the mutex is uncontested on Linux.
745 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
746 as specified by the CLHS. (thanks to Tobias Rittweiler)
747 * bug fix: NaN comparison now works on x86-64.
748 * bug fix: CLOSE :ABORT T on a stream with pending output now
750 * bug fix: instances of non-standard metaclasses using standard
751 instance structure protocol sometimes missed the slot type checks
753 * bug fix: known functions can be inlined in high-debug code.
755 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
756 * minor incompatible change: the MSI installer on Windows no longer
757 associates .lisp and .fasl files with the installed SBCL.
758 * minor incompatible change: :UNIX is no longer present in *FEATURES*
759 on Windows. (thanks to Luis Oliviera)
760 * new feature: SB-CLTL2 contrib module now implements
761 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
762 * optimization: scavenging weak pointers is now more efficient,
763 requiring O(1) instead of O(N) per weak pointer to identify
764 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
765 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
766 method bodies using constant slot-names when the first argument is
767 a specializer parameter for the method.
768 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
769 names, when the class of the instance is a direct instance
770 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
771 as slow as the constant slot-name case.)
772 * optimization: member type construction is now O(N) instead
774 * optimization: UNION and NUNION are now O(N+M) for large
775 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
777 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
778 now more readable in environments like Slime which display it.
779 (thanks to Tobias C. Rittweiler)
780 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
781 was able to infer the correct argument type for the object on which
782 the CAS operation was being performed.
783 * bug fix: copy propagation interfered with parallel assignment
784 semantics in local calls. (reported by Paul Khuong)
785 * bug fix: the signed modular fixnum shift compiled to wrong code on
786 x86 and x86-64. (spotted by a slight modification to some of
789 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
790 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
791 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
792 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
794 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
795 and generic function now print the object identity for anonymous
796 instances. (thanks to Attila Lendvai)
797 * enhancement: as an extension to MOP, SBCL now supports SETF of
798 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
799 (thanks to Attila Lendvai)
800 * enhancement: sb-cover has an alternate annotation mode which puts
801 the form annotations at the car of the form, instead of annotating
802 it completely. The benefit of this mode is that it shows better
803 which forms were instrumented by the compiler.
804 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
805 by printer control variables. (reported by Dan Corkill)
806 * bug fix: system leaked memory when delayed output was performed by
807 the OS in smaller chunks then expected. (thanks to David Smith)
808 * bug fix: system leaked memory when file streams were not closed
810 * bug fix: large objects written to slow streams that were modified
811 after the write could end up with the modified state written to
812 the underlying file descriptor.
813 * bug fix: multiple threads operating in parallel on the same stream
814 could cause buffer-overflows.
815 * bug fix: source location information is stored correctly
816 (broken since 1.0.6). This bug would generally show up as the
817 Slime debugger highlighting the wrong form when the sldb-show-source
819 * bug fix: Branch forms are again annotated as branches in the sb-cover
821 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
822 on x86-64. (reported by Gregory Vanuxem)
823 * bug fix: EQUALP could return wrong results for structures with raw
824 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
825 word). (reported by Vjacheslav Fyodorov)
826 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
829 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
830 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
831 atomic compare-and-swap operations on threaded platforms.
832 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
833 allows assigning a global minimum value to optimization qualities
834 (overriding proclamations and declarations).
835 * enhancement: closed over variables can be stack-allocated on x86
837 * performance bug fix: GETHASH and (SETF GETHASH) are once again
839 * optimization: slot definition lookup is now O(1). This speeds up
840 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
841 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
842 * optimization: ASSOC and MEMBER can now be open-coded for all
843 combinations of keyword arguments when second argument is constant
844 and SPEED >= SPACE. In other cases a specialized version is
846 * bug fix: using obsoleted structure instances with TYPEP and
847 generic functions now signals a sensible error.
848 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
849 (reported by Kristoffer Kvello)
850 * bug fix: backtrace construction is now more careful when making
851 lisp-objects from pointers on the stack, to avoid creating bogus
852 objects that can be seen by the GC.
853 * bug fix: defaulting of values in contexts expecting more than 7
854 variables now works on x86-64. (reported by Christopher Laux)
855 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
857 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
858 as the property-list of a symbol.
859 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
860 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
861 situations requested, are once again file-compileable. (reported
864 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
865 * MOP improvement: support for user-defined subclasses of
866 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
867 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
868 the expansion of DEFMETHOD, is responsible for generating a form
869 which creates a list of specializers when evaluated. Additional
870 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
871 debugging and introspective support.
872 * minor incompatible change: the (unsupported) spinlock interface
873 has changed: free spinlock now has the value NIL, and a held spinlock
874 has the owning thread as its value.
875 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
876 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
878 * enhancement: name of a socket-stream is now "a socket" instead of
880 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
881 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
882 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
883 for generic functions. (Thanks to Tobias C. Rittweiler)
884 * optimization: bignum printing speed has been improved by 20-40%
885 (depending on the bignum size.)
886 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
888 * bug fix: the cache used by the CLOS to store precomputed effective
889 methods, slot offsets, and constant return values is now thread and
891 * bug fix: generic function dispatch function updating is now thread
892 and interrupt safe (in the sense that the known issues have been
894 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
895 * bug fix: interning EQL-specializers is now thread and interrupt safe.
896 * bug fix: asdf systems with dependencies to the SB-POSIX or
897 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
898 * bug fix: interrupt safety of applicable method computation has been
901 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
902 * new contrib: sb-cover, an experimental code coverage tool, is included
904 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
905 significantly faster.
906 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
907 produce more efficient code on x86-64 if the compiler can determine
908 that (+ INDEX <constant>) does not require a bounds check and FOO
909 has an element type at least 8 bits wide.
910 * enhancement: a new, experimental synchronous timeout facility is
911 provided. Refer to SB-SYS:WITH-DEADLINE for details.
912 * enhancement: when a symbol name conflict error arises, the
913 conflicting symbols are always printed with a package prefix.
914 (thanks to Kevin Reid)
915 * enhancement: stepping is now once again supported on the SPARC and
916 MIPS platforms. (It is also now more likely to work on CheneyGC
918 * enhancement: sb-sprof can now also track and report accurate call
920 * bug fixes: the treatment of non-standard subclasses of
921 SB-MOP:SPECIALIZER is more correct.
922 * incompatible change: PURIFY no longer copies the data from the
923 dynamic space into the static and read-only spaces on platforms
924 that use the generational garbage collector
925 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
927 * bug fix: GC race condition occasionally resulting in crashes with
928 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
930 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
932 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
933 a STYLE-WARNING to be signalled (regression from 1.0.4.)
934 * bug fix: an asynchronous interrupt could previously leave the
935 system running with GC inhibited.
936 * bug fix: a DECLARE form evaluated at top-level now causes an error
937 rather than silently (or verbosely) returning NIL.
938 * bug fix: trying to dissassemble functions compiled at high DEBUG could
939 result in a error being signalled due to source form lookup errors.
940 (reported by Peter Graves)
942 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
943 * incompatible change: removed writer methods for host-ent-name,
944 host-ent-addresses -- changing the values did not update the DNS
946 * minor incompatible change: changed experimental JOIN-THREAD interface
947 * documentation: the manual now lists reader and writer methods
948 in class slot documentation sections. (thanks to Richard M Kreuter)
949 * documentation: unwinding from asyncronous events has been
950 documented as unsafe.
951 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
952 in multithreaded application code.
953 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
954 platforms. (thanks to James Anderson for the optimization hint)
955 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
957 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
958 variants no longer cons.
959 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
960 their NOT- variants no longer cons.
961 * optimization: EQUAL hash tables no longer use SXHASH for objects
962 of all data types, but instead use an EQL hash for types for which
963 EQUAL is the same as EQL.
964 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
965 are significantly faster.
966 * optimization: new STRING-OUTPUT-STREAM implementation is much
967 faster for large outputs and conses 30% less on average.
968 * enhancement: XREF information is now collected for references made
969 to global variables using SYMBOL-VALUE and a constant argument.
970 * enhancement: SIGINT now causes a specific condition
971 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
972 * bug fix: FILE-POSITION used to signal an error in some cases where
973 ANSI requires it to return NIL.
974 * bug fix: ADJUST-ARRAY is now interrupt-safe.
975 * bug fix: adding and removing fd-handlers is now interrupt-safe.
976 * bug fix: inlined calls to C now ensure 16byte stack alignment on
978 * bug fix: bad type declaration in the CLOS implementation has
979 been fixed. (reported by James Anderson)
980 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
981 (thanks to Lutz Euler)
982 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
983 required by the CLHS. (reported by Samium Gromoff)
984 * bug fix: dead unbound variable references now signal an error.
985 (reported by Marco Monteiro)
986 * bug fix: / with an unused value was being deleted in safe code.
987 (thanks to Marco Monteiro and Kevin Reid)
988 * bug fix: number of characters that can be written onto a single
989 line in a file is unlimited.
990 * bug fix: some GC deadlocks caused by asynchronous interrupts have
991 been fixed by inhibiting interrupts for when GC is disabled.
992 * bug fix: some interrupt safety issues with GETHASH, (SETF
993 GETHASH), CLRHASH and REMHASH have been fixed.
994 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
995 specifier no longer causes infinite recursion.
996 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
997 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
998 * bug fix: an &environment argument with the correct variable information
999 is passed to macros that are expanded during byte compilation.
1000 (reported by Samium Gromoff)
1001 * bug fix: a logic error was causing the hash function for bignums to
1002 have excessive amounts of collisions. (reported by Faré Rideau)
1003 * bug fix: modifying the contents of an array could change the return
1004 value of SXHASH on that array, which is only allowed for strings
1005 and bit vectors. (bug introduced in 0.9.16)
1006 * improvement: the x86-64/darwin port now passes all tests (except
1007 for the debugger tests) but should still be considered
1008 experimental until this is fixed.
1009 * improvement: a style-warning is signaled for CASE (etc) clauses with
1010 duplicate keys. (thanks to Kevin Reid)
1011 * improvement: macos/x86 and macos/x86-64 now use mach exception
1012 handlers for dealing with illegal instructions (for trapping and
1013 error handling) and memory protection violations (for GC).
1015 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
1016 * new platform: experimental support for x86-64/darwin (MacOS).
1017 * incompatible change: the thread-safe (on most platforms) getaddrinfo
1018 and getnameinfo sockets functions are used instead of gethostbyaddr
1019 and gethostbyname, on platforms where the newer functions are available.
1020 As a result, the ALIASES field of HOST-ENT will always be NIL on these
1022 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
1023 don't print the startup banner, but behave as if --noinform was passed
1024 as a command line argument. (thanks to Kevin Reid)
1025 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1026 * optimization: code using alien values with undeclared types is much faster.
1027 * optimization: the compiler is now able to open code SEARCH in more cases.
1028 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1029 * bug fix: using standardized COMMON-LISP special variables as loop
1030 variables no longer signals bogus package lock violations. (reported
1032 * bug fix: declaring local loop variables to be of a range-limited type
1033 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1034 (reported by Andras Simon)
1035 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1036 bugs remain on x86-64.)
1037 * bug fix: the #= and ## reader macros now interact reasonably with
1038 funcallable instances.
1039 * bug fix: type-checks for function arguments were compiled using the
1040 compiler policy settings of the wrong lexical environment (in 1.0.2
1042 * bug fix: SHADOW accepts characters as string designators, as required
1043 by the spec (thanks to Eric Marsden)
1044 * bug fix: fixed GC safety issues when foreign functions are called with
1045 non-base strings as arguments
1046 * bug fix: more consistent error output for fatal-compiler-warnings, like
1048 * bug fix: functions with proper names were showing up as NIL in
1049 backtraces (reported by Edi Weitz, regression in 1.0.2)
1051 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1052 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1053 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1054 produce more efficient code on the x86 if the compiler can determine
1055 that (+ INDEX <constant>) does not require a bounds check and FOO
1056 has an element type at least 8 bits wide.
1057 * bug fix: references to undefined variables are handled the same way
1058 in toplevel forms as in the normal compiler
1059 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1061 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1062 (thanks to Jon Buller)
1063 * improvement: faster compilation times for complex functions
1064 * improvement: added readlink support to SB-POSIX (thanks to Richard
1067 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1068 * improvement: experimental support for mach exception handling on
1069 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1071 * improvement: support for GBK external format.
1072 (thanks to Chun Tian (binghe))
1073 * improvement: the debugger now displays variables that have been closed
1074 over, in code compiled with (DEBUG 2) or higher
1075 * improvement: support for executable cores on NetBSD (thanks to
1077 * new feature: added a RESTART-FRAME debugger command
1078 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1079 be used to provide an implementation for FILE-POSITION on Gray streams
1080 (thanks to Eric Marsden)
1081 * new feature: add syslog, openlog and closelog support to SB-POSIX
1082 (thanks to Richard Kreuter)
1083 * optimization: the function call overhead in code compiled with
1084 a high DEBUG optimization setting is significantly
1085 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1086 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1087 (thanks to Tony Martinez)
1088 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1089 evaluated once (reported by Kevin Reid)
1090 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1091 works again (thanks to Kilian Sprotte)
1092 * bug fix: an error is signaled for tagbodies with duplicate tags
1093 (thanks to Stephen Wilson)
1094 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1095 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1096 stack frames from alien callbacks.
1097 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1098 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1099 65k pages (thanks to David Woodhouse)
1100 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1102 changes in sbcl-1.0.1 relative to sbcl-1.0:
1103 * new platform: FreeBSD/x86-64, including support for threading.
1104 * new feature: the compiler stores cross-referencing information
1105 about function calls (who-calls), macroexpansion (who-macroexpands)
1106 and special variables (who-binds, who-sets, who-references) for code
1107 compiled with (< SPACE 3). This information is available through the
1108 sb-introspect contrib.
1109 * new feature: users may subclass SEQUENCE, and have instances of
1110 these classes interoperate with standard Common Lisp functions if
1111 a number of methods are defined. (This feature is experimental
1112 and the interface subject to change based on feedback from SBCL
1113 users and the general community)
1114 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1115 x86/x86-64, rather than the previous fixed depth of 8
1116 * improvement: another pthread back-end of mutex "pthread-futex".
1117 * bug fix: non-ascii command-line arguments are processed correctly
1118 (thanks to Yaroslav Kavenchuk)
1119 * bug fix: non-required arguments were not passed correctly when a method
1120 defined using DEFMETHOD was called from a mop-generated method using
1121 CALL-NEXT-METHOD (reported by Pascal Costanza)
1122 * bug fix: recursion is now permitted in accessors through
1123 SLOT-UNBOUND. (reported by Pascal Costanza)
1124 * bug fix: an error was signaled at startup if the HOME environment
1125 variable was defined, but had an empty value (reported by Peter Van Eynde)
1126 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1127 signal a PROGRAM-ERROR, not a BUG.
1128 * bug fix: SB-POSIX:READDIR works when built with large file support.
1129 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1131 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1132 for better automatic indentation support. (thanks to Matt Pillsbury)
1133 * optimization: loading generic functions no longer takes O(n^2) time,
1134 proportional to the amount of methods in the generic function
1135 (reported by Todd Sabin and Jeremy Brown)
1136 * optimization: the FIND and POSITION family of sequence functions
1137 are significantly faster on arrays whose element types have been
1139 * improvements to the Windows port:
1140 ** Intermittent heap corruption problems have been fixed. (thanks
1141 to Alastair Bridgewater)
1142 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1144 ** Lisp is able to unwind foreign exception frames from alien
1145 callbacks. (thanks to Alastair Bridgewater)
1147 changes in sbcl-1.0 relative to sbcl-0.9.18:
1148 * improvement: experimental support for threading on FreeBSD/x86.
1149 (thanks to NIIMI Satoshi)
1150 * improvement: runtime option --dynamic-space-size can be used
1151 to set the size of the dynamic space reserved on startup.
1152 * improvement: floating point modes in effect are now saved in
1153 core, and restored on startup.
1154 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1155 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1156 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1157 threads are not running after *SAVE-HOOKS* have run.
1158 * improvement: writes to CLOS instance slots are type-checked in code
1159 compiled with (SAFETY 3)
1160 * improvement: floating-point exception handling on FreeBSD (thanks to
1162 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1163 (thanks to Zach Beane)
1164 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1166 * improvement: added support for the Shift-JIS external format.
1167 (contributed by NIIMI Satoshi)
1168 * improvement: callbacks are supported on Linux/PPC. (thanks to
1170 * bug fix: compiler bug triggered by a (non-standard) VALUES
1171 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1172 * bug fix: file compiler no longer confuses validated and already
1173 dumped structurres. (reported by Kaersten Poeck)
1174 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1175 fill-pointer signals a type-error as required. (thanks to
1177 * bug fix: disassemly of funcallable instances works.
1178 * bug fix: single stepping on PPC.
1179 * bug fix: fix thread-safety problems in the type system (generally
1180 manifesting as nonsensical errors like "STRING is a bad type specifier
1181 for sequences" or "The value 1 is not of type FIXNUM")
1182 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1183 (reported by Josip Gracin)
1184 * bug fix: an error is signaled for attempts to displace arrays with
1185 incompatible element types (thanks to Mario Mommer)
1186 * bug fix: more correct handling of wide characters in debug info
1187 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1188 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1189 and don't cause extra consing
1190 * optimization: MAP and MAP-INTO are significantly faster on vectors
1191 whose elements types have been declared.
1192 * Improvements to SB-SPROF:
1193 ** Support for allocation profiling
1194 ** Reduced profiling overhead, especially for long profiling runs
1195 * Improvements to the Windows port:
1196 ** floating point exceptions are now reported correctly.
1197 ** stack exhaustion detection works partially.
1198 ** more accurate GET-INTERNAL-REAL-TIME.
1199 ** better RUN-PROGRAM behaviour in case of errors in the child
1201 ** PROBE-FILE now simplifies pathnames correctly.
1202 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1203 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1204 1st 1970 (although time timezone data will be unreliable).
1205 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1207 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1208 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1209 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1211 * bug fix: two potential GC deadlocks affecting threaded builds.
1212 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1213 returns T (reported by Anton Kazennikov)
1214 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1215 with non-variable places
1216 * bug fix: remove a race condition in the setting of
1217 funcallable-instance functions, this should make threaded CLOS
1218 code more stable against memory faults.
1219 * bug fix: corruption of specials when unbinding is interrupted by an
1220 asynchronous unwind (reported by Hannu Koivisto)
1221 * improvement: the debugger will now also display local variables that
1222 are only used once, for code compiled with a DEBUG optimization quality
1225 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1226 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1227 * incompatible change: External-format support for FFI calls. The
1228 SB-ALIEN:C-STRING no longer implies an ASCII
1229 external-format. Instead the string is subject to external-format
1230 conversion. Additionally return values of type SB-ALIEN:C-STRING
1231 are by default of type (SIMPLE-ARRAY CHARACTER), not
1232 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1233 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1234 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1235 * incompatible change: SB-EXT package no longer contains the
1236 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1237 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1239 * incompatible change: the single-stepper is no longer available
1240 on Alpha, Mips and Sparc platforms.
1241 * minor incompatible change: the direct superclasses of
1242 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1243 not (STANDARD-OBJECT FUNCTION). This makes the
1244 class-precedence-lists of GENERIC-FUNCTION and
1245 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1247 * minor incompatible change: the default stream external format on
1248 non-unicode SBCL is no longer detected from the locale, but is always
1250 * new feature: Add a version of evaluator that uses an interpreter instead
1251 of the compiler. EVAL still uses the compiler by default, to switch it
1252 to use the interpreter, set the value of the variable
1253 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1254 * minor incompatible change: the single-stepper REPL has been merged
1255 with the normal debugger (see the "Stepping" heading of the debugger help
1256 for more details). The debugger command STEP will no longer switch
1257 to the single-stepper REPL.
1258 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1259 for a type now works.
1260 * bug fix: loading of default sysinit file works. (thanks to Leonid
1262 * bug fix: better detection of circularities in the file-compiler.
1263 (reported by Marco Monteiro)
1264 * bug fix: the CL pathname functions now work with files that have
1265 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1266 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1267 systems with Unix98 pty semantics.
1268 * bug fix: ASDF-INSTALL will now work with bsd tar.
1269 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1271 * bug fix: timers expiring in dead threads no longer cause a
1272 type-error (reported by Paul "Nonny Mouse").
1273 * bug fix: thanks to more lightweight single-stepper instrumentation,
1274 code compiled with (DEBUG 3) will compile and execute significantly faster,
1275 and will have more accurate type-inferencing than before
1276 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1277 whose bindings are modified
1278 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1279 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1280 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1281 CL:SOFTWARE-VERSION with proper version
1283 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1284 * feature: implemented the READER-METHOD-CLASS and
1285 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1286 as specified by AMOP.
1287 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1289 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1290 (thanks to Lutz Euler)
1291 * optimization: hashing of general arrays and vectors has been
1292 improved. (reported by Any Fingerhut)
1293 * enhancement: SB-INTROSPECT is now able to find definitions of
1294 profiled functions. (thanks to Troels Henriksen)
1295 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1296 as well, allowing compiler-macros for SETF-functions to expand.
1297 * enhancement: step-instrumentation no longer wraps known
1298 single-value functions with multiple-value context, allowing
1299 better type inference.
1300 * fixed bug #337: use of MAKE-METHOD in method combination now works
1301 even in the presence of user-defined method classes. (reported by
1302 Bruno Haible and Pascal Costanza)
1303 * fixed bug #339(c): if there are applicable methods not part of any
1304 long-form method-combination group, call INVALID-METHOD-ERROR.
1305 (reported by Bruno Haible)
1306 * fixed bug #361: the :FUNCTION initarg in the protocol for
1307 initialization of methods can now be used to override
1308 internally-produced optimized functions. (reported by Bruno
1310 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1311 system-provided lambda expression no longer cause warnings about
1312 unbound #:|pv-table| symbols.
