1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 * minor incompatible change: SAVE-LISP-AND-DIE :TOPLEVEL function is now
3 allowed to return, which causes SBCL to quit with exit status 0. Previously
4 if the function returned with a small integer return value, that value
5 was accidentally reused as the exit status.
6 * new feature: SB-EXT:DEFGLOBAL macro allows defining global non-special
8 * new feature: SB-EXT:ALWAYS-BOUND proclamation inhibits MAKUNBOUND, and
9 allows the compiler to safely elide boundedness checks for special
11 * new feature: SB-EXT:GLOBAL proclamation inhibits SPECIAL proclamations for
12 the symbol, prohibits both lexical and dynamic binding. This is mainly an
13 efficiency measure for threaded platforms, but also valueable in
15 * optimization: accesses to potentially non-simple arrays where element type
16 is known are 50% faster.
17 * optimization: compiler now generates faster array typechecking code.
18 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
20 * optimization: multidimensional array accesses in the absence of type
21 information regarding array rank are approximately 10% faster due to
22 open coding of ARRAY-RANK.
23 * improvement: SBCL now emits a compiler note where stack allocation was
24 requested but could not be provided.
25 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
26 * improvement: pretty-printing loop has been implemented properly. (thanks
28 * documentation: CLOS slot typechecing policy has been documented.
29 * bug fix: better floating point exception handling on x86/OpenBSD.
30 (thanks to Josh Elsasser)
31 * bug fix: exit status from QUIT when called under --script was lost
32 (reported by Hubert Kauker)
33 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
34 same implementation of FILL to initialize the array, even if a more
35 efficient one was available (reported by Stas Boukarev, thanks to
37 * bug fix: potential miscompilation of array stack allocation on x86 and
38 x86-64. (reported by Time Tossavainen)
39 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
40 result in their subforms being treated as top level forms. (reported by
42 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
43 than a machine register could leave garbage in the high bits of the
44 result register (bug 316325).
45 * bug fix: disable address space randomization Linux/x86-64 as well,
46 not just x86-64. (reported by Ken Olum)
47 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
48 generate incorrect code.
49 * bug fix: #201; type inference for CONS and ARRAY types could derive
50 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
51 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
52 could not be assigned to or bound with PROGV. (reported by Lorenz
54 * bug fix: the value of CL:- in the inspector was the previous expression
55 evaluated rather than the expression being evaluated.
56 * bug fix: constants can no longer be locally declared special.
57 * bug fix: signals delivered to threads started from foreign land (read:
58 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
59 thread by blocking all signals and resignalling.
61 changes in sbcl-1.0.28 relative to 1.0.27:
62 * a number of bugs in cross-compilation have been fixed, with the ultimate
63 result that building under (at least) clisp should be much more reliable.
64 * minor incompatible changes: echo-streams now propagate unread-char to the
65 underlying input stream, and no longer permit unreading more than one
67 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
68 frames, allowing for instance more reliable backtraces.
69 * improvement: the debugger REPL can now reference lexical variables
70 by name directly for code compiled with (DEBUG 3).
71 * improvement: errors from malformed declarations now have better source
72 paths associated with them. (thanks to Tobias Rittweiler)
73 * optimization: faster local calls on x86/x86-64
74 * bug fix: some error messages for out-of-bound array indexes confused the
75 index and the bound. (thanks to Stas Boukarev)
76 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
78 * bug fix: running regressions tests in shells without OSTYPE set now works.
79 (reported by Harald Hanche-Olsen)
80 * bug fix: more robust static space exhaustion signalling from
81 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
82 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
83 docstring away instead of storing it under names such as (LAMBDA (X)).
84 (reported by Leslie Polzer)
85 * bug fix: timers could go off in the wrong order, be delayed indefinitely
86 (thanks to Ole Arndt for the patch)
87 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
88 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
89 potentially-invalid effective methods in its cache.
90 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
91 instances as well (thanks to Paul Khuong)
92 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
93 SB-POSIX:WAITPID and their C equivalents.
94 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
96 changes in sbcl-1.0.27 relative to 1.0.26:
97 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
98 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
99 * improvement: the system either recovers from stack exhaustion or dies
100 properly as opposed to leaving the user uncertain of whether the handler
101 trampled on some random memory next to the stack or having to rely on
102 --lose-on-corruption (which is still a good idea to use in production
103 because stack exhaustion can happen in signal handlers which will likely
105 * bug fix: fix gc related interrupt handling bug on ppc (regression from
106 1.0.25.37, reported by Harald Hanche-Olsen)
107 * bug fix: work around signal delivery bug in darwin (regression from
108 1.0.25.44, reported by Sidney Markowitz)
109 * bug fix: fix ERROR leaking memory (reported by David Thompson)
111 changes in sbcl-1.0.26 relative to 1.0.25:
112 * incompatible change: an interruption (be it a function passed to
113 INTERRUPT-THREAD or a timer function) runs in an environment where
114 interrupts can be enabled. The interruption can use
115 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
116 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
117 potentially running out of stack. Keep in mind that in the absence
118 of WITHOUT-INTERRUPTS some potentially blocking operation such as
119 acquiring a lock can enable interrupts.
120 * incompatible change: GC-OFF and GC-ON are removed, as they were
121 always unsafe. Use WITHOUT-GCING instead.
122 * new feature: runtime option --disable-ldb
123 * new feature: runtime option --lose-on-corruption to die at the
124 slightest hint of possibly non-recoverable errors: running out of
125 memory, stack, alien stack, binding stack, encountering a memory
126 fault, etc. In the absence of --lose-on-corruption a warning is
128 * enhancement: detect binding stack exhaustion
129 * enhancement: detect alien stack exhaustion on x86/x86-64
130 * improvement: generally more stable and reliable interrupt handling
131 * improvement: there is a per thread interruption queue,
132 interruptions are executed in order of arrival
133 * improvement: a repeating timer reschedules itself when the it has
134 finished, but expiration times are spaced equally. If an
135 expiration time is in the past it will trigger after a short grace
136 period that may give a chance to other things to run.
137 * optimization: slightly faster gc on multithreaded builds
138 * optimization: faster WITHOUT-GCING
139 * bug fix: when JOIN-THREAD signals an error, do it when not holding
140 important locks so that the debugger/handler doesn't produce
141 recursive errors or deadlock.
142 * bug fix: real-time signals are not used anymore, so no more
143 hanging when the system wide real-time signal queue gets full.
144 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
146 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
147 * bug fix: fix random memory faults related to interrupts on alpha
148 * bug fix: fix random memory faults related to interrupts on sparc
149 * bug fix: fix deadlocks related to starting threads
150 * bug fix: fix deadlines on locks on futex platforms
151 * bug fix: restore errno in signal handlers
152 * bug fix: fix deadlocks related to hash tables
153 * bug fix: fix deadlocks in pcl
155 changes in sbcl-1.0.25 relative to 1.0.24:
156 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
157 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
158 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
159 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
160 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
161 better name and does not return values so stale on multiprocessor systems.
162 Also, HOLDING-MUTEX-P was added for about the only sane usage of
164 * improvement: unithread builds keep track of MUTEX-VALUE.
165 * improvement: reading from a TWO-WAY-STREAM does not touch the output
166 stream anymore making it thread safe to have a concurrent reader and
167 a writer, for instance, in a pipe.
168 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
169 arguments, making compiler-macros for SETF-functions able to inspect
170 their constant arguments.
171 * improvement: COMPILE-FILE reports times with millisecond accuracy
172 (thanks to Luis Oliveira)
173 * optimization: CHAR-CODE type derivation has been improved, making
174 TYPEP elimination on subtypes of CHARACTER work better. (reported
175 by Tobias Rittweiler, patch by Paul Khuong)
176 * bug fix: setting alien structure fields of type struct by value now
177 computes the right offset for the memory copy.
178 * bug fix: compilation problem involving inlined calls to aliens with
179 result type VOID. (reported by Ken Olum)
180 * bug fix: #235a; sequential inline expasion in different policies no
181 longer reuses the functional from the previous expansion site.
182 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
183 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
184 * bug fix: named ENUMs in multiply-referenced alien record types no longer
185 cause an implied type redefinition. Regression from 1.0.21.29.
186 * bug fix: sign-extension of small signed return values now works with C
187 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
188 * improvements to the Windows port:
189 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
190 Westfold, thanks to Rudi Schlatte)
192 changes in sbcl-1.0.24 relative to 1.0.23:
193 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
194 vector of a multidimensional SIMPLE-ARRAY.
195 * new feature: the system now signals a continuable error if standard
196 readtable modification is attempted. (thanks to Tobias Rittweiler)
197 * new feature: DIRECTORY has been extended with a non-standard keyword
198 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
199 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
200 has been robustified and documented better. (thanks to Robert Goldman)
201 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
202 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
203 * tradeoff: constant FORMAT control strings are now compiled unless
204 SPACE > SPEED (previously only when SPEED > SPACE.)
205 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
207 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
209 * bug fix: lexical type declarations are now correctly reported by
210 SB-CLTL2. (reported by Larry D'Anna)
211 * bug fix: STRING-TO-OCTETS did not handle :START properly when
212 using UTF-8 as external format. (thanks to Luis Oliveira)
213 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
214 are signalled correctly. (thanks to Stas Boukarev)
215 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
216 designator. (thanks to Tobias Rittweiler)
217 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
218 designator, and returns T instead of the function. (thanks to
220 * bug fix: direct superclasses of STANDARD-CLASS and
221 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
222 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
223 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
224 when no class with the named slot yet exists no longer causes a
225 compile-time style-warning.
226 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
227 in safe code. (reported by Didier Verna)
228 * bug fix: #430; nested structure constructors can stack allocate.
229 * bug fix: on some 64-bit platforms dynamic space size was truncated
230 to #xffff0000 bytes. (reported by Benjamin Lambert)
231 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
232 REPL unusable. (reported by Daniel Herring)
233 * bug fix: return values of READ-SEQUENCE did not take :START into
234 account on file streams, regressions since 1.0.12.22. (reported by
235 Thomas Russ, patch by Paul Khuong)
236 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
237 method specializer used to confuse permuation vector optimization.
238 * bug fix: system inserted bogus implicit type declarations for local
239 special variables in DEFMETHOD bodies.
240 * bug fix: #354; duplicated frames in backtraces due to
241 non-tail-call-optimized XEPs to functions with return type NIL
243 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
244 initializes structure object slots according to DEFSTRUCT initforms,
245 and DEFSTRUCT forms :INCLUDEind structure classes defined using
246 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
247 (reported by Bruno Haible and Stephen Wilson)
248 * bug fix: #395; fill-pointer output streams used now support
249 element-type BASE-CHAR as well.
250 * bug fix: compiler error when attempting to derive return value of
251 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
253 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
255 * bug fix: Mac OS X binaries should now be portable between Leopard
258 changes in sbcl-1.0.23 relative to 1.0.22:
259 * enhancement: when disassembling method functions, disassembly
260 for the associated fast function is also produced.
261 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
263 * optimization: printing with *PRINT-PRETTY* true is now more
264 efficient as long as the object being printed doesn't require
265 special handling by the pretty printer.
266 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
267 now interact correctly with type declarations.
268 * partial bug fix: PCL detects infinite recursion during wrapper
269 validation. (thanks to Attila Lendvai)
270 * bug fix: #426; nested function calls are inlined properly.
271 Previously if FOO was an inline function, in calls of the form
272 (FOO (FOO ...)) the outer call was not inlined.
273 * bug fix: long long arguments passed in registers to alien
274 functions on PowerPC Linux and NetBSD are now handled correctly
275 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
277 changes in sbcl-1.0.22 relative to 1.0.21:
278 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
279 for the shared object in the current directory, but passes the native
280 namestring of the designated pathname to the operation system's shared
281 object loading function as-it.
282 * minor incompatible change: --disable-debugger toplevel option now takes
283 effect before processing of initialization files and --eval or --load
285 * new feature: new commandline argument: --script, which supports
286 shebang lines. See documentation for details. (based on work by
288 * new feature: SAVE-LISP-AND-DIE can save current values of
289 --dynamic-space-size and --control-stack-size in the executable core,
290 causing it to skip normal runtime option processing. See documentation
291 for details. (thanks to Zach Beane)
292 * enhancement: inoccous calls to EVAL or generic functions dispatching
293 on subclasses of eg. STREAM no longer cause compiler notes to appear.
294 * enhancement: the system no longer resignals errors from --load and
295 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
296 associated with the original error to be lost. (thanks to Ariel
298 * enhancement: :DONT-SAVE keyword argument has been added to
299 LOAD-SHARED-OBJECT for controlling interaction with
300 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
301 the effects of an earlier LOAD-SHARED-OBJECT call.
302 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
303 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
304 provided. (reported by Cedric St-Jean)
305 * bug fix: circularity handling in the reader did not treat raw
306 structure slots correctly. (reported by Cedric St-Jean)
307 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
308 file descriptors when there were none.
309 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
310 was broken. (reporteed by Willem Broekema)
311 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
312 pathnames without a directory.
313 * bug fix: COMPILE-FILE did not warn about undefined variable
314 references at toplevel, and LOAD of the resulting fasl did
316 * bug fix: functions with non-required arguments used to end up with
317 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
318 * bug fix: redefining a function with non-required arguments didn't
319 update the system's knowledge about its call signature properly.
320 * bug fix: fixed #431; incompatible alien record type redefinitions
321 are detected and handled. (reported by Neil Haven)
322 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
323 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
325 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
326 after alien stack frames.
327 * bug fix: strings passed to mk{s,d}temp lacked null termination
329 changes in sbcl-1.0.21 relative to 1.0.20:
330 * new feature: the compiler is able to track the effective type of a
331 generic function across method addition and removal.
332 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
333 appropriately typed structure slots without locking.
334 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
335 information like those gathered by TIME using a programming-friendly
337 * new feature: TIME reports time taken even if the form performs a
338 non-local transfer of control.
339 * enhancement: reduced conservativism on GENCGC platforms: on
340 average 45% less pages pinned (measured from SBCL self build).
341 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
342 mutate constant symbols or violate declaimed type of the symbol.
343 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
344 sleeping on the semaphore in heavily contested semaphores.
345 * bug fix: semaphores and condition variables were not interrupt
347 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
348 owned by other threads anymore.
349 * bug fix: FIND on lists called KEY outside the specified
350 subsequence. (reported by budden)
351 * bug fix: LOG doesn't use single-float intermediate results when
352 given mixed integer and double-float arguments, leading to better
353 precision. (reported by Bob Felts)
354 * bug fix: LOG with base zero returned values of inconsistent type.
356 changes in sbcl-1.0.20 relative to 1.0.19:
357 * minor incompatible change: OPTIMIZE qualities
358 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
359 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
360 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
361 * documentation: some slot access efficiency guidelines have been
362 added to the user manual.
363 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
364 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
365 as ASSOC and MEMEBER.
366 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
367 to more efficient EQ-comparison versions more often.
368 * optimization: enhanced derivation of DOLIST iteration variable type
370 * optimization: constant folding of simple (LIST ...) forms as DOLIST
372 * optimization: runtime lookup of function definitions can be
373 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
374 * optimization: compiler is able to derive the return type of
375 (AREF (THE STRING X) Y) as being CHARACTER.
376 * optimization: CLRHASH on empty hash-tables no longer does pointless
377 work. (thanks to Alec Berryman)
378 * optimization: REPLACE deftransforms don't punt when :START1 or
380 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
381 no longer emits a code deletion note. (reported by Andrew
383 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
384 functions with an odd number of &OPTIONAL arguments, a &REST
385 argument, and one or more &KEY arguments at the call site.
386 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
387 arguments appeared at call sites as well.
388 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
389 READ-LINE &co can no longer read from them afterwards. (reported
391 * bug fix: fixed #427: unused local aliens no longer cause compiler
392 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
394 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
395 alien-stack leakage. (reported by Andy Hefner)
396 * bug fix: PROGV signals an error when an attempt to violate declared
397 type of a variable or bind a constant is made.
398 * bug fix: SET signals an error when an attempt to violate declared
399 type of a variable is made.
400 * bug fix: restart computation during the execution of a restart
401 test function no longer causes infinite recursion. (reported by
403 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
404 signals an error. (thanks to Pierre Mai)
405 * bug fix: EXPORT left symbol unexported in conflict situations.
406 (thanks to Michael Weber)
407 * bug fix: correctly handle name conflicts involving CL:NIL.
408 (thanks to Michael Weber)
409 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
410 conflicts arising from USEing package with conflicting symbols
411 correctly. (thanks to Michael Weber)
413 changes in sbcl-1.0.19 relative to 1.0.18:
414 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
415 warnings that go otherwise unhandled will be muffled if they are
416 of the type that's the value of this variable.
417 * optimization: stack allocation is slightly more efficient on x86
419 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
420 where a raw slot always is initialized using the initform whose
421 type is not know sufficiently well a compile-time are now compiled
422 correctly. (reported by John Morrison)
423 * bug fix: compiler no longer makes erronous assumptions in the
424 presence of non-foldable SATISFIES types.
425 * bug fix: stack analysis missed cleanups of dynamic-extent
426 arguments in non-let-converted calls to local functions.
427 * improvements to the Windows port:
428 ** adjusted address spaces for building on both Win32 and
429 Win64. (thanks for John Connors)
430 * fixed some bugs revealed by Paul Dietz' test suite:
431 ** interval arithmetic during type derivation used inexact integer
432 to single-float coercions.
433 ** arithmetic operations involving large integers and single
434 floats give the same results in compiled and interpreted code.
435 ** deriving the result type of COERCE no longer signals an error
436 if the derived type of the second argument is a MEMBER type
437 containing invalid type specifiers.
438 ** ADJOIN with constant NIL as second argument works correctly.
439 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
441 changes in sbcl-1.0.18 relative to 1.0.17:
442 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
443 profiles only the current thread.
444 * minor incompatible change: changes to SYMBOL-VALUE of constants
445 defined with DEFCONSTANT now signal an error.
446 * enhancement: SB-SPROF now has support for wallclock profiling,
447 and is also able to profile specific threads. REPORT output
448 has also additional sorting options.
449 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
451 * optimization: structure allocation has been improved
452 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
453 ** out of line constructors are ~10% faster.
454 ** inline constructors are ~15% faster.
455 ** inline constructors are capable of dynamic extent allocation
456 (generally on x86 and x86-64, in some cases on other platforms
458 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
460 * optimization: file compiler is now able to coalesce non-circular
461 lists, non-base strings, and bit-vectors. Additionally, constants
462 are never referenced through SYMBOL-VALUE at runtime.
463 * optimization: code defining methods on PRINT-OBJECT (and other
464 generic functions in the COMMON-LISP package) now loads faster.
465 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
467 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
468 fasl is now deleted. (reported by Attila Lendvai)
469 * bug fix: READ-LINE always returned NIL for the last line in files.
470 (reported by Yoshinori Tahara)
471 * bug fix: more accurate disassembly annotations of foreign function
472 calls. (thanks to Andy Hefner)
473 * bug fix: trimming non-simple strings and non-string string
474 designators when the there is nothing to trim works properly.
475 (thanks to James Knight)
476 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
477 (thanks to Travis Cross)
478 * fixed some bugs revealed by Paul Dietz' test suite:
479 ** NIL is a valid function name (regression at 1.0.13.38)
480 ** FILL on lists was missing its return value (regression at 1.0.12.27)
481 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
482 fill pointers properly (regression at 1.0.12.23)
483 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
484 displacement indices properly (regression at 1.0.12.23)
486 changes in sbcl-1.0.17 relative to 1.0.16:
487 * temporary regression: user code can no longer allocate closure
488 variable storage on stack, due to bug 419 without explicitly
489 requesting it. Please consult sbcl-devel for advice if you need to
490 use this feature in the meanwhile.
491 * new feature: runtime argument --control-stack-size can be used to
492 adjust thread default control stack size.
493 * enhancement: improved TIME output
494 ** all times are reported using the measured accuracy (milliseconds
495 for real and GC times, microseconds for everything else.)
496 ** processor cycle counts on x86 and x86-64.
497 ** interpreted forms are counted for both evaluator modes.
498 ** number of lambdas converted by the compiler is reported.
499 ** CPU percentage report (computed from real and total run time.)
500 ** more comprehensive run time reporting, using a condenced format
501 ** interperted form, lambda, and page fault counts are omitted
503 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
505 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
506 * optimization: two argument forms of LAST are upto ~10% faster
507 in normal SPEED policies.
508 * optimization: NCONC no longer needs to heap cons its &REST list
509 in normal SPEED policies.
510 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
512 * bug fix: UNION and NUNION work with :TEST-NOT once more,
513 regression since 1.0.9.1. (thanks to Eric Marsden)
514 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
515 since 1.0.15.36. (thanks to Paul Khuong)
516 * bug fix: LAST when always returned the whole list when given a bignum
517 as the second argument.
518 * bug fix: dynamic extent allocation of nested lists and vectors
519 could leak to otherwise accessible parts.
520 * bug fix: invalid optimization of heap-allocated alien variable
522 * bug fix: fasl header checking is less vulnerable to different
523 platform word lengths.
524 * bug fix: more correct assembler syntax for GNU binutils
525 2.18.50.0.4 support. (thanks to Marijn Schouten)
526 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
528 * bug fix: the fopcompiler can handle LOCALLY forms (with no
529 declarations) successfully. (reported by Attila Lendvai)
531 changes in sbcl-1.0.16 relative to 1.0.15:
532 * minor incompatible change: revert the changes to sb-posix's error
533 signaling added in 1.0.14.
534 * minor incompatible change: change PROBE-FILE back to returning
535 NIL whenever we can't get a truename, as was the case before 1.0.14.
536 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
537 inherits from ERROR instead of just CONDITION.
538 * new feature: SB-INTROSPECT can provide source locations for instances
539 as well. (thanks to Tobias Rittweiler)
540 * optimization: binding special variables now generates smaller code
541 on threaded platforms.
542 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
543 and cases where no :TEST is given but the compiler can infer that
544 the element to search is of type (OR FIXNUM (NOT NUMBER)).
545 * optimization: better LOGNOT on fixnums.
546 * optimization: modular arithmetic for a particular requested width
547 is implemented using a tagged representation unless a better
548 representation is available.
549 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
550 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
551 for wellformedness and returns a specialized vector. (reported by
552 Francois-Rene Rideau)
553 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
554 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
555 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
556 unscheduling timer at the same time as another timer fires.
557 * bug fix: don't reschedule timers for dead threads.
558 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
559 * bug fix: copying output from RUN-PROGRAM to a stream signalled
560 bogus errors if select() was interrupted.