1313 * bug fix: improved the handling of type declarations and the
1314 detection of violations for keyword arguments with non-constant
1316 * bug fix: potentially erronous calls to PATHNAME and
1317 MERGE-PATHNAMES were being flushed in some cases.
1318 (reported by Richard Kreuter)
1319 * bug fix: compiled calls to TYPEP were mishandling obsolete
1320 instances. (reported by James Bielman and Attila Lendvai)
1321 * bug fix: format strings with ~> without matching ~< no longer
1322 trigger an AVER, but signal an understandable error instead.
1323 (reported by Antonio Martinez)
1324 * bug fix: specifying an output-file without a pathname-type for
1325 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1326 of input-file instead of "fasl". (reported by Robert Dodier)
1327 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1328 (reported by James Y Knight).
1329 * bug fix: compiler-macros-function did not consider the environment
1330 argument for shadowing by local functions.
1331 * bug fix: compiler-macros expansion was inhibited by local INLINE
1333 * bug fix: inline expansions of known functions were subject to
1334 step-instrumentation in high debug policies, leading to problems
1335 with type-inference.
1336 * bug fix: compiler failed to differentiate between different CONS
1337 types in some cases.
1338 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1339 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1340 * bug fix: #368: incorrect use of expressed vs. upgraded array
1342 * bug fix: #306a: more precise unions of array types.
1343 * thread-safety improvements:
1344 ** CONDITION-WAIT could return early on Linux, if the thread was
1345 interrupted and subsequently continued with SIGCONT.
1346 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1348 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1349 * added support for the ucs-2 external format. (contributed by Ivan
1351 * minor incompatible change: pretty printing of objects of type
1352 (cons symbol) is, in the default pprint-dispatch-table, now
1353 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1355 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1356 called later in a class's lifetime, possibly as late as when the
1357 first instance of the class is created. Previously,
1358 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1359 class became finalizeable.
1360 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1361 for multibyte external-format streams. (thanks to "vbzoli")
1362 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1363 with :ALLOCATION :CLASS inherited from superclasses of the
1365 * fixed bug: anonymous classes can now be created using the :NAME
1366 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1367 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1368 * fixed bug: core-files saved with :EXECUTABLE T can again be
1369 executed when SBCL_HOME isn't set. (reported by James Knight)
1370 * fixed bug: toplevel LOCALLY forms with declarations could
1371 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1372 * fixed bug: printing from several different threads using different
1373 values of *print-case* could cause invalid output, due to
1374 some internal special variables of the printer not being bound
1375 thread-locally (reported by Max Mikhanosha)
1376 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1377 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1378 (reported by Pascal Costanza)
1379 * fixed bug: FINALIZE-INHERITANCE is called from
1380 REINITIALIZE-INSTANCE on classes when the class has previously
1381 been finalized, as required by AMOP.
1382 * minor code generation optimizations:
1383 ** better register allocation in CLOS dispatching functions
1384 ** overflow detection when coercing signed bytes to fixnums on x86-64
1385 ** is now implemented with one IMUL instruction instead of three shifts
1386 ** more efficient bit-vector access on x86 and x86-64
1387 ** more efficient access to raw structure slots on x86 and x86-64
1388 * fixed some bugs revealed by Paul Dietz' test suite:
1389 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1390 return its argument.
1392 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1393 * feature: thread support on Solaris/x86, and experimental thread support
1395 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1397 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1398 default method for SLOT-UNBOUND.
1399 * minor incompatible change: prevent the user from specializing the
1400 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1401 somewhat counter to the intent of the protocol, I (CSR) think, and
1402 additionally it just doesn't work in SBCL as currently
1403 implemented, thanks to optimizations (that are always valid for
1404 the other three SLOT-VALUEish functions, but not for the setter).
1405 * minor incompatible change: the :SB-LDB feature is now enabled by
1406 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1407 the low-level debugger.
1408 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1409 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1410 down to an SB-SYS:FD-STREAM, allowing direct communication in
1411 more cases. Temporary files and pipes are still used as a fallback
1413 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1414 on the non-reentrant C functions localtime(3) and gmtime(3).
1415 * bug fix: global optimization policy was not visible in LOCALLY and
1417 * bug fix: class objects can be used as specializers in methods.
1418 (reported by Pascal Costanza)
1419 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1420 with a type error. (reported by blitz_ on #lisp)
1421 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1422 failed with a type error. (reported by Pascal Bourguignon)
1423 * bug fix: merging pathnames against defaults with :DIRECTORY
1424 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1425 (reported by James Y Knight)
1426 * bug fix: saving large (>2GB) cores on x86-64 now works
1427 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1428 constant index and a (simple-array (signed-byte 32)) array
1429 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1430 error (patch by Robert J. Macomber)
1431 * feature: TIME now displays the amount of run-time spent in GC
1432 * bug fix: The debugger now does a better job of respecting
1433 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1434 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1435 workaround for bug 403.)
1436 * bug fix: floating point exception handling now works on FreeBSD
1437 (thanks to NIIMI Satoshi)
1438 * fixed some bugs revealed by Paul Dietz' test suite:
1439 ** MISC.641: LET-conversion were not supposed to work in late
1441 * improvements to the Win32/x86 port:
1442 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1443 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1445 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1446 * new feature: source path information is generated for macro-expansion
1447 errors for use in IDE's like Slime (thanks to Helmut Eller)
1448 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1449 * bug fix: compiler does not loop forever on an invalid type in
1451 * improvement: compilation of most CLOS applications is significantly
1453 * optimization: added a limited bytecode compiler for simple toplevel
1454 forms, speeding up compilation and FASL loading
1455 * bug fix: the statistical profiler now properly distinguishes anonymous
1458 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1459 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1460 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1461 into subclasses of the finalized class. Now user calls to
1462 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1463 system return before any subclasses are finalized.
1464 * minor incompatible change: The reader no longer ignores errors
1465 regarding non-existent packages in #+ and #- feature tests.
1466 * new feature: command line options --no-sysinit, --no-userinit to
1467 inhibit loading the corresponding init files
1468 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1469 for sending data through UDP sockets (thanks to François-René Rideau)
1470 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1471 error is signalled instead (thanks to François-René Rideau)
1472 * minor incompatible change: Error signalling behaviour of lexical
1473 operations violating package locks has changed slightly. Refer to
1474 documentation on package locks for details.
1475 * enhancement: EVAL can process IF-expressions without resorting to the
1477 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1478 constant-folded properly due to missing out-of-line functions.
1479 (reported by Mika Pihlajamäki)
1480 * bug fix: LISTEN sometimes returned T even in cases where no data was
1481 immediately available from the stream
1482 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1483 were reversed. (reported by Levente Mészáros)
1484 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1485 when given a vector displaced to an adjustable array. (reported
1487 * bug fix: derivation of float boundaries from numbers outside the
1488 appropriate float range (reported by John Wiseman)
1489 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1491 * bug fix: precomputing cache entries for generic functions with
1492 some subclasses of specializers as yet invalid does not attempt to
1493 fill a cache line with a negative offset. (reported by Levente
1495 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1496 allow condition class objects as arguments to DOCUMENTATION and
1497 (SETF DOCUMENTATION); only find and set documentation for
1498 structure names for the STRUCTURE doc-type. (suggested by Gary
1500 * improvements to the Win32/x86 port:
1501 ** (user-homedir-pathname) and default initialization file
1502 locations now know about the user's "Documents and Settings"
1503 directory (thanks to Yaroslav Kavenchuk)
1504 ** run-program is implemented (thanks to Mike Thomas)
1505 ** sockets support (thanks to Timothy Ritchey)
1506 ** better backtrace support (thanks to Alastair Bridgewater)
1507 ** sb-grovel supported
1508 ** asdf-install and sb-posix work somewhat
1509 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1510 * improvements to the Solaris/x86 port:
1511 ** works on Solaris 11/Solaris Express
1512 ** floating-point exception handling support
1513 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1514 * fixed some bugs revealed by Paul Dietz' test suite:
1515 ** REMOVE-METHOD returns its generic function argument even when
1516 the method is not one of the generic functions' methods.
1517 ** objects declared as MEMBER types can be admissible arguments to
1518 structure accessors.
1519 ** printing characters should simply be printed by the FORMAT ~:C
1521 ** compiler failure when compiling functions with hairy constant
1522 defaults for optional parameters.
1523 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1524 function, which is already optimized.
1526 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1527 * new platform: experimental support for SBCL x86/Darwin, including
1528 MacOS X 10.4.5 on Intel.
1529 * new feature: Unicode character names are now known to the system
1530 (through CHAR-NAME and NAME-CHAR).
1531 * new feature: the filesystem location of SBCL's core file is
1532 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1533 * minor incompatible change: the contrib modules SB-POSIX and
1534 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1535 this change is to make it easier to distribute
1536 location-independent binaries.
1537 * bug fix: as implied by AMOP, standardized classes no longer have
1538 slots named by external symbols of public packages. (reported by
1540 * optimization: calling functions via constant symbols -- as in
1541 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1542 via the function object as in (FUNCALL #'FOO).
1543 * optimization: on x86 and x86-64, the calling convention for the
1544 case of dealing with an unknown number of values has been altered
1545 to be friendlier to the prediction heuristics implemented,
1546 particularly with reference to CALL and RET pairing. (thanks to
1547 Alastair Bridgewater)
1548 * optimization: on x86, the code for access to thread-local symbol
1549 values and binding/unbinding of thread-local symbols is smaller.
1550 (thanks to Alastair Bridgewater)
1551 * enhancement: CONSTANTP is now able to determine constantness of
1552 more complex forms, including calls to constant-foldable standardized
1553 functions and some special forms beyond QUOTE.
1554 * fixed bug: occasional GC crashes on Solaris/x86
1555 * optimization: x86-64 supports stack allocation of results of simple
1556 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1557 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1558 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1559 (thanks to James Knight)
1560 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1561 by the standard (thanks to Stephen Compall)
1563 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1564 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1565 be used for bundling the runtime and the core file into one
1566 executable binary. This feature is not currently supported on all SBCL
1567 platforms. (thanks to James Bielman and NIIMI Satoshi)
1568 * new feature: a generational or ephemeral garbage collector is now
1569 the default on the PowerPC platform (both Linux and Darwin). The
1570 old Cheney (stop and copy) collector is a build-time option.
1571 (thanks to Cyrus Harmon, after Raymond Toy)
1572 * minor incompatible change: the method by which SBCL finds its
1573 contributed modules has changed; it no longer relies on symbolic
1574 links from an $SBCL_HOME/systems directory, but searches directly
1575 in the subdirectories of $SBCL_HOME.
1576 * enhancement: the dynamic heap size on the Linux/PPC platform is
1577 markedly larger, even using the older Cheney garbage collector.
1578 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1579 certain complicated string types. (reported by Gary King)
1580 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1581 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1582 * fixed bug: DOCUMENTATION on structures no longer signals an
1583 error if no structure type of the right name exists. (reported by
1585 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1586 applicable non-standard primary or :AROUND method on
1587 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1588 no longer cause unbound variable errors. (reported by Kevin Reid)
1589 * fixed bug: metacircle resolution in cases where methods have slots
1590 added before the slots from STANDARD-METHOD. (reported by Jean
1592 * fixed bug: the Power PC assembly code for calling into lisp
1593 sometimes computed its return address wrongly (depending on the
1594 15th bit of the address, fixed at link time). This used to
1595 manifest itself in a segmentation violation while building PCL.
1596 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1597 many others over the years)
1598 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1599 the stack pointer (treated as a signed integer) is negative.
1600 (thanks to Peter van Eynde)
1602 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1603 * new platform: experimental support for the Windows operating
1604 system has been added. (thanks to Alastair Bridgewater)
1605 * fixed several bugs in and robustified the PPC FFI (including
1606 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1607 * bug fix: allow non-simple string symbol names (reported by Paul
1609 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1611 * bug fix: saving a core corrupted callbacks on x86/x86-64
1612 * bug fix: closed a loophole in metacircularity detection and
1613 grounding in the PCL implementation of CLOS.
1614 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1615 (reported by tomppa on #lisp)
1616 * bug fix: clear the direction flag on Lisp -> C transitions, as
1617 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1618 (reported by Andrej Grozin and Hendrik Maryns)
1619 * optimization: major improvements to GC efficiency on GENCGC platforms
1620 * optimization: faster implementation of EQUAL
1621 * optimization: emit more efficient opcodes for some common
1622 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1623 * optimization: several other minor code-generation improvements on x86-64
1624 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1626 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1627 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1628 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1629 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1630 by ANSI as a generic function, and for consistency (SETF
1631 GENERIC-FUNCTION-NAME) is treated likewise.
1632 * fixed bug #233.b: make constraint propagation notice when a variable
1633 value is changed after it is referenced but before it is used
1634 * fixed bug #296: no more arbitrary behaviour from filenames with
1635 odd characters as --load arguments. (reported by Adam Warner)
1636 * fixed bug #390: :CHARACTER-SET pathname components now work as
1637 expected. (reported by Tim Daly Jr)
1638 * fixed bug #391: complicated :TYPE intersections in slot
1639 definitions no longer cause an error in PCL internals.
1640 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1641 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1642 returning the number of octets which would be written to the
1643 file-stream. (thanks to Robert J. Macomber)
1644 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1645 arguments to the internal stream functions. (thanks to David
1647 * bug fix: stack exhaustion is now handled on the x86/SunOS
1648 platform. (thanks to Daisuke Homma)
1649 * bug fix: more accurate ROOM results on GENCGC platforms
1650 * optimization: improved type inference for arithmetic-for
1651 index variables in LOOP
1652 * optimization: faster floating-point SQRT on x86-64
1653 * fixed some bugs revealed by Paul Dietz' test suite:
1654 ** DOCUMENTATION returns NIL instead of "" for method combinations
1655 that don't have a docstring
1657 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1658 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1659 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1660 therefore are not customizeable by user code (as seems to be at
1661 least permitted and maybe required by AMOP). As a consolation,
1662 however, the SBCL implementation of these functions now calls
1663 REINITIALIZE-INSTANCE as specified by AMOP.
1664 * bug fix: slot-definition documentation is propagated as per ANSI
1665 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1666 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1667 argument, as do eslotd initialization methods. (from Pascal
1668 Costanza's "Closer" project)
1669 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1670 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1672 * bug fix: it is now possible to have more than one subclass of
1673 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1674 (reported by Bruno Haible, Pascal Costanza and others)
1675 * bug fix: the dependent update protocol now works for generic
1676 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1677 and Pascal Costanza)
1678 * bug fix: condition-class instances corresponding to
1679 DEFINE-CONDITION forms are now created eagerly. (reported by
1680 Kalle Olavi Niemitalo on comp.lang.lisp)
1681 * bug fix: floating point printing is more accurate in some
1682 circumstances. (thanks to Simon Alexander)
1683 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1684 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1685 * bug fix: callbacks on OS X now preserve stack-alignment.
1686 (thanks to Cyrus Harmon)
1687 * enhancement: the x86-64 disassembler is much better at
1688 disassembling SSE instructions. (thanks to Lutz Euler)
1689 * enhancement: CHECK-TYPE on a variable now allows type-inference
1690 (thanks to Vincent Arkesteijn)
1691 * optimization: improved performance of EUC-JP external format.
1692 (thanks to NIIMI Satoshi)
1693 * optimization: performance improvements to IO on file streams of
1694 :ELEMENT-TYPE CHARACTER
1695 * optimization: much faster memory allocation on x86-64
1696 * optimization: faster garbage collections (latency and throughput)
1698 * optimization: faster bignum subtraction and division on x86 and x86-64
1699 (thanks to Lutz Euler)
1700 * optimization: more accurate type-inference for loops that use a
1701 floating point index variable or a negative step.
1703 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1704 * bug fix: add a workaround to SBCL looping infinitely at startup on
1705 Linux kernels with apparently buggy implementations of personality().
1706 (thanks to Svein Ove Aas)
1707 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1708 (thanks to David Lichteblau)
1709 * bug fix: local bindings shadow global symbol macros for
1710 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1711 * optimization: non-open coded uses of numeric comparison operators
1712 (e.g. >) no longer cons when called with more than one parameter
1713 on platforms supporting dynamic-extent allocation.
1714 * enhancement: saving cores with foreign code loaded is now supported
1715 on MIPS/Linux in addition to the previously supported platforms.
1716 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1717 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1718 * bug fix: division by zero in sb-sprof when no samples were collected
1719 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1720 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1721 (thanks to Alexander Kjeldaas)
1722 * new feature: ignoring the return values of destructive functions
1723 like NREVERSE signals a compile-time style-warning.
1724 (thanks to Kevin Reid)
1725 * new feature: the alignment of alien structure fields can be
1726 explicitly specified. (thanks to Cyrus Harmon)
1727 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1728 and dump core on SIGQUIT
1730 ** incompatible change: threads do not inherit values of specials
1731 from their parents (see manual)
1732 ** bug fix: threads stacks belonging to dead threads are freed by the
1733 next exiting thread, no need to gc to collect thread stacks anymore
1734 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1735 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1736 ** bug fix: dynamic variable and thread start related gc lossage
1737 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1739 * fixed some bugs revealed by Paul Dietz' test suite:
1740 ** SUBTYPEP is slightly more accurate on heinously complicated
1741 CONS types where some of the members have uncertain (in the
1742 NIL, NIL sense) type relationships to each other.
1744 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1745 no longer purified unless :PURIFY T is explicitly specified.
1746 ** Non-purified cores are significantly smaller than before
1748 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1749 * new feature: timers based on Zach Beane's excellent timer package
1750 * added support for the following external formats: koi8-u,
1751 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1752 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1753 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1754 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1755 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1756 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1757 * incompatible change: a threaded SBCL will no longer revert to
1758 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1759 * bug fix: interrupts are disabled until startup is complete; no
1760 more sigsegvs when receiving a signal to soon
1761 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1763 * bug fix: add a workaround for the memory randomization features in
1764 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1765 workaround will only be in effect on systems with the proc filesystem
1767 * bug fix: printing objects of type HASH-TABLE signals a
1768 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1770 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1771 have been running for over 50 days. (reported by Gilbert Baumann)
1772 * bug fix: the logic for getting names of functions gets less
1773 confused when confronded with alternate-metaclass
1774 funcallable-instances. (reported by Cyrus Harmon)
1775 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1776 consistent, even on internal alternate-metaclass objects.
1777 * bug fix: flush closure information collected by physical
1778 environment analysis prepass before the main pass. (bug reported
1780 * bug fix: compiler pack phase does not modify a hashtable, which is
1781 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1782 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1783 may be expanded. (reported by Alexander <malishev> on c.l.l)
1784 * bug fix: no more style-warnings when compiling code using LAST
1785 under high-SPEED low-SPACE settings. (reported by David Wallin
1787 * fixed some incompatibilities between SBCL's MOP and the MOP
1789 ** the METAOBJECT class is now implemented;
1790 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1791 STANDARD-OBJECT, as required;
1792 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1793 now compatible (as required by VALIDATE-SUPERCLASS); there
1794 remains a consistency requirement over the presence or absence
1795 of the FUNCTION class in the superclass list of finalized
1796 classes; see the manual for more details;
1797 ** the :AROUND method for COMPUTE-SLOTS on
1798 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1799 requested slot ordering.
1801 ** bug fix: parent thread now can be gc'ed even with a live
1803 ** bug fix: binding a special with PROGV to no value is not
1804 the same as that symbol not having been bound (thanks to
1806 ** bug fix: binding specials is thread safe (thanks to
1808 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1809 and SB-SPROF do not die with 'no handler for signal XX in
1810 interrupt_handle_now(..)' anymore
1811 ** bug fix: WITH-TIMEOUT works with multiple running threads
1812 * fixed some bugs revealed by Paul Dietz' test suite:
1813 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1814 the :method-class keyword argument.
1816 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1817 * new port: the Solaris operating system on x86 processors is now
1818 mostly supported, though some rough edges in the environment
1819 remain. (thanks to Daisuke Homma)
1820 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1821 space for its heap. (thanks to Thiemo Seufer)
1822 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1823 closures. (thanks to Thiemo Seufer)
1824 * minor incompatible change: eof selects abort in the debugger.
1825 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1826 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1828 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1829 deprecated, and will go away in a future revision of SBCL.
1830 * minor incompatible change: GC-ON and GC-OFF are no longer
1831 implemented with a counter, it does not matter how many times gc
1832 is switched on or off
1833 * bug fix: discriminating functions for generic function classes
1834 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1835 make invalid assumptions about method precedence order. (reported
1837 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1838 platforms. (thanks to Thiemo Seufer)
1839 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1840 * bug fix: correct pathname printing with printer escaping is on.
1841 (thanks to Kevin Reid)
1842 * bug fix: complex VOP definitions in "user-space" no longer trigger
1843 package locks. (reported by Zach Beane)
1844 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1845 for funcallable-instance functions, and is no different from
1847 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1848 characters in the current readtable. (reported by Nicholas Neuss)
1849 * bug fix: SUBTYPEP on various CONS types returns more a more
1850 accurate acknowledgment of its certainty.
1851 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1852 lists in some cases. This partially fixes bug 384.
1853 * flush all standard streams before prompting in the REPL and the
1855 * bug fix: signal handling and triggering gc do not conflict
1856 directly anymore, in particular a high frequency sb-sprof does
1857 not prevent gc from running
1858 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1859 approximation for timezone and DST information between the
1860 universal time epoch and the smallest negative 32-bit time_t.
1861 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1862 year 1899 so long as the encoded time is non-negative.
1863 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1864 the correct answer for non-constant rotate amounts when performing
1865 an inline 32-bit rotation.
1867 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1868 there is only one thread in the session
1869 ** bug fix: memory leak for streams created in one thread and
1870 written to in another
1871 ** bug fix: lockup when compiled with gcc4
1872 ** bug fix: race that allows the gc to be triggered when gc is
1874 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1875 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1877 ** bug fix: finished threads are gc'ed properly
1878 * fixed some bugs revealed by Paul Dietz' test suite:
1879 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1880 has arguments with a different set of applicable methods from
1881 the orignal arguments.
1882 ** The type error thrown by MAP now has a correct expected-type
1884 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1885 name a compiled function.
1886 ** Three MISC tests where a large bignum was improperly coerced to
1887 a float that couldn't represent that bignum during type
1888 derivation were fixed.
1889 ** SUBTYPEP can now handle types involving the negation of a
1890 list-form FUNCTION type.
1891 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1892 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1893 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1895 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1896 * New feature: Experimental support for bivalent streams: streams
1897 opened with :element-type :default now allow character and binary
1898 (unsigned-byte 8) I/O
1899 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1900 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1902 * bug fix: no more highly sporadic "couldn't check whether ~S is
1903 readable" when reading a stream and an interrupt hits in the middle
1904 of a select system call
1905 * compiler better recognizes complex arrays (reported by Eduardo
1907 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1908 for variables on SBCL built with linkage-tables. (reported by Luis
1910 * various error reporting improvements.
1911 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1912 (Thanks to Thiemo Seufer)
1913 * bug fix: nested reader invokations work correctly
1914 * bug fix: it is possible to have simultaneous references to foreign
1915 code and foreign data with the same name.
1917 ** added x86-64 support
1918 ** incompatible change: the threading api now works with thread
1919 objects instead of thread ids
1920 ** bug fix: threads are protected from signals and interruption when
1921 starting up or going down
1922 ** bug fix: a race where an exiting thread could lose its stack to gc
1923 ** bug fix: don't halt on infinite error in threads if possible
1924 ** fixed numerous gc deadlocks introduced in the pthread merge
1925 ** bug fix: fixed thread safety issues in read and print
1926 ** bug fix: debugger doesn't hang on session lock if interrupted at
1927 an inappropriate moment
1928 ** bug fix: run-program is now thread safe(r)
1929 ** bug fix: inner with-recursive-lock no longer releases the mutex
1930 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1931 * fixed some bugs revealed by Paul Dietz' test suite:
1932 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1933 EXPECTED-TYPE slots filled.
1934 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1936 ** MAKE-CONDITION accepts classes as type-designators.
1937 ** COMPILE may never return NIL.
1938 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1939 range before calling Unix time functions
1941 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1942 * numerous signal handling fixes to increase stability
1943 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1944 * minor incompatible change: we now correctly canonize default
1945 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1946 as per the MOP, rather than the historical (INITARG INITFUNCTION
1947 INITFORM). (reported by Bruno Haible)
1948 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1949 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1950 for more information.
1951 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1952 pathname is a directory pathname.
1953 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1954 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1956 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1957 :METACLASS argument in addition to a class name. (reported by
1958 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1959 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1960 (reported by Bruno Haible; patch from Gerd Moellmann)
1961 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1963 * bug fix: more cleanups to the floating point exception handling on
1964 x86-64 (thanks to James Knight)
1965 * bug fix: the compiler does not try to constant fold calls of
1966 COERCE and BIT-* functions when they return freshly constructed
1967 objects. (reported by Nikodemus Siivola and Paul Dietz)
1968 * optimization: improved the MIPS versions of generic (in the
1969 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1970 * optimization: direct conversion of (unsigned-byte 32) to floats on
1971 the PowerPC platform.
1972 * optimization: structure instances with raw slots now use less
1973 memory, and probably show better memory locality. (thanks to
1975 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1976 are treated as requests for stack allocation on the x86-64,
1977 Alpha-32, PPC and SPARC platforms.
1978 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1979 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1980 (thanks to Tony Martinez)
1982 ** gcing a dead thread can no longer lead to lockups
1983 ** threads block signals until they are set up properly
1984 ** errno is no longer shared by threads
1985 ** interrupt-thread restores the eflags register on x86
1986 ** fixed some lockups due to gc/thread interaction
1987 * fixed some bugs revealed by Paul Dietz' test suite:
1988 ** invalid dotted lists no longer raise a read error when
1989 *READ-SUPPRESS* is T
1990 ** signal an error if a symbol that names a declaration is used
1991 as the name of a type, or vice versa
1992 ** allow using the (declare (typespec var*)) abbreviation for
1993 (declare (type typespec var*)) with all type specifiers
1994 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1995 ** always return NIL from PPRINT-POP when OBJECT is NIL
1996 ** don't signal errors when pretty-printing malformed LABELS,
1997 FLET or MACROLET forms
1998 ** declarations in a DOLIST body are also in scope for the
2000 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
2002 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
2005 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
2006 * fixed cross-compiler leakages that prevented building a 32-bit
2007 target with a 64-bit host compiler.
2008 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
2009 opened with :IF-EXISTS :OVERWRITE.
2010 * fixed bug 281: error for an invalid qualifier in a short-form method
2011 combination method is not signalled until the faulty method is called.
2012 * bug fix: iteration variable type inferrer failed to deal with open
2013 intervals. (reported by Alan Shields)
2014 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
2015 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
2016 or more runtime options were provided to the sbcl binary.
2017 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
2018 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
2019 * optimization: there's now a fast-path for fixnum arguments in the
2020 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
2021 * optimization: the code generated on x86-64 is more compact thanks
2022 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
2023 * bug fix: floating-point exception handling now partly works on
2024 x86-64. (Thanks to James Knight)
2025 * improvement to the MIPS backend from Thiemo Seufer: C-style
2026 64-bit long long arguments and return values to alien functions
2028 * fixed some bugs revealed by Paul Dietz' test suite:
2029 ** the type-error signalled from WARN has a filled-in DATUM slot.
2030 ** the type-error required when a stream is not associated with
2031 a file has the stream as its datum.
2032 ** type-errors on single-floats on x86-64 no longer have
2033 :INVALID-OBJECT as the datum
2034 ** the type-errors signalled for invalid function names now have
2035 a correct expected type
2036 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2037 for structures defined with a :TYPE. Documentation strings for
2038 typed structures are no longer immediately discarded
2039 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2040 broadcast streams and synonym streams. FILE-LENGTH now also works
2041 on broadcast streams.
2043 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2044 * incompatible change: the --noprogrammer option, deprecated since
2045 version 0.7.5, has been removed. Please use the equivalent
2046 --disable-debugger option instead.
2047 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2049 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2050 inoperational for a while now) has been completely removed.
2051 * null lexical environments are now printed as #<NULL-LEXENV>,
2052 significantly reducing the amount of clutter in typical backtraces.
2053 * documentation on weak pointers, finalization, and after GC hooks
2054 has been added to the manual.
2055 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2056 as other specialized array types, is much faster. SUBSEQ and
2057 COPY-SEQ on such arrays have also been sped up.
2058 * optimization: EQL is now more efficient when at least other argument
2059 is known to be of type (OR FIXNUM (NOT NUMBER)).
2060 * fixed bug: the runtime is now less vulnerable to changes in the
2061 size of the SBCL object on OS X, and virtual memory is reserved for
2062 all spaces that need to be at a fixed address.
2063 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2064 * fixed bug: finalizers and after GC hooks that cause consing are now
2066 * fixed bug: compiler error messages and summaries are now printed to
2067 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2068 * fixed inference of the upper bound of an iteration variable.
2069 (reported by Rajat Datta).
2070 * fixed bug 211e: calling local functions with duplicated constant
2071 keyword argument no longer causes a bogus style warning about an
2073 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2074 ftype declaration to be disregarded. (reported by Dave Roberts)
2075 * fixed bug 373: caused by erronous compilation of references to alien
2076 variables in the runtime on ppc/darwin.
2077 * fixed bug 376: CONJUGATE type deriver.
2078 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2079 (reported by Baughn on #lisp)
2080 * a cross-compiler bug on non-x86 platforms has been identified and
2081 fixed. (thanks to Bruno Haible)
2082 * improvements to the MIPS runtime code for increased stability.
2083 (thanks to Thiemo Seufer)
2084 * increased the maximimum compact environment size to allow
2085 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2086 * partial workaround for bug 135: don't name the function we're
2087 calling for hairy cases of EVAL, so as not to accumulate one environment
2088 entry per call to EVAL. (reported by Kevin Reid)
2089 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2090 * optimization: type testing for non-vector arrays should be faster.
2091 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2093 * bug fix: setting 31st element of a bit vector to zero did not work
2095 * bug fix: redefining a class definition which failed due to a
2096 previous accessor / function clash now works (but see BUGS entry
2097 #380 for more problems in this area). (thanks to Zach Beane)
2098 * the long form of DEFINE-METHOD-COMBINATION disables method group
2099 checking when given a single method group with pattern *. (thanks
2101 * on x86 compiler supports stack allocation of results of simple
2102 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2103 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2104 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2105 * fixed some bugs related to Unicode integration:
2106 ** the restarts for recovering from input and output encoding
2107 errors only appear when there is in fact such an error to
2109 * fixed some bugs revealed by Paul Dietz' test suite:
2110 ** MISC.549 and similar: late transformation of unsafe type
2111 assertions into derived types caused unexpected code
2113 ** SCALE-FLOAT type deriver is less wrong.
2114 ** type derivers for EXP, LOG and similar functions compute result
2115 types for complex arguments better.
2116 ** (MISC.563) CONJUGATE type deriver works for very restricted
2118 ** out-of-line type testers for character strings are available.
2119 ** EQUAL compiler transform understands specialness of objects
2121 ** accessing double-floats stored on the stack now works on x86-64.
2122 ** debugger internals could sometimes create invalid lispobjs,
2123 resulting in GC crashes.
2124 ** MISC.548: type check weakening can convert required type into
2126 ** initialization forms for bindings are not in scope of free special
2129 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2130 * incompatible change: thread support for non-NPTL systems has
2131 been removed - locking is buggy and unreliable. A threaded
2132 SBCL build will now warn at startup and refuse to create
2133 new threads, unless futex support is detected in the kernel
2134 * incompatible change: the top level REPL now has only an ABORT
2135 restart associated with it, not TOPLEVEL and ABORT as it used to.
2136 TOP and TOPLEVEL are now available as debugger commands for
2137 returning to the top level.
2138 * incompatible change: forms evaluated in the REPL now use the
2139 global optimization policy.
2140 * incompatible change: user- and system-initialization files are
2141 no longer processed with LOAD, but by READ and EVAL; hence the
2142 global optimization policy, startup package, readtable, etc,
2144 * The .fasl file format number has been incremented because of
2145 various incompatible changes.
2146 * internal entry point details and argument counts no longer appear
2147 in backtraces unless explicitly requested by setting
2148 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2149 * built-in and standard functions no longer have names like "top
2150 level local call to FOO".
2151 * fixed bug 32: functions defined in non-null lexical environments
2152 now have more legible printed representation
2153 * fixed bug 33: functions defined in non-null lexical environemnts
2154 are now more amenable to inspection by INSPECT.
2155 * workaround for bug 354: XEPs no longer appear in backtraces unless
2156 explicitly requested.
2157 * fixed bug: receiving the signal which results from attempting to
2158 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2159 SIGBUS on other *BSDs) on architectures where the C stack is also
2160 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2161 to debugger confusion or wild instability. (reported by Cheuksan
2163 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2164 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2165 (reported by Lutz Euler)
2166 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2167 compiled code. (reported by Paul Dietz)
2168 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2169 specialized parameter to a method to something that is not TYPEP
2170 the specializer is now possible.
2171 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2172 face of package deletion.
2173 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2174 pathname. (reported by David Tolpin on c.l.l)
2175 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2176 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2177 STRING-STREAM and FILE-STREAM, respectively.
2178 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2179 than previously; in addition, its test suite is now run on build.
2180 (thanks to Luke Gorrie)
2181 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2182 * optimization: added a immediate representation for single-floats
2184 * fixed some bugs revealed by Paul Dietz' test suite:
2185 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2186 ** attempting to create a package with a colliding nickname causes
2187 correctable errors to be signalled.
2188 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2189 ** several x86-64 backend bugs related to sign-extension of immediate
2192 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2193 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2194 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2196 * fixed loading of multiply forward-referenced layouts.
2197 (thanks to Cheuksan Wang)
2198 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2199 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2200 (see COMPILE-FILE documentation for details of :PRINT option)
2201 * fixed bugs 19 and 317: fixed-format floating point printing is
2202 more accurate. This also fixes a bug reported by Adam Warner
2203 related to the ~@F format directive.
2204 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2206 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2207 dispatch table if the from-char is a dispatch macro character.
2208 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2209 by the wordsize no longer ignore the last word. (reported by Lutz
2211 * fixed bug in type checking of dynamic-extent variables. (reported
2213 * optimization: sequence traversal functions use their freedom to
2214 coerce function designators to functions.
2215 * optimization: code with many calls to CLOS methods specialized on
2216 CLOS classes has had redundant type checks removed.
2217 * optimization: added declarations to speed up operations that access
2218 the internal character database (for example STRING-UPCASE)
2219 * optimization: comparison operations between floats and sufficiently small
2220 fixnums no longer create extra rationals
2221 * fixed some bugs related to Unicode integration:
2222 ** portions of multibyte characters at the end of buffers for
2223 character-based file input are correctly transferred to the
2224 start of the buffer at the next read.
2225 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2226 passing it through to OPEN.
2227 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2228 argument given to internal calls to OPEN.
2229 ** the built-in comment readers (introduced by character sequences
2230 ";" and "#|") are more forgiving to encoding errors; they will
2231 STYLE-WARN and then attempt to resync the stream at a character
2232 boundary. (thanks to Teemu Kalvas)
2233 * fixed some bugs in the x86-64 port:
2234 ** Negative short int return values from c-calls are sign-extended
2236 ** The stack is aligned to 16-bytes for c-calls, as required by
2237 the ABI. (thanks to Cheuksan Wang)
2238 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2239 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2240 for enums. (thanks to Vincent Arkesteijn)
2241 ** Multiple small optimizations and bugfixes for floating point
2243 * fixed some bugs revealed by Paul Dietz' test suite:
2244 ** Space, Tab, Linefeed, Return and Page have the invalid
2245 secondary constituent character trait.
2246 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2248 ** WITH-INPUT-FROM-STRING should only update the index place on
2250 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2251 works more reliably.
2252 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2253 with one-dimensional arrays with fill pointers.
2254 ** TYPE-OF failed on a complex with an integer realpart and a
2256 ** compiler failure during type inference for the code of form
2257 (IF (EQL X (THE ...)) ...) (MISC.535).
2259 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2260 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2261 platform. The port supports 61-bit fixnums, large memory spaces
2262 and reloading shared object files.
2263 * enhancement: saving cores with foreign code loaded is now
2264 supported on ppc/Darwin in addition to the previously supported
2266 * enhancement: the statistical profiler now walks deeper into the
2267 call stack for more meaningful call-graphs and accrued time
2268 reports (x86/x86-64 only). It also now reports time spent in
2270 * enhancement: it is now possible to trace most individual methods
2271 of a generic function in addition to tracing the generic function
2273 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2275 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2276 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2277 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2278 produces an error. (thanks to Vincent Arkesteijn)
2279 * bug fix: NAMESTRING on pathnames with :WILD components in their
2280 directories works correctly. (thanks to Artem V. Andreev)
2281 * fixed bug 125: compiler preserves identity of closures. (reported
2283 * bug fixed: functions with &REST arguments sometimes failed with
2284 "Undefined function" when compiled with (DEBUG 3). (reported by
2286 * bug fix: overflow during compiling of setting element of a bit
2287 vector with constant index and value. (reported by Timmy Douglas)
2288 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2289 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2290 * on x86 compiler partially supports stack allocation of dynamic-extent
2292 * GO and RETURN-FROM do not check the extent of their exit points
2293 when compiled with SAFETY 0.
2294 * fixed some bugs related to Unicode integration:
2295 ** encoding and decoding errors are now much more robustly
2296 handled; it should now be possible to recover even from invalid
2297 input or output to the terminal. (thanks to Teemu Kalvas)
2298 ** provided a first cut at implementing STRING-TO-OCTETS and
2299 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2300 ** altered the SB-MD5 contributed module slightly, changing the
2301 interface just enough for it to be supportable for builds where
2302 lisp characters are not eight bits.
2303 * fixed some bugs revealed by Paul Dietz' test suite:
2304 ** the FORMATTER-generated functions for ~V[ conditionals require
2305 the correct number of arguments.
2306 ** READ-FROM-STRING returns the mandated second value when applied
2307 to displaced strings.
2308 ** the #\Rubout and #\Backspace characters are treated as invalid
2309 constituent characters by the tokenizer.
2311 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2312 * new feature: reloading changed shared object files with
2313 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2314 * new feature: references to foreign variables and functions
2315 can now be compiled and loaded before the corresponding shared
2316 object file is loaded, as long as the foreign definitions are
2317 available at runtime.
2318 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2319 supported. (thanks to Dan Debertin)
2320 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2321 just keywords. (thanks to Vincent Arkesteijn)
2322 * fixed bug #331: structure-class instances corresponding to
2323 DEFSTRUCT forms are now created eagerly.
2324 * fixed bug #345: backtraces from calls to undefined functions work
2325 on x86 as well. Related bug #61 is now also partially fixed on x86
2326 (backtraces from throws to unknown catch tags.)
2327 * bug fix: lambda-list parsing is now stricter vrt. order and number
2328 of lambda-list keywords.
2329 * bug fix: as specified by AMOP, an error is signalled if a
2330 class-option appears multiple times in a DEFCLASS form. (reported
2332 * bug fix: RANDOM can be compiled when the compiler derives the type
2333 of its numeric argument as a disjoint set of small integers.
2334 (reported by Paul Dietz)
2335 * bug fix: starting a core saved with shared objects loaded when
2336 those objects are not available no longer causes threaded SBCL to
2337 hang. (reported by Sean Ross)
2338 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2340 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2341 argument, which must always be NIL. (reported by Kalle Niemitalo)
2342 * bug fix: printing 1.0d+23 no longer results in an error.
2343 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2344 * bug fix: structure slot setters preserve evaluation order. (thanks
2346 * fixed some bugs related to Unicode integration:
2347 ** RUN-PROGRAM can allow its child to take input from a Lisp
2348 stream. (reported by Stefan Scholl)
2349 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2350 locales. (reported by Ken Causey)
2351 * fixed some bugs revealed by Paul Dietz' test suite:
2352 ** INCF, DECF and REMF evaluate their place form as specified in
2354 ** FORMATTER expands ~{ ~} iteration directives with V or #
2355 parameters correctly.
2356 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2357 consequent uses no arguments correctly.
2358 ** the system has a partial understanding of the (COMPLEX RATIO)
2360 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2363 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2364 * new feature: a build-time option (controlled by the :SB-UNICODE
2365 keyword feature, enabled by default) for building the system with
2366 support for the entire 21-bit character space defined by the
2368 * new feature: the system now has rudimentary external-format
2369 support; the primary user-visible change at this time is that
2370 characters with the high bit set (such as lower-case-e-acute) will
2371 print correctly to a terminal in a UTF-8 environment.
2372 * minor incompatible change: BASE-CHAR no longer names a class;
2373 however, CHARACTER continues to do so, as required by ANSI.
2374 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2375 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2376 has been moved to the SB-EXT package (temporarily re-exported from
2378 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2379 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2381 * fixed bug #351: better error-handlind and reporting for malformed
2383 * fixed bug #350: bignum-printing is now more memory-efficient,
2384 allowing printing of very large bignums, eg. (expt 2 10000000).
2385 (reported by Bruno Haible)
2386 * fixed bug #302: better primitive-type selection for intersection
2388 * fixed bug #308: non-graphic characters now all have names, as
2389 required. (reported by Bruno Haible)
2390 * bug fix: redefining a class with different superclasses now correctly
2391 removes it from the direct-subclasses of its previous superclasses.
2392 (reported by David Morse)
2393 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2394 new value now works. (reported by Bruno Haible)
2395 * bug fix: correct canonicalization of multiple non-standard slot
2396 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2397 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2398 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2399 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2400 now exists, an signals an error.
2401 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2402 during macroexpansion for non-iteration variables as well. (reported
2403 by Bruno Haible for CMUCL)
2404 * bug fix: Cyclic structures and unprintable objects in compiler
2405 messages no longer cause errors. (reported by Bruno Haible)
2406 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2407 and READ-FROM-STRING all now return a primary value of NIL if
2408 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2409 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2410 (reported by Bruno Haible for CMUCL)
2411 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2412 do not designate a condition. (reported by Bruno Haible for
2414 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2415 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2416 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2417 permitting resolution in favour of any of the conflicting symbols.
2418 (reported by Bruno Haible for CMUCL)
2419 * bug fix: EQUAL compiler optimizations is less aggressive on
2420 strings which can potentially compare true despite having distinct
2421 specialized array element types.
2422 * bug fix: unit enumerations can be defined without dividing by
2423 zero. (thanks to Vincent Arkesteijn)
2424 * FORMAT compile-time argument count checking has been enhanced.
2425 (report from Bruno Haible for CMUCL)
2426 * a partial workaround for the bug 262: the compiler does not try to
2427 inline-expand a local function doing RETURN-FROM from a deleted
2428 BLOCK. (thanks to Peter Denno for the bug report and to David
2429 Wragg for the simple test case)
2430 * fixed some bugs revealed by Paul Dietz' test suite:
2431 ** RENAME-PACKAGE allows all package designators as new package
2433 ** constraint propagation and lambda variable substitution are
2434 more cautious in dealing with partially deleted code.
2435 ** compiler sometimes lost reoptimization passes.