561 * enhancement: add support for fcntl's struct flock to SB-POSIX.
563 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
564 * enhancement: cleaner backtraces for interactive interrupts, as
565 well as other cases where the interesting frames used to be
566 obscured by interrupt handling frames.
567 * enhancement: untracing a whole package using (UNTRACE "FOO") is
568 now supported, and tracing a whole package using (TRACE "FOO") now
569 traces SETF-functions as well.
570 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
571 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
572 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
573 * unreadably printed representation of hash-tables now includes
575 * bug fix: partially fixed #188: type propagation from assignments
576 is now more efficient.
577 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
578 'DOUBLE-FLOAT) are not flushable.
579 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
580 full address of the object, and none of the tag bits.
581 * bug fix: readably printing hash-tables now respects other printer
582 control variables. (reported by Cedric St-Jean)
583 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
584 keyword with MAKE-HASH-TABLE.
585 * bug fix: export SB-POSIX:MKSTEMP.
586 * bug fix: SORT was not interrupt safe.
587 * bug fix: XREF accounts for the last node of each basic-block as
589 * bug fix: MAKE-INSTANCE optimizations interacted badly with
590 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
591 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
592 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
593 create obsolete instances in the case of redefinition or
594 obsoletion of a superclass. (thanks to Andy Hefner)
595 * bug fix: Support for the Alpha architecture has been revived; it had
596 suffered somewhat from lack of maintenance since sbcl-1.0.
597 * improvements to the Windows port:
598 ** The system detects the codepage to use at startup, instead of
599 using the value from the saved core. (thanks to Kei Suzuki)
601 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
602 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
603 (see documentation for details.)
604 * revived support for OpenBSD (contributed by Josh Elsasser)
605 * partially fixed bug #108: ROOM no longer suffers from occasional
606 (AVER (SAP= CURRENT END)) failures.
607 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
608 bodies no longer cause a WARNING to be signalled. (reported by
610 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
611 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
612 the bug in TRUNCATE.) (reported by Michael Weber)
613 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
614 no samples. (reported by Andy Hefner)
615 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
616 no longer appear as (NIL ...) frames in backtraces.
617 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
618 is now actually exported from SB-EXT as documented. (reported by
620 * bug fix: sb-aclrepl now correctly understands how to inspect
621 single-floats on 64-bit platforms where single-floats are not boxed.
622 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
623 yet been finalized. (reported by Levente Meszaros)
624 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
625 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
626 SBCL-specific optimize qualities.
628 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
629 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
630 an executable in the search path, and does so in the child
631 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
632 been removed; it can be found in the CVS history, for somebody who
633 needs that search behavior (see the manual).
634 * minor incompatible change: compiler policy re. weakening type
635 checks has changed: now type checks are weakened only if SAFETY < 2
637 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
638 unparsing of directory pathnames as files. Analogously,
639 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
640 filename to parse into a directory pathname.
641 * enhancement: implicit generic function creation now signals a
642 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
643 users can bind handlers for and muffle around calls to LOAD.
644 * enhancement: RUN-PROGRAM allows unicode arguments and environments
645 to be used (using the default stream external format), and allows
646 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
647 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
648 strings and vectors whose element-type or simplicity is not fully
649 known at compile-time.
650 * optimization: STRING-TRIM and related functions no longer allocate
651 a new string when no trimming needs to be performed. These functions
652 are also faster than before when the input string has been declared
654 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
655 * optimization: READ-LINE is significantly faster for files containing
657 * optimization: non-open coded uses of character comparison operators
658 (e.g. char=) no longer cons when called with more than one parameter
659 on platforms supporting dynamic-extent allocation.
660 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
661 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
662 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
664 * bug fix: some sequence functions elided bounds checking when
666 * bug fix: too liberal weakening of union-type checks when SPEED >
668 * bug fix: more bogus fixnum declarations in ROOM implementation
671 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
672 * new feature: MAKE-HASH-TABLE now experimentally accepts a
673 :SYNCHRONIZED argument, which makes the hash-table safe for
674 concurrent accesses (but not iteration.) See also:
675 SB-EXT:WITH-LOCKED-HASH-TABLE, and
676 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
677 * optimization: CONCATENATE on strings is an order of magnitude faster
678 in code compiled with (> SPEED SPACE).
679 * optimization: SUBSEQ is ~50% faster on lists.
680 * bug fix: bug 417 fixed -- source location reporting is now more robust.
681 * bug fix: SUBSEQ on a list will now correctly signal an error if if
682 END is smaller then START.
683 * bug fix: SB-PROFILE will no longer report extra consing for nested
684 calls to profiled functions.
685 * bug fix: ROOM implementation had bogus fixnum declarations which
686 could cause type-errors when calling ROOM on large images.
687 * bug fix: if file compilation is aborted, the partial fasl is now
688 deleted, and COMPILE-FILE returns NIL as the primary value.
689 * bug fix: number of thread safety issues relating to SBCL's internal
690 hash-table usage have been fixed.
691 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
692 be returned from its body when the values were being returned
693 using unknown-values return convection and the W-P-O was wrapped
694 inside an UNWIND-PROTECT.
695 * bug fix: sb-posix should now compile again under Windows, enabling
698 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
699 * incompatible change: hash-table accessor functions are no longer
700 automatically protected by locks. Concurrent accesses on the same hash-table
701 from multiple threads can give inconsistent results or even corrupt the
702 hash-table completely. Multi-threaded applications should do their own
703 locking at the correct granularity. In the current implementation it is
704 still safe to have multiple readers access the same table, but it's not
705 guaranteed that this property will be maintained in future releases.
706 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
707 and will signal an error at runtime.
708 * enhancement: SB-THREAD package now exports a semaphore interface.
709 * enhancement: CONS can now stack-allocate on x86 and
710 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
712 * enhancement: nested lists can now be stack allocated on
713 platforms providing stack allocation support.
714 * enhancement: dynamic-extent support has been extended to support
715 cases where there are multiple possible sources for the stack
717 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
718 if the mutex is uncontested on Linux.
719 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
720 as specified by the CLHS. (thanks to Tobias Rittweiler)
721 * bug fix: NaN comparison now works on x86-64.
722 * bug fix: CLOSE :ABORT T on a stream with pending output now
724 * bug fix: instances of non-standard metaclasses using standard
725 instance structure protocol sometimes missed the slot type checks
727 * bug fix: known functions can be inlined in high-debug code.
729 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
730 * minor incompatible change: the MSI installer on Windows no longer
731 associates .lisp and .fasl files with the installed SBCL.
732 * minor incompatible change: :UNIX is no longer present in *FEATURES*
733 on Windows. (thanks to Luis Oliviera)
734 * new feature: SB-CLTL2 contrib module now implements
735 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
736 * optimization: scavenging weak pointers is now more efficient,
737 requiring O(1) instead of O(N) per weak pointer to identify
738 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
739 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
740 method bodies using constant slot-names when the first argument is
741 a specializer parameter for the method.
742 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
743 names, when the class of the instance is a direct instance
744 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
745 as slow as the constant slot-name case.)
746 * optimization: member type construction is now O(N) instead
748 * optimization: UNION and NUNION are now O(N+M) for large
749 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
751 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
752 now more readable in environments like Slime which display it.
753 (thanks to Tobias C. Rittweiler)
754 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
755 was able to infer the correct argument type for the object on which
756 the CAS operation was being performed.
757 * bug fix: copy propagation interfered with parallel assignment
758 semantics in local calls. (reported by Paul Khuong)
759 * bug fix: the signed modular fixnum shift compiled to wrong code on
760 x86 and x86-64. (spotted by a slight modification to some of
763 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
764 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
765 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
766 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
768 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
769 and generic function now print the object identity for anonymous
770 instances. (thanks to Attila Lendvai)
771 * enhancement: as an extension to MOP, SBCL now supports SETF of
772 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
773 (thanks to Attila Lendvai)
774 * enhancement: sb-cover has an alternate annotation mode which puts
775 the form annotations at the car of the form, instead of annotating
776 it completely. The benefit of this mode is that it shows better
777 which forms were instrumented by the compiler.
778 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
779 by printer control variables. (reported by Dan Corkill)
780 * bug fix: system leaked memory when delayed output was performed by
781 the OS in smaller chunks then expected. (thanks to David Smith)
782 * bug fix: system leaked memory when file streams were not closed
784 * bug fix: large objects written to slow streams that were modified
785 after the write could end up with the modified state written to
786 the underlying file descriptor.
787 * bug fix: multiple threads operating in parallel on the same stream
788 could cause buffer-overflows.
789 * bug fix: source location information is stored correctly
790 (broken since 1.0.6). This bug would generally show up as the
791 Slime debugger highlighting the wrong form when the sldb-show-source
793 * bug fix: Branch forms are again annotated as branches in the sb-cover
795 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
796 on x86-64. (reported by Gregory Vanuxem)
797 * bug fix: EQUALP could return wrong results for structures with raw
798 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
799 word). (reported by Vjacheslav Fyodorov)
800 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
803 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
804 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
805 atomic compare-and-swap operations on threaded platforms.
806 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
807 allows assigning a global minimum value to optimization qualities
808 (overriding proclamations and declarations).
809 * enhancement: closed over variables can be stack-allocated on x86
811 * performance bug fix: GETHASH and (SETF GETHASH) are once again
813 * optimization: slot definition lookup is now O(1). This speeds up
814 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
815 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
816 * optimization: ASSOC and MEMBER can now be open-coded for all
817 combinations of keyword arguments when second argument is constant
818 and SPEED >= SPACE. In other cases a specialized version is
820 * bug fix: using obsoleted structure instances with TYPEP and
821 generic functions now signals a sensible error.
822 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
823 (reported by Kristoffer Kvello)
824 * bug fix: backtrace construction is now more careful when making
825 lisp-objects from pointers on the stack, to avoid creating bogus
826 objects that can be seen by the GC.
827 * bug fix: defaulting of values in contexts expecting more than 7
828 variables now works on x86-64. (reported by Christopher Laux)
829 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
831 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
832 as the property-list of a symbol.
833 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
834 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
835 situations requested, are once again file-compileable. (reported
838 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
839 * MOP improvement: support for user-defined subclasses of
840 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
841 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
842 the expansion of DEFMETHOD, is responsible for generating a form
843 which creates a list of specializers when evaluated. Additional
844 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
845 debugging and introspective support.
846 * minor incompatible change: the (unsupported) spinlock interface
847 has changed: free spinlock now has the value NIL, and a held spinlock
848 has the owning thread as its value.
849 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
850 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
852 * enhancement: name of a socket-stream is now "a socket" instead of
854 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
855 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
856 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
857 for generic functions. (Thanks to Tobias C. Rittweiler)
858 * optimization: bignum printing speed has been improved by 20-40%
859 (depending on the bignum size.)
860 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
862 * bug fix: the cache used by the CLOS to store precomputed effective
863 methods, slot offsets, and constant return values is now thread and
865 * bug fix: generic function dispatch function updating is now thread
866 and interrupt safe (in the sense that the known issues have been
868 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
869 * bug fix: interning EQL-specializers is now thread and interrupt safe.
870 * bug fix: asdf systems with dependencies to the SB-POSIX or
871 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
872 * bug fix: interrupt safety of applicable method computation has been
875 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
876 * new contrib: sb-cover, an experimental code coverage tool, is included
878 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
879 significantly faster.
880 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
881 produce more efficient code on x86-64 if the compiler can determine
882 that (+ INDEX <constant>) does not require a bounds check and FOO
883 has an element type at least 8 bits wide.
884 * enhancement: a new, experimental synchronous timeout facility is
885 provided. Refer to SB-SYS:WITH-DEADLINE for details.
886 * enhancement: when a symbol name conflict error arises, the
887 conflicting symbols are always printed with a package prefix.
888 (thanks to Kevin Reid)
889 * enhancement: stepping is now once again supported on the SPARC and
890 MIPS platforms. (It is also now more likely to work on CheneyGC
892 * enhancement: sb-sprof can now also track and report accurate call
894 * bug fixes: the treatment of non-standard subclasses of
895 SB-MOP:SPECIALIZER is more correct.
896 * incompatible change: PURIFY no longer copies the data from the
897 dynamic space into the static and read-only spaces on platforms
898 that use the generational garbage collector
899 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
901 * bug fix: GC race condition occasionally resulting in crashes with
902 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
904 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
906 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
907 a STYLE-WARNING to be signalled (regression from 1.0.4.)
908 * bug fix: an asynchronous interrupt could previously leave the
909 system running with GC inhibited.
910 * bug fix: a DECLARE form evaluated at top-level now causes an error
911 rather than silently (or verbosely) returning NIL.
912 * bug fix: trying to dissassemble functions compiled at high DEBUG could
913 result in a error being signalled due to source form lookup errors.
914 (reported by Peter Graves)
916 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
917 * incompatible change: removed writer methods for host-ent-name,
918 host-ent-addresses -- changing the values did not update the DNS
920 * minor incompatible change: changed experimental JOIN-THREAD interface
921 * documentation: the manual now lists reader and writer methods
922 in class slot documentation sections. (thanks to Richard M Kreuter)
923 * documentation: unwinding from asyncronous events has been
924 documented as unsafe.
925 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
926 in multithreaded application code.
927 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
928 platforms. (thanks to James Anderson for the optimization hint)
929 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
931 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
932 variants no longer cons.
933 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
934 their NOT- variants no longer cons.
935 * optimization: EQUAL hash tables no longer use SXHASH for objects
936 of all data types, but instead use an EQL hash for types for which
937 EQUAL is the same as EQL.
938 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
939 are significantly faster.
940 * optimization: new STRING-OUTPUT-STREAM implementation is much
941 faster for large outputs and conses 30% less on average.
942 * enhancement: XREF information is now collected for references made
943 to global variables using SYMBOL-VALUE and a constant argument.
944 * enhancement: SIGINT now causes a specific condition
945 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
946 * bug fix: FILE-POSITION used to signal an error in some cases where
947 ANSI requires it to return NIL.
948 * bug fix: ADJUST-ARRAY is now interrupt-safe.
949 * bug fix: adding and removing fd-handlers is now interrupt-safe.
950 * bug fix: inlined calls to C now ensure 16byte stack alignment on
952 * bug fix: bad type declaration in the CLOS implementation has
953 been fixed. (reported by James Anderson)
954 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
955 (thanks to Lutz Euler)
956 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
957 required by the CLHS. (reported by Samium Gromoff)
958 * bug fix: dead unbound variable references now signal an error.
959 (reported by Marco Monteiro)
960 * bug fix: / with an unused value was being deleted in safe code.
961 (thanks to Marco Monteiro and Kevin Reid)
962 * bug fix: number of characters that can be written onto a single
963 line in a file is unlimited.
964 * bug fix: some GC deadlocks caused by asynchronous interrupts have
965 been fixed by inhibiting interrupts for when GC is disabled.
966 * bug fix: some interrupt safety issues with GETHASH, (SETF
967 GETHASH), CLRHASH and REMHASH have been fixed.
968 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
969 specifier no longer causes infinite recursion.
970 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
971 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
972 * bug fix: an &environment argument with the correct variable information
973 is passed to macros that are expanded during byte compilation.
974 (reported by Samium Gromoff)
975 * bug fix: a logic error was causing the hash function for bignums to
976 have excessive amounts of collisions. (reported by Faré Rideau)
977 * bug fix: modifying the contents of an array could change the return
978 value of SXHASH on that array, which is only allowed for strings
979 and bit vectors. (bug introduced in 0.9.16)
980 * improvement: the x86-64/darwin port now passes all tests (except
981 for the debugger tests) but should still be considered
982 experimental until this is fixed.
983 * improvement: a style-warning is signaled for CASE (etc) clauses with
984 duplicate keys. (thanks to Kevin Reid)
985 * improvement: macos/x86 and macos/x86-64 now use mach exception
986 handlers for dealing with illegal instructions (for trapping and
987 error handling) and memory protection violations (for GC).
989 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
990 * new platform: experimental support for x86-64/darwin (MacOS).
991 * incompatible change: the thread-safe (on most platforms) getaddrinfo
992 and getnameinfo sockets functions are used instead of gethostbyaddr
993 and gethostbyname, on platforms where the newer functions are available.
994 As a result, the ALIASES field of HOST-ENT will always be NIL on these
996 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
997 don't print the startup banner, but behave as if --noinform was passed
998 as a command line argument. (thanks to Kevin Reid)
999 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
1000 * optimization: code using alien values with undeclared types is much faster.
1001 * optimization: the compiler is now able to open code SEARCH in more cases.
1002 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1003 * bug fix: using standardized COMMON-LISP special variables as loop
1004 variables no longer signals bogus package lock violations. (reported
1006 * bug fix: declaring local loop variables to be of a range-limited type
1007 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1008 (reported by Andras Simon)
1009 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1010 bugs remain on x86-64.)
1011 * bug fix: the #= and ## reader macros now interact reasonably with
1012 funcallable instances.
1013 * bug fix: type-checks for function arguments were compiled using the
1014 compiler policy settings of the wrong lexical environment (in 1.0.2
1016 * bug fix: SHADOW accepts characters as string designators, as required
1017 by the spec (thanks to Eric Marsden)
1018 * bug fix: fixed GC safety issues when foreign functions are called with
1019 non-base strings as arguments
1020 * bug fix: more consistent error output for fatal-compiler-warnings, like
1022 * bug fix: functions with proper names were showing up as NIL in
1023 backtraces (reported by Edi Weitz, regression in 1.0.2)
1025 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1026 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1027 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1028 produce more efficient code on the x86 if the compiler can determine
1029 that (+ INDEX <constant>) does not require a bounds check and FOO
1030 has an element type at least 8 bits wide.
1031 * bug fix: references to undefined variables are handled the same way
1032 in toplevel forms as in the normal compiler
1033 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1035 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1036 (thanks to Jon Buller)
1037 * improvement: faster compilation times for complex functions
1038 * improvement: added readlink support to SB-POSIX (thanks to Richard
1041 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1042 * improvement: experimental support for mach exception handling on
1043 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1045 * improvement: support for GBK external format.
1046 (thanks to Chun Tian (binghe))
1047 * improvement: the debugger now displays variables that have been closed
1048 over, in code compiled with (DEBUG 2) or higher
1049 * improvement: support for executable cores on NetBSD (thanks to
1051 * new feature: added a RESTART-FRAME debugger command
1052 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1053 be used to provide an implementation for FILE-POSITION on Gray streams
1054 (thanks to Eric Marsden)
1055 * new feature: add syslog, openlog and closelog support to SB-POSIX
1056 (thanks to Richard Kreuter)
1057 * optimization: the function call overhead in code compiled with
1058 a high DEBUG optimization setting is significantly
1059 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1060 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1061 (thanks to Tony Martinez)
1062 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1063 evaluated once (reported by Kevin Reid)
1064 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1065 works again (thanks to Kilian Sprotte)
1066 * bug fix: an error is signaled for tagbodies with duplicate tags
1067 (thanks to Stephen Wilson)
1068 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1069 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1070 stack frames from alien callbacks.
1071 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1072 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1073 65k pages (thanks to David Woodhouse)
1074 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1076 changes in sbcl-1.0.1 relative to sbcl-1.0:
1077 * new platform: FreeBSD/x86-64, including support for threading.
1078 * new feature: the compiler stores cross-referencing information
1079 about function calls (who-calls), macroexpansion (who-macroexpands)
1080 and special variables (who-binds, who-sets, who-references) for code
1081 compiled with (< SPACE 3). This information is available through the
1082 sb-introspect contrib.
1083 * new feature: users may subclass SEQUENCE, and have instances of
1084 these classes interoperate with standard Common Lisp functions if
1085 a number of methods are defined. (This feature is experimental
1086 and the interface subject to change based on feedback from SBCL
1087 users and the general community)
1088 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1089 x86/x86-64, rather than the previous fixed depth of 8
1090 * improvement: another pthread back-end of mutex "pthread-futex".
1091 * bug fix: non-ascii command-line arguments are processed correctly
1092 (thanks to Yaroslav Kavenchuk)
1093 * bug fix: non-required arguments were not passed correctly when a method
1094 defined using DEFMETHOD was called from a mop-generated method using
1095 CALL-NEXT-METHOD (reported by Pascal Costanza)
1096 * bug fix: recursion is now permitted in accessors through
1097 SLOT-UNBOUND. (reported by Pascal Costanza)
1098 * bug fix: an error was signaled at startup if the HOME environment
1099 variable was defined, but had an empty value (reported by Peter Van Eynde)
1100 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1101 signal a PROGRAM-ERROR, not a BUG.
1102 * bug fix: SB-POSIX:READDIR works when built with large file support.
1103 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1105 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1106 for better automatic indentation support. (thanks to Matt Pillsbury)
1107 * optimization: loading generic functions no longer takes O(n^2) time,
1108 proportional to the amount of methods in the generic function
1109 (reported by Todd Sabin and Jeremy Brown)
1110 * optimization: the FIND and POSITION family of sequence functions
1111 are significantly faster on arrays whose element types have been
1113 * improvements to the Windows port:
1114 ** Intermittent heap corruption problems have been fixed. (thanks
1115 to Alastair Bridgewater)
1116 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1118 ** Lisp is able to unwind foreign exception frames from alien
1119 callbacks. (thanks to Alastair Bridgewater)
1121 changes in sbcl-1.0 relative to sbcl-0.9.18:
1122 * improvement: experimental support for threading on FreeBSD/x86.
1123 (thanks to NIIMI Satoshi)
1124 * improvement: runtime option --dynamic-space-size can be used
1125 to set the size of the dynamic space reserved on startup.
1126 * improvement: floating point modes in effect are now saved in
1127 core, and restored on startup.
1128 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1129 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1130 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1131 threads are not running after *SAVE-HOOKS* have run.
1132 * improvement: writes to CLOS instance slots are type-checked in code
1133 compiled with (SAFETY 3)
1134 * improvement: floating-point exception handling on FreeBSD (thanks to
1136 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1137 (thanks to Zach Beane)
1138 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1140 * improvement: added support for the Shift-JIS external format.
1141 (contributed by NIIMI Satoshi)
1142 * improvement: callbacks are supported on Linux/PPC. (thanks to
1144 * bug fix: compiler bug triggered by a (non-standard) VALUES
1145 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1146 * bug fix: file compiler no longer confuses validated and already
1147 dumped structurres. (reported by Kaersten Poeck)
1148 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1149 fill-pointer signals a type-error as required. (thanks to
1151 * bug fix: disassemly of funcallable instances works.
1152 * bug fix: single stepping on PPC.
1153 * bug fix: fix thread-safety problems in the type system (generally
1154 manifesting as nonsensical errors like "STRING is a bad type specifier
1155 for sequences" or "The value 1 is not of type FIXNUM")
1156 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1157 (reported by Josip Gracin)
1158 * bug fix: an error is signaled for attempts to displace arrays with
1159 incompatible element types (thanks to Mario Mommer)
1160 * bug fix: more correct handling of wide characters in debug info
1161 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1162 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1163 and don't cause extra consing
1164 * optimization: MAP and MAP-INTO are significantly faster on vectors
1165 whose elements types have been declared.
1166 * Improvements to SB-SPROF:
1167 ** Support for allocation profiling
1168 ** Reduced profiling overhead, especially for long profiling runs
1169 * Improvements to the Windows port:
1170 ** floating point exceptions are now reported correctly.
1171 ** stack exhaustion detection works partially.
1172 ** more accurate GET-INTERNAL-REAL-TIME.
1173 ** better RUN-PROGRAM behaviour in case of errors in the child
1175 ** PROBE-FILE now simplifies pathnames correctly.
1176 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1177 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1178 1st 1970 (although time timezone data will be unreliable).
1179 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1181 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1182 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1183 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1185 * bug fix: two potential GC deadlocks affecting threaded builds.
1186 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1187 returns T (reported by Anton Kazennikov)
1188 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1189 with non-variable places
1190 * bug fix: remove a race condition in the setting of
1191 funcallable-instance functions, this should make threaded CLOS
1192 code more stable against memory faults.
1193 * bug fix: corruption of specials when unbinding is interrupted by an
1194 asynchronous unwind (reported by Hannu Koivisto)
1195 * improvement: the debugger will now also display local variables that
1196 are only used once, for code compiled with a DEBUG optimization quality
1199 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1200 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1201 * incompatible change: External-format support for FFI calls. The
1202 SB-ALIEN:C-STRING no longer implies an ASCII
1203 external-format. Instead the string is subject to external-format
1204 conversion. Additionally return values of type SB-ALIEN:C-STRING
1205 are by default of type (SIMPLE-ARRAY CHARACTER), not
1206 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1207 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1208 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1209 * incompatible change: SB-EXT package no longer contains the
1210 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1211 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1213 * incompatible change: the single-stepper is no longer available
1214 on Alpha, Mips and Sparc platforms.
1215 * minor incompatible change: the direct superclasses of
1216 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1217 not (STANDARD-OBJECT FUNCTION). This makes the
1218 class-precedence-lists of GENERIC-FUNCTION and
1219 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1221 * minor incompatible change: the default stream external format on
1222 non-unicode SBCL is no longer detected from the locale, but is always
1224 * new feature: Add a version of evaluator that uses an interpreter instead
1225 of the compiler. EVAL still uses the compiler by default, to switch it
1226 to use the interpreter, set the value of the variable
1227 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1228 * minor incompatible change: the single-stepper REPL has been merged
1229 with the normal debugger (see the "Stepping" heading of the debugger help
1230 for more details). The debugger command STEP will no longer switch
1231 to the single-stepper REPL.
1232 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1233 for a type now works.
1234 * bug fix: loading of default sysinit file works. (thanks to Leonid
1236 * bug fix: better detection of circularities in the file-compiler.
1237 (reported by Marco Monteiro)
1238 * bug fix: the CL pathname functions now work with files that have
1239 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1240 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1241 systems with Unix98 pty semantics.
1242 * bug fix: ASDF-INSTALL will now work with bsd tar.
1243 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1245 * bug fix: timers expiring in dead threads no longer cause a
1246 type-error (reported by Paul "Nonny Mouse").
1247 * bug fix: thanks to more lightweight single-stepper instrumentation,
1248 code compiled with (DEBUG 3) will compile and execute significantly faster,
1249 and will have more accurate type-inferencing than before
1250 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1251 whose bindings are modified
1252 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1253 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1254 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1255 CL:SOFTWARE-VERSION with proper version
1257 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1258 * feature: implemented the READER-METHOD-CLASS and
1259 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1260 as specified by AMOP.
1261 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1263 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1264 (thanks to Lutz Euler)
1265 * optimization: hashing of general arrays and vectors has been
1266 improved. (reported by Any Fingerhut)
1267 * enhancement: SB-INTROSPECT is now able to find definitions of
1268 profiled functions. (thanks to Troels Henriksen)
1269 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1270 as well, allowing compiler-macros for SETF-functions to expand.
1271 * enhancement: step-instrumentation no longer wraps known
1272 single-value functions with multiple-value context, allowing
1273 better type inference.
1274 * fixed bug #337: use of MAKE-METHOD in method combination now works
1275 even in the presence of user-defined method classes. (reported by
1276 Bruno Haible and Pascal Costanza)
1277 * fixed bug #339(c): if there are applicable methods not part of any
1278 long-form method-combination group, call INVALID-METHOD-ERROR.
1279 (reported by Bruno Haible)
1280 * fixed bug #361: the :FUNCTION initarg in the protocol for
1281 initialization of methods can now be used to override
1282 internally-produced optimized functions. (reported by Bruno
1284 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1285 system-provided lambda expression no longer cause warnings about
1286 unbound #:|pv-table| symbols.
1287 * bug fix: improved the handling of type declarations and the
1288 detection of violations for keyword arguments with non-constant
1290 * bug fix: potentially erronous calls to PATHNAME and
1291 MERGE-PATHNAMES were being flushed in some cases.
1292 (reported by Richard Kreuter)
1293 * bug fix: compiled calls to TYPEP were mishandling obsolete
1294 instances. (reported by James Bielman and Attila Lendvai)
1295 * bug fix: format strings with ~> without matching ~< no longer
1296 trigger an AVER, but signal an understandable error instead.
1297 (reported by Antonio Martinez)
1298 * bug fix: specifying an output-file without a pathname-type for
1299 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1300 of input-file instead of "fasl". (reported by Robert Dodier)
1301 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1302 (reported by James Y Knight).
1303 * bug fix: compiler-macros-function did not consider the environment
1304 argument for shadowing by local functions.
1305 * bug fix: compiler-macros expansion was inhibited by local INLINE
1307 * bug fix: inline expansions of known functions were subject to
1308 step-instrumentation in high debug policies, leading to problems
1309 with type-inference.
1310 * bug fix: compiler failed to differentiate between different CONS
1311 types in some cases.
1312 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1313 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1314 * bug fix: #368: incorrect use of expressed vs. upgraded array
1316 * bug fix: #306a: more precise unions of array types.
1317 * thread-safety improvements:
1318 ** CONDITION-WAIT could return early on Linux, if the thread was
1319 interrupted and subsequently continued with SIGCONT.
1320 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1322 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1323 * added support for the ucs-2 external format. (contributed by Ivan
1325 * minor incompatible change: pretty printing of objects of type
1326 (cons symbol) is, in the default pprint-dispatch-table, now
1327 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1329 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1330 called later in a class's lifetime, possibly as late as when the
1331 first instance of the class is created. Previously,
1332 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1333 class became finalizeable.
1334 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1335 for multibyte external-format streams. (thanks to "vbzoli")
1336 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1337 with :ALLOCATION :CLASS inherited from superclasses of the
1339 * fixed bug: anonymous classes can now be created using the :NAME
1340 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1341 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1342 * fixed bug: core-files saved with :EXECUTABLE T can again be
1343 executed when SBCL_HOME isn't set. (reported by James Knight)
1344 * fixed bug: toplevel LOCALLY forms with declarations could
1345 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1346 * fixed bug: printing from several different threads using different
1347 values of *print-case* could cause invalid output, due to
1348 some internal special variables of the printer not being bound
1349 thread-locally (reported by Max Mikhanosha)
1350 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1351 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1352 (reported by Pascal Costanza)
1353 * fixed bug: FINALIZE-INHERITANCE is called from
1354 REINITIALIZE-INSTANCE on classes when the class has previously
1355 been finalized, as required by AMOP.
1356 * minor code generation optimizations:
1357 ** better register allocation in CLOS dispatching functions
1358 ** overflow detection when coercing signed bytes to fixnums on x86-64
1359 ** is now implemented with one IMUL instruction instead of three shifts
1360 ** more efficient bit-vector access on x86 and x86-64
1361 ** more efficient access to raw structure slots on x86 and x86-64
1362 * fixed some bugs revealed by Paul Dietz' test suite:
1363 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1364 return its argument.
1366 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1367 * feature: thread support on Solaris/x86, and experimental thread support
1369 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1371 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1372 default method for SLOT-UNBOUND.
1373 * minor incompatible change: prevent the user from specializing the
1374 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1375 somewhat counter to the intent of the protocol, I (CSR) think, and
1376 additionally it just doesn't work in SBCL as currently
1377 implemented, thanks to optimizations (that are always valid for
1378 the other three SLOT-VALUEish functions, but not for the setter).
1379 * minor incompatible change: the :SB-LDB feature is now enabled by
1380 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1381 the low-level debugger.
1382 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1383 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1384 down to an SB-SYS:FD-STREAM, allowing direct communication in
1385 more cases. Temporary files and pipes are still used as a fallback
1387 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1388 on the non-reentrant C functions localtime(3) and gmtime(3).
1389 * bug fix: global optimization policy was not visible in LOCALLY and
1391 * bug fix: class objects can be used as specializers in methods.
1392 (reported by Pascal Costanza)
1393 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1394 with a type error. (reported by blitz_ on #lisp)
1395 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1396 failed with a type error. (reported by Pascal Bourguignon)
1397 * bug fix: merging pathnames against defaults with :DIRECTORY
1398 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1399 (reported by James Y Knight)
1400 * bug fix: saving large (>2GB) cores on x86-64 now works
1401 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1402 constant index and a (simple-array (signed-byte 32)) array
1403 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1404 error (patch by Robert J. Macomber)
1405 * feature: TIME now displays the amount of run-time spent in GC
1406 * bug fix: The debugger now does a better job of respecting
1407 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1408 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1409 workaround for bug 403.)
1410 * bug fix: floating point exception handling now works on FreeBSD
1411 (thanks to NIIMI Satoshi)
1412 * fixed some bugs revealed by Paul Dietz' test suite:
1413 ** MISC.641: LET-conversion were not supposed to work in late
1415 * improvements to the Win32/x86 port:
1416 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1417 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1419 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1420 * new feature: source path information is generated for macro-expansion
1421 errors for use in IDE's like Slime (thanks to Helmut Eller)
1422 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1423 * bug fix: compiler does not loop forever on an invalid type in
1425 * improvement: compilation of most CLOS applications is significantly
1427 * optimization: added a limited bytecode compiler for simple toplevel
1428 forms, speeding up compilation and FASL loading
1429 * bug fix: the statistical profiler now properly distinguishes anonymous
1432 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1433 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1434 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1435 into subclasses of the finalized class. Now user calls to
1436 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1437 system return before any subclasses are finalized.
1438 * minor incompatible change: The reader no longer ignores errors
1439 regarding non-existent packages in #+ and #- feature tests.
1440 * new feature: command line options --no-sysinit, --no-userinit to
1441 inhibit loading the corresponding init files
1442 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1443 for sending data through UDP sockets (thanks to François-René Rideau)
1444 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1445 error is signalled instead (thanks to François-René Rideau)
1446 * minor incompatible change: Error signalling behaviour of lexical
1447 operations violating package locks has changed slightly. Refer to
1448 documentation on package locks for details.
1449 * enhancement: EVAL can process IF-expressions without resorting to the
1451 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1452 constant-folded properly due to missing out-of-line functions.
1453 (reported by Mika Pihlajamäki)
1454 * bug fix: LISTEN sometimes returned T even in cases where no data was
1455 immediately available from the stream
1456 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1457 were reversed. (reported by Levente Mészáros)
1458 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1459 when given a vector displaced to an adjustable array. (reported
1461 * bug fix: derivation of float boundaries from numbers outside the
1462 appropriate float range (reported by John Wiseman)
1463 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1465 * bug fix: precomputing cache entries for generic functions with
1466 some subclasses of specializers as yet invalid does not attempt to
1467 fill a cache line with a negative offset. (reported by Levente
1469 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1470 allow condition class objects as arguments to DOCUMENTATION and
1471 (SETF DOCUMENTATION); only find and set documentation for
1472 structure names for the STRUCTURE doc-type. (suggested by Gary
1474 * improvements to the Win32/x86 port:
1475 ** (user-homedir-pathname) and default initialization file
1476 locations now know about the user's "Documents and Settings"
1477 directory (thanks to Yaroslav Kavenchuk)
1478 ** run-program is implemented (thanks to Mike Thomas)
1479 ** sockets support (thanks to Timothy Ritchey)
1480 ** better backtrace support (thanks to Alastair Bridgewater)
1481 ** sb-grovel supported
1482 ** asdf-install and sb-posix work somewhat
1483 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1484 * improvements to the Solaris/x86 port:
1485 ** works on Solaris 11/Solaris Express
1486 ** floating-point exception handling support
1487 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1488 * fixed some bugs revealed by Paul Dietz' test suite:
1489 ** REMOVE-METHOD returns its generic function argument even when
1490 the method is not one of the generic functions' methods.
1491 ** objects declared as MEMBER types can be admissible arguments to
1492 structure accessors.
1493 ** printing characters should simply be printed by the FORMAT ~:C
1495 ** compiler failure when compiling functions with hairy constant
1496 defaults for optional parameters.
1497 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1498 function, which is already optimized.
1500 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1501 * new platform: experimental support for SBCL x86/Darwin, including
1502 MacOS X 10.4.5 on Intel.
1503 * new feature: Unicode character names are now known to the system
1504 (through CHAR-NAME and NAME-CHAR).
1505 * new feature: the filesystem location of SBCL's core file is
1506 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1507 * minor incompatible change: the contrib modules SB-POSIX and
1508 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1509 this change is to make it easier to distribute
1510 location-independent binaries.
1511 * bug fix: as implied by AMOP, standardized classes no longer have
1512 slots named by external symbols of public packages. (reported by
1514 * optimization: calling functions via constant symbols -- as in
1515 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1516 via the function object as in (FUNCALL #'FOO).
1517 * optimization: on x86 and x86-64, the calling convention for the
1518 case of dealing with an unknown number of values has been altered
1519 to be friendlier to the prediction heuristics implemented,
1520 particularly with reference to CALL and RET pairing. (thanks to
1521 Alastair Bridgewater)
1522 * optimization: on x86, the code for access to thread-local symbol
1523 values and binding/unbinding of thread-local symbols is smaller.
1524 (thanks to Alastair Bridgewater)
1525 * enhancement: CONSTANTP is now able to determine constantness of
1526 more complex forms, including calls to constant-foldable standardized
1527 functions and some special forms beyond QUOTE.
1528 * fixed bug: occasional GC crashes on Solaris/x86
1529 * optimization: x86-64 supports stack allocation of results of simple
1530 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1531 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1532 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1533 (thanks to James Knight)
1534 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1535 by the standard (thanks to Stephen Compall)
1537 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1538 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1539 be used for bundling the runtime and the core file into one
1540 executable binary. This feature is not currently supported on all SBCL
1541 platforms. (thanks to James Bielman and NIIMI Satoshi)
1542 * new feature: a generational or ephemeral garbage collector is now
1543 the default on the PowerPC platform (both Linux and Darwin). The
1544 old Cheney (stop and copy) collector is a build-time option.
1545 (thanks to Cyrus Harmon, after Raymond Toy)
1546 * minor incompatible change: the method by which SBCL finds its
1547 contributed modules has changed; it no longer relies on symbolic
1548 links from an $SBCL_HOME/systems directory, but searches directly
1549 in the subdirectories of $SBCL_HOME.
1550 * enhancement: the dynamic heap size on the Linux/PPC platform is
1551 markedly larger, even using the older Cheney garbage collector.
1552 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1553 certain complicated string types. (reported by Gary King)
1554 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1555 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1556 * fixed bug: DOCUMENTATION on structures no longer signals an
1557 error if no structure type of the right name exists. (reported by
1559 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1560 applicable non-standard primary or :AROUND method on
1561 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1562 no longer cause unbound variable errors. (reported by Kevin Reid)
1563 * fixed bug: metacircle resolution in cases where methods have slots
1564 added before the slots from STANDARD-METHOD. (reported by Jean
1566 * fixed bug: the Power PC assembly code for calling into lisp
1567 sometimes computed its return address wrongly (depending on the
1568 15th bit of the address, fixed at link time). This used to
1569 manifest itself in a segmentation violation while building PCL.
1570 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1571 many others over the years)
1572 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1573 the stack pointer (treated as a signed integer) is negative.
1574 (thanks to Peter van Eynde)
1576 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1577 * new platform: experimental support for the Windows operating
1578 system has been added. (thanks to Alastair Bridgewater)
1579 * fixed several bugs in and robustified the PPC FFI (including
1580 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1581 * bug fix: allow non-simple string symbol names (reported by Paul
1583 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1585 * bug fix: saving a core corrupted callbacks on x86/x86-64
1586 * bug fix: closed a loophole in metacircularity detection and
1587 grounding in the PCL implementation of CLOS.
1588 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1589 (reported by tomppa on #lisp)
1590 * bug fix: clear the direction flag on Lisp -> C transitions, as
1591 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1592 (reported by Andrej Grozin and Hendrik Maryns)
1593 * optimization: major improvements to GC efficiency on GENCGC platforms
1594 * optimization: faster implementation of EQUAL
1595 * optimization: emit more efficient opcodes for some common
1596 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1597 * optimization: several other minor code-generation improvements on x86-64
1598 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1600 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1601 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1602 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1603 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1604 by ANSI as a generic function, and for consistency (SETF
1605 GENERIC-FUNCTION-NAME) is treated likewise.
1606 * fixed bug #233.b: make constraint propagation notice when a variable
1607 value is changed after it is referenced but before it is used
1608 * fixed bug #296: no more arbitrary behaviour from filenames with
1609 odd characters as --load arguments. (reported by Adam Warner)
1610 * fixed bug #390: :CHARACTER-SET pathname components now work as
1611 expected. (reported by Tim Daly Jr)
1612 * fixed bug #391: complicated :TYPE intersections in slot
1613 definitions no longer cause an error in PCL internals.
1614 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1615 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1616 returning the number of octets which would be written to the
1617 file-stream. (thanks to Robert J. Macomber)
1618 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1619 arguments to the internal stream functions. (thanks to David
1621 * bug fix: stack exhaustion is now handled on the x86/SunOS
1622 platform. (thanks to Daisuke Homma)
1623 * bug fix: more accurate ROOM results on GENCGC platforms
1624 * optimization: improved type inference for arithmetic-for
1625 index variables in LOOP
1626 * optimization: faster floating-point SQRT on x86-64
1627 * fixed some bugs revealed by Paul Dietz' test suite:
1628 ** DOCUMENTATION returns NIL instead of "" for method combinations
1629 that don't have a docstring
1631 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1632 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1633 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1634 therefore are not customizeable by user code (as seems to be at
1635 least permitted and maybe required by AMOP). As a consolation,
1636 however, the SBCL implementation of these functions now calls
1637 REINITIALIZE-INSTANCE as specified by AMOP.
1638 * bug fix: slot-definition documentation is propagated as per ANSI
1639 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1640 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1641 argument, as do eslotd initialization methods. (from Pascal
1642 Costanza's "Closer" project)
1643 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1644 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1646 * bug fix: it is now possible to have more than one subclass of
1647 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1648 (reported by Bruno Haible, Pascal Costanza and others)
1649 * bug fix: the dependent update protocol now works for generic
1650 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1651 and Pascal Costanza)
1652 * bug fix: condition-class instances corresponding to
1653 DEFINE-CONDITION forms are now created eagerly. (reported by
1654 Kalle Olavi Niemitalo on comp.lang.lisp)
1655 * bug fix: floating point printing is more accurate in some
1656 circumstances. (thanks to Simon Alexander)
1657 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1658 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1659 * bug fix: callbacks on OS X now preserve stack-alignment.
1660 (thanks to Cyrus Harmon)
1661 * enhancement: the x86-64 disassembler is much better at
1662 disassembling SSE instructions. (thanks to Lutz Euler)
1663 * enhancement: CHECK-TYPE on a variable now allows type-inference
1664 (thanks to Vincent Arkesteijn)
1665 * optimization: improved performance of EUC-JP external format.
1666 (thanks to NIIMI Satoshi)
1667 * optimization: performance improvements to IO on file streams of
1668 :ELEMENT-TYPE CHARACTER
1669 * optimization: much faster memory allocation on x86-64
1670 * optimization: faster garbage collections (latency and throughput)
1672 * optimization: faster bignum subtraction and division on x86 and x86-64
1673 (thanks to Lutz Euler)
1674 * optimization: more accurate type-inference for loops that use a
1675 floating point index variable or a negative step.
1677 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1678 * bug fix: add a workaround to SBCL looping infinitely at startup on
1679 Linux kernels with apparently buggy implementations of personality().
1680 (thanks to Svein Ove Aas)
1681 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1682 (thanks to David Lichteblau)
1683 * bug fix: local bindings shadow global symbol macros for
1684 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1685 * optimization: non-open coded uses of numeric comparison operators
1686 (e.g. >) no longer cons when called with more than one parameter
1687 on platforms supporting dynamic-extent allocation.
1688 * enhancement: saving cores with foreign code loaded is now supported
1689 on MIPS/Linux in addition to the previously supported platforms.
1690 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1691 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1692 * bug fix: division by zero in sb-sprof when no samples were collected
1693 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1694 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1695 (thanks to Alexander Kjeldaas)
1696 * new feature: ignoring the return values of destructive functions
1697 like NREVERSE signals a compile-time style-warning.
1698 (thanks to Kevin Reid)
1699 * new feature: the alignment of alien structure fields can be
1700 explicitly specified. (thanks to Cyrus Harmon)
1701 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1702 and dump core on SIGQUIT
1704 ** incompatible change: threads do not inherit values of specials
1705 from their parents (see manual)
1706 ** bug fix: threads stacks belonging to dead threads are freed by the
1707 next exiting thread, no need to gc to collect thread stacks anymore
1708 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1709 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1710 ** bug fix: dynamic variable and thread start related gc lossage
1711 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1713 * fixed some bugs revealed by Paul Dietz' test suite:
1714 ** SUBTYPEP is slightly more accurate on heinously complicated
1715 CONS types where some of the members have uncertain (in the
1716 NIL, NIL sense) type relationships to each other.
1718 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1719 no longer purified unless :PURIFY T is explicitly specified.