2436 ** CERROR, given a condition as condition designator, passes its
2437 remaining arguments to the continue format control without
2439 ** Case-altering FORMAT directives work correctly on non-ASCII
2441 ** The REAL type specifier handles bounds outside the floating
2442 point ranges without signalling FLOATING-POINT-OVERFLOW.
2443 ** Functions with IR1-transformations can create intercomponent
2444 references to global functions.
2445 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2447 ** FORMAT ~R treats a nil value for its first parameter correctly.
2449 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2450 * enhancement: saving cores with foreign code loaded is now
2451 supported on x86/NetBSD and sparc/Linux in addition to the previously
2452 supported platforms.
2453 * bug fix: on some platforms repeated installations caused multiple
2454 copies of HTML documentation to be installed -- should not happen
2455 any more. (reported by Stefan Scholl)
2456 * bug fix: parsing self-recursive alien record types multiple times
2457 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2458 original patch by Helmut Eller for CMUCL)
2459 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2460 (thanks to Richard Kreuter)
2461 * bug fix: defining classes whose accessors are methods on existing
2462 generic functions in other (locked) packages no longer signals
2463 bogus package lock violations. (reported by François-René Rideau)
2464 * bug fix: special variables as DEFMETHOD parameters no longer have
2465 associated bogus type declarations. (reported by David Wragg and
2467 * bug fix: read-write consistency on streams of element-type
2468 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2469 * bug fix: redefiniton of the only method of a generic function with
2470 no DEFGENERIC no longer emits a full WARNING. In addition,
2471 redefinition of generic functions with no DEFGENERIC to an
2472 incompatible lambda list now signals an error. (thanks to Zach
2474 * bug fix: DEFGENERIC now works even when there's a function of the
2475 same name in an enclosing lexical environment. (thanks to Zach
2477 * fixed compiler failure, caused by instrumenting code during
2478 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2479 * optimization: added loop analysis and improved register allocation
2480 * fixed some bugs revealed by Paul Dietz' test suite:
2481 ** POSITION on displaced vectors with non-zero displacement
2482 returns the right answer.
2483 ** (SIMPLE-STRING) is a valid type specifier for sequence
2485 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2487 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2488 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2490 ** The justification version of the FORMAT ~< directive treats
2491 non-zero minpad parameter correctly.
2493 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2494 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2495 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2496 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2497 the supported interface.
2498 * new feature: Single-stepping of code compiled with DEBUG 2 or
2499 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2500 * new feature: saving cores with foreign code loaded is now
2501 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2502 Timothy Moore's work for CMUCL)
2503 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2504 parameters to * instead of NIL if no initform is supplied.
2505 (reported by Johan Bockgaard)
2506 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2507 correctly when FUNCALL appears as the car of the form. Note:
2508 despite this FUNCALL forms are not currently subject to
2509 compiler-macro expansion. (port of Raymond Toy's fix for the
2510 same from CMUCL, reported by Johan Bockgaard)
2511 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2512 (thanks for Teemu Kalvas)
2513 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2514 case instead of terminating the iteration (thanks for Julian
2515 Squires, Sean Champ and Raymond Toy)
2516 * bug fix: incorrect expansion of defgeneric that caused a style
2517 warning. (thanks for Zach Beane)
2518 * bug fix: slot accessor effective method computation works properly
2519 for classes with multiple non-standard applicable methods on
2520 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2521 * on x86 compiler supports stack allocation of results of LIST and
2522 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2523 CMUCL implementation by Gerd Moellmann)
2524 * fixed some bugs revealed by Paul Dietz' test suite:
2525 ** FORMAT strings with both the ~<~:;~> form of the justification
2526 directive and pretty-printing directives cause an error.
2528 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2529 * incompatible change: the internal functions
2530 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2531 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2532 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2533 instead of the old functions.
2534 * new feature: on platforms where "dladdr" is available foreign
2535 function names now appear in backtraces. (based on Helmut Eller's
2537 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2538 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2540 * documentation: ASDF manual is now distributed with SBCL.
2541 * bug fix: SBCL can now load its contributed modules with REQUIRE
2542 even if the system-provided entries have been removed from
2543 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2545 * bug fix: backtraces involving undefined functions or assembly
2546 routines are more informative. (thanks to Brian Downing)
2547 * bug fix: mutually referent alien structures now work correctly.
2548 (reported by Rick Taube)
2549 * bug fix: structures defined by WITH-ALIEN can be referred to
2550 within other definitions in the same WITH-ALIEN.
2551 * bug fix: division operators (MOD, TRUNCATE and the like) with
2552 constant zero divisors and integer dividends no longer generate
2554 * bug fix: provide default methods for INPUT-STREAM-P and
2555 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2556 * bug fix: improve the detection and resolution of MOP metacycles
2557 (where effective method computation depends on the generic function
2558 having its effective method computed). (reported by Bruno Haible)
2559 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2560 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2561 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2562 * bug fix: pathnames with relative directory components can be
2563 represented relative to default pathnames.
2564 * optimization: in taking the GCD of bignums, reduce the two bignums
2565 to approximately the same size (using Euclid's algorithm) before
2566 applying the more sophisticated binary GCD. (thanks to Juho
2568 * optimization: COUNT on bitvectors now operates word-at-a-time.
2569 * optimization: ASH with a positive, but not necessarily constant,
2570 (leftwards) shift, when in a modular context, is compiled to a
2572 * fixed some bugs revealed by Paul Dietz' test suite:
2573 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2574 if the corresponding argument is NIL.
2575 ** FORMAT directives accepting a minpad parameter treat negative
2577 ** literal commas as character parameters to FORMAT directives are
2579 ** literal spaces directly after ~<Newline> directives within a
2580 format-logical-block (~:< ~@:>) do not induce :FILL-style
2581 conditional newlines.
2582 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2583 as specified: it no longer includes conditional newlines).
2584 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2586 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2587 :PER-LINE-PREFIX argument does not evaluate to a string.
2589 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2590 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2591 the manual for details; for now, package locks can be disabled by
2592 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2593 if no major problems are found then it is likely that they will be
2594 compiled in unconditionally.
2595 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2596 unsupported operators on all platforms. To load a shared library
2597 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2598 object file, link it into a shared library outside of SBCL and
2599 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2601 * bug workaround: ROOM T and the GC were not working together
2602 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2603 reporting facility aren't preserved. That reporting has been
2604 disabled, so now until and unless someone figures out how to make it
2605 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2606 * minor incompatible change: as threatened around sbcl-0.8.0, the
2607 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2608 The symbols are also exported from SB-PCL for backwards
2609 compatibility, but more so than before SB-PCL should be treated as
2610 an implementation-internal package.
2611 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2613 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2614 It is known as of this release not to work on the Alpha, however.
2615 * fixed bug #167: errors signalled due to illegal syntax in method
2616 bodies are now more legible.
2617 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2618 designators and can hence be used with TYPEP.
2619 * fixed bug #333: CHECK-TYPE now ensures that the type error
2620 signalled, if any, has the right object to be accessed by
2621 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2622 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2623 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2625 * fixed bug #334: programmatic addition of slots using specialized
2626 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2627 and :ALLOCATION :CLASS slots.
2628 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2629 float exponents. (rereported by Peter Seibel)
2630 * fixed a bug: #\Space (and other whitespace characters) are no
2631 longer considered to be macro characters in standard syntax by
2632 GET-MACRO-CHARACTER.
2633 * fixed bug: initialization of condition class metaobjects no longer
2634 causes an instance of the condition to be created. (reported by
2636 * fixed bug: it is now possible to have slots such that
2637 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2638 description is neither :INSTANCE nor :CLASS.
2639 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2640 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2641 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2642 non-standard SLOT-DEFINITION-ALLOCATION values.
2643 * fixed bugs: various race conditions handling exiting threads.
2644 CL-PPCRE's thread torture test now passes.
2645 * fixed arguably-a-bug: GC time proportional to number of threads in
2646 system even when most of them are idle
2647 * optimization: improved performance of BIT and SBIT on bit-vectors.
2648 * .fasl file incompatibility: The fasl file version number has
2649 been incremented because of changes associated with package locks.
2651 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2652 * minor incompatible change: the system no longer provides
2653 optimization hints (conditions which are TYPEP
2654 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2655 mode; these hints are still emitted when higher SPEED optimization
2657 * new contrib module: a sampling profiler (profiling by statistical
2658 sampling, rather than by instrumenting functions) is available as
2659 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2660 * the behaviour of the standard function ED is now customizeable by
2661 third parties through a hook variable: see ED's documentation
2662 string for information on the protocol.
2663 * the compiler no longer emits efficiency notes for (FUNCALL X)
2664 when the type of X is uncertain under default optimization
2666 * fixed bug 276: mutating a binding of a specialized parameter to a
2667 method to something that is not TYPEP the specializer is now
2669 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2670 exhibit the required behaviour on the x86 platform. (thanks to
2671 Peter van Eynde, Eric Marsden and Bruno Haible)
2672 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2673 even for difficult arguments. (reported by Peter Graves)
2674 * fixed bug 141a: the backquote printer now descends quoted
2676 * fixed another bug in backquote printing: no more destructive
2677 modification of the form's list structure. (reported by Brian
2679 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2680 move between its address being taken and the call to
2681 interrupt_thread, fixing a crashing race condition.
2682 * the SB-POSIX contrib implementation has been adjusted so that it
2683 no longer exhibits ridiculously poor performance when constructing
2684 instances corresponding to C structs.
2686 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2687 * minor incompatible change: the sb-grovel contrib now treats C
2688 structures as alien (in the sense of SB-ALIEN) objects rather than
2689 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2690 has implications for memory management of client code
2691 (sb-grovel-returned objects must now be manually managed) and for
2692 type safety (alien objects now have full types).
2693 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2694 used to control emission of compiler diagnostics, rather than the
2695 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2696 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2697 quality should be considered deprecated.
2698 * install.sh now installs the user manual as well
2699 * (not quite a new documentable feature, but worth considering in
2700 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2701 beginnings of a semantically meaningful condition hierarchy is
2702 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2704 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2705 designator as the defaults argument.
2706 * fixed bug: Displaced arrays whose displaced-to array has become
2707 too small now cause ARRAY-DIMENSION to signal an error, providing
2708 for safer bounds-checking. (reported by Bruno Haible)
2709 * fixed bug: DEFCLASS slot definitions with identical :READER and
2710 :WRITER names now signal a reasonable error. (reported by Thomas
2712 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2714 * fixed bug: Invalid slot specification errors now print correctly.
2715 (thanks to Zach Beane)
2716 * fixed bug 320: Shared to local slot value transfers in class
2717 redefinitions now happen corrently from superclasses as
2718 well. (reported by Bruno Haible)
2719 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2721 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2722 improper lists correctly. (reported by Bruno Haible)
2723 * fixed bug 313: source-transform for <fun-name> was erroneously
2724 applied to a call of a value of a variable with name <fun-name>.
2725 (reported by Antonio Menezes Leitao)
2726 * fixed bug 307: The obsolete instance protocol ensures that
2727 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2728 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2729 remove dead unknown-values globs from the middle of the stack.
2730 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2731 easier to resume long computations after using *BREAK-ON-SIGNALS*
2732 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2733 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2734 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2735 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2736 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2737 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2739 * fixed a (fairly theoretical) bug in string printing: if
2740 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2741 does not have array-element-type equal to the most general string
2743 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2744 function name. (thanks to Nikodemus Siivola)
2745 * fixed bug: compile-time format string checker failed on
2746 non-closed ~{. (reported by Thomas F Burdick)
2747 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2748 don't warn on legal loop code involving more than one aggregate
2749 boolean. (thanks to Nikodemus Siivola)
2750 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2751 in keyword position to MAKE-PACKAGE was not being reported as
2753 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2754 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2756 * fixed bugs 280 and 312: the checking for multiple definitions in a
2757 file is less likely to become confused by uses of inline
2759 * fixed bug: the #S reader macro performs the keyword coercion
2760 specified for slot names. (reported by Kalle Niemitalo)
2761 * fixed bug: lambda lists may contain symbols whose names start with
2762 & but are not lambda-list-keywords; their occurrence triggers a
2764 * fixed bug 321: define-method-combination argument lambda lists do
2765 not cause invalid code to be generated when &optional or &aux
2766 variables have default values. (reported by Bruno Haible)
2767 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2768 have CLOS classes; this appears also to have fixed reported
2769 instability in stack exhaustion detection.
2770 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2771 MUFFLE-WARNING when no associated restart is present is now
2773 * optimization: rearranged the expansion of various defining macros
2774 so that each expands into only one top-level form in a
2775 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2777 * optimization: used a previously little-used slot in symbols to
2778 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2780 * fixed some bugs revealed by Paul Dietz' test suite:
2781 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2783 ** Fixed an optimization bug related to inheritance of initforms
2784 from local to shared slots.
2785 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2786 ** CAST optimizer forgot to flush argument derived type.
2787 ** print/read consistency on floats is now orders of magnitude
2788 more likely. (thanks also to Bruno Haible for a similar report
2790 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2791 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2792 for objects of type REAL. Make it so.
2793 ** SXHASH is specified (infelicitously) to respect similarity,
2794 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2795 it so. (thanks to Markus Ziegler)
2796 ** on the Alpha, the compiler succeeds in compiling functions
2797 returning a known number of arguments greater than 63.
2798 ** fixed handling of invalid NIL arguments in keyword position
2800 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2802 ** provided a readably-printable representation for RANDOM-STATE
2804 ** ensured that pathnames, where they have a namestring, always
2805 print using #P"..." syntax.
2807 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2808 * Support for the forthcoming 2.0 version of the NetBSD kernel
2809 running on x86 hardware has been added. (thanks to Perry
2810 E. Metzger most immediately, and others for their past work)
2811 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2812 current release version; SBCL's OpenBSD support had been broken
2813 since about the time of OpenBSD's switch to ELF binary format.)
2814 * [placeholder for DX summary]
2815 ** user code with &REST lists declared dynamic-extent, under high
2816 speed or space and low safety and debug optimization policy.
2817 * The manual has been converted to Texinfo format and the debugger
2818 chapter from the cmucl manual has been added.
2819 * A facility has been added to extract documentation strings from
2820 sbcl and store them as Texinfo-formatted snippets for inclusion in
2821 the manual (via Texinfo's @include directive)
2822 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2823 the test case to Dave Roberts)
2824 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2825 pointers. (reported by Sean Ross)
2826 * bug fix: PROFILE output is printed nicely even for large numerical
2827 values. (thanks to Zach Beane)
2828 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2829 greater than 32 handle EOF correctly.
2830 * bug fix: on X86 an immediate argument of the IMUL instruction is
2831 correctly printed in disassembly. (thanks to Lutz Euler)
2832 * bug fix: class slots in redefined classes preserve their old
2833 values. (thanks to Bruno Haible and Nikodemus Siivola)
2834 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2835 succeeds. (reported by Marco Baringer)
2836 * fixed some bugs revealed by Paul Dietz' test suite:
2837 ** READ-SEQUENCE now works on ECHO-STREAMs.
2838 ** RATIONALIZE works more according to its specification. (thanks
2841 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2842 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2843 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2844 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2845 should matter to you only if you rebind the printer control
2846 variables and then find you want different bindings in the
2847 debugger than in the ordinary execution of your program.)
2848 * The runtime build system has been tweaked to support building
2849 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2850 assembler and linker. (thanks to Nikodemus Siivola)
2851 * Unbound, undefined, undeclared variables now trigger full
2852 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2853 more often programmer error than deliberate exploitation of undefined
2855 * optimization: the hash algorithm for strings has changed to one
2856 that is less vulnerable to spurious collisions. (thanks to Juho
2858 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2859 needless bounds checking. (thanks to Juho Snellman)
2860 * optimization: implemented multiplication as a modular
2861 (UNSIGNED-BYTE 32) operation on the PPC backend.
2862 * fixed some bugs revealed by Paul Dietz' test suite:
2863 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2865 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2866 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2867 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2869 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2871 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2873 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2874 their output stream on EOF from read.
2875 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2876 have been read to end-of-file.
2877 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2879 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2880 description of determination of which consecutive characters
2882 ** Printing the "Space" character with escaping on now yields "#\\ ",
2883 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2884 ** Reading floating-point numbers with *READ-BASE* set to a number
2885 less than 10 works correctly.
2886 ** Reading floating-point numbers with *READ-BASE* set to a number
2887 more than 10 works correctly.
2888 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2889 the readtable currently in effect.
2891 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2892 * minor incompatible change: parsing of namestrings on a physical
2893 (Unix) host has changed; numbers after the final #\. in a
2894 namestring are no longer interpreted as a version field. This is
2895 intented to be largely invisible to the user, except that the
2896 meaning of the namestring "*.*.*" has changed: it now refers to a
2897 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2898 should usually be replaced by
2899 (make-pathname :name :wild :type :wild :version :wild)
2900 with the added benefit that this is more likely to be portable.
2901 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2902 option to OPEN now signals an error if the file being opened
2903 exists; this may have an impact on existing code.
2904 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2905 signals received, on the PowerPC platforms (both Linux and
2906 Darwin). (thanks to Pierre Mai for pointing out the location of
2908 * several fixes on OS X: The system now builds and runs cleanly on
2909 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2910 (thanks to Brian Mastenbrook)
2911 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2912 second-resolution: integer multiples of 1/3600 between -24 and 24.
2913 (thanks to Vincent Arkesteijn)
2914 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2915 type when called with 1 argument; PEEK-CHAR checked type of
2916 PEEK-TYPE only after having read first character from a
2917 stream. (reported by Peter Graves)
2918 * bug fix: the garbage collector now has much better locality
2919 behaviour, and in particular no longer treats all memory as being
2920 exhausted when many small objects point to each other in a deeply
2922 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2923 recognized as being TYPEP their class.
2924 * bug fix: the PUSHNEW documentation string has been corrected.
2925 (thanks to Vincent Arkesteijn)
2926 * bug fix: defaulting of the value for the last of an atypically
2927 large number of multiple values being bound was not being
2928 performed correctly on the Alpha or PPC platforms
2929 * optimization: implemented multiplication as a modular
2930 (UNSIGNED-BYTE 32) operation on the x86 backend.
2931 * optimization: SEARCH on simple-base-strings can now be open-coded.
2932 (see also contrib/compiler-extras.lisp for inspiration for
2933 teaching the compiler about the Boyer-Moore algorithm).
2934 * value, returned by MAX (and MIN) called with several EQUALP, but
2935 not EQL, arguments now does not depend on compiler settings.
2936 * fixed some bugs revealed by Paul Dietz' test suite:
2937 ** in stack analysis liveness information is propagated from
2938 non-local entry points.
2939 ** pathwise CAST removing failed when the CAST node did not start
2941 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2942 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2944 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2945 host is already defined.
2946 ** RENAME-FILE works on streams instead of signalling an internal
2948 ** PEEK-CHAR uses the current readtable when determining whether
2949 or not a character is whitespace.
2950 ** MERGE-PATHNAMES handles the case when the pathname does not
2951 specify a name while the default-pathname specifies a version.
2952 ** Pathnames now stand a better chance of respecting print/read
2954 ** Attempting to use standardized file system operators with a
2955 pathname with invalid :DIRECTORY components signals a
2957 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2959 ** DIRECTORY on logical pathnames is more correct.
2960 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2961 signal a TYPE-ERROR if their argument is not a stream.
2962 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2963 designator argument does not designate a stream.
2964 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2965 examining the synonym.
2966 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2968 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2969 element-type larger than ([UN]SIGNED-BYTE 32).
2971 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2972 * When built with the :SB-FUTEX feature, threaded builds now take
2973 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2974 for faster/more reliable mutex and condition variable support.
2975 * Incompatible change (but one you probably shouldn't have been using
2976 anyway): the interface and code for arbitrating between multiple
2977 threads in the same user session has been redesigned.
2978 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2979 error when called without an explicit environment argument.
2980 (thanks to Nikodemus Siivola)
2981 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2982 confused about their position. (thanks to Adam Warner and Gerd
2984 * bug fix: Pretty printing backquoted forms with unquotations in the
2985 argument list position of various code constructs such as LAMBDA
2986 now works correctly. (reported by Paul Dietz)
2987 * bug fix: Pretty printing unquotations no longer loses all
2988 stream position information.
2989 * optimization: performance of string output streams is now less
2990 poor for multiple small sequence writes.
2991 * optimization: performance of CSUBTYPEP in the presence of complex
2992 expressions involving CONS and NOT many times has been improved.
2993 (reported by Paul Dietz)
2994 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2996 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2997 streams for the REPL, for the debugger, and so on. (thanks to
2999 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
3000 * fixed some bugs revealed by Paul Dietz' test suite:
3001 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
3002 is ignored if it is too small, rather than propagating through
3003 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
3004 ** extremely complex negations of CONS types were not being
3005 sufficiently canonicalized, leading to inconsistencies in
3007 ** VALUES tranformer lost derived type.
3009 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
3010 * fixed a bootstrapping bug: the build process no longer assumes
3011 that the various BOOLE-related constants have the same value in
3012 host and target lisps. (noted by Paul Dietz' test suite on an
3013 SBCL binary built from CLISP)
3014 * The system can now be dynamically linked on the MIPS platform,
3015 which enables dynamic loading of foreign code from Lisp. (thanks
3016 to Ralf Baechle for discussions on the MIPS ABI)
3017 * The system now records debugging information for its own source
3018 files in a filesystem-position-independent manner, relative to
3019 the "SYS" logical host.
3020 * fixed a compiler bug: MV-LET convertion did not check references
3021 to the "max args" entry point. (reported by Brian Downing)
3022 * tweaked disassembly notes to be less confident about proclaiming
3023 some instruction as an LRA. (thanks to Brian Downing)
3024 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
3025 now each have their own history, command character, and other
3026 characteristics. (thanks to David Lichteblau)
3027 * fixed some bugs revealed by Paul Dietz' test suite:
3028 ** compiler failure in compiling LOGAND expressions including a
3030 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3031 shift greater than 32.