1720 ** Non-purified cores are significantly smaller than before
1722 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1723 * new feature: timers based on Zach Beane's excellent timer package
1724 * added support for the following external formats: koi8-u,
1725 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1726 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1727 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1728 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1729 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1730 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1731 * incompatible change: a threaded SBCL will no longer revert to
1732 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1733 * bug fix: interrupts are disabled until startup is complete; no
1734 more sigsegvs when receiving a signal to soon
1735 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1737 * bug fix: add a workaround for the memory randomization features in
1738 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1739 workaround will only be in effect on systems with the proc filesystem
1741 * bug fix: printing objects of type HASH-TABLE signals a
1742 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1744 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1745 have been running for over 50 days. (reported by Gilbert Baumann)
1746 * bug fix: the logic for getting names of functions gets less
1747 confused when confronded with alternate-metaclass
1748 funcallable-instances. (reported by Cyrus Harmon)
1749 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1750 consistent, even on internal alternate-metaclass objects.
1751 * bug fix: flush closure information collected by physical
1752 environment analysis prepass before the main pass. (bug reported
1754 * bug fix: compiler pack phase does not modify a hashtable, which is
1755 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1756 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1757 may be expanded. (reported by Alexander <malishev> on c.l.l)
1758 * bug fix: no more style-warnings when compiling code using LAST
1759 under high-SPEED low-SPACE settings. (reported by David Wallin
1761 * fixed some incompatibilities between SBCL's MOP and the MOP
1763 ** the METAOBJECT class is now implemented;
1764 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1765 STANDARD-OBJECT, as required;
1766 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1767 now compatible (as required by VALIDATE-SUPERCLASS); there
1768 remains a consistency requirement over the presence or absence
1769 of the FUNCTION class in the superclass list of finalized
1770 classes; see the manual for more details;
1771 ** the :AROUND method for COMPUTE-SLOTS on
1772 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1773 requested slot ordering.
1775 ** bug fix: parent thread now can be gc'ed even with a live
1777 ** bug fix: binding a special with PROGV to no value is not
1778 the same as that symbol not having been bound (thanks to
1780 ** bug fix: binding specials is thread safe (thanks to
1782 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1783 and SB-SPROF do not die with 'no handler for signal XX in
1784 interrupt_handle_now(..)' anymore
1785 ** bug fix: WITH-TIMEOUT works with multiple running threads
1786 * fixed some bugs revealed by Paul Dietz' test suite:
1787 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1788 the :method-class keyword argument.
1790 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1791 * new port: the Solaris operating system on x86 processors is now
1792 mostly supported, though some rough edges in the environment
1793 remain. (thanks to Daisuke Homma)
1794 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1795 space for its heap. (thanks to Thiemo Seufer)
1796 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1797 closures. (thanks to Thiemo Seufer)
1798 * minor incompatible change: eof selects abort in the debugger.
1799 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1800 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1802 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1803 deprecated, and will go away in a future revision of SBCL.
1804 * minor incompatible change: GC-ON and GC-OFF are no longer
1805 implemented with a counter, it does not matter how many times gc
1806 is switched on or off
1807 * bug fix: discriminating functions for generic function classes
1808 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1809 make invalid assumptions about method precedence order. (reported
1811 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1812 platforms. (thanks to Thiemo Seufer)
1813 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1814 * bug fix: correct pathname printing with printer escaping is on.
1815 (thanks to Kevin Reid)
1816 * bug fix: complex VOP definitions in "user-space" no longer trigger
1817 package locks. (reported by Zach Beane)
1818 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1819 for funcallable-instance functions, and is no different from
1821 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1822 characters in the current readtable. (reported by Nicholas Neuss)
1823 * bug fix: SUBTYPEP on various CONS types returns more a more
1824 accurate acknowledgment of its certainty.
1825 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1826 lists in some cases. This partially fixes bug 384.
1827 * flush all standard streams before prompting in the REPL and the
1829 * bug fix: signal handling and triggering gc do not conflict
1830 directly anymore, in particular a high frequency sb-sprof does
1831 not prevent gc from running
1832 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1833 approximation for timezone and DST information between the
1834 universal time epoch and the smallest negative 32-bit time_t.
1835 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1836 year 1899 so long as the encoded time is non-negative.
1837 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1838 the correct answer for non-constant rotate amounts when performing
1839 an inline 32-bit rotation.
1841 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1842 there is only one thread in the session
1843 ** bug fix: memory leak for streams created in one thread and
1844 written to in another
1845 ** bug fix: lockup when compiled with gcc4
1846 ** bug fix: race that allows the gc to be triggered when gc is
1848 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1849 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1851 ** bug fix: finished threads are gc'ed properly
1852 * fixed some bugs revealed by Paul Dietz' test suite:
1853 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1854 has arguments with a different set of applicable methods from
1855 the orignal arguments.
1856 ** The type error thrown by MAP now has a correct expected-type
1858 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1859 name a compiled function.
1860 ** Three MISC tests where a large bignum was improperly coerced to
1861 a float that couldn't represent that bignum during type
1862 derivation were fixed.
1863 ** SUBTYPEP can now handle types involving the negation of a
1864 list-form FUNCTION type.
1865 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1866 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1867 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1869 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1870 * New feature: Experimental support for bivalent streams: streams
1871 opened with :element-type :default now allow character and binary
1872 (unsigned-byte 8) I/O
1873 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1874 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1876 * bug fix: no more highly sporadic "couldn't check whether ~S is
1877 readable" when reading a stream and an interrupt hits in the middle
1878 of a select system call
1879 * compiler better recognizes complex arrays (reported by Eduardo
1881 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1882 for variables on SBCL built with linkage-tables. (reported by Luis
1884 * various error reporting improvements.
1885 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1886 (Thanks to Thiemo Seufer)
1887 * bug fix: nested reader invokations work correctly
1888 * bug fix: it is possible to have simultaneous references to foreign
1889 code and foreign data with the same name.
1891 ** added x86-64 support
1892 ** incompatible change: the threading api now works with thread
1893 objects instead of thread ids
1894 ** bug fix: threads are protected from signals and interruption when
1895 starting up or going down
1896 ** bug fix: a race where an exiting thread could lose its stack to gc
1897 ** bug fix: don't halt on infinite error in threads if possible
1898 ** fixed numerous gc deadlocks introduced in the pthread merge
1899 ** bug fix: fixed thread safety issues in read and print
1900 ** bug fix: debugger doesn't hang on session lock if interrupted at
1901 an inappropriate moment
1902 ** bug fix: run-program is now thread safe(r)
1903 ** bug fix: inner with-recursive-lock no longer releases the mutex
1904 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1905 * fixed some bugs revealed by Paul Dietz' test suite:
1906 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1907 EXPECTED-TYPE slots filled.
1908 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1910 ** MAKE-CONDITION accepts classes as type-designators.
1911 ** COMPILE may never return NIL.
1912 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1913 range before calling Unix time functions
1915 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1916 * numerous signal handling fixes to increase stability
1917 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1918 * minor incompatible change: we now correctly canonize default
1919 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1920 as per the MOP, rather than the historical (INITARG INITFUNCTION
1921 INITFORM). (reported by Bruno Haible)
1922 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1923 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1924 for more information.
1925 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1926 pathname is a directory pathname.
1927 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1928 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1930 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1931 :METACLASS argument in addition to a class name. (reported by
1932 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1933 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1934 (reported by Bruno Haible; patch from Gerd Moellmann)
1935 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1937 * bug fix: more cleanups to the floating point exception handling on
1938 x86-64 (thanks to James Knight)
1939 * bug fix: the compiler does not try to constant fold calls of
1940 COERCE and BIT-* functions when they return freshly constructed
1941 objects. (reported by Nikodemus Siivola and Paul Dietz)
1942 * optimization: improved the MIPS versions of generic (in the
1943 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1944 * optimization: direct conversion of (unsigned-byte 32) to floats on
1945 the PowerPC platform.
1946 * optimization: structure instances with raw slots now use less
1947 memory, and probably show better memory locality. (thanks to
1949 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1950 are treated as requests for stack allocation on the x86-64,
1951 Alpha-32, PPC and SPARC platforms.
1952 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1953 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1954 (thanks to Tony Martinez)
1956 ** gcing a dead thread can no longer lead to lockups
1957 ** threads block signals until they are set up properly
1958 ** errno is no longer shared by threads
1959 ** interrupt-thread restores the eflags register on x86
1960 ** fixed some lockups due to gc/thread interaction
1961 * fixed some bugs revealed by Paul Dietz' test suite:
1962 ** invalid dotted lists no longer raise a read error when
1963 *READ-SUPPRESS* is T
1964 ** signal an error if a symbol that names a declaration is used
1965 as the name of a type, or vice versa
1966 ** allow using the (declare (typespec var*)) abbreviation for
1967 (declare (type typespec var*)) with all type specifiers
1968 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1969 ** always return NIL from PPRINT-POP when OBJECT is NIL
1970 ** don't signal errors when pretty-printing malformed LABELS,
1971 FLET or MACROLET forms
1972 ** declarations in a DOLIST body are also in scope for the
1974 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1976 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1979 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1980 * fixed cross-compiler leakages that prevented building a 32-bit
1981 target with a 64-bit host compiler.
1982 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1983 opened with :IF-EXISTS :OVERWRITE.
1984 * fixed bug 281: error for an invalid qualifier in a short-form method
1985 combination method is not signalled until the faulty method is called.
1986 * bug fix: iteration variable type inferrer failed to deal with open
1987 intervals. (reported by Alan Shields)
1988 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1989 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1990 or more runtime options were provided to the sbcl binary.
1991 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1992 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1993 * optimization: there's now a fast-path for fixnum arguments in the
1994 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1995 * optimization: the code generated on x86-64 is more compact thanks
1996 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1997 * bug fix: floating-point exception handling now partly works on
1998 x86-64. (Thanks to James Knight)
1999 * improvement to the MIPS backend from Thiemo Seufer: C-style
2000 64-bit long long arguments and return values to alien functions
2002 * fixed some bugs revealed by Paul Dietz' test suite:
2003 ** the type-error signalled from WARN has a filled-in DATUM slot.
2004 ** the type-error required when a stream is not associated with
2005 a file has the stream as its datum.
2006 ** type-errors on single-floats on x86-64 no longer have
2007 :INVALID-OBJECT as the datum
2008 ** the type-errors signalled for invalid function names now have
2009 a correct expected type
2010 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2011 for structures defined with a :TYPE. Documentation strings for
2012 typed structures are no longer immediately discarded
2013 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2014 broadcast streams and synonym streams. FILE-LENGTH now also works
2015 on broadcast streams.
2017 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2018 * incompatible change: the --noprogrammer option, deprecated since
2019 version 0.7.5, has been removed. Please use the equivalent
2020 --disable-debugger option instead.
2021 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2023 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2024 inoperational for a while now) has been completely removed.
2025 * null lexical environments are now printed as #<NULL-LEXENV>,
2026 significantly reducing the amount of clutter in typical backtraces.
2027 * documentation on weak pointers, finalization, and after GC hooks
2028 has been added to the manual.
2029 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2030 as other specialized array types, is much faster. SUBSEQ and
2031 COPY-SEQ on such arrays have also been sped up.
2032 * optimization: EQL is now more efficient when at least other argument
2033 is known to be of type (OR FIXNUM (NOT NUMBER)).
2034 * fixed bug: the runtime is now less vulnerable to changes in the
2035 size of the SBCL object on OS X, and virtual memory is reserved for
2036 all spaces that need to be at a fixed address.
2037 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2038 * fixed bug: finalizers and after GC hooks that cause consing are now
2040 * fixed bug: compiler error messages and summaries are now printed to
2041 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2042 * fixed inference of the upper bound of an iteration variable.
2043 (reported by Rajat Datta).
2044 * fixed bug 211e: calling local functions with duplicated constant
2045 keyword argument no longer causes a bogus style warning about an
2047 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2048 ftype declaration to be disregarded. (reported by Dave Roberts)
2049 * fixed bug 373: caused by erronous compilation of references to alien
2050 variables in the runtime on ppc/darwin.
2051 * fixed bug 376: CONJUGATE type deriver.
2052 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2053 (reported by Baughn on #lisp)
2054 * a cross-compiler bug on non-x86 platforms has been identified and
2055 fixed. (thanks to Bruno Haible)
2056 * improvements to the MIPS runtime code for increased stability.
2057 (thanks to Thiemo Seufer)
2058 * increased the maximimum compact environment size to allow
2059 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2060 * partial workaround for bug 135: don't name the function we're
2061 calling for hairy cases of EVAL, so as not to accumulate one environment
2062 entry per call to EVAL. (reported by Kevin Reid)
2063 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2064 * optimization: type testing for non-vector arrays should be faster.
2065 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2067 * bug fix: setting 31st element of a bit vector to zero did not work
2069 * bug fix: redefining a class definition which failed due to a
2070 previous accessor / function clash now works (but see BUGS entry
2071 #380 for more problems in this area). (thanks to Zach Beane)
2072 * the long form of DEFINE-METHOD-COMBINATION disables method group
2073 checking when given a single method group with pattern *. (thanks
2075 * on x86 compiler supports stack allocation of results of simple
2076 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2077 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2078 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2079 * fixed some bugs related to Unicode integration:
2080 ** the restarts for recovering from input and output encoding
2081 errors only appear when there is in fact such an error to
2083 * fixed some bugs revealed by Paul Dietz' test suite:
2084 ** MISC.549 and similar: late transformation of unsafe type
2085 assertions into derived types caused unexpected code
2087 ** SCALE-FLOAT type deriver is less wrong.
2088 ** type derivers for EXP, LOG and similar functions compute result
2089 types for complex arguments better.
2090 ** (MISC.563) CONJUGATE type deriver works for very restricted
2092 ** out-of-line type testers for character strings are available.
2093 ** EQUAL compiler transform understands specialness of objects
2095 ** accessing double-floats stored on the stack now works on x86-64.
2096 ** debugger internals could sometimes create invalid lispobjs,
2097 resulting in GC crashes.
2098 ** MISC.548: type check weakening can convert required type into
2100 ** initialization forms for bindings are not in scope of free special
2103 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2104 * incompatible change: thread support for non-NPTL systems has
2105 been removed - locking is buggy and unreliable. A threaded
2106 SBCL build will now warn at startup and refuse to create
2107 new threads, unless futex support is detected in the kernel
2108 * incompatible change: the top level REPL now has only an ABORT
2109 restart associated with it, not TOPLEVEL and ABORT as it used to.
2110 TOP and TOPLEVEL are now available as debugger commands for
2111 returning to the top level.
2112 * incompatible change: forms evaluated in the REPL now use the
2113 global optimization policy.
2114 * incompatible change: user- and system-initialization files are
2115 no longer processed with LOAD, but by READ and EVAL; hence the
2116 global optimization policy, startup package, readtable, etc,
2118 * The .fasl file format number has been incremented because of
2119 various incompatible changes.
2120 * internal entry point details and argument counts no longer appear
2121 in backtraces unless explicitly requested by setting
2122 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2123 * built-in and standard functions no longer have names like "top
2124 level local call to FOO".
2125 * fixed bug 32: functions defined in non-null lexical environments
2126 now have more legible printed representation
2127 * fixed bug 33: functions defined in non-null lexical environemnts
2128 are now more amenable to inspection by INSPECT.
2129 * workaround for bug 354: XEPs no longer appear in backtraces unless
2130 explicitly requested.
2131 * fixed bug: receiving the signal which results from attempting to
2132 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2133 SIGBUS on other *BSDs) on architectures where the C stack is also
2134 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2135 to debugger confusion or wild instability. (reported by Cheuksan
2137 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2138 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2139 (reported by Lutz Euler)
2140 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2141 compiled code. (reported by Paul Dietz)
2142 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2143 specialized parameter to a method to something that is not TYPEP
2144 the specializer is now possible.
2145 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2146 face of package deletion.
2147 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2148 pathname. (reported by David Tolpin on c.l.l)
2149 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2150 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2151 STRING-STREAM and FILE-STREAM, respectively.
2152 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2153 than previously; in addition, its test suite is now run on build.
2154 (thanks to Luke Gorrie)
2155 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2156 * optimization: added a immediate representation for single-floats
2158 * fixed some bugs revealed by Paul Dietz' test suite:
2159 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2160 ** attempting to create a package with a colliding nickname causes
2161 correctable errors to be signalled.
2162 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2163 ** several x86-64 backend bugs related to sign-extension of immediate
2166 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2167 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2168 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2170 * fixed loading of multiply forward-referenced layouts.
2171 (thanks to Cheuksan Wang)
2172 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2173 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2174 (see COMPILE-FILE documentation for details of :PRINT option)
2175 * fixed bugs 19 and 317: fixed-format floating point printing is
2176 more accurate. This also fixes a bug reported by Adam Warner
2177 related to the ~@F format directive.
2178 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2180 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2181 dispatch table if the from-char is a dispatch macro character.
2182 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2183 by the wordsize no longer ignore the last word. (reported by Lutz
2185 * fixed bug in type checking of dynamic-extent variables. (reported
2187 * optimization: sequence traversal functions use their freedom to
2188 coerce function designators to functions.
2189 * optimization: code with many calls to CLOS methods specialized on
2190 CLOS classes has had redundant type checks removed.
2191 * optimization: added declarations to speed up operations that access
2192 the internal character database (for example STRING-UPCASE)
2193 * optimization: comparison operations between floats and sufficiently small
2194 fixnums no longer create extra rationals
2195 * fixed some bugs related to Unicode integration:
2196 ** portions of multibyte characters at the end of buffers for
2197 character-based file input are correctly transferred to the
2198 start of the buffer at the next read.
2199 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2200 passing it through to OPEN.
2201 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2202 argument given to internal calls to OPEN.
2203 ** the built-in comment readers (introduced by character sequences
2204 ";" and "#|") are more forgiving to encoding errors; they will
2205 STYLE-WARN and then attempt to resync the stream at a character
2206 boundary. (thanks to Teemu Kalvas)
2207 * fixed some bugs in the x86-64 port:
2208 ** Negative short int return values from c-calls are sign-extended
2210 ** The stack is aligned to 16-bytes for c-calls, as required by
2211 the ABI. (thanks to Cheuksan Wang)
2212 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2213 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2214 for enums. (thanks to Vincent Arkesteijn)
2215 ** Multiple small optimizations and bugfixes for floating point
2217 * fixed some bugs revealed by Paul Dietz' test suite:
2218 ** Space, Tab, Linefeed, Return and Page have the invalid
2219 secondary constituent character trait.
2220 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2222 ** WITH-INPUT-FROM-STRING should only update the index place on
2224 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2225 works more reliably.
2226 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2227 with one-dimensional arrays with fill pointers.
2228 ** TYPE-OF failed on a complex with an integer realpart and a
2230 ** compiler failure during type inference for the code of form
2231 (IF (EQL X (THE ...)) ...) (MISC.535).
2233 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2234 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2235 platform. The port supports 61-bit fixnums, large memory spaces
2236 and reloading shared object files.
2237 * enhancement: saving cores with foreign code loaded is now
2238 supported on ppc/Darwin in addition to the previously supported
2240 * enhancement: the statistical profiler now walks deeper into the
2241 call stack for more meaningful call-graphs and accrued time
2242 reports (x86/x86-64 only). It also now reports time spent in
2244 * enhancement: it is now possible to trace most individual methods
2245 of a generic function in addition to tracing the generic function
2247 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2249 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2250 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2251 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2252 produces an error. (thanks to Vincent Arkesteijn)
2253 * bug fix: NAMESTRING on pathnames with :WILD components in their
2254 directories works correctly. (thanks to Artem V. Andreev)
2255 * fixed bug 125: compiler preserves identity of closures. (reported
2257 * bug fixed: functions with &REST arguments sometimes failed with
2258 "Undefined function" when compiled with (DEBUG 3). (reported by
2260 * bug fix: overflow during compiling of setting element of a bit
2261 vector with constant index and value. (reported by Timmy Douglas)
2262 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2263 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2264 * on x86 compiler partially supports stack allocation of dynamic-extent
2266 * GO and RETURN-FROM do not check the extent of their exit points
2267 when compiled with SAFETY 0.
2268 * fixed some bugs related to Unicode integration:
2269 ** encoding and decoding errors are now much more robustly
2270 handled; it should now be possible to recover even from invalid
2271 input or output to the terminal. (thanks to Teemu Kalvas)
2272 ** provided a first cut at implementing STRING-TO-OCTETS and
2273 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2274 ** altered the SB-MD5 contributed module slightly, changing the
2275 interface just enough for it to be supportable for builds where
2276 lisp characters are not eight bits.
2277 * fixed some bugs revealed by Paul Dietz' test suite:
2278 ** the FORMATTER-generated functions for ~V[ conditionals require
2279 the correct number of arguments.
2280 ** READ-FROM-STRING returns the mandated second value when applied
2281 to displaced strings.
2282 ** the #\Rubout and #\Backspace characters are treated as invalid
2283 constituent characters by the tokenizer.
2285 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2286 * new feature: reloading changed shared object files with
2287 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2288 * new feature: references to foreign variables and functions
2289 can now be compiled and loaded before the corresponding shared
2290 object file is loaded, as long as the foreign definitions are
2291 available at runtime.
2292 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2293 supported. (thanks to Dan Debertin)
2294 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2295 just keywords. (thanks to Vincent Arkesteijn)
2296 * fixed bug #331: structure-class instances corresponding to
2297 DEFSTRUCT forms are now created eagerly.
2298 * fixed bug #345: backtraces from calls to undefined functions work
2299 on x86 as well. Related bug #61 is now also partially fixed on x86
2300 (backtraces from throws to unknown catch tags.)
2301 * bug fix: lambda-list parsing is now stricter vrt. order and number
2302 of lambda-list keywords.
2303 * bug fix: as specified by AMOP, an error is signalled if a
2304 class-option appears multiple times in a DEFCLASS form. (reported
2306 * bug fix: RANDOM can be compiled when the compiler derives the type
2307 of its numeric argument as a disjoint set of small integers.
2308 (reported by Paul Dietz)
2309 * bug fix: starting a core saved with shared objects loaded when
2310 those objects are not available no longer causes threaded SBCL to
2311 hang. (reported by Sean Ross)
2312 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2314 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2315 argument, which must always be NIL. (reported by Kalle Niemitalo)
2316 * bug fix: printing 1.0d+23 no longer results in an error.