3032 ** FLUSH-DEST did not mark blocks for type check regeneration.
3033 ** HANDLER-CASE failed to accept declarations in handler clauses
3034 in some circumstances.
3036 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3037 * New code in contrib/sb-introspect (still probably not entirely
3038 stable yet) provides some support for smart Lisp development
3039 environments like SLIME.
3040 * The conditions signalled for errors occurring when loading .fasl
3041 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3042 in a way which should help ASDF recover gracefully.
3043 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3044 stuff has been cleaned up. If you code contrib/ stuff, this might
3045 affect you, and you can look at contrib/README, contrib/STANDARDS,
3046 and/or the 0.8.4.27 diff to check.
3047 * In full calls the compiler now does not generate checks for declared
3048 argument types for all arguments.
3049 * various threading fixes
3050 ** and some experimental patches which didn't make it into
3051 the main tree for this release, but which are shipped in
3052 contrib/experimental-thread.patch as a possible fix for some
3053 failures (deadlock, spinning...) in GC-intensive multithreaded
3055 * fixed PPC build problem (source code incompatibility of different
3056 library versions): added offsetof() hackery which attempts to divine
3057 where glibc maintainers put uc_mcontext today
3058 * fixed bug 282: compiler does not trust type assertions while passing
3059 arguments to a full call.
3060 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3061 &OPTIONAL VALUES type parameter.
3062 * fix bug 214: algorithm for noting rejected templates is now more
3063 similar to that of template seletion. (also reported by rydis on
3065 * fixed bug 141b: printing backquoted information readably and prettily
3066 inserts a space where necessary.
3067 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3068 (CHARACTER 10) are now reported as errors, rather than propagated
3069 as unknown types. (reported by piso on #lisp)
3070 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3071 with values NIL and :ERROR. (thanks to Milan Zamazal)
3072 * fixed bug 191c: CLOS now does proper keyword argument checking as
3073 described in CLHS 7.6.5 and 7.6.5.1.
3074 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3075 longer raise an error; further, using a list as a for-as-arithmetic
3076 counter now raises a meaningful error.
3077 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3078 now understood by sequence creation functions such as MAKE-SEQUENCE
3080 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3081 a STRING-INPUT-STREAM.
3082 * compiler enhancement: SIGNUM is now better able to derive the type
3084 * type declarations inside WITH-SLOTS are checked. (reported by
3086 * fixed some bugs revealed by Paul Dietz' test suite:
3087 ** incorrect optimization of TRUNCATE for a positive first
3088 argument and negative second.
3089 ** compiler failure in let-convertion during flushing dead code.
3090 ** compiler failure while deriving type of TRUNCATE on an
3091 interval, containing 0.
3092 ** ASH of a negative bignum by a negative bignum count now returns
3094 ** intersection of CONS types now canonicalizes properly, fixing
3095 inconsistencies in SUBTYPEP.
3097 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3098 * incompatible change: The --disable-debugger command line
3099 option now clobbers the debugger at a more fundamental
3100 level, by redefining #'INVOKE-DEBUGGER instead of by
3101 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3102 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3103 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3104 enter the debugger REPL and then suffer recursive errors
3105 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3106 while under the new system, BREAK in --disable-debugger mode
3107 terminates the system just as an unhandled error would.
3108 * fixed compiler performance when processing loops with a step >1;
3109 * bug fix: DOCUMENTATION now retrieves generic function
3110 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3111 support has been systematized, and now supports the methods
3112 specified by ANSI, along with a default method and a method for
3113 slot documentation. (reported by Nathan Froyd)
3114 * bug fix: effective methods associated with a generic function are
3115 no longer cached over a change of that generic function's method
3116 combination. (reported by Andreas Fuchs)
3117 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3118 names a non-existent directory. (thanks to Andreas Fuchs)
3119 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3120 the PPC platform, lead to stack corruption; this has been fixed.
3121 (reported by Rainer Joswig)
3122 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3123 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3124 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3125 arithmetic in general, is now much more reliable on the PPC
3127 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3128 platform now returns the right answer.
3129 * optimization: restored some effective method precomputation in
3130 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3131 precomputation is now tunable.
3132 * optimization: compiler-internal data structure use has been
3133 reviewed, and changes have been made that should improve the
3134 performance of the compiler by about 20%.
3135 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3136 simple-base-strings and simple-bit-vectors is improved.
3137 * optimization: the optimization of 32-bit logical and arithmetic
3138 functions introduced in version 0.8.3 on the x86 has been
3139 implemented on the mips, ppc and sparc platforms; an
3140 implementation of the same facility, but for 64-bit arithmetic,
3141 has been added for the alpha.
3142 * microoptimization: the compiler is better able to make use of the
3143 x86 LEA instruction for multiplication by constants.
3144 * bug fix: in some situations compiler did not report usage of
3145 generic arithmetic in (SPEED 3) policy.
3146 * bug 145b fix: compiler used wrong type specifier while converting
3147 MEMBER-types to numeric.
3148 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3150 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3151 index argument now works properly on the Alpha platform.
3152 * bug fix: floating point exception treatment on the Alpha platform
3154 * bug fix: FILE-POSITION works much better on string input and
3155 output streams. (thanks to Nikodemus Siivola)
3156 * bug fix: many threading/garbage collection symptoms sorted.
3157 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3158 might be pseudo-atomic.
3159 * internal change: Stopping for GC is now done with signals not
3160 ptrace. GC is now done in whichever thread wanted it, instead of
3162 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3164 * bug fix: result form in DO is not contained in the implicit
3166 * incompatible change: ICR structure is changed; the value part of
3167 CONTINUATION is now called LVAR; corresponding functions are
3168 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3169 * added type deriver for ISQRT (thanks to Robert E. Brown).
3170 * bug fix: better support for loading from the command line when an
3171 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3173 * fixed some bugs revealed by Paul Dietz' test suite:
3174 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3175 ** ROUND and FROUND now give the right answer when given very
3176 small float arguments.
3177 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3179 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3180 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3181 ** LCM should return a non-negative integer.
3182 ** PARSE-INTEGER returned the index of a terminator instead of the
3183 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3184 ** PARSE-INTEGER returned an incorrect index being applied to a
3186 ** LCM with two arguments of 0 returns 0 rather than signalling
3188 ** unsigned addition of a 32-bit constant with the high bit set no
3189 longer causes an internal compiler error on the x86.
3190 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3191 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3192 with negative last argument.
3193 ** byte specifiers with zero size and position no longer cause
3194 an error during type derivation.
3195 ** bignum multiplication on the Alpha platform now returns the
3197 * porting: The system now builds on SuSE AMD64, although it still
3198 generates a 32-bit binary.
3199 * .fasl file incompatibility: The fasl file version number has
3200 been incremented (because of the changes to internal compiler
3201 data structures referred to above).
3203 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3204 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3205 more accurately, on the Darwin kernel running on PowerPC hardware.
3206 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3207 * Compiler code deletion notes now signal a condition of type
3208 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3209 an associated MUFFLE-WARNING restart.
3210 * The compiler now performs limited argument count validation of
3211 constant format strings in FORMAT, and where appropriate in ERROR,
3212 CERROR and WARN. (thanks to Gerd Moellmann)
3213 * New ASDF-INSTALL contrib can be used for automatic download and
3214 installation of third-party Lisp code from CCLAN or other sites
3216 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3217 which forces another thread to execute a function supplied by the
3219 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3220 now accept and act upon their :ELEMENT-TYPE keyword argument.
3221 (reported by Martin Atzmueller, Edi Weitz)
3222 * bug fix: FILE-POSITION now accepts position designators up to
3223 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3224 is the greater. (thanks to Patrik Nordebo)
3225 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3226 Kalle Olavi Niemitalo)
3227 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3228 printable. (reported by Eric Marsden)
3229 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3230 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3231 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3232 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3233 (thanks to Stig Sandoe)
3234 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3235 UNREAD-CHAR now correctly includes the unread character in the
3236 target sequence. (thanks to Gerd Moellmann)
3237 * bug fix in threaded builds: the system can now be suspended and
3238 resumed by shell job control with minimal disruption.
3239 * bug fixes in times and timezones >2038AD
3240 * better handling of "where is GNU make?" problem in build scripts
3241 (thanks to Nikodemus Siivola)
3242 * new optimization: inside a named function any reference to a
3243 function with the same name is considered to be a self-reference;
3244 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3245 optimization quality.
3246 * new optimization on x86: logical functions and + now have
3247 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3248 used when the result is truncated to 32 bits.
3249 * VALUES declaration is partially enabled.
3250 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3251 * fixed some bugs revealed by Paul Dietz' test suite:
3252 ** The system now obeys the constraint imposed by
3253 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3254 types form a lattice under type intersection.
3255 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3256 ** ASSOC now ignores NIL elements in an alist.
3257 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3258 and (1+ MOST-POSITIVE-FIXNUM) answers.
3259 ** The addition of a method with invalid qualifiers to a generic
3260 function does not cause an error to be signalled immediately;
3261 a warning is signalled, and the error is generated only on
3262 calling the generic function.
3263 * changed .fasl file version number, in part to add type codes for
3264 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3265 obscure ANSI requirements
3267 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3268 * fixed bug 148: failure to inline-expand a local function left
3269 garbage, confusing the compiler.
3270 * fixed bugs 3cd: structure slot readers perform type check if the
3271 slot can have an invalid value (i.e. it is either not initialized
3272 or can be written with a less specific slot writer).
3273 * bug fix: the compiler now traps array references to elements off
3274 the end of an array; previously, the bounds checking in some
3275 circumstances could go off-by-one.
3276 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3278 * type declarations for array element types now obey the description
3279 on the CLHS page "Declaration TYPE", as per discussions on
3280 sbcl-help around 2003-05-08. This means that a declaration
3281 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3282 declaration, all references to BAR will be asserted or assumed
3283 (with THE, so dependent on compiler policy) to involve objects of
3284 type FOO. Note that no such declaration is implied in
3285 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3286 * declared types of functions from the "Conditions"
3287 chapter. (reported by Paul Dietz)
3288 * bug fix: CERROR accepts a function as its first argument.
3289 * bug fix: NTH an NTHCDR accept a bignum as index
3290 arguments. (reported by Adam Warner)
3291 * optimization: character compare routines now optimize comparing
3292 against a constant character. (reported by Gilbert Baumann)
3293 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3294 argument now works properly on the MIPS platform.
3295 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3296 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3297 declared to have a functional type, does not cause an error
3299 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3300 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3301 * fixed bug: if last continuation of a deleted block has a
3302 destination, this destination should be deleted too. (reported by
3304 * fixed a bug in the bootstrap process: the host compiler's values
3305 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3306 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3307 test case from Patrik Nordebo)
3308 * improved the ability of the disassembler on the PPC platform to
3309 provide helpful disassembly notes.
3310 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3311 the class in more cases than previously.
3312 * bug fix: FILE-POSITION now understands :START and :END for
3313 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3314 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3315 8) data. (thanks to David Lichteblau)
3316 * bug fix: it is possible to add a method to a generic function
3317 without lambda list.
3318 * bug fix: reader failed to signal END-OF-FILE inside an
3319 object representation. (reported by Nikodemus Siivola)
3320 * fixed some bugs revealed by Paul Dietz' test suite:
3321 ** LAST and [N]BUTLAST should accept a bignum.
3322 ** condition slot accessors are methods.
3323 ** (VECTOR NIL) is a subtype of STRING.
3325 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3326 * minor incompatible change: some nonsensical specialized lambda
3327 lists (used in DEFMETHOD) which were previously ignored now signal
3329 * minor incompatible change: the system is now aware of the types of
3330 variables in the COMMON-LISP package, and will signal errors for
3331 most violations of these type constraints (where previously they
3332 were silently accepted).
3333 * minor incompatible change: COMPILE-FILE now uses the freedom
3334 afforded (ANSI 3.2.2.3) to use derived function types for
3335 functions defined in the same file. This also permits the system
3336 to warn on static type mismatches and function
3337 redefinition. (Currently it does not work with high DEBUG level.)
3338 * minor incompatible change: VALUES declaration is disabled.
3339 * When issuing notes, the compiler now signals a condition of type
3340 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3341 restart for use in user handlers. It is expected that the
3342 COMPILER-NOTE condition will eventually become a condition
3343 supertype to a hierarchy of note types, which will then be
3344 handleable in a similar fashion. However, at the moment, no such
3345 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3346 but it's an implementation detail, not a classification for the
3348 * Changes in type checking closed the following bugs:
3349 ** type checking of unused values (192b, 194d, 203);
3350 ** template selection based on unsafe type assertions (192c, 236);
3351 ** type checking in branches (194bc).
3352 * A short form of VALUES type specifier has ANSI meaning (it has
3353 increased the number of situations when SBCL cannot perform type
3355 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3356 keywords or constants is permissible.
3357 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3358 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3359 classes. (thanks to Antonio Martinez)
3360 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3361 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3362 argument to a pretty-print stream. Instead, it leaves any such
3363 operations to DESCRIBE-OBJECT methods.
3364 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3366 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3367 lambda lists. (thanks to David Lichteblau)
3368 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3369 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3370 (thanks to Gerd Moellmann)
3371 * fixed bug 246: increased compilation speed of long
3372 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3374 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3376 * DEFCONSTANT now throws a condition of type
3377 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3378 constant to a non-EQL value; CONTINUE and ABORT restarts
3379 respectively change and preserve the value.
3380 * fixed bug 63: The code walker, part of the implementation of CLOS,
3381 is now better at handling symbol macros.
3382 * bug fix: There is no longer an internal implementation type named
3383 CL:LENGTH. (reported by Raymond Toy)
3384 * bug fix: In macro-like defining macros/special operators the
3385 implicit block does not enclose the lambda list.
3386 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3387 enclosed in parentheses) are not suitable as type specifiers, and
3388 their use properly signals an error now.
3389 * bug fix: An argument count mismatch for a type specifier in code
3390 being compiled no longer causes an unhandled error at compile
3391 time, but signals a compile-time warning.
3392 * fixed simple vector readable printing
3393 * bug fix: DESCRIBE takes more care over whether the class
3394 precedence list slot of a class is bound before accessing it.
3395 (reported by Markus Krummenacker)
3396 * bug fix: FORMATTER can successfully compile pretty-printer format
3397 strings which use variants of the ~* directive inside.
3398 * bug fix: SEARCH now applies its TEST predicate to the elements of
3399 the arguments in the correct order. (thanks to Wolfhard Buss)
3400 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3401 in type inference. (thanks to Robert E. Brown)
3402 * bug fix: Reading in symbols with an explicit package name of ""
3403 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3404 * fixed some bugs revealed by Paul Dietz' test suite:
3405 ** NIL is now allowed as a structure slot name.
3406 ** Arbitrary numbers, not just REALs, are allowed in certain
3407 circumstances in LOOP for-as-arithmetic clauses.
3408 ** Multiple class redefinitions before slot access no longer
3409 causes a type error.
3410 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3411 association between the name and a class.
3412 ** Generic functions with non-standard method-combination and over
3413 five methods all of which return constants no longer return NIL
3414 after the first few invocations. (thanks to Gerd Moellmann)
3415 ** CALL-NEXT-METHOD with no arguments now passes the original
3416 values of the arguments, even in the presence of assignment.
3417 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3419 ** DISASSEMBLE works with closures and funcallable instances.
3420 ** ADD-METHOD now returns the generic function, not the new method.
3421 ** FIND-METHOD signals an error if the lengths of the specializers
3422 is incompatible with the generic function, even if the ERRORP
3424 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3425 which its argument is a member.
3426 ** DEFCLASS only redefines the class named by its class-name
3427 argument if that name is the proper name of the class;
3428 otherwise, it creates a new class.
3429 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3430 of the UNBOUND-SLOT condition to the name of the slot.
3431 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3432 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3433 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3434 SLOT-MAKUNBOUND in the specified fashion.
3436 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3437 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3438 cross-compilation host. As a consequence, we can now bootstrap our
3439 way up to SBCL starting with a bare gcc toolchain and human-readable
3440 source code (first the source to CLISP, then the source to SBCL).
3441 * A contributed module containing a partial implementation of the
3442 simple-streams interface has been included. (thanks to Rudi
3444 * A contributed module implementing the RFC1321 Message Digest
3445 Algorithm, known as MD5, has been included.
3446 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3447 no longer has any effect, as the code controlled by this feature
3448 has been deleted. (As far as we know, no-one has ever built using
3449 this feature, and its semantics were confused in any case).
3450 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3451 work on conditions (as required by the ANSI specification),
3452 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3453 expected behaviour on conditions. Users should note, however,
3454 that such behaviour is not required by the ANSI specification,
3455 and so use of this behaviour may render their code unportable.
3456 * fixed some bugs revealed by Paul Dietz' test suite:
3457 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3459 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3460 ** SXHASH on condition objects no longer returns NIL.
3461 ** :ALLOCATION :CLASS slots are better treated; their values are
3462 updated on class redefinition, and initforms inherited from
3463 superclasses are applied.
3464 ** REMOVE-METHOD returns its generic function argument even when
3465 no method was removed.
3466 ** SHARED-INITIALIZE now initializes the values of the requested
3467 slots, including those with :ALLOCATION :CLASS.
3468 ** ALLOCATE-INSTANCE now works on structure classes defined via
3469 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3471 ** SLOT-EXISTS-P now works on conditions, as well as structures
3473 ** MAKE-LOAD-FORM now has the required methods on
3474 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3475 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3476 keyword, and now implements the SLOT-NAMES argument.
3477 ** methods with &OPTIONAL arguments no longer allow too many
3478 arguments to be passed in the call without error.
3479 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3480 option is consistent with the required arguments of the generic
3481 function lambda list.
3482 * bug fix: REQUIRE accepts a string designator. (Thanks to
3484 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3485 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3486 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3487 * bug fix: The compiler now checks for duplicated variables in macro
3489 * bug fix: SETQ on globals returns the correct value.
3490 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3491 not a COMPILER-ERROR (followed by some other strange error on
3492 choosing the CONTINUE restart).
3493 * bug fix: make.sh and friends are now more consistent in the way that
3494 they look for GNU "make".
3496 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3497 * experimental native threads support (on x86 Linux >=2.4 only).
3498 This is not compiled in by default: you need to add :SB-THREAD to
3499 the target features. See the "Beyond ANSI" chapter of the manual
3501 * fix for longstanding nonANSIism: The old distinction between
3502 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3503 The return value from CL:FIND-CLASS is now a CLOS class, and
3504 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3505 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3506 * An interface to the MetaObject Protocol, as described in Kiczales,
3507 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3508 MIT Press, 1991, is available from the SB-MOP package.
3509 * incompatible change: the SB-PCL package should now be considered
3510 a private implementation detail, and no longer a semi-private MOP
3512 * minor incompatible change: due to rearrangement for threads, the
3513 control stack and binding stack are now allocated at arbitrary
3514 addresses instead of being hardcoded per-port. Users affected by
3515 this probably have to be doing advanced things with shared
3516 libraries, and will know who they are.
3517 * minor incompatible change: Previously, all --eval forms used were
3518 processed with READ before any of them were processed with EVAL.
3519 Now each --eval form is processed with both READ and EVAL before
3520 the next --eval form is processed. (Thus package operations like
3521 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3522 work as the user might reasonably expect.)
3523 * minor incompatible change: *STANDARD-INPUT* is now only an
3524 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3526 * minor incompatible change: Y-OR-N-P is now character-oriented,
3527 not line oriented. Also, YES-OR-NO-P now works without errors.
3528 (thanks to Antonio Martinez)
3529 * sb-aclrepl module improvements: an integrated inspector, added
3530 repl features, and a bug fix to :trace command.
3531 * Known functions, which cannot be open coded by the backend, are
3532 considered to be able to check types of their arguments. (fixing
3533 a bug report by Nathan J. Froyd)
3534 * fixed a bug in computing method discriminating functions: It is
3535 now possible to define methods specialized on classes which have
3536 forward-referenced superclasses. (thanks to Gerd Moellmann)
3537 * fixed evaluation order in optional entries (reported by Gilbert
3539 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3540 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3541 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3542 required-by-AMOP NAME argument, as well as CLASS and
3543 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3544 * fixed bug 20: DEFMETHOD can define methods using names that are
3545 not the proper names of classes to designate class specializers.
3546 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3547 Unix file descriptors, instead of blowing up. (thanks to Antonio
3549 * Garbage collection refactoring: user-visible change is that a
3550 call to the GC function during WITHOUT-GCING will not do garbage
3551 collection until the end of the WITHOUT-GCING. If you were doing
3552 this you were probably losing anyway.
3553 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3554 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3555 TYPEP the latter but not the former.
3556 * The compiler issues a full WARNING for calls to undefined functions
3557 with names from the CL package.
3558 * MAP-INTO for a vector destination is open coded. (reported by
3559 Brian Downing on c.l.l)
3560 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3561 documentation string.
3562 * fixed some bugs revealed by Paul Dietz' test suite:
3563 ** COPY-ALIST now signals an error if its argument is a dotted
3565 ** Condition slots are now accessed more correctly in the presence
3566 of multiple initargs for a given slot.
3567 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3568 exclude restarts of the same name associated with a different
3570 ** DEFCLASS of forward-referenced classes with another
3571 forward-referenced class in the superclasses list no longer
3573 ** Condition slots are now initialized once each, not multiple
3574 times. (thanks to Gerd Moellmann)
3575 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3577 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3578 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3579 arguments contain duplicated elements.
3580 ** RESTART-CASE understands local macros.