2317 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2318 * bug fix: structure slot setters preserve evaluation order. (thanks
2320 * fixed some bugs related to Unicode integration:
2321 ** RUN-PROGRAM can allow its child to take input from a Lisp
2322 stream. (reported by Stefan Scholl)
2323 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2324 locales. (reported by Ken Causey)
2325 * fixed some bugs revealed by Paul Dietz' test suite:
2326 ** INCF, DECF and REMF evaluate their place form as specified in
2328 ** FORMATTER expands ~{ ~} iteration directives with V or #
2329 parameters correctly.
2330 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2331 consequent uses no arguments correctly.
2332 ** the system has a partial understanding of the (COMPLEX RATIO)
2334 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2337 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2338 * new feature: a build-time option (controlled by the :SB-UNICODE
2339 keyword feature, enabled by default) for building the system with
2340 support for the entire 21-bit character space defined by the
2342 * new feature: the system now has rudimentary external-format
2343 support; the primary user-visible change at this time is that
2344 characters with the high bit set (such as lower-case-e-acute) will
2345 print correctly to a terminal in a UTF-8 environment.
2346 * minor incompatible change: BASE-CHAR no longer names a class;
2347 however, CHARACTER continues to do so, as required by ANSI.
2348 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2349 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2350 has been moved to the SB-EXT package (temporarily re-exported from
2352 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2353 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2355 * fixed bug #351: better error-handlind and reporting for malformed
2357 * fixed bug #350: bignum-printing is now more memory-efficient,
2358 allowing printing of very large bignums, eg. (expt 2 10000000).
2359 (reported by Bruno Haible)
2360 * fixed bug #302: better primitive-type selection for intersection
2362 * fixed bug #308: non-graphic characters now all have names, as
2363 required. (reported by Bruno Haible)
2364 * bug fix: redefining a class with different superclasses now correctly
2365 removes it from the direct-subclasses of its previous superclasses.
2366 (reported by David Morse)
2367 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2368 new value now works. (reported by Bruno Haible)
2369 * bug fix: correct canonicalization of multiple non-standard slot
2370 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2371 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2372 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2373 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2374 now exists, an signals an error.
2375 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2376 during macroexpansion for non-iteration variables as well. (reported
2377 by Bruno Haible for CMUCL)
2378 * bug fix: Cyclic structures and unprintable objects in compiler
2379 messages no longer cause errors. (reported by Bruno Haible)
2380 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2381 and READ-FROM-STRING all now return a primary value of NIL if
2382 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2383 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2384 (reported by Bruno Haible for CMUCL)
2385 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2386 do not designate a condition. (reported by Bruno Haible for
2388 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2389 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2390 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2391 permitting resolution in favour of any of the conflicting symbols.
2392 (reported by Bruno Haible for CMUCL)
2393 * bug fix: EQUAL compiler optimizations is less aggressive on
2394 strings which can potentially compare true despite having distinct
2395 specialized array element types.
2396 * bug fix: unit enumerations can be defined without dividing by
2397 zero. (thanks to Vincent Arkesteijn)
2398 * FORMAT compile-time argument count checking has been enhanced.
2399 (report from Bruno Haible for CMUCL)
2400 * a partial workaround for the bug 262: the compiler does not try to
2401 inline-expand a local function doing RETURN-FROM from a deleted
2402 BLOCK. (thanks to Peter Denno for the bug report and to David
2403 Wragg for the simple test case)
2404 * fixed some bugs revealed by Paul Dietz' test suite:
2405 ** RENAME-PACKAGE allows all package designators as new package
2407 ** constraint propagation and lambda variable substitution are
2408 more cautious in dealing with partially deleted code.
2409 ** compiler sometimes lost reoptimization passes.
2410 ** CERROR, given a condition as condition designator, passes its
2411 remaining arguments to the continue format control without
2413 ** Case-altering FORMAT directives work correctly on non-ASCII
2415 ** The REAL type specifier handles bounds outside the floating
2416 point ranges without signalling FLOATING-POINT-OVERFLOW.
2417 ** Functions with IR1-transformations can create intercomponent
2418 references to global functions.
2419 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2421 ** FORMAT ~R treats a nil value for its first parameter correctly.
2423 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2424 * enhancement: saving cores with foreign code loaded is now
2425 supported on x86/NetBSD and sparc/Linux in addition to the previously
2426 supported platforms.
2427 * bug fix: on some platforms repeated installations caused multiple
2428 copies of HTML documentation to be installed -- should not happen
2429 any more. (reported by Stefan Scholl)
2430 * bug fix: parsing self-recursive alien record types multiple times
2431 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2432 original patch by Helmut Eller for CMUCL)
2433 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2434 (thanks to Richard Kreuter)
2435 * bug fix: defining classes whose accessors are methods on existing
2436 generic functions in other (locked) packages no longer signals
2437 bogus package lock violations. (reported by François-René Rideau)
2438 * bug fix: special variables as DEFMETHOD parameters no longer have
2439 associated bogus type declarations. (reported by David Wragg and
2441 * bug fix: read-write consistency on streams of element-type
2442 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2443 * bug fix: redefiniton of the only method of a generic function with
2444 no DEFGENERIC no longer emits a full WARNING. In addition,
2445 redefinition of generic functions with no DEFGENERIC to an
2446 incompatible lambda list now signals an error. (thanks to Zach
2448 * bug fix: DEFGENERIC now works even when there's a function of the
2449 same name in an enclosing lexical environment. (thanks to Zach
2451 * fixed compiler failure, caused by instrumenting code during
2452 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2453 * optimization: added loop analysis and improved register allocation
2454 * fixed some bugs revealed by Paul Dietz' test suite:
2455 ** POSITION on displaced vectors with non-zero displacement
2456 returns the right answer.
2457 ** (SIMPLE-STRING) is a valid type specifier for sequence
2459 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2461 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2462 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2464 ** The justification version of the FORMAT ~< directive treats
2465 non-zero minpad parameter correctly.
2467 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2468 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2469 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2470 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2471 the supported interface.
2472 * new feature: Single-stepping of code compiled with DEBUG 2 or
2473 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2474 * new feature: saving cores with foreign code loaded is now
2475 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2476 Timothy Moore's work for CMUCL)
2477 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2478 parameters to * instead of NIL if no initform is supplied.
2479 (reported by Johan Bockgaard)
2480 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2481 correctly when FUNCALL appears as the car of the form. Note:
2482 despite this FUNCALL forms are not currently subject to
2483 compiler-macro expansion. (port of Raymond Toy's fix for the
2484 same from CMUCL, reported by Johan Bockgaard)
2485 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2486 (thanks for Teemu Kalvas)
2487 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2488 case instead of terminating the iteration (thanks for Julian
2489 Squires, Sean Champ and Raymond Toy)
2490 * bug fix: incorrect expansion of defgeneric that caused a style
2491 warning. (thanks for Zach Beane)
2492 * bug fix: slot accessor effective method computation works properly
2493 for classes with multiple non-standard applicable methods on
2494 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2495 * on x86 compiler supports stack allocation of results of LIST and
2496 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2497 CMUCL implementation by Gerd Moellmann)
2498 * fixed some bugs revealed by Paul Dietz' test suite:
2499 ** FORMAT strings with both the ~<~:;~> form of the justification
2500 directive and pretty-printing directives cause an error.
2502 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2503 * incompatible change: the internal functions
2504 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2505 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2506 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2507 instead of the old functions.
2508 * new feature: on platforms where "dladdr" is available foreign
2509 function names now appear in backtraces. (based on Helmut Eller's
2511 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2512 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2514 * documentation: ASDF manual is now distributed with SBCL.
2515 * bug fix: SBCL can now load its contributed modules with REQUIRE
2516 even if the system-provided entries have been removed from
2517 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2519 * bug fix: backtraces involving undefined functions or assembly
2520 routines are more informative. (thanks to Brian Downing)
2521 * bug fix: mutually referent alien structures now work correctly.
2522 (reported by Rick Taube)
2523 * bug fix: structures defined by WITH-ALIEN can be referred to
2524 within other definitions in the same WITH-ALIEN.
2525 * bug fix: division operators (MOD, TRUNCATE and the like) with
2526 constant zero divisors and integer dividends no longer generate
2528 * bug fix: provide default methods for INPUT-STREAM-P and
2529 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2530 * bug fix: improve the detection and resolution of MOP metacycles
2531 (where effective method computation depends on the generic function
2532 having its effective method computed). (reported by Bruno Haible)
2533 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2534 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2535 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2536 * bug fix: pathnames with relative directory components can be
2537 represented relative to default pathnames.
2538 * optimization: in taking the GCD of bignums, reduce the two bignums
2539 to approximately the same size (using Euclid's algorithm) before
2540 applying the more sophisticated binary GCD. (thanks to Juho
2542 * optimization: COUNT on bitvectors now operates word-at-a-time.
2543 * optimization: ASH with a positive, but not necessarily constant,
2544 (leftwards) shift, when in a modular context, is compiled to a
2546 * fixed some bugs revealed by Paul Dietz' test suite:
2547 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2548 if the corresponding argument is NIL.
2549 ** FORMAT directives accepting a minpad parameter treat negative
2551 ** literal commas as character parameters to FORMAT directives are
2553 ** literal spaces directly after ~<Newline> directives within a
2554 format-logical-block (~:< ~@:>) do not induce :FILL-style
2555 conditional newlines.
2556 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2557 as specified: it no longer includes conditional newlines).
2558 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2560 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2561 :PER-LINE-PREFIX argument does not evaluate to a string.
2563 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2564 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2565 the manual for details; for now, package locks can be disabled by
2566 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2567 if no major problems are found then it is likely that they will be
2568 compiled in unconditionally.
2569 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2570 unsupported operators on all platforms. To load a shared library
2571 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2572 object file, link it into a shared library outside of SBCL and
2573 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2575 * bug workaround: ROOM T and the GC were not working together
2576 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2577 reporting facility aren't preserved. That reporting has been
2578 disabled, so now until and unless someone figures out how to make it
2579 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2580 * minor incompatible change: as threatened around sbcl-0.8.0, the
2581 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2582 The symbols are also exported from SB-PCL for backwards
2583 compatibility, but more so than before SB-PCL should be treated as
2584 an implementation-internal package.
2585 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2587 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2588 It is known as of this release not to work on the Alpha, however.
2589 * fixed bug #167: errors signalled due to illegal syntax in method
2590 bodies are now more legible.
2591 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2592 designators and can hence be used with TYPEP.
2593 * fixed bug #333: CHECK-TYPE now ensures that the type error
2594 signalled, if any, has the right object to be accessed by
2595 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2596 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2597 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2599 * fixed bug #334: programmatic addition of slots using specialized
2600 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2601 and :ALLOCATION :CLASS slots.
2602 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2603 float exponents. (rereported by Peter Seibel)
2604 * fixed a bug: #\Space (and other whitespace characters) are no
2605 longer considered to be macro characters in standard syntax by
2606 GET-MACRO-CHARACTER.
2607 * fixed bug: initialization of condition class metaobjects no longer
2608 causes an instance of the condition to be created. (reported by
2610 * fixed bug: it is now possible to have slots such that
2611 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2612 description is neither :INSTANCE nor :CLASS.
2613 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2614 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2615 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2616 non-standard SLOT-DEFINITION-ALLOCATION values.
2617 * fixed bugs: various race conditions handling exiting threads.
2618 CL-PPCRE's thread torture test now passes.
2619 * fixed arguably-a-bug: GC time proportional to number of threads in
2620 system even when most of them are idle
2621 * optimization: improved performance of BIT and SBIT on bit-vectors.
2622 * .fasl file incompatibility: The fasl file version number has
2623 been incremented because of changes associated with package locks.
2625 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2626 * minor incompatible change: the system no longer provides
2627 optimization hints (conditions which are TYPEP
2628 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2629 mode; these hints are still emitted when higher SPEED optimization
2631 * new contrib module: a sampling profiler (profiling by statistical
2632 sampling, rather than by instrumenting functions) is available as
2633 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2634 * the behaviour of the standard function ED is now customizeable by
2635 third parties through a hook variable: see ED's documentation
2636 string for information on the protocol.
2637 * the compiler no longer emits efficiency notes for (FUNCALL X)
2638 when the type of X is uncertain under default optimization
2640 * fixed bug 276: mutating a binding of a specialized parameter to a
2641 method to something that is not TYPEP the specializer is now
2643 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2644 exhibit the required behaviour on the x86 platform. (thanks to
2645 Peter van Eynde, Eric Marsden and Bruno Haible)
2646 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2647 even for difficult arguments. (reported by Peter Graves)
2648 * fixed bug 141a: the backquote printer now descends quoted
2650 * fixed another bug in backquote printing: no more destructive
2651 modification of the form's list structure. (reported by Brian
2653 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2654 move between its address being taken and the call to
2655 interrupt_thread, fixing a crashing race condition.
2656 * the SB-POSIX contrib implementation has been adjusted so that it
2657 no longer exhibits ridiculously poor performance when constructing
2658 instances corresponding to C structs.
2660 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2661 * minor incompatible change: the sb-grovel contrib now treats C
2662 structures as alien (in the sense of SB-ALIEN) objects rather than
2663 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2664 has implications for memory management of client code
2665 (sb-grovel-returned objects must now be manually managed) and for
2666 type safety (alien objects now have full types).
2667 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2668 used to control emission of compiler diagnostics, rather than the
2669 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2670 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2671 quality should be considered deprecated.
2672 * install.sh now installs the user manual as well
2673 * (not quite a new documentable feature, but worth considering in
2674 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2675 beginnings of a semantically meaningful condition hierarchy is
2676 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2678 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2679 designator as the defaults argument.
2680 * fixed bug: Displaced arrays whose displaced-to array has become
2681 too small now cause ARRAY-DIMENSION to signal an error, providing
2682 for safer bounds-checking. (reported by Bruno Haible)
2683 * fixed bug: DEFCLASS slot definitions with identical :READER and
2684 :WRITER names now signal a reasonable error. (reported by Thomas
2686 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2688 * fixed bug: Invalid slot specification errors now print correctly.
2689 (thanks to Zach Beane)
2690 * fixed bug 320: Shared to local slot value transfers in class
2691 redefinitions now happen corrently from superclasses as
2692 well. (reported by Bruno Haible)
2693 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2695 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2696 improper lists correctly. (reported by Bruno Haible)
2697 * fixed bug 313: source-transform for <fun-name> was erroneously
2698 applied to a call of a value of a variable with name <fun-name>.
2699 (reported by Antonio Menezes Leitao)
2700 * fixed bug 307: The obsolete instance protocol ensures that
2701 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2702 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2703 remove dead unknown-values globs from the middle of the stack.
2704 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2705 easier to resume long computations after using *BREAK-ON-SIGNALS*
2706 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2707 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2708 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2709 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2710 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2711 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2713 * fixed a (fairly theoretical) bug in string printing: if
2714 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2715 does not have array-element-type equal to the most general string
2717 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2718 function name. (thanks to Nikodemus Siivola)
2719 * fixed bug: compile-time format string checker failed on
2720 non-closed ~{. (reported by Thomas F Burdick)
2721 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2722 don't warn on legal loop code involving more than one aggregate
2723 boolean. (thanks to Nikodemus Siivola)
2724 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2725 in keyword position to MAKE-PACKAGE was not being reported as
2727 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2728 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2730 * fixed bugs 280 and 312: the checking for multiple definitions in a
2731 file is less likely to become confused by uses of inline
2733 * fixed bug: the #S reader macro performs the keyword coercion
2734 specified for slot names. (reported by Kalle Niemitalo)
2735 * fixed bug: lambda lists may contain symbols whose names start with
2736 & but are not lambda-list-keywords; their occurrence triggers a
2738 * fixed bug 321: define-method-combination argument lambda lists do
2739 not cause invalid code to be generated when &optional or &aux
2740 variables have default values. (reported by Bruno Haible)
2741 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2742 have CLOS classes; this appears also to have fixed reported
2743 instability in stack exhaustion detection.
2744 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2745 MUFFLE-WARNING when no associated restart is present is now
2747 * optimization: rearranged the expansion of various defining macros
2748 so that each expands into only one top-level form in a
2749 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2751 * optimization: used a previously little-used slot in symbols to
2752 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2754 * fixed some bugs revealed by Paul Dietz' test suite:
2755 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2757 ** Fixed an optimization bug related to inheritance of initforms
2758 from local to shared slots.
2759 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2760 ** CAST optimizer forgot to flush argument derived type.
2761 ** print/read consistency on floats is now orders of magnitude
2762 more likely. (thanks also to Bruno Haible for a similar report
2764 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2765 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2766 for objects of type REAL. Make it so.
2767 ** SXHASH is specified (infelicitously) to respect similarity,
2768 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2769 it so. (thanks to Markus Ziegler)
2770 ** on the Alpha, the compiler succeeds in compiling functions
2771 returning a known number of arguments greater than 63.
2772 ** fixed handling of invalid NIL arguments in keyword position
2774 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2776 ** provided a readably-printable representation for RANDOM-STATE
2778 ** ensured that pathnames, where they have a namestring, always
2779 print using #P"..." syntax.
2781 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2782 * Support for the forthcoming 2.0 version of the NetBSD kernel
2783 running on x86 hardware has been added. (thanks to Perry
2784 E. Metzger most immediately, and others for their past work)
2785 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2786 current release version; SBCL's OpenBSD support had been broken
2787 since about the time of OpenBSD's switch to ELF binary format.)
2788 * [placeholder for DX summary]
2789 ** user code with &REST lists declared dynamic-extent, under high
2790 speed or space and low safety and debug optimization policy.
2791 * The manual has been converted to Texinfo format and the debugger
2792 chapter from the cmucl manual has been added.
2793 * A facility has been added to extract documentation strings from
2794 sbcl and store them as Texinfo-formatted snippets for inclusion in
2795 the manual (via Texinfo's @include directive)
2796 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2797 the test case to Dave Roberts)
2798 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2799 pointers. (reported by Sean Ross)
2800 * bug fix: PROFILE output is printed nicely even for large numerical
2801 values. (thanks to Zach Beane)
2802 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2803 greater than 32 handle EOF correctly.
2804 * bug fix: on X86 an immediate argument of the IMUL instruction is
2805 correctly printed in disassembly. (thanks to Lutz Euler)
2806 * bug fix: class slots in redefined classes preserve their old
2807 values. (thanks to Bruno Haible and Nikodemus Siivola)
2808 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2809 succeeds. (reported by Marco Baringer)
2810 * fixed some bugs revealed by Paul Dietz' test suite:
2811 ** READ-SEQUENCE now works on ECHO-STREAMs.
2812 ** RATIONALIZE works more according to its specification. (thanks
2815 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2816 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2817 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2818 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2819 should matter to you only if you rebind the printer control
2820 variables and then find you want different bindings in the
2821 debugger than in the ordinary execution of your program.)
2822 * The runtime build system has been tweaked to support building
2823 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2824 assembler and linker. (thanks to Nikodemus Siivola)
2825 * Unbound, undefined, undeclared variables now trigger full
2826 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2827 more often programmer error than deliberate exploitation of undefined
2829 * optimization: the hash algorithm for strings has changed to one
2830 that is less vulnerable to spurious collisions. (thanks to Juho
2832 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2833 needless bounds checking. (thanks to Juho Snellman)
2834 * optimization: implemented multiplication as a modular
2835 (UNSIGNED-BYTE 32) operation on the PPC backend.
2836 * fixed some bugs revealed by Paul Dietz' test suite:
2837 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2839 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2840 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2841 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2843 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2845 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2847 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2848 their output stream on EOF from read.
2849 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2850 have been read to end-of-file.
2851 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2853 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2854 description of determination of which consecutive characters
2856 ** Printing the "Space" character with escaping on now yields "#\\ ",
2857 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2858 ** Reading floating-point numbers with *READ-BASE* set to a number
2859 less than 10 works correctly.
2860 ** Reading floating-point numbers with *READ-BASE* set to a number
2861 more than 10 works correctly.
2862 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2863 the readtable currently in effect.
2865 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2866 * minor incompatible change: parsing of namestrings on a physical
2867 (Unix) host has changed; numbers after the final #\. in a
2868 namestring are no longer interpreted as a version field. This is
2869 intented to be largely invisible to the user, except that the
2870 meaning of the namestring "*.*.*" has changed: it now refers to a
2871 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2872 should usually be replaced by
2873 (make-pathname :name :wild :type :wild :version :wild)
2874 with the added benefit that this is more likely to be portable.
2875 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2876 option to OPEN now signals an error if the file being opened
2877 exists; this may have an impact on existing code.
2878 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2879 signals received, on the PowerPC platforms (both Linux and
2880 Darwin). (thanks to Pierre Mai for pointing out the location of
2882 * several fixes on OS X: The system now builds and runs cleanly on
2883 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2884 (thanks to Brian Mastenbrook)
2885 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2886 second-resolution: integer multiples of 1/3600 between -24 and 24.
2887 (thanks to Vincent Arkesteijn)
2888 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2889 type when called with 1 argument; PEEK-CHAR checked type of
2890 PEEK-TYPE only after having read first character from a
2891 stream. (reported by Peter Graves)
2892 * bug fix: the garbage collector now has much better locality
2893 behaviour, and in particular no longer treats all memory as being
2894 exhausted when many small objects point to each other in a deeply
2896 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2897 recognized as being TYPEP their class.
2898 * bug fix: the PUSHNEW documentation string has been corrected.
2899 (thanks to Vincent Arkesteijn)
2900 * bug fix: defaulting of the value for the last of an atypically
2901 large number of multiple values being bound was not being
2902 performed correctly on the Alpha or PPC platforms
2903 * optimization: implemented multiplication as a modular
2904 (UNSIGNED-BYTE 32) operation on the x86 backend.
2905 * optimization: SEARCH on simple-base-strings can now be open-coded.
2906 (see also contrib/compiler-extras.lisp for inspiration for
2907 teaching the compiler about the Boyer-Moore algorithm).
2908 * value, returned by MAX (and MIN) called with several EQUALP, but
2909 not EQL, arguments now does not depend on compiler settings.
2910 * fixed some bugs revealed by Paul Dietz' test suite:
2911 ** in stack analysis liveness information is propagated from
2912 non-local entry points.
2913 ** pathwise CAST removing failed when the CAST node did not start
2915 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2916 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2918 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2919 host is already defined.
2920 ** RENAME-FILE works on streams instead of signalling an internal
2922 ** PEEK-CHAR uses the current readtable when determining whether
2923 or not a character is whitespace.
2924 ** MERGE-PATHNAMES handles the case when the pathname does not
2925 specify a name while the default-pathname specifies a version.
2926 ** Pathnames now stand a better chance of respecting print/read
2928 ** Attempting to use standardized file system operators with a
2929 pathname with invalid :DIRECTORY components signals a
2931 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2933 ** DIRECTORY on logical pathnames is more correct.