3581 ** RESTART-CASE associates exactly its own restarts with a condition.
3582 ** ENDP in safe mode checks its argument to be of type LIST.
3583 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3584 in question is unbound.
3585 ** Optimized MAKE-INSTANCE functions no longer cause internal
3586 assertion failures in the presence of duplicate initargs.
3587 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3588 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3589 propagated even in the ctor optimized implementation of
3591 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3593 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3594 * a better implementation of SXHASH on (simple) bit vectors,
3595 measured both in execution speed and in distribution of results
3596 over the positive fixnums, has been installed. Likewise, a better
3597 implementation of EQUAL for simple bit vectors is now available.
3598 * fixed CEILING optimization for a divisor of form 2^k.
3599 * fixed bug 240 (emitting extra style warnings "using the lexical
3600 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3601 by Antonio Martinez)
3602 * fixed SXHASH, giving different results for NIL depending on type
3603 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3604 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3605 variable. (found by Rolf Wester)
3606 * SBCL does not ignore type declarations for special
3607 variables. (reported by rif on c.l.l 2003-03-05)
3608 * some bug fixes in contrib/sb-aclrepl/
3609 * fixed some bugs revealed by Paul Dietz' test suite:
3610 ** a bug in the CONS type specifier, whereby the CAR and CDR
3611 types got intertwined, has been fixed;
3612 ** the type system is now able to reason about the interaction
3613 between INTEGER and RATIO types more completely;
3614 ** APPEND, [N]REVERSE and NRECONC check that those their
3615 arguments, which must be proper lists, are really so;
3616 ** An array specialized to be unable to hold elements has been
3617 implemented, as required -- yes, really -- by ANSI;
3618 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3619 on malformed property lists;
3621 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3622 * incompatible packaging change: in line with Unix convention,
3623 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3624 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3625 * REQUIRE and PROVIDE are now optionally capable of doing something
3626 useful. See the documentation string for REQUIRE.
3627 * infrastructure for a managed SBCL contrib system: contributed
3628 modules in this release include:
3629 ** the ASDF system definition facility;
3630 ** an interface to the BSD Sockets API;
3631 ** an ACL-like convenience interface to the repl;
3632 (thanks to Kevin Rosenberg)
3633 ** an implementation of ROTATE-BYTE, with efficient implementation
3635 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3636 gives a closer approximation to the right answer than previously.
3637 (thanks to Raymond Toy)
3638 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3639 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3640 optional environment arguments, as required by ANSI.
3641 * fixed bugs in other functions taking environment objects, allowing
3642 calls with an explicit NIL environment argument to be compiled
3644 * fixed bug 228: primary return values from
3645 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3646 COMPILE or FUNCTION.
3647 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3648 :TYPE will no longer signal errors on innocuous objects.
3649 * fixed bug 231b: SETQ is better at respecting type declarations in
3650 the lexical environment.
3651 * fixed a bug in DEFCLASS: classes named by symbols with no or
3652 unprintable packages can now be defined.
3653 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3654 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3655 * fixed compiler failure related to checking types of functions.
3656 (reported by Robert E. Brown)
3657 * the compiler is now much more consistent in its error-checking
3658 treatment of bounding index arguments to sequence functions: in
3659 (SAFETY 3) code, errors will be signalled in almost all cases if
3660 invalid sequence bounding indices are passed to functions defined
3661 by ANSI to operate on sequences.
3662 * fixed a bug in the build procedure: documentation of SBCL-specific
3663 packages is now preserved and available in the final Lisp image.
3664 * lifted FDEFINITION lookup out of loops in the implementation of
3665 many list operations. (thanks to Robert E. Brown)
3666 * fixed a bug in the reader: the #n# reader macro now works for
3667 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3668 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3669 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3670 E. Brown and Tony Martinez respectively)
3671 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3672 SLOT-BOUNDP: in particular, it is now able to optimize them much
3673 better, and is now not vulnerable to having packages renamed.
3674 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3675 always called when a slot is not present in an instance. (thanks
3677 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3678 longer signal an error on unreading a character following EOF on
3679 the previous constituent stream. (thanks to Tony Martinez)
3680 * fixed some bugs revealed by Paul Dietz' test suite:
3681 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3682 not just nonnegative fixnums;
3683 ** the logical bit-array operators such as BIT-AND now accept an
3684 explicit NIL for their "opt-arg" argument (to indicate a
3685 freshly-consed result bit-array);
3686 ** ELT now signals an error on an invalid sequence index in safe
3688 ** the type system is now cleverer about negations of numeric
3689 types, and consequently understands the BIGNUM and RATIO types
3691 ** the type system is now cleverer about the interaction between
3692 INTEGER and RATIO types: while bugs still remain, many more
3693 cases are accurately computed;
3694 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3695 if it is in the last clause;
3696 ** CONSTANTLY now correctly returns a side-effect-free function in
3698 ** DECLARE is no longer treated as a special-operator; in
3699 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3700 * incremented fasl file version number due to the change in the
3701 DEFSTRUCT-SLOT-DESCRIPTION structure.
3703 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3704 * minor incompatible change: code processed by the "interpreter" or
3705 EVAL now has a compilation optimization policy of (DEBUG 2)
3706 (changed from (DEBUG 1)) to improve debuggability of interactive
3707 development, and to allow the use of the debug RETURN command in
3709 * an experimental implementation of the RETURN command for the
3710 debugger has been included. (thanks to Frederik Kuivinen)
3711 * fixed bug 62: constraints were not propagated into a loop.
3712 * fixed bug in embedded calls of SORT (reported and investigated by
3714 * fixed some bugs revealed by Paul Dietz' test suite:
3715 ** printing and reading of arrays with some dimensions having
3716 length 0 (thanks to Gerd Moellmann);
3717 ** BOA constructor with &AUX argument without a default value does
3718 not cause a type error;
3719 ** CONSTANTP now returns true for all self-evaluating objects.
3721 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3722 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3723 accessors that are related by inheritance, as specified in the
3724 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3725 Valtteri Vuorikoski)
3726 * The compiler is now able to inline functions that were defined in
3727 a complex lexical environment (e.g. inside a MACROLET).
3728 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3730 * Support for the upcoming FreeBSD-5.0 release has been included.
3731 (thanks to Dag-Erling Smorgrav)
3732 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3733 effect when it is not in a toplevel context.
3734 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3735 stand a better chance of being correct. (thanks to Gerd
3737 * fixed bug in COERCE, which now signals an error on coercing a
3738 rational to a bounded real type which excludes the expected
3740 * The compiler is now able to derive types more accurately from the
3741 COERCE and COMPILE functions.
3742 * fixed bug 223: functional binding is considered to be constant
3743 only for symbols in the CL package.
3744 * fixed bug 231: SETQ did not check the type of a variable being set
3745 (reported by Robert E. Brown)
3746 * A new optimization for MAKE-INSTANCE has been included, fixing
3747 various bugs (including relating to :ALLOCATION :CLASS slots and
3748 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3750 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3751 ** As required by ANSI, LOOP now disallows anonymous collection
3752 clauses such as COLLECT I in conjunction with aggregate boolean
3753 clauses such as THEREIS (= I 1);
3754 ** LOOP now signals an error when any variable is reused in the
3755 same loop (including the potentially useful construct analogous
3756 to WITH A = 1 WITH A = (1+ A);
3757 ** IT is only a special loop symbol within the first clause of a
3758 conditional loop clause;
3759 ** LOOP with a typed iteration variable over a hashtable now
3760 signals a type error iff it should.
3761 * fixed some other bugs revealed by Paul Dietz' test suite:
3762 ** FILE-STREAM now names the class previously known as FD-STREAM;
3763 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3764 argument) no longer signals an error;
3765 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3766 of the slot symbol, rather than using the current package
3767 ((:CONC-NAME "") continues to intern the slot's name in the
3769 * incremented fasl file version number, because of the incompatible
3770 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3771 of the new implementation of DEFINE-COMPILER-MACRO.
3773 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3774 * Support for building SBCL for MIPS platforms running in
3775 little-endian mode has now been checked in, and basic
3776 functionality on said platforms verified.
3777 * minor incompatible change: PCL now records the pathname of a file
3778 in which methods and the like are defined, rather than its
3780 * minor incompatible change: TRUENAME now considers the truename of
3781 a file naming a directory to be the pathname with :DIRECTORY
3782 component indicating that directory.
3783 * minor incompatible change: a NAMED clause in the extended form of
3784 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3785 reason for the previous behaviour is unclear.
3786 * more systematization and improvement of CLOS and MOP conformance
3787 in PCL (thanks to Gerd Moellman and Pierre Mai):
3788 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3790 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3791 primary methods with no specializers;
3792 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3794 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3795 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3796 FORWARD-REFERENCED-CLASSes; error reporting on
3797 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3799 ** SXHASH on CLOS instances now uses a slot internal to the
3800 instance to return different numbers on distinct instances,
3801 while preserving the same return value through invocations of
3803 ** DEFMETHOD signals errors when methods with longer incongruent
3804 lambda lists are added to generic functions;
3805 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3806 CLASS, as specified in AMOP;
3807 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3808 on the order returned by the primary method for classes of
3809 class STANDARD-CLASS;
3810 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3811 * fixed some bugs shown by Paul Dietz' test suite:
3812 ** DOLIST puts its body in TAGBODY;
3813 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3815 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3816 value producing form;
3817 ** if more variables are given to PROGV than values, extra
3818 variables are bound and made to have no value;
3819 ** NSUBSTITUTE on list arguments gets the right answer with
3821 ** ELT signals an error of type TYPE-ERROR when the index argument
3822 is not a valid sequence index;
3823 ** LOOP signals (at macroexpansion time) an error of type
3824 PROGRAM-ERROR when duplicate variable names are found;
3825 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3826 ** FUNCALL of special-operators now cause an error of type
3828 ** PSETQ now works as required in the presence of side-effecting
3829 symbol-macro places;
3830 ** NCONC accepts any object as its last argument;
3831 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3833 ** loop-for-as-package does not require a package to be explicitely
3835 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3837 * fixed bug 166: compiler preserves "there is a way to go"
3838 invariant when deleting code.
3839 * fixed bug 172: macro lambda lists with required arguments after
3840 &REST arguments now cause an error to be signalled. (thanks to
3842 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3843 bug. (thanks to Matthew Danish)
3844 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3846 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3847 arguments are lexically rebound. (thanks to Gerd Moellmann and
3849 * fixed bug 194: error messages are now more informative when there
3850 is no primary method applicable in a call to a generic
3851 function. (thanks to Gerd Moellmann)
3852 * fixed bug in command line argument checking (thanks to Julian
3854 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3855 (thanks to Matthew Danish)
3856 * incremented fasl file version number, because of the
3857 SXHASH-related changes in the layout of CLOS data structures
3859 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3860 * minor incompatible change: The runtime (the Unix executable named
3861 "sbcl") is now much pickier about the .core files it will load.
3862 Essentially it now requires .core files to descend from the same
3863 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3864 as the runtime does. (The intent is to prevent the crashes which
3865 can occur, and which can even be reported as mysterious failures,
3866 when people patch the sources or change the build parameters
3867 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3868 sbcl and .core files.)
3869 * fixed bug: VALUES-LIST is no longer optimized away.
3870 * fixed bug 142: The FFI conversion of C string values to Lisp
3871 string values no longer conses excessively. (thanks to Nathan
3872 Froyd porting Raymond Toy's fix to CMU CL)
3873 * began to systematize and improve MOP conformance in PCL (thanks to
3874 Nathan Froyd, Gerd Moellman and Pierre Mai):
3875 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3877 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3878 ** FINALIZE-INHERITANCE is now called on class finalization;
3879 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3880 argument precedence order.
3881 * fixed bug 202: The compiler no longer fails on functions whose
3882 derived types contradict their declared type.
3883 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3884 so it can be non-toplevel.
3885 * The fasl file version number has changed (because of the new
3886 implementation of DEFMACRO).
3887 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3888 safe code, that any length requirement by their type-specifier
3889 argument is valid. The exceptions to this are described in bug
3891 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3892 functions now check the types of their inputs as required by ANSI.
3893 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3894 introduced symbol is DECLAREd to be SPECIAL.
3895 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3896 * fixed bug: :COUNT argument to sequence functions may be negative
3897 * fixed bug: body of DO-SYMBOLS may contain declarations
3898 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3899 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3900 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3901 symbol macro only once
3902 * fixed printing of call frame when argument list is unavailable
3903 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3904 * compiler no longer signals WARNING on unknown keyword
3907 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3908 * A beta-quality port to the mips architecture running Linux,
3909 based on the old CMUCL backend, has been made. It has been tested
3910 on a big-endian kernel, and works sufficiently well to be able to
3911 rebuild itself; it has not been tested in little-endian mode.
3912 * fixed an inconsistency between gencgc.c and purify.c which made
3913 dumping/loading .core files unreliable
3914 * fixed bug 120a: The compiler now deals correctly with IFs where
3915 the consequent is the same as the alternative, instead of
3916 misderiving the return type. (thanks to Alexey Dejneka)
3917 * fixed bug 113: Logical pathnames are now dumpable (the logical
3918 host is resolved at load-time, throwing an error if it is not
3920 * fixed bug 174: FORMAT's error message is slightly clearer when a
3921 non-printing character is used in a format directive.
3922 * fixed several bugs in compiler checking of type declarations, i.e.
3923 violations of the Python "declarations are assertions" principle
3924 (thanks to Alexey Dejneka)
3925 * fixed several bugs in PCL's error checking (thanks to Gerd
3927 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3929 * fixed bug in compilation of functions as first class values
3930 (thanks to Antonio Martinez-Shotton)
3931 * The compiler's handling TYPE-ERRORs which it can prove will
3932 inevitably happen at runtime has been cleaned up and corrected
3933 in several ways. (thanks to Alexey Dejneka)
3934 * improved argument type checking for various basic arithmetic
3935 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3936 lavished on them in the past that they can be compiled in many
3937 ways in different special cases
3938 * fixed bug 181: compiler checks validity of user supplied type
3940 * cleaned up code flushing in optimization: Function calls which
3941 should signal errors for safety purposes (e.g. which ANSI says
3942 should signal errors when their arguments are of incorrect type)
3943 are no longer optimized away.
3944 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3945 * incremented fasl file version number, because changes in the
3946 implementation of sequence functions like COERCE caused
3947 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3948 in old inline expansions) to become undefined. (Actually these
3949 changes were later undone, so we might very well be binary
3950 compatible with 0.7.7 after all, but leaving the version number
3951 incremented seemed like the simplest and most conservative
3954 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3955 * An alpha-quality port to the parisc architecture running Linux,
3956 based on the old CMUCL backend, has been made. This, even more so
3957 than the other backends, should be considered still a work in
3958 progress; known problems include that the Linux kernel in 64-bit
3959 mode does not propagate the correct sigcontext structure to
3960 userspace, and consequently SBCL on a parisc64 kernel will not
3962 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3963 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3964 suppressed.) Also now that the compiler is looking at declarations
3965 in the environment, it checks optimization declarations as well,
3966 and suppresses inlining when (> DEBUG SPEED).
3967 * More fixes have been made to treatment of floating point exception
3968 treatment and other Unix signals. In particular, floating point
3969 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3970 * The detection and handling of control stack exhaustion (infinite
3971 or very deeply nested recursion) has changed. Stack exhaustion
3972 detection is now done by write-protecting pages at the OS level
3973 and applies at all optimization settings; when found, a
3974 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3975 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3976 be caught using IGNORE-ERRORS.
3977 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3978 correctly when transforming recursive LABELS functions to LETs.
3979 (thanks to Alexey Dejneka)
3980 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3981 that are names of constants or global variables.
3982 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3983 alien routines with docstrings.
3984 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3985 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3987 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3988 (thanks to Eric Marsden)
3989 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3990 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3991 * Bug fix: The SPARC backend can now compile functions involving
3992 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3993 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3994 to a functions expecting a PATHNAME-DESIGNATOR.
3995 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3996 lambda lists. (thanks to Alexey Dejneka)
3997 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3998 functions so that EXPORTing the name of the class doesn't cause
3999 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
4000 to Antonio Martinez for reporting this)
4001 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
4002 constant-folded without failing an assertion. (thanks to Einar
4003 Floystad Dorum for reporting this)
4004 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
4005 types are now dealt with more correctly by the compiler.
4006 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
4007 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
4008 behaviour analogous to RENAME-FILE. This puts its behaviour more
4009 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
4011 * The fasl file version number has changed again. (because of the
4012 bug fix involving the names of PCL MAKE-INSTANCE functions)
4014 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
4015 * bug fix: Floating point exceptions are treated much more
4016 consistently on the x86/Linux and PPC/Linux platforms.
4017 * Array initialization with :INITIAL-ELEMENT is now much faster for
4018 cases when the compiler cannot open code the array creation, but
4019 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
4020 array accesses have also seen a speed increase.
4021 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
4022 specified. (This was at the root of some bad interactions between
4023 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
4025 * bug fix: Internal error arguments for undefined functions are now
4026 computed correctly on the PPC/Linux platform.
4027 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4028 Raymond Toy's patch for CMU CL)
4029 * Support for the Solaris 9 operating environment has been included
4030 (thanks to Daniel Merritt)
4031 * A very ugly but hopefully complete draft of the missing FFI chapter
4032 of the manual has been created by reformatting the corresponding
4033 CMU CL manual chapter into (currently very ugly and incoherent)
4034 DocBook and bringing it up to date for SBCL behavior. Thus, the
4035 manual is now essentially complete, at least by my extreme
4036 once-and-only-once standards, whereby it's acceptable to refer to
4037 the doc strings of SB-EXT functions as the primary documentation.
4038 * The fasl file version number has changed again, due to cleanup of
4039 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4040 is no longer a static symbol.)
4042 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4043 * SBCL now builds with OpenMCL (version 0.12) as the
4044 cross-compilation host; also, more progress has been made toward
4045 bootstrapping under CLISP.
4046 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4048 * bug 158 fixed: The compiler can now deal with integer loop
4049 increments different from 1; fixing this turned out also to fix
4051 * bug 169 fixed: no more bogus warnings about using lexical bindings
4052 despite the presence of perfectly good SPECIAL declarations (thanks
4053 to David Lichteblau)
4054 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4055 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4056 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4058 * bug 180 fixed: Method combination specifications no longer ignore
4059 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4060 * bug fix: Structure type predicate functions now check their argument
4061 count as they should.
4062 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4063 correctly. (thanks to Pierre Mai)
4064 * minor incompatible change: The --noprogrammer option is deprecated
4065 in favor of the new --disable-debugger option, which is very similar.
4066 (The major difference is that it takes effect at a slightly different
4067 time at startup, causing handling of errors in --sysinit and
4068 --userinit files will be affected differently.) The
4069 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4070 been added to allow this functionality to be controlled from ordinary
4071 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4072 maintainers, who might want to run non-interactive scripts to
4073 build SBCL cores which will later be used interactively.)
4074 * minor incompatible change: The LOAD function no longer, when given
4075 a wild pathname to load, loads all files matching that pathname.
4076 Instead, an error of type FILE-ERROR is signalled.
4078 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4079 * bug 147 fixed: The compiler preserves its block link/count
4080 invariants more correctly now so that it doesn't crash. (thanks
4082 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4084 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4085 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4086 no-value-for-this-character as NIL (as specified by ANSI).
4087 * HOST-NAMESTRING on physical pathnames now returns a string that is
4088 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4089 (thanks to Christophe Rhodes)
4090 * The Alpha port handles icache flushing more correctly. (thanks to
4092 * More progress has been made toward bootstrapping under CLISP. (thanks
4093 to Christophe Rhodes)
4094 * The fasl file format has changed again, because dynamic loading
4095 on OpenBSD (which has non-ELF object files) motivated some cleanups
4096 in the way that foreign symbols are transformed and passed around.
4097 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4098 is no longer treated as whitespace by the reader, but instead as
4099 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4100 |A
\7fB|, instead of A as it used to.
4102 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4103 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4104 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4105 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4106 this as of 0.7.1.45, but was left out of the previous news section
4107 (thanks to Dan Barlow)
4108 * SBCL now runs on the Solaris operating system on SPARC architectures
4109 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4110 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4111 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4112 * SPARC backend cleanups, allowing builds of cores optimized for V8
4113 and V9 SPARCS, and also emission of code targeted to a particular
4114 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4116 * SBCL is closer to bootstrapping under CLISP, thanks to various
4117 fixes by Christophe Rhodes.
4118 * The fasl file format has changed again, to allow the compiler's
4119 INFO database to support symbol macros.
4120 * The user manual (in doc/) is formatted into HTML more nicely.
4121 (thanks to coreythomas)
4122 * The system is smarter about SUBTYPEP relationships, especially
4123 those involving NOT types (including types such as ATOM which are
4124 represented internally using NOT types). Thus SUBTYPEP is less
4125 likely to return (VALUES NIL NIL) in general, and in particular
4126 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4128 * The fasl file format has changed again, because the internal
4129 representation of types now includes a new slot to support the new
4130 SUBTYPEP-of-NOT-types logic.
4131 * (not a change in the main branch of SBCL, but a related prototype
4132 which can hopefully be merged into the main branch of SBCL in the
4133 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4134 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4135 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4136 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4137 dispatch macro character combinations. (thanks to Alexey Dejneka)
4138 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4139 that superficially look like logical namestrings correctly.
4140 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4142 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4143 default supplied-p) for &optional and &key arguments. (thanks to
4146 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4147 * incompatible change: The compiler is now less aggressive about
4148 tail call optimization, doing it only when (> SPACE DEBUG) or
4149 (> SPEED DEBUG). (This is an incompatible change because there are
4150 programs which relied on the old CMU-CL-style behavior to optimize
4151 away their unbounded recursion which will now die of stack overflow.)
4152 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4153 for non-GENCGC systems has been increased to 20M (since that
4154 seems much closer to the likely performance optimum for modern
4155 systems than the old 4M value was)
4156 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4157 and *DEBUG-PRINT-LEVEL*
4158 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4159 of CMU CL's support for SPARC, and various endianness and other
4160 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4161 * new syntactic sugar for the Unix command line: --load foo.bar is now
4162 an alternate notation for --eval '(load "foo.bar")'.
4164 ** The system now detects stack overflow and handles it gracefully,
4165 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4166 optimization settings. (This is a good thing in general, and
4167 its introduction in this version should be particularly timely
4168 for anyone whose code fails because of suppression of tail
4170 ** The system now hunts for the C variable "environ" in a more
4171 devious way, to avoid segfaults when the C library version differs
4172 between compile time and run time. (thanks to Christophe Rhodes)
4173 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4174 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4175 ** The compiler no longer issues bogus style warnings for undefined
4176 classes in the same source file as the DEFCLASSes which defined
4177 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4179 ** fixes in CONDITION class precedence list for undefined function
4180 errors (thanks to Alexei Dejneka)
4181 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4182 correctly. (thanks to Dan Barlow)
4183 ** portability fixes aiming at bootstrapping under CLISP (thanks
4184 to Dave McDonald and Christophe Rhodes)
4185 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4186 ** fixes in type translation and and type inference (thanks to
4188 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4189 ** various fixes in the new ports (thanks to Dan Barlow)
4190 * several changes related to debugging:
4191 ** suppression of tail recursion, as noted above
4192 ** stack overflow detection, as noted above
4193 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4194 is now the default. (For some time encapsulation has been more
4195 reliable than the breakpoint-based :ENCAPSULATE NIL
4196 implementation, at least on X86 systems; and I just noticed that
4197 encapsulation also seems closer to the spirit of the ANSI
4200 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4202 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4203 up properly again. (There was a packaging bug in 0.7.0 which
4204 left their definitions in SB-SYS::LOAD-FOREIGN and
4205 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4206 vital for most things which interface to C-level interfaces,
4207 like extensions working with sockets or databases or
4208 Perl-compatible regexes or whatever, and the need to fix
4209 this bug is the main reason that 0.7.1 was released so
4211 ** DEFGENERIC is now choosier about the methods it redefines, so that
4212 reLOADing a previously-LOADed file containing DEFGENERICs does
4213 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4214 cycle now works as it should. (thanks to Alexey Dejneka)
4215 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4216 fixed by Christophe Rhodes. (He actually submitted this patch
4217 months ago, and I delayed until after 0.7.0.)
4218 ** Bug 111 (internal compiler confusion about runtime checks on
4219 FUNCTION types) was fixed by Alexey Dejneka.
4220 * Some internal cleanups (getting rid of variables which aren't
4221 needed now that the byte interpreter is gone) caused the fasl
4222 file format number to change again.
4224 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4225 * major incompatible change: The default fasl file extension, i.e. the
4226 default extension for files produced by COMPILE-FILE, has changed
4227 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4229 ** There are many changes in the implementation of the compiler.
4230 SBCL is now essentially a compiler-only implementation of ANSI
4231 Common Lisp. EVAL still "interprets" a few special cases, but
4232 almost all the interesting cases are handled by creating
4233 a LAMBDA expression, calling COMPILE on it, then calling
4234 FUNCALL on the result.
4235 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4236 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4237 Since the code is newer, there might still be some new bugs
4238 (though not as many as before Martin Atzmueller's fixes:-). But
4239 the new code is substantially simpler and clearer, and hopefully
4240 any remaining bugs will be simpler, less fundamental, and more
4241 fixable then the bugs in the old code.
4242 ** The revised compiler is still a little unsteady on its feet.
4244 *** The debugging information it produces (particularly the names
4245 of FUNCTION objects) is sometimes much less useful than what
4246 the old compiler produced.
4247 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4248 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4249 within a MACROLET) has been temporarily weakened.
4250 ** There are new compiler optimizations for various functions:
4251 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4252 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4253 *** the math functions TRUNCATE, FLOOR, and CEILING
4254 *** the function-of-all-trades COERCE
4255 Mostly these should be transparent, but there's one
4256 potentially-annoying problem (bug 117): when the compiler
4257 inline-expands a function and does type analysis on the result,
4258 it can create control paths which have type mismatches, and
4259 when it can't prove that those control paths aren't taken,
4260 it will issue WARNINGs about the type mismatches. This is
4261 a particular problem in practice for the new sequence functions.
4262 It's not clear how this should be fixed, and for now, a
4263 workaround is given in the entry for 117 in the BUGS file.
4264 ** (Because of the interaction between the two previous items --
4265 occasional inlining problems and new inline expansions -- some
4266 of the new sequence function optimizations won't really kick in
4267 completely until debugging information, and then inlining, are
4268 straightened out in some future version.)
4269 * minor incompatible changes:
4270 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4271 DIRECTORY no longer implicitly promotes NIL slots of its
4272 pathname argument to :WILD. In particular, when you ask for the
4273 contents of a directory (which you used to be able to do without
4274 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4275 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4276 ** changes in behavior that ANSI explicitly defines to be
4277 implementation dependent:
4278 *** The new compiler-only implementation still conforms with ANSI,
4279 but acts a little different than before. Besides the obvious
4280 changes in performance tradeoffs (that the cost per form passed
4281 to EVAL has gone up, and the cost per form executed by EVAL
4282 has gone down), the behavior of the system changes a little
4283 because there are no longer any interpreted function objects.
4284 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4285 e.g. doing COMPILE on the output of interactive DEFUN is
4287 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4289 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4290 has changed from (:CL) to NIL.
4291 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4292 e.g. #\Newline and #\Tab, don't have names specified in the
4293 ANSI Common Lisp standard, is now based on their ASCII symbolic
4294 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4295 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4296 are no longer used for output.
4297 ** changes in internal implementation constants:
4298 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4299 4 million. (If your application spends a lot of time GCing and
4300 you have a lot of RAM, you might want to experiment with
4301 increasing it even more.)
4302 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4303 However, almost all old code should still continue to work without
4304 immediate update, as SB-C-CALL is now a (deprecated) nickname
4306 ** Old operator names in the style DEF-FOO are now deprecated in
4307 favor of new corresponding names DEFINE-FOO, for consistency with
4308 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4309 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4310 internal symbols, but a few supported extensions like
4311 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4312 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4313 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4314 etc. as you get deeper into recursive calls to the debugger
4315 command loop, instead of the old "5]", "5]]", "5]]]"
4316 sequence. (I was motivated to do this when squabbles between
4317 ILISP and SBCL left me very deeply nested in the debugger. In the
4318 short term, this change will probably provoke more ILISP/SBCL
4319 squabbles, but hopefully it will be an improvement in the long run.)
4320 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4321 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4322 function, since declaiming return types involving aliens is
4323 (1) annoyingly messy to do by hand and (2) vital to efficient
4324 compilation of code which calls such functions.
4325 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4326 longer reexported by the SB-EXT package. They're solely useful
4327 for alien code, so it seems more logical that you should get
4328 them from the SB-ALIEN package, not in SB-EXT.
4329 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4330 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4331 features. Instead, the code that they used to control is always
4332 built into the system.
4333 * many other bug fixes
4334 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4335 advantage of the new EVAL-WHEN stuff and to clean them up in
4336 general, and they are now more ANSI-compliant in a number of
4337 ways. Martin Atzmueller is responsible for a lot of this.
4338 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4340 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4341 *** correct ERROR type for various file operations
4342 *** some fixes for Lisp streams
4343 *** DEFMETHOD syntax checking
4344 *** changing old weird representation of debug information as
4345 strings (which, among their other deficiencies, don't transform
4346 correctly when you rename packages, and don't change their
4347 print representation when you change things like *PACKAGE*
4348 and *PRINT-LENGTH*) to symbols and lists of symbols
4349 He also made several improvements and fixed several bugs in DESCRIBE.
4350 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4352 *** misbehavior of WRITE-STRING/WRITE-LINE
4353 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4354 and several other LOOP problems as well
4355 *** DIRECTORY when similar filenames are present
4356 *** DEFGENERIC with :METHOD options
4357 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4358 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4359 *** argument ordering in FIND with :TEST option
4360 *** mishandled package designator argument in APROPOS-LIST
4361 *** various problems in the backquote readmacro
4362 *** a bug in APROPOS
4363 *** probably some others that I'm not describing very well here,
4364 since the CVS log documents them by reference to sbcl-devel
4365 messages, and the SourceForge archives aren't working well.:-(
4366 ** Dan Barlow improved the Alpha port (and is making progress on the
4367 PPC port, for those of you who think different).
4368 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4369 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4370 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4371 continues to work on the SPARC port (for those of you in a position
4372 to look down upon our little PC-compatible boxes from a great height).
4373 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4374 and uses that copy, rather than the current dynamic value, when
4375 it's trying to decide whether to truncate output. Thus e.g.
4376 (let ((*print-lines* 50))
4377 (pprint-logical-block (stream nil)
4379 (let ((*print-lines* 8))
4380 (print (aref possiblybigthings i) stream)))))
4381 should now truncate the logical block only at 50 lines, instead of
4382 often truncating it at 8 lines, as it did before.
4383 * The doc/cmucl/ directory, containing old CMU CL documentation
4384 from the time of the fork, is no longer part of the base system.
4385 SourceForge has shut down its anonymous FTP service, and with it
4386 my original plan for distributing the old CMU CL documentation
4387 there. For now, if you need these files you can download an old
4388 SBCL source release and extract them from it.
4389 * The fasl file version number changed again, for dozens of reasons,
4390 some of which are apparent above.
4392 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4393 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4394 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4395 patch, so that DISASSEMBLE output is much nicer.
4396 * The code in the SB-PROFILE package now seems reasonably stable.
4397 I still haven't decided what the final interface should look like
4398 (I'd like PROFILE to interact cleanly with TRACE, since both
4399 facilities use function encapsulation) but if you have a need
4400 for profiling now, you can probably use it successfully with
4401 the current CMU-CL-style interface.
4402 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4403 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4404 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4405 link now returns the dangling link itself, and for similar
4406 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4407 link itself. (In these cases the old code signalled an error and
4408 looped endlessly, respectively.) Thus, DIRECTORY now works even
4409 in the presence of dangling and cyclic symbolic links.
4410 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4411 is now a supported extension again, since the consensus on
4412 sbcl-devel was that it can be useful for ordinary development
4413 work, not just for debugging SBCL itself.
4414 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4415 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4416 that currently-defined functions might be redefined later with
4417 different return types.
4418 * Hash tables can be printed readably, as inspired by CMU CL code
4419 of Eric Marsden and SBCL code of Martin Atzmueller.
4420 * better error handling in CLOS method combination, thanks to
4421 Martin Atzmueller porting Pierre Mai's CMU CL patches
4422 * more overflow fixes for >16Mbyte I/O buffers
4423 * A bug in READ has been fixed, so that now a single Ctrl-D
4424 character suffices to cause end-of-file on character streams.
4425 In particular, now you only need one Ctrl-D at the command
4426 line (not two) to exit SBCL.
4427 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4429 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4430 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4431 does the right thing.
4432 * The implementation of some type tests, especially for CONDITION
4433 types, is now tidier and maybe faster, due to CMU CL code
4434 originally by Douglas Crosher, ported by Martin Atzmueller.
4435 * Some math functions have been fixed, and there are new
4436 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4437 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4438 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4439 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4440 likely to be built into sbcl-0.7.0.)
4441 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4442 CL, ported to SBCL by Martin Atzmueller)
4443 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4444 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4445 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4446 * There's a new slam.sh hack to shorten the edit/compile/debug
4447 cycle for low-level changes to SBCL itself, and a new
4448 :SB-AFTER-XC-CORE target feature to control the generation of
4449 the after-xc.core file needed by slam.sh.
4450 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4451 COMPILE-FILE is no longer supported, so that now every function
4452 gets an entry point, so that block compilation looks a little
4453 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4454 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4455 returns the number of bytes consed since the system started,
4456 rather than the number consed since the first time the function
4457 was called. (The new definition parallels ANSI functions like
4458 CL:GET-INTERNAL-RUN-TIME.)
4459 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4460 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4461 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4462 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4463 * Fasl file version numbers are now independent of the target CPU,
4464 since historically most system changes which required version
4465 number changes have affected all CPUs equally. Similarly,
4466 the byte fasl file version is now equal to the ordinary
4469 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4470 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4471 is no longer recognized. I apologize for this, because it was
4472 listed in SB-EXT as a supported extension, but I found that
4473 its existing behavior was poorly specified, as well as incorrectly
4474 specified, and it looked like too much of a mess to straighten it
4475 out. I have enough on my hands trying to get ANSI stuff to work..
4476 * many patches ported from CMU CL by Martin Atzmueller, with
4477 half a dozen bug fixes in pretty-printing and the debugger, and
4478 half a dozen others elsewhere
4479 * fixed bug 13: Floating point infinities are now supported again.
4480 They might still be a little bit flaky, but thanks to bug reports
4481 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4482 as flaky as they were.
4483 * The --noprogrammer command line option is now supported. (Its
4484 behavior is slightly different in detail from what the old man
4485 page claimed it would do, but it's still appropriate under the
4486 same circumstances that the man page talks about.)
4487 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4488 are now supported, and enabled by default. Thus, the compiler can
4489 handle many floating point and complex operations much less
4490 inefficiently. (Thus e.g. you can implement a complex FFT
4492 * The compiler now detects type mismatches between DECLAIM FTYPE
4493 and DEFUN better, and implements CHECK-TYPE more correctly, and
4494 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4495 thanks to patches from Martin Atzmueller.
4496 * various fixes to make the cross-compiler more portable to
4497 ANSI-conforming-but-different cross-compilation hosts (notably
4498 Lispworks for Windows, following bug reports from Arthur Lemmens)
4499 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4500 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4501 fixed thanks to Pierre Mai's CMU CL patches.
4502 * fixes to keep the system from overflowing internal counters when
4503 it tries to use i/o buffers larger than 16M bytes
4504 * fixed bug 45a: Various internal functions required to support
4505 complex special functions have been merged from CMU CL sources.
4506 (When I was first setting up SBCL, I misunderstood a compile-time
4507 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4508 * improved support for type intersection and union, fixing bug 12
4509 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4510 more obscure bugs as well
4511 * some steps toward byte-compiling non-performance-critical
4512 parts of the system, courtesy of patches from Martin Atzmueller
4513 * Christophe Rhodes has made some debian packages of sbcl at
4514 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4515 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4516 stable, but are nonetheless usable. When he's ready, I'd be happy
4517 to add them to the SourceForge "File Releases" section. (And if
4518 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4519 * new fasl file format version number (because of changes in
4520 internal representation of (OR ..) types to accommodate the new
4521 support for (AND ..) types, among other things)
4523 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4524 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4526 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4528 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4529 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4530 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4531 * DESCRIBE now works on structure objects again.
4532 * Most function call argument type mismatches are now handled as
4533 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4534 know whether the function will be redefined before the call is
4535 executed. (The compiler could flag local calls with full WARNINGs,
4536 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4537 it doesn't keep track of enough information to know whether calls
4538 are local in this sense.)
4539 * Compiler output is now more verbose, with messages truncated
4540 later than before. (There should be some supported way for users
4541 to override the default verbosity, but I haven't decided how to
4542 provide it yet, so this behavior is still controlled by the internal
4543 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4544 src/compiler/ir1util.lisp.)
4545 * Fasl file format version numbers have increased again, because
4546 support for the Gray streams extension changes the layout of the
4547 system's STREAM objects.
4548 * The Gray subclassable streams extension now works, thanks to a
4549 patch from Martin Atzmueller.
4550 * The full LOAD-FOREIGN extension (not just the primitive
4551 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4552 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4553 but like most other programs, it defaults to copying the Unix
4554 environment from the original process instead of starting the
4555 new process in an empty environment.
4556 * Extensions which manipulate the Unix environment now support
4557 an :ENVIRONMENT keyword option which doesn't smash case or
4558 do other bad things. The CMU-CL-style :ENV option is retained
4559 for porting convenience.
4560 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4561 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4563 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4565 * A patch from Martin Atzmueller seems to have solved the SIGINT
4566 problem, and as far as we know, signal-handling now works cleanly.
4567 (If you find any new bugs, please report them!)
4568 * The system no longer defaults Lisp source file names to types
4569 ".l", ".cl", or ".lsp", but only to ".lisp".
4570 * The compiler no longer uses special default file extensions for
4571 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4572 seems to expect a single default extension for all compiled code,
4573 and there's no compelling reason to try to stretch the standard
4574 to allow two different extensions.) Instead, byte-compiled files
4575 default to the same extension as native-compiled files.
4576 * Fasl file format version numbers have increased again, because
4577 a rearrangement of internal implementation packages made some
4578 dumped symbols in old fasl files unreadable in new cores.
4579 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4580 many fewer weird special cases.
4581 * Bug #17 (differing COMPILE-FILE behavior between logical and
4582 physical pathnames) has been fixed, and some related misbehavior too,
4583 thanks to a patch from Martin Atzmueller.
4584 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4585 by Tim Moore, ported to SBCL by Martin Atzmueller.
4586 * Martin Atzmueller fixed several filesystem-related problems,
4587 including bug #36, in part by porting CMU CL patches, which were
4588 written in part by Paul Werkowski.
4589 * More compiler warnings in src/runtime/ are gone, thanks to
4590 more patches from Martin Atzmueller.
4591 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4594 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4596 * DESCRIBE now works on CONDITION objects.
4597 * The debugger now handles errors which arise when trying to print
4598 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4600 * The build system now uses an additional file, customize-target-features.lisp,
4601 to allow local modifications to the target *FEATURES* list. (The point of
4602 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4603 debugging features enabled, without having to worry about propagating it
4604 into everyone's system when I do a "cvs update".) When no
4605 customize-target-features.lisp file exists, the target *FEATURES* list
4606 should be constructed the same way as before.
4607 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4608 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4609 warn about behavior which is undefined under the ANSI spec. Note: This
4610 is specified by ANSI, but it's not very popular with programmers.
4611 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4612 macro in the SBCL sources for an example of a workaround which you
4613 might use to make portable ANSI-standard code which does what you want.
4614 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4615 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4616 when it's used as a non-toplevel form. (This is required in order
4617 to implement the DEFCONSTANT-EQX macro.)
4618 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4619 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4620 funny things, due to bugs in the implementation of EVAL-WHEN
4621 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4622 non-toplevel DEFCONSTANTs are uncommon.)
4623 * The core file version number and fasl file version number have been
4624 incremented, because the old noncompliant DEFCONSTANT behavior involved
4625 calling functions which no longer exist, and because I also took the
4626 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4627 * fixed bug 1 (error handling before read-eval-print loop starts), and
4628 redid debugger restarts and related debugger commands somewhat while
4630 ** The QUIT debugger command is gone, since it did something
4631 rather different than the SB-EXT:QUIT command, and since it never
4632 worked properly outside the main toplevel read/eval/print loop.
4633 Invoking the new TOPLEVEL restart provides the same functionality.
4634 ** The GO debugger command is also gone, since you can just invoke
4635 the CONTINUE restart directly instead.
4636 ** The TOP debugger command is also gone, since it's redundant with the
4637 FRAME 0 command, and since it interfered with abbreviations for the
4639 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4641 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4642 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4643 some porting work by Martin Atzmueller)
4644 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4645 with a boatload of patches which he ported to SBCL. Now that those
4647 ** The system tries to make sure that its low-priority messages
4648 are prefixed by semicolons, to help people who like to use
4649 syntax highlighting in their ILISP buffer. (This patch
4650 was originally due to Raymond Toy.)
4651 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4652 patches originally written by Raymond Toy.
4653 ** The compiler understands coercion between single-value and
4654 multiple-VALUES type expressions better, getting rid of some very
4655 weird behavior, thanks to patches originally by Robert MacLachlan
4656 and Douglas Crosher.
4657 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4658 lambda lists, thanks to a patch originally by Pierre Mai.
4659 ** The system no longer bogusly warns about "abbreviated type
4661 ** The compiler gets less confused by inlining and RETURN-FROM,
4662 thanks to some patches originally by Tim Moore.
4663 ** The system no longer hangs when dumping circular lists to fasl
4664 files, thanks to a patch originally from Douglas Crosher.
4665 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4666 undefined function error.
4667 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4668 in structure slot initforms) for now, documented workaround instead:-|
4669 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4670 * fixed bug 5: added stubs for various Gray stream functions called
4671 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4672 installed, at least appropriate type errors are generated
4673 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4674 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4675 consistently in DEFMETHOD forms.
4676 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4677 it doesn't seem to affect SBCL after all
4678 * The C runtime system now builds with better optimization and many
4679 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4681 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4683 * The system is now under CVS at SourceForge (instead of the
4684 CVS repository on my home machine).
4685 * The new signal handling code has been tweaked to treat register
4686 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4687 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4688 downstream, e.g. in debug-int.lisp, has implicit dependencies
4689 on the unsignedness of integer representation of machine words,
4690 and that caused the system to bomb out with infinite regress
4691 when trying to recover from type errors involving signed values,
4692 e.g. (BUTLAST '(1 2 3) -1).)
4693 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4694 (This was one of the bugs Peter Van Eynde reported back in July.)