2934 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2935 signal a TYPE-ERROR if their argument is not a stream.
2936 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2937 designator argument does not designate a stream.
2938 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2939 examining the synonym.
2940 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2942 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2943 element-type larger than ([UN]SIGNED-BYTE 32).
2945 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2946 * When built with the :SB-FUTEX feature, threaded builds now take
2947 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2948 for faster/more reliable mutex and condition variable support.
2949 * Incompatible change (but one you probably shouldn't have been using
2950 anyway): the interface and code for arbitrating between multiple
2951 threads in the same user session has been redesigned.
2952 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2953 error when called without an explicit environment argument.
2954 (thanks to Nikodemus Siivola)
2955 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2956 confused about their position. (thanks to Adam Warner and Gerd
2958 * bug fix: Pretty printing backquoted forms with unquotations in the
2959 argument list position of various code constructs such as LAMBDA
2960 now works correctly. (reported by Paul Dietz)
2961 * bug fix: Pretty printing unquotations no longer loses all
2962 stream position information.
2963 * optimization: performance of string output streams is now less
2964 poor for multiple small sequence writes.
2965 * optimization: performance of CSUBTYPEP in the presence of complex
2966 expressions involving CONS and NOT many times has been improved.
2967 (reported by Paul Dietz)
2968 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2970 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2971 streams for the REPL, for the debugger, and so on. (thanks to
2973 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2974 * fixed some bugs revealed by Paul Dietz' test suite:
2975 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2976 is ignored if it is too small, rather than propagating through
2977 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2978 ** extremely complex negations of CONS types were not being
2979 sufficiently canonicalized, leading to inconsistencies in
2981 ** VALUES tranformer lost derived type.
2983 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2984 * fixed a bootstrapping bug: the build process no longer assumes
2985 that the various BOOLE-related constants have the same value in
2986 host and target lisps. (noted by Paul Dietz' test suite on an
2987 SBCL binary built from CLISP)
2988 * The system can now be dynamically linked on the MIPS platform,
2989 which enables dynamic loading of foreign code from Lisp. (thanks
2990 to Ralf Baechle for discussions on the MIPS ABI)
2991 * The system now records debugging information for its own source
2992 files in a filesystem-position-independent manner, relative to
2993 the "SYS" logical host.
2994 * fixed a compiler bug: MV-LET convertion did not check references
2995 to the "max args" entry point. (reported by Brian Downing)
2996 * tweaked disassembly notes to be less confident about proclaiming
2997 some instruction as an LRA. (thanks to Brian Downing)
2998 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2999 now each have their own history, command character, and other
3000 characteristics. (thanks to David Lichteblau)
3001 * fixed some bugs revealed by Paul Dietz' test suite:
3002 ** compiler failure in compiling LOGAND expressions including a
3004 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3005 shift greater than 32.
3006 ** FLUSH-DEST did not mark blocks for type check regeneration.
3007 ** HANDLER-CASE failed to accept declarations in handler clauses
3008 in some circumstances.
3010 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3011 * New code in contrib/sb-introspect (still probably not entirely
3012 stable yet) provides some support for smart Lisp development
3013 environments like SLIME.
3014 * The conditions signalled for errors occurring when loading .fasl
3015 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3016 in a way which should help ASDF recover gracefully.
3017 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3018 stuff has been cleaned up. If you code contrib/ stuff, this might
3019 affect you, and you can look at contrib/README, contrib/STANDARDS,
3020 and/or the 0.8.4.27 diff to check.
3021 * In full calls the compiler now does not generate checks for declared
3022 argument types for all arguments.
3023 * various threading fixes
3024 ** and some experimental patches which didn't make it into
3025 the main tree for this release, but which are shipped in
3026 contrib/experimental-thread.patch as a possible fix for some
3027 failures (deadlock, spinning...) in GC-intensive multithreaded
3029 * fixed PPC build problem (source code incompatibility of different
3030 library versions): added offsetof() hackery which attempts to divine
3031 where glibc maintainers put uc_mcontext today
3032 * fixed bug 282: compiler does not trust type assertions while passing
3033 arguments to a full call.
3034 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3035 &OPTIONAL VALUES type parameter.
3036 * fix bug 214: algorithm for noting rejected templates is now more
3037 similar to that of template seletion. (also reported by rydis on
3039 * fixed bug 141b: printing backquoted information readably and prettily
3040 inserts a space where necessary.
3041 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3042 (CHARACTER 10) are now reported as errors, rather than propagated
3043 as unknown types. (reported by piso on #lisp)
3044 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3045 with values NIL and :ERROR. (thanks to Milan Zamazal)
3046 * fixed bug 191c: CLOS now does proper keyword argument checking as
3047 described in CLHS 7.6.5 and 7.6.5.1.
3048 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3049 longer raise an error; further, using a list as a for-as-arithmetic
3050 counter now raises a meaningful error.
3051 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3052 now understood by sequence creation functions such as MAKE-SEQUENCE
3054 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3055 a STRING-INPUT-STREAM.
3056 * compiler enhancement: SIGNUM is now better able to derive the type
3058 * type declarations inside WITH-SLOTS are checked. (reported by
3060 * fixed some bugs revealed by Paul Dietz' test suite:
3061 ** incorrect optimization of TRUNCATE for a positive first
3062 argument and negative second.
3063 ** compiler failure in let-convertion during flushing dead code.
3064 ** compiler failure while deriving type of TRUNCATE on an
3065 interval, containing 0.
3066 ** ASH of a negative bignum by a negative bignum count now returns
3068 ** intersection of CONS types now canonicalizes properly, fixing
3069 inconsistencies in SUBTYPEP.
3071 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3072 * incompatible change: The --disable-debugger command line
3073 option now clobbers the debugger at a more fundamental
3074 level, by redefining #'INVOKE-DEBUGGER instead of by
3075 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3076 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3077 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3078 enter the debugger REPL and then suffer recursive errors
3079 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3080 while under the new system, BREAK in --disable-debugger mode
3081 terminates the system just as an unhandled error would.
3082 * fixed compiler performance when processing loops with a step >1;
3083 * bug fix: DOCUMENTATION now retrieves generic function
3084 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3085 support has been systematized, and now supports the methods
3086 specified by ANSI, along with a default method and a method for
3087 slot documentation. (reported by Nathan Froyd)
3088 * bug fix: effective methods associated with a generic function are
3089 no longer cached over a change of that generic function's method
3090 combination. (reported by Andreas Fuchs)
3091 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3092 names a non-existent directory. (thanks to Andreas Fuchs)
3093 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3094 the PPC platform, lead to stack corruption; this has been fixed.
3095 (reported by Rainer Joswig)
3096 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3097 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3098 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3099 arithmetic in general, is now much more reliable on the PPC
3101 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3102 platform now returns the right answer.
3103 * optimization: restored some effective method precomputation in
3104 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3105 precomputation is now tunable.
3106 * optimization: compiler-internal data structure use has been
3107 reviewed, and changes have been made that should improve the
3108 performance of the compiler by about 20%.
3109 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3110 simple-base-strings and simple-bit-vectors is improved.
3111 * optimization: the optimization of 32-bit logical and arithmetic
3112 functions introduced in version 0.8.3 on the x86 has been
3113 implemented on the mips, ppc and sparc platforms; an
3114 implementation of the same facility, but for 64-bit arithmetic,
3115 has been added for the alpha.
3116 * microoptimization: the compiler is better able to make use of the
3117 x86 LEA instruction for multiplication by constants.
3118 * bug fix: in some situations compiler did not report usage of
3119 generic arithmetic in (SPEED 3) policy.
3120 * bug 145b fix: compiler used wrong type specifier while converting
3121 MEMBER-types to numeric.
3122 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3124 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3125 index argument now works properly on the Alpha platform.
3126 * bug fix: floating point exception treatment on the Alpha platform
3128 * bug fix: FILE-POSITION works much better on string input and
3129 output streams. (thanks to Nikodemus Siivola)
3130 * bug fix: many threading/garbage collection symptoms sorted.
3131 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3132 might be pseudo-atomic.
3133 * internal change: Stopping for GC is now done with signals not
3134 ptrace. GC is now done in whichever thread wanted it, instead of
3136 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3138 * bug fix: result form in DO is not contained in the implicit
3140 * incompatible change: ICR structure is changed; the value part of
3141 CONTINUATION is now called LVAR; corresponding functions are
3142 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3143 * added type deriver for ISQRT (thanks to Robert E. Brown).
3144 * bug fix: better support for loading from the command line when an
3145 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3147 * fixed some bugs revealed by Paul Dietz' test suite:
3148 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3149 ** ROUND and FROUND now give the right answer when given very
3150 small float arguments.
3151 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3153 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3154 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3155 ** LCM should return a non-negative integer.
3156 ** PARSE-INTEGER returned the index of a terminator instead of the
3157 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3158 ** PARSE-INTEGER returned an incorrect index being applied to a
3160 ** LCM with two arguments of 0 returns 0 rather than signalling
3162 ** unsigned addition of a 32-bit constant with the high bit set no
3163 longer causes an internal compiler error on the x86.
3164 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3165 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3166 with negative last argument.
3167 ** byte specifiers with zero size and position no longer cause
3168 an error during type derivation.
3169 ** bignum multiplication on the Alpha platform now returns the
3171 * porting: The system now builds on SuSE AMD64, although it still
3172 generates a 32-bit binary.
3173 * .fasl file incompatibility: The fasl file version number has
3174 been incremented (because of the changes to internal compiler
3175 data structures referred to above).
3177 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3178 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3179 more accurately, on the Darwin kernel running on PowerPC hardware.
3180 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3181 * Compiler code deletion notes now signal a condition of type
3182 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3183 an associated MUFFLE-WARNING restart.
3184 * The compiler now performs limited argument count validation of
3185 constant format strings in FORMAT, and where appropriate in ERROR,
3186 CERROR and WARN. (thanks to Gerd Moellmann)
3187 * New ASDF-INSTALL contrib can be used for automatic download and
3188 installation of third-party Lisp code from CCLAN or other sites
3190 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3191 which forces another thread to execute a function supplied by the
3193 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3194 now accept and act upon their :ELEMENT-TYPE keyword argument.
3195 (reported by Martin Atzmueller, Edi Weitz)
3196 * bug fix: FILE-POSITION now accepts position designators up to
3197 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3198 is the greater. (thanks to Patrik Nordebo)
3199 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3200 Kalle Olavi Niemitalo)
3201 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3202 printable. (reported by Eric Marsden)
3203 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3204 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3205 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3206 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3207 (thanks to Stig Sandoe)
3208 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3209 UNREAD-CHAR now correctly includes the unread character in the
3210 target sequence. (thanks to Gerd Moellmann)
3211 * bug fix in threaded builds: the system can now be suspended and
3212 resumed by shell job control with minimal disruption.
3213 * bug fixes in times and timezones >2038AD
3214 * better handling of "where is GNU make?" problem in build scripts
3215 (thanks to Nikodemus Siivola)
3216 * new optimization: inside a named function any reference to a
3217 function with the same name is considered to be a self-reference;
3218 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3219 optimization quality.
3220 * new optimization on x86: logical functions and + now have
3221 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3222 used when the result is truncated to 32 bits.
3223 * VALUES declaration is partially enabled.
3224 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3225 * fixed some bugs revealed by Paul Dietz' test suite:
3226 ** The system now obeys the constraint imposed by
3227 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3228 types form a lattice under type intersection.
3229 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3230 ** ASSOC now ignores NIL elements in an alist.
3231 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3232 and (1+ MOST-POSITIVE-FIXNUM) answers.
3233 ** The addition of a method with invalid qualifiers to a generic
3234 function does not cause an error to be signalled immediately;
3235 a warning is signalled, and the error is generated only on
3236 calling the generic function.
3237 * changed .fasl file version number, in part to add type codes for
3238 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3239 obscure ANSI requirements
3241 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3242 * fixed bug 148: failure to inline-expand a local function left
3243 garbage, confusing the compiler.
3244 * fixed bugs 3cd: structure slot readers perform type check if the
3245 slot can have an invalid value (i.e. it is either not initialized
3246 or can be written with a less specific slot writer).
3247 * bug fix: the compiler now traps array references to elements off
3248 the end of an array; previously, the bounds checking in some
3249 circumstances could go off-by-one.
3250 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3252 * type declarations for array element types now obey the description
3253 on the CLHS page "Declaration TYPE", as per discussions on
3254 sbcl-help around 2003-05-08. This means that a declaration
3255 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3256 declaration, all references to BAR will be asserted or assumed
3257 (with THE, so dependent on compiler policy) to involve objects of
3258 type FOO. Note that no such declaration is implied in
3259 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3260 * declared types of functions from the "Conditions"
3261 chapter. (reported by Paul Dietz)
3262 * bug fix: CERROR accepts a function as its first argument.
3263 * bug fix: NTH an NTHCDR accept a bignum as index
3264 arguments. (reported by Adam Warner)
3265 * optimization: character compare routines now optimize comparing
3266 against a constant character. (reported by Gilbert Baumann)
3267 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3268 argument now works properly on the MIPS platform.
3269 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3270 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3271 declared to have a functional type, does not cause an error
3273 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3274 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3275 * fixed bug: if last continuation of a deleted block has a
3276 destination, this destination should be deleted too. (reported by
3278 * fixed a bug in the bootstrap process: the host compiler's values
3279 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3280 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3281 test case from Patrik Nordebo)
3282 * improved the ability of the disassembler on the PPC platform to
3283 provide helpful disassembly notes.
3284 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3285 the class in more cases than previously.
3286 * bug fix: FILE-POSITION now understands :START and :END for
3287 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3288 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3289 8) data. (thanks to David Lichteblau)
3290 * bug fix: it is possible to add a method to a generic function
3291 without lambda list.
3292 * bug fix: reader failed to signal END-OF-FILE inside an
3293 object representation. (reported by Nikodemus Siivola)
3294 * fixed some bugs revealed by Paul Dietz' test suite:
3295 ** LAST and [N]BUTLAST should accept a bignum.
3296 ** condition slot accessors are methods.
3297 ** (VECTOR NIL) is a subtype of STRING.
3299 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3300 * minor incompatible change: some nonsensical specialized lambda
3301 lists (used in DEFMETHOD) which were previously ignored now signal
3303 * minor incompatible change: the system is now aware of the types of
3304 variables in the COMMON-LISP package, and will signal errors for
3305 most violations of these type constraints (where previously they
3306 were silently accepted).
3307 * minor incompatible change: COMPILE-FILE now uses the freedom
3308 afforded (ANSI 3.2.2.3) to use derived function types for
3309 functions defined in the same file. This also permits the system
3310 to warn on static type mismatches and function
3311 redefinition. (Currently it does not work with high DEBUG level.)
3312 * minor incompatible change: VALUES declaration is disabled.
3313 * When issuing notes, the compiler now signals a condition of type
3314 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3315 restart for use in user handlers. It is expected that the
3316 COMPILER-NOTE condition will eventually become a condition
3317 supertype to a hierarchy of note types, which will then be
3318 handleable in a similar fashion. However, at the moment, no such
3319 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3320 but it's an implementation detail, not a classification for the
3322 * Changes in type checking closed the following bugs:
3323 ** type checking of unused values (192b, 194d, 203);
3324 ** template selection based on unsafe type assertions (192c, 236);
3325 ** type checking in branches (194bc).
3326 * A short form of VALUES type specifier has ANSI meaning (it has
3327 increased the number of situations when SBCL cannot perform type
3329 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3330 keywords or constants is permissible.
3331 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3332 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3333 classes. (thanks to Antonio Martinez)
3334 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3335 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3336 argument to a pretty-print stream. Instead, it leaves any such
3337 operations to DESCRIBE-OBJECT methods.
3338 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3340 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3341 lambda lists. (thanks to David Lichteblau)
3342 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3343 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3344 (thanks to Gerd Moellmann)
3345 * fixed bug 246: increased compilation speed of long
3346 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3348 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3350 * DEFCONSTANT now throws a condition of type
3351 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3352 constant to a non-EQL value; CONTINUE and ABORT restarts
3353 respectively change and preserve the value.
3354 * fixed bug 63: The code walker, part of the implementation of CLOS,
3355 is now better at handling symbol macros.
3356 * bug fix: There is no longer an internal implementation type named
3357 CL:LENGTH. (reported by Raymond Toy)
3358 * bug fix: In macro-like defining macros/special operators the
3359 implicit block does not enclose the lambda list.
3360 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3361 enclosed in parentheses) are not suitable as type specifiers, and
3362 their use properly signals an error now.
3363 * bug fix: An argument count mismatch for a type specifier in code
3364 being compiled no longer causes an unhandled error at compile
3365 time, but signals a compile-time warning.
3366 * fixed simple vector readable printing
3367 * bug fix: DESCRIBE takes more care over whether the class
3368 precedence list slot of a class is bound before accessing it.
3369 (reported by Markus Krummenacker)
3370 * bug fix: FORMATTER can successfully compile pretty-printer format
3371 strings which use variants of the ~* directive inside.
3372 * bug fix: SEARCH now applies its TEST predicate to the elements of
3373 the arguments in the correct order. (thanks to Wolfhard Buss)
3374 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3375 in type inference. (thanks to Robert E. Brown)
3376 * bug fix: Reading in symbols with an explicit package name of ""
3377 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3378 * fixed some bugs revealed by Paul Dietz' test suite:
3379 ** NIL is now allowed as a structure slot name.
3380 ** Arbitrary numbers, not just REALs, are allowed in certain
3381 circumstances in LOOP for-as-arithmetic clauses.
3382 ** Multiple class redefinitions before slot access no longer
3383 causes a type error.
3384 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3385 association between the name and a class.
3386 ** Generic functions with non-standard method-combination and over
3387 five methods all of which return constants no longer return NIL
3388 after the first few invocations. (thanks to Gerd Moellmann)
3389 ** CALL-NEXT-METHOD with no arguments now passes the original
3390 values of the arguments, even in the presence of assignment.
3391 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3393 ** DISASSEMBLE works with closures and funcallable instances.
3394 ** ADD-METHOD now returns the generic function, not the new method.
3395 ** FIND-METHOD signals an error if the lengths of the specializers
3396 is incompatible with the generic function, even if the ERRORP
3398 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3399 which its argument is a member.
3400 ** DEFCLASS only redefines the class named by its class-name
3401 argument if that name is the proper name of the class;
3402 otherwise, it creates a new class.
3403 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3404 of the UNBOUND-SLOT condition to the name of the slot.
3405 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3406 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3407 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3408 SLOT-MAKUNBOUND in the specified fashion.
3410 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3411 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3412 cross-compilation host. As a consequence, we can now bootstrap our
3413 way up to SBCL starting with a bare gcc toolchain and human-readable
3414 source code (first the source to CLISP, then the source to SBCL).
3415 * A contributed module containing a partial implementation of the
3416 simple-streams interface has been included. (thanks to Rudi
3418 * A contributed module implementing the RFC1321 Message Digest
3419 Algorithm, known as MD5, has been included.
3420 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3421 no longer has any effect, as the code controlled by this feature
3422 has been deleted. (As far as we know, no-one has ever built using
3423 this feature, and its semantics were confused in any case).
3424 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3425 work on conditions (as required by the ANSI specification),
3426 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3427 expected behaviour on conditions. Users should note, however,
3428 that such behaviour is not required by the ANSI specification,
3429 and so use of this behaviour may render their code unportable.
3430 * fixed some bugs revealed by Paul Dietz' test suite:
3431 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3433 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3434 ** SXHASH on condition objects no longer returns NIL.
3435 ** :ALLOCATION :CLASS slots are better treated; their values are
3436 updated on class redefinition, and initforms inherited from
3437 superclasses are applied.
3438 ** REMOVE-METHOD returns its generic function argument even when
3439 no method was removed.
3440 ** SHARED-INITIALIZE now initializes the values of the requested
3441 slots, including those with :ALLOCATION :CLASS.
3442 ** ALLOCATE-INSTANCE now works on structure classes defined via
3443 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3445 ** SLOT-EXISTS-P now works on conditions, as well as structures
3447 ** MAKE-LOAD-FORM now has the required methods on
3448 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3449 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3450 keyword, and now implements the SLOT-NAMES argument.
3451 ** methods with &OPTIONAL arguments no longer allow too many
3452 arguments to be passed in the call without error.
3453 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3454 option is consistent with the required arguments of the generic
3455 function lambda list.
3456 * bug fix: REQUIRE accepts a string designator. (Thanks to
3458 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3459 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3460 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3461 * bug fix: The compiler now checks for duplicated variables in macro
3463 * bug fix: SETQ on globals returns the correct value.
3464 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3465 not a COMPILER-ERROR (followed by some other strange error on
3466 choosing the CONTINUE restart).
3467 * bug fix: make.sh and friends are now more consistent in the way that
3468 they look for GNU "make".
3470 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3471 * experimental native threads support (on x86 Linux >=2.4 only).
3472 This is not compiled in by default: you need to add :SB-THREAD to
3473 the target features. See the "Beyond ANSI" chapter of the manual
3475 * fix for longstanding nonANSIism: The old distinction between
3476 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3477 The return value from CL:FIND-CLASS is now a CLOS class, and
3478 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3479 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3480 * An interface to the MetaObject Protocol, as described in Kiczales,
3481 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3482 MIT Press, 1991, is available from the SB-MOP package.
3483 * incompatible change: the SB-PCL package should now be considered
3484 a private implementation detail, and no longer a semi-private MOP
3486 * minor incompatible change: due to rearrangement for threads, the
3487 control stack and binding stack are now allocated at arbitrary
3488 addresses instead of being hardcoded per-port. Users affected by
3489 this probably have to be doing advanced things with shared
3490 libraries, and will know who they are.
3491 * minor incompatible change: Previously, all --eval forms used were
3492 processed with READ before any of them were processed with EVAL.
3493 Now each --eval form is processed with both READ and EVAL before
3494 the next --eval form is processed. (Thus package operations like
3495 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3496 work as the user might reasonably expect.)
3497 * minor incompatible change: *STANDARD-INPUT* is now only an
3498 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3500 * minor incompatible change: Y-OR-N-P is now character-oriented,
3501 not line oriented. Also, YES-OR-NO-P now works without errors.
3502 (thanks to Antonio Martinez)
3503 * sb-aclrepl module improvements: an integrated inspector, added
3504 repl features, and a bug fix to :trace command.
3505 * Known functions, which cannot be open coded by the backend, are
3506 considered to be able to check types of their arguments. (fixing
3507 a bug report by Nathan J. Froyd)
3508 * fixed a bug in computing method discriminating functions: It is
3509 now possible to define methods specialized on classes which have
3510 forward-referenced superclasses. (thanks to Gerd Moellmann)
3511 * fixed evaluation order in optional entries (reported by Gilbert
3513 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3514 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3515 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3516 required-by-AMOP NAME argument, as well as CLASS and
3517 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3518 * fixed bug 20: DEFMETHOD can define methods using names that are
3519 not the proper names of classes to designate class specializers.
3520 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3521 Unix file descriptors, instead of blowing up. (thanks to Antonio
3523 * Garbage collection refactoring: user-visible change is that a
3524 call to the GC function during WITHOUT-GCING will not do garbage
3525 collection until the end of the WITHOUT-GCING. If you were doing
3526 this you were probably losing anyway.
3527 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3528 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3529 TYPEP the latter but not the former.
3530 * The compiler issues a full WARNING for calls to undefined functions
3531 with names from the CL package.
3532 * MAP-INTO for a vector destination is open coded. (reported by
3533 Brian Downing on c.l.l)
3534 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3535 documentation string.
3536 * fixed some bugs revealed by Paul Dietz' test suite:
3537 ** COPY-ALIST now signals an error if its argument is a dotted
3539 ** Condition slots are now accessed more correctly in the presence
3540 of multiple initargs for a given slot.
3541 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3542 exclude restarts of the same name associated with a different
3544 ** DEFCLASS of forward-referenced classes with another
3545 forward-referenced class in the superclasses list no longer
3547 ** Condition slots are now initialized once each, not multiple
3548 times. (thanks to Gerd Moellmann)
3549 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3551 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3552 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3553 arguments contain duplicated elements.
3554 ** RESTART-CASE understands local macros.
3555 ** RESTART-CASE associates exactly its own restarts with a condition.
3556 ** ENDP in safe mode checks its argument to be of type LIST.
3557 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3558 in question is unbound.
3559 ** Optimized MAKE-INSTANCE functions no longer cause internal
3560 assertion failures in the presence of duplicate initargs.
3561 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3562 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3563 propagated even in the ctor optimized implementation of
3565 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3567 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3568 * a better implementation of SXHASH on (simple) bit vectors,
3569 measured both in execution speed and in distribution of results
3570 over the positive fixnums, has been installed. Likewise, a better
3571 implementation of EQUAL for simple bit vectors is now available.
3572 * fixed CEILING optimization for a divisor of form 2^k.
3573 * fixed bug 240 (emitting extra style warnings "using the lexical
3574 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3575 by Antonio Martinez)
3576 * fixed SXHASH, giving different results for NIL depending on type
3577 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3578 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3579 variable. (found by Rolf Wester)
3580 * SBCL does not ignore type declarations for special
3581 variables. (reported by rif on c.l.l 2003-03-05)
3582 * some bug fixes in contrib/sb-aclrepl/
3583 * fixed some bugs revealed by Paul Dietz' test suite:
3584 ** a bug in the CONS type specifier, whereby the CAR and CDR
3585 types got intertwined, has been fixed;
3586 ** the type system is now able to reason about the interaction
3587 between INTEGER and RATIO types more completely;
3588 ** APPEND, [N]REVERSE and NRECONC check that those their
3589 arguments, which must be proper lists, are really so;
3590 ** An array specialized to be unable to hold elements has been
3591 implemented, as required -- yes, really -- by ANSI;
3592 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3593 on malformed property lists;
3595 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3596 * incompatible packaging change: in line with Unix convention,
3597 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3598 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3599 * REQUIRE and PROVIDE are now optionally capable of doing something
3600 useful. See the documentation string for REQUIRE.
3601 * infrastructure for a managed SBCL contrib system: contributed
3602 modules in this release include:
3603 ** the ASDF system definition facility;
3604 ** an interface to the BSD Sockets API;
3605 ** an ACL-like convenience interface to the repl;
3606 (thanks to Kevin Rosenberg)
3607 ** an implementation of ROTATE-BYTE, with efficient implementation
3609 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3610 gives a closer approximation to the right answer than previously.
3611 (thanks to Raymond Toy)
3612 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3613 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3614 optional environment arguments, as required by ANSI.
3615 * fixed bugs in other functions taking environment objects, allowing
3616 calls with an explicit NIL environment argument to be compiled
3618 * fixed bug 228: primary return values from
3619 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3620 COMPILE or FUNCTION.
3621 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3622 :TYPE will no longer signal errors on innocuous objects.
3623 * fixed bug 231b: SETQ is better at respecting type declarations in
3624 the lexical environment.
3625 * fixed a bug in DEFCLASS: classes named by symbols with no or
3626 unprintable packages can now be defined.
3627 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3628 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3629 * fixed compiler failure related to checking types of functions.
3630 (reported by Robert E. Brown)
3631 * the compiler is now much more consistent in its error-checking
3632 treatment of bounding index arguments to sequence functions: in
3633 (SAFETY 3) code, errors will be signalled in almost all cases if
3634 invalid sequence bounding indices are passed to functions defined
3635 by ANSI to operate on sequences.
3636 * fixed a bug in the build procedure: documentation of SBCL-specific
3637 packages is now preserved and available in the final Lisp image.
3638 * lifted FDEFINITION lookup out of loops in the implementation of
3639 many list operations. (thanks to Robert E. Brown)
3640 * fixed a bug in the reader: the #n# reader macro now works for
3641 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3642 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3643 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3644 E. Brown and Tony Martinez respectively)
3645 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3646 SLOT-BOUNDP: in particular, it is now able to optimize them much
3647 better, and is now not vulnerable to having packages renamed.
3648 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3649 always called when a slot is not present in an instance. (thanks
3651 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3652 longer signal an error on unreading a character following EOF on
3653 the previous constituent stream. (thanks to Tony Martinez)
3654 * fixed some bugs revealed by Paul Dietz' test suite:
3655 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3656 not just nonnegative fixnums;
3657 ** the logical bit-array operators such as BIT-AND now accept an
3658 explicit NIL for their "opt-arg" argument (to indicate a
3659 freshly-consed result bit-array);
3660 ** ELT now signals an error on an invalid sequence index in safe
3662 ** the type system is now cleverer about negations of numeric
3663 types, and consequently understands the BIGNUM and RATIO types
3665 ** the type system is now cleverer about the interaction between
3666 INTEGER and RATIO types: while bugs still remain, many more
3667 cases are accurately computed;
3668 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3669 if it is in the last clause;
3670 ** CONSTANTLY now correctly returns a side-effect-free function in
3672 ** DECLARE is no longer treated as a special-operator; in
3673 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3674 * incremented fasl file version number due to the change in the
3675 DEFSTRUCT-SLOT-DESCRIPTION structure.
3677 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3678 * minor incompatible change: code processed by the "interpreter" or
3679 EVAL now has a compilation optimization policy of (DEBUG 2)
3680 (changed from (DEBUG 1)) to improve debuggability of interactive
3681 development, and to allow the use of the debug RETURN command in
3683 * an experimental implementation of the RETURN command for the
3684 debugger has been included. (thanks to Frederik Kuivinen)
3685 * fixed bug 62: constraints were not propagated into a loop.
3686 * fixed bug in embedded calls of SORT (reported and investigated by
3688 * fixed some bugs revealed by Paul Dietz' test suite:
3689 ** printing and reading of arrays with some dimensions having
3690 length 0 (thanks to Gerd Moellmann);
3691 ** BOA constructor with &AUX argument without a default value does
3692 not cause a type error;
3693 ** CONSTANTP now returns true for all self-evaluating objects.
3695 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3696 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3697 accessors that are related by inheritance, as specified in the
3698 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3699 Valtteri Vuorikoski)
3700 * The compiler is now able to inline functions that were defined in
3701 a complex lexical environment (e.g. inside a MACROLET).
3702 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3704 * Support for the upcoming FreeBSD-5.0 release has been included.
3705 (thanks to Dag-Erling Smorgrav)
3706 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3707 effect when it is not in a toplevel context.
3708 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3709 stand a better chance of being correct. (thanks to Gerd
3711 * fixed bug in COERCE, which now signals an error on coercing a
3712 rational to a bounded real type which excludes the expected
3714 * The compiler is now able to derive types more accurately from the
3715 COERCE and COMPILE functions.
3716 * fixed bug 223: functional binding is considered to be constant
3717 only for symbols in the CL package.
3718 * fixed bug 231: SETQ did not check the type of a variable being set
3719 (reported by Robert E. Brown)
3720 * A new optimization for MAKE-INSTANCE has been included, fixing
3721 various bugs (including relating to :ALLOCATION :CLASS slots and
3722 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3724 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3725 ** As required by ANSI, LOOP now disallows anonymous collection
3726 clauses such as COLLECT I in conjunction with aggregate boolean
3727 clauses such as THEREIS (= I 1);
3728 ** LOOP now signals an error when any variable is reused in the
3729 same loop (including the potentially useful construct analogous
3730 to WITH A = 1 WITH A = (1+ A);
3731 ** IT is only a special loop symbol within the first clause of a
3732 conditional loop clause;
3733 ** LOOP with a typed iteration variable over a hashtable now
3734 signals a type error iff it should.
3735 * fixed some other bugs revealed by Paul Dietz' test suite:
3736 ** FILE-STREAM now names the class previously known as FD-STREAM;
3737 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3738 argument) no longer signals an error;
3739 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3740 of the slot symbol, rather than using the current package
3741 ((:CONC-NAME "") continues to intern the slot's name in the
3743 * incremented fasl file version number, because of the incompatible
3744 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3745 of the new implementation of DEFINE-COMPILER-MACRO.
3747 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3748 * Support for building SBCL for MIPS platforms running in
3749 little-endian mode has now been checked in, and basic
3750 functionality on said platforms verified.
3751 * minor incompatible change: PCL now records the pathname of a file
3752 in which methods and the like are defined, rather than its
3754 * minor incompatible change: TRUENAME now considers the truename of
3755 a file naming a directory to be the pathname with :DIRECTORY
3756 component indicating that directory.
3757 * minor incompatible change: a NAMED clause in the extended form of
3758 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3759 reason for the previous behaviour is unclear.
3760 * more systematization and improvement of CLOS and MOP conformance
3761 in PCL (thanks to Gerd Moellman and Pierre Mai):
3762 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3764 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3765 primary methods with no specializers;
3766 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3768 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3769 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3770 FORWARD-REFERENCED-CLASSes; error reporting on
3771 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3773 ** SXHASH on CLOS instances now uses a slot internal to the
3774 instance to return different numbers on distinct instances,
3775 while preserving the same return value through invocations of
3777 ** DEFMETHOD signals errors when methods with longer incongruent
3778 lambda lists are added to generic functions;
3779 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3780 CLASS, as specified in AMOP;
3781 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3782 on the order returned by the primary method for classes of
3783 class STANDARD-CLASS;
3784 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3785 * fixed some bugs shown by Paul Dietz' test suite:
3786 ** DOLIST puts its body in TAGBODY;
3787 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3789 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3790 value producing form;
3791 ** if more variables are given to PROGV than values, extra
3792 variables are bound and made to have no value;
3793 ** NSUBSTITUTE on list arguments gets the right answer with
3795 ** ELT signals an error of type TYPE-ERROR when the index argument
3796 is not a valid sequence index;
3797 ** LOOP signals (at macroexpansion time) an error of type
3798 PROGRAM-ERROR when duplicate variable names are found;
3799 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3800 ** FUNCALL of special-operators now cause an error of type
3802 ** PSETQ now works as required in the presence of side-effecting
3803 symbol-macro places;
3804 ** NCONC accepts any object as its last argument;
3805 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3807 ** loop-for-as-package does not require a package to be explicitely
3809 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3811 * fixed bug 166: compiler preserves "there is a way to go"
3812 invariant when deleting code.
3813 * fixed bug 172: macro lambda lists with required arguments after
3814 &REST arguments now cause an error to be signalled. (thanks to
3816 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3817 bug. (thanks to Matthew Danish)
3818 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3820 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3821 arguments are lexically rebound. (thanks to Gerd Moellmann and
3823 * fixed bug 194: error messages are now more informative when there
3824 is no primary method applicable in a call to a generic
3825 function. (thanks to Gerd Moellmann)
3826 * fixed bug in command line argument checking (thanks to Julian
3828 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3829 (thanks to Matthew Danish)
3830 * incremented fasl file version number, because of the
3831 SXHASH-related changes in the layout of CLOS data structures
3833 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3834 * minor incompatible change: The runtime (the Unix executable named
3835 "sbcl") is now much pickier about the .core files it will load.
3836 Essentially it now requires .core files to descend from the same
3837 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3838 as the runtime does. (The intent is to prevent the crashes which
3839 can occur, and which can even be reported as mysterious failures,
3840 when people patch the sources or change the build parameters
3841 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3842 sbcl and .core files.)
3843 * fixed bug: VALUES-LIST is no longer optimized away.
3844 * fixed bug 142: The FFI conversion of C string values to Lisp
3845 string values no longer conses excessively. (thanks to Nathan
3846 Froyd porting Raymond Toy's fix to CMU CL)
3847 * began to systematize and improve MOP conformance in PCL (thanks to
3848 Nathan Froyd, Gerd Moellman and Pierre Mai):
3849 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3851 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3852 ** FINALIZE-INHERITANCE is now called on class finalization;
3853 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3854 argument precedence order.
3855 * fixed bug 202: The compiler no longer fails on functions whose
3856 derived types contradict their declared type.
3857 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3858 so it can be non-toplevel.
3859 * The fasl file version number has changed (because of the new
3860 implementation of DEFMACRO).
3861 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3862 safe code, that any length requirement by their type-specifier
3863 argument is valid. The exceptions to this are described in bug
3865 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3866 functions now check the types of their inputs as required by ANSI.
3867 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3868 introduced symbol is DECLAREd to be SPECIAL.
3869 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3870 * fixed bug: :COUNT argument to sequence functions may be negative
3871 * fixed bug: body of DO-SYMBOLS may contain declarations
3872 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3873 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3874 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3875 symbol macro only once
3876 * fixed printing of call frame when argument list is unavailable
3877 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3878 * compiler no longer signals WARNING on unknown keyword
3881 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3882 * A beta-quality port to the mips architecture running Linux,
3883 based on the old CMUCL backend, has been made. It has been tested
3884 on a big-endian kernel, and works sufficiently well to be able to
3885 rebuild itself; it has not been tested in little-endian mode.
3886 * fixed an inconsistency between gencgc.c and purify.c which made
3887 dumping/loading .core files unreliable
3888 * fixed bug 120a: The compiler now deals correctly with IFs where
3889 the consequent is the same as the alternative, instead of
3890 misderiving the return type. (thanks to Alexey Dejneka)
3891 * fixed bug 113: Logical pathnames are now dumpable (the logical
3892 host is resolved at load-time, throwing an error if it is not
3894 * fixed bug 174: FORMAT's error message is slightly clearer when a
3895 non-printing character is used in a format directive.
3896 * fixed several bugs in compiler checking of type declarations, i.e.
3897 violations of the Python "declarations are assertions" principle
3898 (thanks to Alexey Dejneka)
3899 * fixed several bugs in PCL's error checking (thanks to Gerd
3901 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3903 * fixed bug in compilation of functions as first class values
3904 (thanks to Antonio Martinez-Shotton)
3905 * The compiler's handling TYPE-ERRORs which it can prove will
3906 inevitably happen at runtime has been cleaned up and corrected
3907 in several ways. (thanks to Alexey Dejneka)
3908 * improved argument type checking for various basic arithmetic
3909 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3910 lavished on them in the past that they can be compiled in many
3911 ways in different special cases
3912 * fixed bug 181: compiler checks validity of user supplied type
3914 * cleaned up code flushing in optimization: Function calls which
3915 should signal errors for safety purposes (e.g. which ANSI says
3916 should signal errors when their arguments are of incorrect type)
3917 are no longer optimized away.
3918 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3919 * incremented fasl file version number, because changes in the
3920 implementation of sequence functions like COERCE caused
3921 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3922 in old inline expansions) to become undefined. (Actually these
3923 changes were later undone, so we might very well be binary
3924 compatible with 0.7.7 after all, but leaving the version number
3925 incremented seemed like the simplest and most conservative
3928 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3929 * An alpha-quality port to the parisc architecture running Linux,
3930 based on the old CMUCL backend, has been made. This, even more so
3931 than the other backends, should be considered still a work in
3932 progress; known problems include that the Linux kernel in 64-bit
3933 mode does not propagate the correct sigcontext structure to
3934 userspace, and consequently SBCL on a parisc64 kernel will not
3936 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3937 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3938 suppressed.) Also now that the compiler is looking at declarations
3939 in the environment, it checks optimization declarations as well,
3940 and suppresses inlining when (> DEBUG SPEED).
3941 * More fixes have been made to treatment of floating point exception
3942 treatment and other Unix signals. In particular, floating point
3943 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3944 * The detection and handling of control stack exhaustion (infinite
3945 or very deeply nested recursion) has changed. Stack exhaustion
3946 detection is now done by write-protecting pages at the OS level
3947 and applies at all optimization settings; when found, a
3948 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3949 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3950 be caught using IGNORE-ERRORS.
3951 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3952 correctly when transforming recursive LABELS functions to LETs.
3953 (thanks to Alexey Dejneka)
3954 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3955 that are names of constants or global variables.
3956 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3957 alien routines with docstrings.
3958 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3959 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3961 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3962 (thanks to Eric Marsden)
3963 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3964 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3965 * Bug fix: The SPARC backend can now compile functions involving
3966 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3967 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3968 to a functions expecting a PATHNAME-DESIGNATOR.
3969 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3970 lambda lists. (thanks to Alexey Dejneka)
3971 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3972 functions so that EXPORTing the name of the class doesn't cause
3973 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3974 to Antonio Martinez for reporting this)
3975 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3976 constant-folded without failing an assertion. (thanks to Einar
3977 Floystad Dorum for reporting this)
3978 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3979 types are now dealt with more correctly by the compiler.
3980 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3981 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3982 behaviour analogous to RENAME-FILE. This puts its behaviour more
3983 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3985 * The fasl file version number has changed again. (because of the
3986 bug fix involving the names of PCL MAKE-INSTANCE functions)
3988 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3989 * bug fix: Floating point exceptions are treated much more
3990 consistently on the x86/Linux and PPC/Linux platforms.
3991 * Array initialization with :INITIAL-ELEMENT is now much faster for
3992 cases when the compiler cannot open code the array creation, but
3993 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3994 array accesses have also seen a speed increase.
3995 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3996 specified. (This was at the root of some bad interactions between
3997 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3999 * bug fix: Internal error arguments for undefined functions are now
4000 computed correctly on the PPC/Linux platform.
4001 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4002 Raymond Toy's patch for CMU CL)
4003 * Support for the Solaris 9 operating environment has been included
4004 (thanks to Daniel Merritt)
4005 * A very ugly but hopefully complete draft of the missing FFI chapter
4006 of the manual has been created by reformatting the corresponding
4007 CMU CL manual chapter into (currently very ugly and incoherent)
4008 DocBook and bringing it up to date for SBCL behavior. Thus, the
4009 manual is now essentially complete, at least by my extreme
4010 once-and-only-once standards, whereby it's acceptable to refer to
4011 the doc strings of SB-EXT functions as the primary documentation.
4012 * The fasl file version number has changed again, due to cleanup of
4013 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4014 is no longer a static symbol.)
4016 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4017 * SBCL now builds with OpenMCL (version 0.12) as the
4018 cross-compilation host; also, more progress has been made toward
4019 bootstrapping under CLISP.
4020 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4022 * bug 158 fixed: The compiler can now deal with integer loop
4023 increments different from 1; fixing this turned out also to fix
4025 * bug 169 fixed: no more bogus warnings about using lexical bindings
4026 despite the presence of perfectly good SPECIAL declarations (thanks
4027 to David Lichteblau)
4028 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4029 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4030 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4032 * bug 180 fixed: Method combination specifications no longer ignore
4033 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4034 * bug fix: Structure type predicate functions now check their argument
4035 count as they should.
4036 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4037 correctly. (thanks to Pierre Mai)
4038 * minor incompatible change: The --noprogrammer option is deprecated
4039 in favor of the new --disable-debugger option, which is very similar.
4040 (The major difference is that it takes effect at a slightly different
4041 time at startup, causing handling of errors in --sysinit and
4042 --userinit files will be affected differently.) The
4043 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4044 been added to allow this functionality to be controlled from ordinary
4045 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4046 maintainers, who might want to run non-interactive scripts to
4047 build SBCL cores which will later be used interactively.)
4048 * minor incompatible change: The LOAD function no longer, when given
4049 a wild pathname to load, loads all files matching that pathname.
4050 Instead, an error of type FILE-ERROR is signalled.
4052 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4053 * bug 147 fixed: The compiler preserves its block link/count
4054 invariants more correctly now so that it doesn't crash. (thanks
4056 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4058 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4059 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4060 no-value-for-this-character as NIL (as specified by ANSI).
4061 * HOST-NAMESTRING on physical pathnames now returns a string that is
4062 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4063 (thanks to Christophe Rhodes)
4064 * The Alpha port handles icache flushing more correctly. (thanks to
4066 * More progress has been made toward bootstrapping under CLISP. (thanks
4067 to Christophe Rhodes)
4068 * The fasl file format has changed again, because dynamic loading
4069 on OpenBSD (which has non-ELF object files) motivated some cleanups
4070 in the way that foreign symbols are transformed and passed around.
4071 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4072 is no longer treated as whitespace by the reader, but instead as
4073 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4074 |A
\7fB|, instead of A as it used to.
4076 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4077 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4078 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4079 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4080 this as of 0.7.1.45, but was left out of the previous news section
4081 (thanks to Dan Barlow)
4082 * SBCL now runs on the Solaris operating system on SPARC architectures
4083 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4084 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4085 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4086 * SPARC backend cleanups, allowing builds of cores optimized for V8
4087 and V9 SPARCS, and also emission of code targeted to a particular
4088 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4090 * SBCL is closer to bootstrapping under CLISP, thanks to various
4091 fixes by Christophe Rhodes.
4092 * The fasl file format has changed again, to allow the compiler's
4093 INFO database to support symbol macros.
4094 * The user manual (in doc/) is formatted into HTML more nicely.