4695 * The system now uses code inspired by Colin Walters' O(N)
4696 implementation of MAP (from the cmucl-imp@cons.org mailing
4697 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4698 inline the MAP operation, and there is more than one
4699 sequence argument to the MAP call (so that it can't just
4700 do ETYPECASE once and for all based on the type of the
4701 single sequence argument). (The old non-inline implementation
4702 of the general M-argument sequence-of-length-N case required
4703 O(M*N*N) time when any of the sequence arguments were LISTs.)
4704 * The QUIT :UNIX-CODE keyword argument has been renamed to
4705 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4706 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4708 * Raymond Wiker's patches to port dynamic loading from Linux to
4709 FreeBSD have been added.
4710 * The BUGS file is now more nearly up to date, thanks in large part
4711 to Martin Atzmueller's review of it.
4712 * The debugger now flushes standard output streams before it begins
4713 its output ("debugger invoked" and so forth).
4714 * The core version number and fasl file version number have both
4715 been incremented, because of incompatible changes in the layout
4717 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4718 used to not be called for a saved Lisp image.)
4719 * Martin Atzmueller's version of a patch to fix a compiler crash,
4720 as posted on sbcl-devel 13 September 2000, has been installed.
4721 * Instead of installing Martin Atzmueller's patch for the
4722 compiler transform for SUBSEQ, I deleted the compiler transform,
4723 and transforms for some similar consing operations.
4724 * A bug in signal handling which kept TRACE from working on OpenBSD
4726 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4727 away by constant folding
4728 * The system now defines its address space constants in one place
4729 (in the Lisp sources), and propagates them automatically elsewhere
4730 (through GENESIS and the sbcl.h file). Therefore, patching the
4731 address map is less unnecessarily tedious and error-prone. The
4732 Lisp names of address space constants have also been systematized.
4733 * CVS tags like dollar-Header-dollar have been removed from
4734 the sources, because they have never saved me trouble and
4735 they've been source of trouble working with patches and other
4736 diff-related operations.
4737 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4738 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4740 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4742 * The system has been ported to OpenBSD.
4743 * The system now compiles with a simple "sh make.sh" on the systems
4744 that it's supported on. I.e., now you no longer need to tweak
4745 text in the target-features.lisp-expr and symlinks in src/runtime/
4746 by hand, the make.sh takes care of it for you.
4747 * The system is no longer so grossly inefficient when compiling code
4748 involving vectors implemented as general (not simple) vectors (VECTOR T),
4749 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4750 which dares to use the various sequence functions on non-simple
4751 vectors, takes less of a performance hit.
4752 * There is now a primitive type predicate VECTOR-T-P
4753 to test for the (VECTOR T) type, so that e.g.
4754 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4755 can now be compiled with some semblance of efficiency. (The old code
4756 turned the type declaration into a full call to %TYPEP at runtime!)
4757 * AREF on (VECTOR T) is still not fast, since it's still compiled
4758 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4759 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4760 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4761 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4762 without making full calls to SUBTYPEP at runtime.
4763 (Some analogous efficiency issues for non-simple vectors specialized to
4764 element types other than T, or for non-simple multidimensional arrays,
4765 have not been addressed. They could almost certainly be handled the
4766 same way if anyone is motivated to do so.)
4767 * The changes in array handling break binary compatibility, so
4768 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4769 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4770 instead of (VALUES T T).
4771 * By following the instructions that Dan Barlow posted to sbcl-devel
4772 on 2 July 2000, I was able to enable primitive dynamic object
4773 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4774 functionality is not implemented (since it calls ld to resolve
4775 library references automatically, requiring RUN-PROGRAM for its
4776 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4777 try to resolve library references) is now supported.
4778 * The system now flushes the standard output streams when it terminates,
4779 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4780 them at several other probably-convenient times, e.g. in each pass of
4781 the toplevel read-eval-print loop, and after evaluating a form given
4782 as an "--eval" command-line option. (These changes were motivated by a
4783 discussion of stream flushing issues on cmucl-imp in August 2000.)
4784 * The source transform for TYPEP of array types no longer assumes
4785 that an array whose element type is a not-yet-defined type
4786 is implemented as an array of T, but instead punts, so that the
4787 type will be interpreted at runtime.
4788 * There is now some support for cross-compiling in make.sh: each of
4789 the phases of make.sh has its own script. (This should be transparent
4790 to people doing ordinary, non-cross-compile builds.)
4791 * Since my laptop doesn't have hundreds of megabytes of memory like
4792 my desktop machine, I became more motivated to do some items on
4793 my to-do list in order to reduce the size of the system a little:
4794 ** Arrange for various needed-only-at-cold-init things to be
4795 uninterned after cold init. To support this, those things have
4796 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4797 symbols with such names are now uninterned after cold init).
4798 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4799 fasl files for cold load.
4800 ** Remove the old compiler structure pooling code (which used to
4801 be conditional on the target feature :SB-ALLOC) completely.
4802 ** Redo the representation of some data in cold init to be more compact.
4803 (I also looked into supporting byte compiled code at bootstrap time,
4804 which would probably reduce the size of the system a lot, but that
4805 looked too complicated, so I punted for now.)
4806 * The maximum signal nesting depth in the src/runtime/ support code has
4807 been reduced from 4096 to 256. (I don't know any reason for the very
4808 large old value. If the new smaller value turns out to break something,
4809 I'll probably just bump it back up.)
4810 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4812 * Many, many bugs reported by Peter Van Eynde have been added to
4813 the BUGS list; some have even been fixed.
4814 * While enabling dynamic object file loading, I tried to make the
4815 code easier to understand, renaming various functions and variables
4816 with less ambiguous names, and changing some function calling
4817 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4818 * While trying to figure out how to do the OpenBSD port, I tried to
4819 clean up some of the code in src/runtime/. In particular, I dropped
4820 support for non-POSIX signal handling, added various comments,
4821 tweaked the code to reduce the number of compilation warnings, and
4822 renamed some files to increase consistency.
4823 * To support the new automatic configuration functionality in make.sh,
4824 the source file target-features.lisp-expr has been replaced with the
4825 source file base-target-features.lisp-expr and the machine-generated
4826 file local-target-features.lisp-expr.
4827 * fixed a stupid quoting error in make.sh so that using CMU CL
4828 "lisp -batch" as cross-compilation host works again
4830 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4832 * DESCRIBE no longer tries to call itself recursively to describe
4833 bound/fbound values, so that it no longer fails on symbols which are
4834 bound to themselves (like keywords, T, and NIL).
4835 * DESCRIBE now works on generic functions.
4836 * The printer now prints less-screwed-up representations of closures
4837 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4838 * A private symbol is used instead of the :EMPTY keyword previously
4839 used to mark empty slots in hash tables. Thus
4840 (DEFVAR *HT* (MAKE-HASH-TABLE))
4841 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4842 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4843 now does what ANSI says that it should. (You can still get
4844 similar noncompliant behavior if bang on the hash table
4845 implementation with all the symbols you get back from
4846 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4847 This breaks binary compatibility, since tests for equality to
4848 :EMPTY are wired into things like the macroexpansion of
4849 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4851 * There's now a minimal placeholder implementation for CL:STEP,
4852 as required by ANSI.
4853 * An obscure bug in the interaction of the normal compiler, the byte
4854 compiler, inlining, and structure predicates has been patched
4855 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4856 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4858 * Missing ordinary arguments in a macro call are now detected even
4859 when the macro lambda list contains &KEY or &REST.
4860 * The debugger no longer complains about encountering the top of the
4861 stack when you type "FRAME 0" to explicitly instruct it to go to
4862 the top of the stack. And it now prints the frame you request even
4863 if it's the current frame (instead of saying "You are here.").
4864 * As specified by ANSI, the system now always prints keywords
4865 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4867 * The default initial SIZE of HASH-TABLEs is now smaller.
4868 * Type information from CLOS class dispatch is now propagated
4869 into DEFMETHOD bodies, so that e.g.
4870 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4872 is now basically equivalent to
4873 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4874 (DECLARE (TYPE SINGLE-FLOAT X))
4876 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4877 operation, without having to do run-time type dispatch.
4878 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4879 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4880 * Since I don't seem to be making big file reorganizations very often
4881 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4882 is overflowing my ability to conveniently back them up), I've finally
4883 checked the system into CVS. (The CVS repository is on my home system,
4884 not at SourceForge -- putting it on SourceForge might come later.)
4885 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4886 high-level GC-NOTIFY-FOO functions send their output. (There's
4887 still very little control of where low-level verbose GC functions
4888 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4889 less than it used to -- the GC-NOTIFY-FOO functions are now under
4890 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4891 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4892 in only one place in the source code, and propagates it automatically
4893 everywhere that it's needed. Thus e.g. when I bump the version from
4894 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4895 * The C source files now include boilerplate legalese and documentation
4896 at the head of each file (just as the Lisp source files already did).
4897 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4898 to his page will be replaced with a link to his new CLiki service.
4900 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4902 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4903 * The build process now looks for GNU make under the default name "gmake",
4904 instead of "make" as it used to. If GNU make is not available as "gmake"
4905 on your system, you can change this default behavior by setting the
4906 GNUMAKE environment variable.
4907 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4908 can build without error under CMU CL.
4910 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4912 * There is now a partial SBCL user manual (with some new text and some
4913 text cribbed from the CMU CL manual).
4914 * The beginnings of a profiler have been added (starting with the
4915 CMU CL profiler and simplifying and cleaning up). Eventually the
4916 main interface should be through the TRACE macro, but for now,
4917 it's still accessed through vaguely CMU-CL-style functions and macros
4918 exported from the package SB-PROFILE.
4919 * Some problems left over from porting CMU CL to the new
4920 cross-compilation bootstrap process have been cleaned up:
4921 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4922 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4923 part when getting rid of the compiler *BACKEND* stuff.)
4924 ** The value of *NULL-TYPE* was screwed up, because it was
4925 being initialized before the type system knew the final
4926 definition of the 'NULL type. This screwed up several key
4927 optimizations in the compiler, causing inefficiency in all sorts
4928 of places. (I found it because I wanted to understand why
4929 GET-INTERNAL-RUN-TIME was consing.)
4930 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4931 PROCLAIM FTYPE information. Unfortunately this broke binary
4932 compatibility again, since now the forms output by DEFGENERIC
4933 to refer to functions which didn't exist in 0.6.3.
4934 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4935 can use the new (as of 0.6.3) transform for SOME into MAP into
4937 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4938 values to (INTEGER 0 1000000), so that the time code will no longer
4939 occasionally get blown up by Linux returning 1000000 microseconds
4940 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4941 its output conform to the ANSI spec. (Alas, this makes its output
4942 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4943 * A full call to MAP NIL with a single sequence argument no longer conses.
4944 * fixes to problems pointed out by Martin Atzmueller:
4945 * The manual page no longer talks about multiprocessing as though
4946 it were currently supported.
4947 * The ILISP support patches have been removed from the distribution,
4948 because as of version 5.10.1, ILISP now supports SBCL without us
4949 having to maintain patches.
4950 * added a modified version of Raymond Toy's recent CMU CL patch for
4951 EQUALP comparison of HASH-TABLE
4953 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4955 * The system still can't cross-compile itself with
4956 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4957 entails), but at least it can compile more of itself that way
4958 than it used to be able to, and various buglets which were uncovered
4959 by trying to cross-compile itself that way have now been fixed.
4960 * This release breaks binary compatibility again. This time
4961 at least I've incremented the FASL file format version to 2, so that the
4962 problem can be detected reliably instead of just causing weird errors.
4963 * various new style warnings:
4964 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4965 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4966 ** using the lexical binding of a variable named in the *FOO* style
4967 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4968 as specified by ANSI.
4969 * *RANDOM-STATE* is no longer automatically initialized from
4970 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4971 default behavior of the system is to repeat its behavior every time
4972 it's run. If you'd like to change this behavior, you can always
4973 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4974 old convention there was no comparably easy way to get the system to
4975 repeat its behavior every time it was run.
4976 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4977 FTYPE declarations has been removed, in favor of their ANSI
4978 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4979 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4980 the types of their sequence arguments can be determined at compile time.
4981 This is done through a new open code expansion for MAP which eliminates
4982 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4983 argument types can be determined at compile time.
4984 * The optimizer now transforms COERCE into an identity operation when it
4985 can prove that the coerced object is already of the correct type. (This
4986 can be a win for machine generated code, including the output of other
4987 optimization transforms, such as the MAP transform above.)
4988 * Credit information has been moved from source file headers into CREDITS.
4989 * Source file headers have been made more standard.
4990 * The CASE macro now compiles without complaining even when it has
4993 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4995 * (Note that the way that the PCL macroexpansions were rewritten
4996 to accommodate the change in DEFGENERIC below breaks binary
4997 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4998 not run under sbcl-0.6.2. Once we get out of alpha releases,
4999 i.e. hit release 1.0.0, we'll probably try to maintain binary
5000 compatibility between maintenance releases, e.g. between sbcl-1.4.3
5001 and sbcl-1.4.4. Until then, however, it might be fairly common
5002 for maintenance releases to break binary compatibility.)
5003 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
5005 * The reporting of unhandled signals has been changed to print some
5006 explanatory text as well as the report form. (Previously only
5007 the report form was printed.)
5008 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
5009 it defines, so that the compiler no longer issues undefined function
5010 warnings for compiled-but-not-yet-loaded generic functions.
5011 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
5012 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
5013 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
5014 section of the ANSI spec are supported.
5015 * The "" nickname for the "KEYWORD" package has been removed.
5016 The reader still handles symbol tokens which begin with a package marker
5017 as keywords, but it doesn't expose its mechanism for doing so in the
5018 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
5019 * The system now issues STYLE-WARNINGs for contradictory TYPE
5020 proclamations. (Warnings for contradictory FTYPE proclamations would
5021 be nice too, but those can't be done usefully unless the type system
5022 is made smarter about FUNCTION types.)
5023 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
5024 systematized, so that "*target-*.lisp is supposed to exist only on the
5025 target and imply that there's a related file which exists on the
5026 host, and *host-*.lisp is supposed to exist only on the host and imply
5027 that there's a related file which exists on the target. This involves a
5028 lot of renaming. Hopefully the acute confusion caused by the renaming
5029 will be justified by the reduction in chronic confusion..
5030 ** runtime-type.lisp -> early-target-type.lisp
5031 ** target-type.lisp -> late-target-type.lisp
5032 ** early-host-format.lisp -> early-format.lisp
5033 ** late-host-format.lisp -> late-format.lisp
5034 ** host-error.lisp -> misc-error.lisp
5035 ** early-error.lisp -> early-target-error.lisp
5036 ** late-error.lisp -> late-target-error.lisp
5037 ** host-defboot.lisp -> early-defboot.lisp
5038 ** code/misc.lisp -> code/target-misc.lisp
5039 ** code/host-misc.lisp -> code/misc.lisp
5040 ** code/numbers.lisp -> code/target-numbers.lisp
5041 ** code/early-numbers.lisp -> numbers.lisp
5042 ** early-host-type.lisp -> early-type.lisp
5043 ** late-host-type.lisp -> late-type.lisp
5044 ** host-typep.lisp -> typep.lisp
5045 ** load.lisp -> target-load.lisp
5046 ** host-load.lisp -> load.lisp
5047 ** host-disassem.lisp -> disassem.lisp
5048 ** host-insts.lisp -> insts.lisp
5049 ** byte-comp.lisp -> target-byte-comp.lisp
5050 ** host-byte-comp.lisp -> byte-comp.lisp
5051 ** host-signal.lisp -> signal.lisp
5052 ** host-defstruct.lisp -> defstruct.lisp
5053 ** late-target-type.lisp -> deftypes-for-target.lisp
5054 Furthermore, several other previously target-only files foo.lisp (e.g.
5055 hash-table.lisp and random.lisp) have been split into a target-and-host
5056 foo.lisp file and a target-only target-foo.lisp file, with their key type
5057 definitions in the target-and-host part, so that the cross-compiler will
5058 know more about target types.
5059 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5060 gone away. In their place are various *BACKEND-FOO* variables
5061 corresponding to the slots of the old structure.
5062 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5063 into the target SBCL has been fixed.
5064 * The chill.lisp system for loading cold code into a running SBCL
5066 * Support for the CMU CL "scavenger hook" extension has been removed.
5067 (It was undocumented and unused in the CMU CL sources that SBCL was
5068 derived from, and stale in sbcl-0.6.1.)
5069 * Various errors in the cross-compiler type system were detected
5070 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5071 (enabling various consistency checks). Many of them were fixed,
5072 but some hard problems remain, so the compiler is back to
5073 running without *TYPE-SYSTEM-INITIALIZED* for now.
5074 * As part of the cross-compiler type system cleanup, I implemented
5075 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5076 * I have started adding UNCROSS calls throughout the type system
5077 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5078 on cross-compiler input files will be able to go away, and various
5080 * CONSTANTP now returns true for quoted forms (as explicitly required
5083 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5085 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5086 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5087 (MAKE-LIST -1). In the longer term, it ought to become true
5088 that declarations are assertions even at SAFETY 1. For now, it's not
5089 quite true even at SAFETY 3, but it's at least more nearly true..
5090 (Note that this change seems to increases the size of the system by
5091 O(5%) and to decrease the speed of the compiler by 20% or more.)
5092 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5093 for the problem of printing dozens of lines of distracting information
5094 about low-level system machinery as part of the top stack frame
5095 on entry to the debugger when an undefined function was called.
5096 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5097 is not reset to COMMON-LISP-USER.
5098 * Compilation of stuff related to dyncount.lisp has been made conditional
5099 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5100 smaller. The various dyncount-related symbols have been moved into
5101 a new "SB-DYNCOUNT" package.
5102 * tty-inspect.lisp has been renamed to inspect.lisp.
5103 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5104 feature has gone away. (When we eventually port to other flavors of
5105 libc and/or Unix, we'll try to make the differences between flavors
5106 invisible at the user level.)
5107 * Various other *FEATURES* tags, and/or their associated conditionals,
5108 have been removed if obsolescent, or given better documentation, or
5109 sometimes given more-mnemonic names.
5111 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5113 * tidied up "make.sh" script
5114 * tidied up system directory structure
5115 * better "clean.sh" behavior
5116 * added doc/FOR-CMUCL-DEVELOPERS
5117 * many many small tweaks to output format, e.g. removing possibly-confusing
5118 trailing #\. character in DESCRIBE-INSTANCE
5119 * (EQUALP #\A 'A) no longer signals an error.
5120 * new hashing code, including EQUALP hashing
5121 * tidied up Lisp initialization and toplevel
5122 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5123 * command line argument processing
5124 * added POSIX-GETENV function to deal with Unix-ish environment variables
5125 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5126 terminating SBCL on EOF
5127 * non-verbose GC by default
5128 * There is no more "sbcl" shell script; the sbcl file is now the C
5129 runtime executable (just like CMU CL).
5130 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5132 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5133 symbol and package information as Lisp native symbol and package objects
5134 instead of strings naming symbols and strings naming packages. This way,
5135 whenever packages are renamed (as in warm init), debug information is
5136 transformed along with everything else.
5137 * tweaked the optimization policy declarations which control the building
5138 of SBCL itself. Now, among other things, the system no longer saves
5139 source location debugging information. (This helps two problems at once
5140 by reducing SBCL size and by keeping SBCL from trying to look for its
5141 sources -- which may not exist -- when reporting errors.)
5142 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5143 debugging and testing purposes
5144 * cleaned up printing, making the printer call PRINT-OBJECT for
5145 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5146 methods, giving nearly-ANSI behavior
5147 * converted almost all special variables to use *FOO* naming convention
5148 * deleted PARSE-TIME functionality, since it can be done portably
5149 * moved some files out of cold init into warm init
5150 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5152 * regularized formatting of source files
5153 * added an install.sh script
5154 * fixed ridiculous memory usage of cross-compiler by making
5155 compiler/alloc.lisp not try to do pooling unless it can hook
5156 itself into the GC of the cross-compilation host. Now the system
5157 builds nicely on my old laptop.
5158 * added :SB-ALLOC in target-features.lisp-expr
5159 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5160 was not implemented)
5161 * re-did condition handling and note reporting in the compiler. Notes
5162 are no longer handled by signalling conditions. Style warnings
5163 and warnings are handled more correctly and reported in such a way
5164 that it's easy to find one or the other in your output (so that you
5165 can e.g. figure out which of many problems caused COMPILE-FILE to
5167 * changed the severity of several compiler warnings from full WARNING
5168 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5169 compiler note reporting so that it doesn't use the condition system
5170 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5171 COMPILE-FILE command)
5172 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5173 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5174 no longer supported, since their implementation was deeply intertwingled
5175 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5176 * removed bogus "support" for compiler macros named (SETF FOO), and
5177 removed the compiler macro for SETF INFO (but only after making a fool
5178 of myself on the cmucl-imp mailing list by posting a bogus patch for
5179 DEFINE-COMPILER-MACRO..)
5180 * Compiled files containing forms which have side effects on the Lisp
5181 reader (such as DEFPACKAGE forms) are now handled more correctly.
5182 (Compiler queuing of top level lambdas has been suppressed by setting
5183 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5184 * deleted various currently-unused source files, e.g. gengc.lisp. They
5185 may be added back at some point e.g. when porting to other architectures,
5186 but until they are it's distracting to distribute them and to try to
5188 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5189 there were so many of them they're just distractions, and UNCROSS is
5190 known to be able to handle the current sources
5191 * moved PROFILE functionality into TRACE, so that it will be clear
5192 how the wrapping and unwrapping of functions when you profile them
5193 interacts with the wrapping and unwrapping of functions when you
5194 trace them. (Actually, the functionality isn't there yet, but at least
5195 the interface specification is there. Hopefully, the functionality will
5196 arrive with some future maintenance release.)
5197 * removed host-oops.lisp
5198 * changed signature of QUIT function to allow UNIX-CODE argument
5199 * fixed READ-SEQUENCE bug
5200 * tweaked verbose GC output so that it looks more like the progress
5201 output that ANSI specifies for functions like LOAD
5202 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5203 * added <http://sbcl.sourceforge.com> to the banner information printed by