4095 (thanks to coreythomas)
4096 * The system is smarter about SUBTYPEP relationships, especially
4097 those involving NOT types (including types such as ATOM which are
4098 represented internally using NOT types). Thus SUBTYPEP is less
4099 likely to return (VALUES NIL NIL) in general, and in particular
4100 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4102 * The fasl file format has changed again, because the internal
4103 representation of types now includes a new slot to support the new
4104 SUBTYPEP-of-NOT-types logic.
4105 * (not a change in the main branch of SBCL, but a related prototype
4106 which can hopefully be merged into the main branch of SBCL in the
4107 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4108 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4109 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4110 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4111 dispatch macro character combinations. (thanks to Alexey Dejneka)
4112 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4113 that superficially look like logical namestrings correctly.
4114 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4116 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4117 default supplied-p) for &optional and &key arguments. (thanks to
4120 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4121 * incompatible change: The compiler is now less aggressive about
4122 tail call optimization, doing it only when (> SPACE DEBUG) or
4123 (> SPEED DEBUG). (This is an incompatible change because there are
4124 programs which relied on the old CMU-CL-style behavior to optimize
4125 away their unbounded recursion which will now die of stack overflow.)
4126 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4127 for non-GENCGC systems has been increased to 20M (since that
4128 seems much closer to the likely performance optimum for modern
4129 systems than the old 4M value was)
4130 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4131 and *DEBUG-PRINT-LEVEL*
4132 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4133 of CMU CL's support for SPARC, and various endianness and other
4134 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4135 * new syntactic sugar for the Unix command line: --load foo.bar is now
4136 an alternate notation for --eval '(load "foo.bar")'.
4138 ** The system now detects stack overflow and handles it gracefully,
4139 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4140 optimization settings. (This is a good thing in general, and
4141 its introduction in this version should be particularly timely
4142 for anyone whose code fails because of suppression of tail
4144 ** The system now hunts for the C variable "environ" in a more
4145 devious way, to avoid segfaults when the C library version differs
4146 between compile time and run time. (thanks to Christophe Rhodes)
4147 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4148 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4149 ** The compiler no longer issues bogus style warnings for undefined
4150 classes in the same source file as the DEFCLASSes which defined
4151 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4153 ** fixes in CONDITION class precedence list for undefined function
4154 errors (thanks to Alexei Dejneka)
4155 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4156 correctly. (thanks to Dan Barlow)
4157 ** portability fixes aiming at bootstrapping under CLISP (thanks
4158 to Dave McDonald and Christophe Rhodes)
4159 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4160 ** fixes in type translation and and type inference (thanks to
4162 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4163 ** various fixes in the new ports (thanks to Dan Barlow)
4164 * several changes related to debugging:
4165 ** suppression of tail recursion, as noted above
4166 ** stack overflow detection, as noted above
4167 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4168 is now the default. (For some time encapsulation has been more
4169 reliable than the breakpoint-based :ENCAPSULATE NIL
4170 implementation, at least on X86 systems; and I just noticed that
4171 encapsulation also seems closer to the spirit of the ANSI
4174 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4176 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4177 up properly again. (There was a packaging bug in 0.7.0 which
4178 left their definitions in SB-SYS::LOAD-FOREIGN and
4179 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4180 vital for most things which interface to C-level interfaces,
4181 like extensions working with sockets or databases or
4182 Perl-compatible regexes or whatever, and the need to fix
4183 this bug is the main reason that 0.7.1 was released so
4185 ** DEFGENERIC is now choosier about the methods it redefines, so that
4186 reLOADing a previously-LOADed file containing DEFGENERICs does
4187 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4188 cycle now works as it should. (thanks to Alexey Dejneka)
4189 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4190 fixed by Christophe Rhodes. (He actually submitted this patch
4191 months ago, and I delayed until after 0.7.0.)
4192 ** Bug 111 (internal compiler confusion about runtime checks on
4193 FUNCTION types) was fixed by Alexey Dejneka.
4194 * Some internal cleanups (getting rid of variables which aren't
4195 needed now that the byte interpreter is gone) caused the fasl
4196 file format number to change again.
4198 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4199 * major incompatible change: The default fasl file extension, i.e. the
4200 default extension for files produced by COMPILE-FILE, has changed
4201 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4203 ** There are many changes in the implementation of the compiler.
4204 SBCL is now essentially a compiler-only implementation of ANSI
4205 Common Lisp. EVAL still "interprets" a few special cases, but
4206 almost all the interesting cases are handled by creating
4207 a LAMBDA expression, calling COMPILE on it, then calling
4208 FUNCALL on the result.
4209 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4210 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4211 Since the code is newer, there might still be some new bugs
4212 (though not as many as before Martin Atzmueller's fixes:-). But
4213 the new code is substantially simpler and clearer, and hopefully
4214 any remaining bugs will be simpler, less fundamental, and more
4215 fixable then the bugs in the old code.
4216 ** The revised compiler is still a little unsteady on its feet.
4218 *** The debugging information it produces (particularly the names
4219 of FUNCTION objects) is sometimes much less useful than what
4220 the old compiler produced.
4221 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4222 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4223 within a MACROLET) has been temporarily weakened.
4224 ** There are new compiler optimizations for various functions:
4225 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4226 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4227 *** the math functions TRUNCATE, FLOOR, and CEILING
4228 *** the function-of-all-trades COERCE
4229 Mostly these should be transparent, but there's one
4230 potentially-annoying problem (bug 117): when the compiler
4231 inline-expands a function and does type analysis on the result,
4232 it can create control paths which have type mismatches, and
4233 when it can't prove that those control paths aren't taken,
4234 it will issue WARNINGs about the type mismatches. This is
4235 a particular problem in practice for the new sequence functions.
4236 It's not clear how this should be fixed, and for now, a
4237 workaround is given in the entry for 117 in the BUGS file.
4238 ** (Because of the interaction between the two previous items --
4239 occasional inlining problems and new inline expansions -- some
4240 of the new sequence function optimizations won't really kick in
4241 completely until debugging information, and then inlining, are
4242 straightened out in some future version.)
4243 * minor incompatible changes:
4244 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4245 DIRECTORY no longer implicitly promotes NIL slots of its
4246 pathname argument to :WILD. In particular, when you ask for the
4247 contents of a directory (which you used to be able to do without
4248 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4249 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4250 ** changes in behavior that ANSI explicitly defines to be
4251 implementation dependent:
4252 *** The new compiler-only implementation still conforms with ANSI,
4253 but acts a little different than before. Besides the obvious
4254 changes in performance tradeoffs (that the cost per form passed
4255 to EVAL has gone up, and the cost per form executed by EVAL
4256 has gone down), the behavior of the system changes a little
4257 because there are no longer any interpreted function objects.
4258 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4259 e.g. doing COMPILE on the output of interactive DEFUN is
4261 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4263 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4264 has changed from (:CL) to NIL.
4265 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4266 e.g. #\Newline and #\Tab, don't have names specified in the
4267 ANSI Common Lisp standard, is now based on their ASCII symbolic
4268 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4269 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4270 are no longer used for output.
4271 ** changes in internal implementation constants:
4272 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4273 4 million. (If your application spends a lot of time GCing and
4274 you have a lot of RAM, you might want to experiment with
4275 increasing it even more.)
4276 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4277 However, almost all old code should still continue to work without
4278 immediate update, as SB-C-CALL is now a (deprecated) nickname
4280 ** Old operator names in the style DEF-FOO are now deprecated in
4281 favor of new corresponding names DEFINE-FOO, for consistency with
4282 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4283 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4284 internal symbols, but a few supported extensions like
4285 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4286 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4287 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4288 etc. as you get deeper into recursive calls to the debugger
4289 command loop, instead of the old "5]", "5]]", "5]]]"
4290 sequence. (I was motivated to do this when squabbles between
4291 ILISP and SBCL left me very deeply nested in the debugger. In the
4292 short term, this change will probably provoke more ILISP/SBCL
4293 squabbles, but hopefully it will be an improvement in the long run.)
4294 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4295 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4296 function, since declaiming return types involving aliens is
4297 (1) annoyingly messy to do by hand and (2) vital to efficient
4298 compilation of code which calls such functions.
4299 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4300 longer reexported by the SB-EXT package. They're solely useful
4301 for alien code, so it seems more logical that you should get
4302 them from the SB-ALIEN package, not in SB-EXT.
4303 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4304 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4305 features. Instead, the code that they used to control is always
4306 built into the system.
4307 * many other bug fixes
4308 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4309 advantage of the new EVAL-WHEN stuff and to clean them up in
4310 general, and they are now more ANSI-compliant in a number of
4311 ways. Martin Atzmueller is responsible for a lot of this.
4312 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4314 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4315 *** correct ERROR type for various file operations
4316 *** some fixes for Lisp streams
4317 *** DEFMETHOD syntax checking
4318 *** changing old weird representation of debug information as
4319 strings (which, among their other deficiencies, don't transform
4320 correctly when you rename packages, and don't change their
4321 print representation when you change things like *PACKAGE*
4322 and *PRINT-LENGTH*) to symbols and lists of symbols
4323 He also made several improvements and fixed several bugs in DESCRIBE.
4324 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4326 *** misbehavior of WRITE-STRING/WRITE-LINE
4327 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4328 and several other LOOP problems as well
4329 *** DIRECTORY when similar filenames are present
4330 *** DEFGENERIC with :METHOD options
4331 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4332 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4333 *** argument ordering in FIND with :TEST option
4334 *** mishandled package designator argument in APROPOS-LIST
4335 *** various problems in the backquote readmacro
4336 *** a bug in APROPOS
4337 *** probably some others that I'm not describing very well here,
4338 since the CVS log documents them by reference to sbcl-devel
4339 messages, and the SourceForge archives aren't working well.:-(
4340 ** Dan Barlow improved the Alpha port (and is making progress on the
4341 PPC port, for those of you who think different).
4342 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4343 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4344 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4345 continues to work on the SPARC port (for those of you in a position
4346 to look down upon our little PC-compatible boxes from a great height).
4347 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4348 and uses that copy, rather than the current dynamic value, when
4349 it's trying to decide whether to truncate output. Thus e.g.
4350 (let ((*print-lines* 50))
4351 (pprint-logical-block (stream nil)
4353 (let ((*print-lines* 8))
4354 (print (aref possiblybigthings i) stream)))))
4355 should now truncate the logical block only at 50 lines, instead of
4356 often truncating it at 8 lines, as it did before.
4357 * The doc/cmucl/ directory, containing old CMU CL documentation
4358 from the time of the fork, is no longer part of the base system.
4359 SourceForge has shut down its anonymous FTP service, and with it
4360 my original plan for distributing the old CMU CL documentation
4361 there. For now, if you need these files you can download an old
4362 SBCL source release and extract them from it.
4363 * The fasl file version number changed again, for dozens of reasons,
4364 some of which are apparent above.
4366 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4367 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4368 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4369 patch, so that DISASSEMBLE output is much nicer.
4370 * The code in the SB-PROFILE package now seems reasonably stable.
4371 I still haven't decided what the final interface should look like
4372 (I'd like PROFILE to interact cleanly with TRACE, since both
4373 facilities use function encapsulation) but if you have a need
4374 for profiling now, you can probably use it successfully with
4375 the current CMU-CL-style interface.
4376 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4377 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4378 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4379 link now returns the dangling link itself, and for similar
4380 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4381 link itself. (In these cases the old code signalled an error and
4382 looped endlessly, respectively.) Thus, DIRECTORY now works even
4383 in the presence of dangling and cyclic symbolic links.
4384 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4385 is now a supported extension again, since the consensus on
4386 sbcl-devel was that it can be useful for ordinary development
4387 work, not just for debugging SBCL itself.
4388 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4389 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4390 that currently-defined functions might be redefined later with
4391 different return types.
4392 * Hash tables can be printed readably, as inspired by CMU CL code
4393 of Eric Marsden and SBCL code of Martin Atzmueller.
4394 * better error handling in CLOS method combination, thanks to
4395 Martin Atzmueller porting Pierre Mai's CMU CL patches
4396 * more overflow fixes for >16Mbyte I/O buffers
4397 * A bug in READ has been fixed, so that now a single Ctrl-D
4398 character suffices to cause end-of-file on character streams.
4399 In particular, now you only need one Ctrl-D at the command
4400 line (not two) to exit SBCL.
4401 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4403 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4404 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4405 does the right thing.
4406 * The implementation of some type tests, especially for CONDITION
4407 types, is now tidier and maybe faster, due to CMU CL code
4408 originally by Douglas Crosher, ported by Martin Atzmueller.
4409 * Some math functions have been fixed, and there are new
4410 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4411 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4412 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4413 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4414 likely to be built into sbcl-0.7.0.)
4415 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4416 CL, ported to SBCL by Martin Atzmueller)
4417 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4418 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4419 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4420 * There's a new slam.sh hack to shorten the edit/compile/debug
4421 cycle for low-level changes to SBCL itself, and a new
4422 :SB-AFTER-XC-CORE target feature to control the generation of
4423 the after-xc.core file needed by slam.sh.
4424 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4425 COMPILE-FILE is no longer supported, so that now every function
4426 gets an entry point, so that block compilation looks a little
4427 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4428 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4429 returns the number of bytes consed since the system started,
4430 rather than the number consed since the first time the function
4431 was called. (The new definition parallels ANSI functions like
4432 CL:GET-INTERNAL-RUN-TIME.)
4433 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4434 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4435 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4436 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4437 * Fasl file version numbers are now independent of the target CPU,
4438 since historically most system changes which required version
4439 number changes have affected all CPUs equally. Similarly,
4440 the byte fasl file version is now equal to the ordinary
4443 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4444 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4445 is no longer recognized. I apologize for this, because it was
4446 listed in SB-EXT as a supported extension, but I found that
4447 its existing behavior was poorly specified, as well as incorrectly
4448 specified, and it looked like too much of a mess to straighten it
4449 out. I have enough on my hands trying to get ANSI stuff to work..
4450 * many patches ported from CMU CL by Martin Atzmueller, with
4451 half a dozen bug fixes in pretty-printing and the debugger, and
4452 half a dozen others elsewhere
4453 * fixed bug 13: Floating point infinities are now supported again.
4454 They might still be a little bit flaky, but thanks to bug reports
4455 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4456 as flaky as they were.
4457 * The --noprogrammer command line option is now supported. (Its
4458 behavior is slightly different in detail from what the old man
4459 page claimed it would do, but it's still appropriate under the
4460 same circumstances that the man page talks about.)
4461 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4462 are now supported, and enabled by default. Thus, the compiler can
4463 handle many floating point and complex operations much less
4464 inefficiently. (Thus e.g. you can implement a complex FFT
4466 * The compiler now detects type mismatches between DECLAIM FTYPE
4467 and DEFUN better, and implements CHECK-TYPE more correctly, and
4468 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4469 thanks to patches from Martin Atzmueller.
4470 * various fixes to make the cross-compiler more portable to
4471 ANSI-conforming-but-different cross-compilation hosts (notably
4472 Lispworks for Windows, following bug reports from Arthur Lemmens)
4473 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4474 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4475 fixed thanks to Pierre Mai's CMU CL patches.
4476 * fixes to keep the system from overflowing internal counters when
4477 it tries to use i/o buffers larger than 16M bytes
4478 * fixed bug 45a: Various internal functions required to support
4479 complex special functions have been merged from CMU CL sources.
4480 (When I was first setting up SBCL, I misunderstood a compile-time
4481 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4482 * improved support for type intersection and union, fixing bug 12
4483 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4484 more obscure bugs as well
4485 * some steps toward byte-compiling non-performance-critical
4486 parts of the system, courtesy of patches from Martin Atzmueller
4487 * Christophe Rhodes has made some debian packages of sbcl at
4488 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4489 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4490 stable, but are nonetheless usable. When he's ready, I'd be happy
4491 to add them to the SourceForge "File Releases" section. (And if
4492 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4493 * new fasl file format version number (because of changes in
4494 internal representation of (OR ..) types to accommodate the new
4495 support for (AND ..) types, among other things)
4497 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4498 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4500 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4502 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4503 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4504 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4505 * DESCRIBE now works on structure objects again.
4506 * Most function call argument type mismatches are now handled as
4507 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4508 know whether the function will be redefined before the call is
4509 executed. (The compiler could flag local calls with full WARNINGs,
4510 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4511 it doesn't keep track of enough information to know whether calls
4512 are local in this sense.)
4513 * Compiler output is now more verbose, with messages truncated
4514 later than before. (There should be some supported way for users
4515 to override the default verbosity, but I haven't decided how to
4516 provide it yet, so this behavior is still controlled by the internal
4517 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4518 src/compiler/ir1util.lisp.)
4519 * Fasl file format version numbers have increased again, because
4520 support for the Gray streams extension changes the layout of the
4521 system's STREAM objects.
4522 * The Gray subclassable streams extension now works, thanks to a
4523 patch from Martin Atzmueller.
4524 * The full LOAD-FOREIGN extension (not just the primitive
4525 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4526 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4527 but like most other programs, it defaults to copying the Unix
4528 environment from the original process instead of starting the
4529 new process in an empty environment.
4530 * Extensions which manipulate the Unix environment now support
4531 an :ENVIRONMENT keyword option which doesn't smash case or
4532 do other bad things. The CMU-CL-style :ENV option is retained
4533 for porting convenience.
4534 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4535 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4537 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4539 * A patch from Martin Atzmueller seems to have solved the SIGINT
4540 problem, and as far as we know, signal-handling now works cleanly.
4541 (If you find any new bugs, please report them!)
4542 * The system no longer defaults Lisp source file names to types
4543 ".l", ".cl", or ".lsp", but only to ".lisp".
4544 * The compiler no longer uses special default file extensions for
4545 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4546 seems to expect a single default extension for all compiled code,
4547 and there's no compelling reason to try to stretch the standard
4548 to allow two different extensions.) Instead, byte-compiled files
4549 default to the same extension as native-compiled files.
4550 * Fasl file format version numbers have increased again, because
4551 a rearrangement of internal implementation packages made some
4552 dumped symbols in old fasl files unreadable in new cores.
4553 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4554 many fewer weird special cases.
4555 * Bug #17 (differing COMPILE-FILE behavior between logical and
4556 physical pathnames) has been fixed, and some related misbehavior too,
4557 thanks to a patch from Martin Atzmueller.
4558 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4559 by Tim Moore, ported to SBCL by Martin Atzmueller.
4560 * Martin Atzmueller fixed several filesystem-related problems,
4561 including bug #36, in part by porting CMU CL patches, which were
4562 written in part by Paul Werkowski.
4563 * More compiler warnings in src/runtime/ are gone, thanks to
4564 more patches from Martin Atzmueller.
4565 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4568 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4570 * DESCRIBE now works on CONDITION objects.
4571 * The debugger now handles errors which arise when trying to print
4572 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4574 * The build system now uses an additional file, customize-target-features.lisp,
4575 to allow local modifications to the target *FEATURES* list. (The point of
4576 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4577 debugging features enabled, without having to worry about propagating it
4578 into everyone's system when I do a "cvs update".) When no
4579 customize-target-features.lisp file exists, the target *FEATURES* list
4580 should be constructed the same way as before.
4581 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4582 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4583 warn about behavior which is undefined under the ANSI spec. Note: This
4584 is specified by ANSI, but it's not very popular with programmers.
4585 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4586 macro in the SBCL sources for an example of a workaround which you
4587 might use to make portable ANSI-standard code which does what you want.
4588 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4589 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4590 when it's used as a non-toplevel form. (This is required in order
4591 to implement the DEFCONSTANT-EQX macro.)
4592 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4593 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4594 funny things, due to bugs in the implementation of EVAL-WHEN
4595 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4596 non-toplevel DEFCONSTANTs are uncommon.)
4597 * The core file version number and fasl file version number have been
4598 incremented, because the old noncompliant DEFCONSTANT behavior involved
4599 calling functions which no longer exist, and because I also took the
4600 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4601 * fixed bug 1 (error handling before read-eval-print loop starts), and
4602 redid debugger restarts and related debugger commands somewhat while
4604 ** The QUIT debugger command is gone, since it did something
4605 rather different than the SB-EXT:QUIT command, and since it never
4606 worked properly outside the main toplevel read/eval/print loop.
4607 Invoking the new TOPLEVEL restart provides the same functionality.
4608 ** The GO debugger command is also gone, since you can just invoke
4609 the CONTINUE restart directly instead.
4610 ** The TOP debugger command is also gone, since it's redundant with the
4611 FRAME 0 command, and since it interfered with abbreviations for the
4613 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4615 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4616 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4617 some porting work by Martin Atzmueller)
4618 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4619 with a boatload of patches which he ported to SBCL. Now that those
4621 ** The system tries to make sure that its low-priority messages
4622 are prefixed by semicolons, to help people who like to use
4623 syntax highlighting in their ILISP buffer. (This patch
4624 was originally due to Raymond Toy.)
4625 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4626 patches originally written by Raymond Toy.
4627 ** The compiler understands coercion between single-value and
4628 multiple-VALUES type expressions better, getting rid of some very
4629 weird behavior, thanks to patches originally by Robert MacLachlan
4630 and Douglas Crosher.
4631 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4632 lambda lists, thanks to a patch originally by Pierre Mai.
4633 ** The system no longer bogusly warns about "abbreviated type
4635 ** The compiler gets less confused by inlining and RETURN-FROM,
4636 thanks to some patches originally by Tim Moore.
4637 ** The system no longer hangs when dumping circular lists to fasl
4638 files, thanks to a patch originally from Douglas Crosher.
4639 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4640 undefined function error.
4641 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4642 in structure slot initforms) for now, documented workaround instead:-|
4643 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4644 * fixed bug 5: added stubs for various Gray stream functions called
4645 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4646 installed, at least appropriate type errors are generated
4647 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4648 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4649 consistently in DEFMETHOD forms.
4650 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4651 it doesn't seem to affect SBCL after all
4652 * The C runtime system now builds with better optimization and many
4653 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4655 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4657 * The system is now under CVS at SourceForge (instead of the
4658 CVS repository on my home machine).
4659 * The new signal handling code has been tweaked to treat register
4660 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4661 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4662 downstream, e.g. in debug-int.lisp, has implicit dependencies
4663 on the unsignedness of integer representation of machine words,
4664 and that caused the system to bomb out with infinite regress
4665 when trying to recover from type errors involving signed values,
4666 e.g. (BUTLAST '(1 2 3) -1).)
4667 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4668 (This was one of the bugs Peter Van Eynde reported back in July.)
4669 * The system now uses code inspired by Colin Walters' O(N)
4670 implementation of MAP (from the cmucl-imp@cons.org mailing
4671 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4672 inline the MAP operation, and there is more than one
4673 sequence argument to the MAP call (so that it can't just
4674 do ETYPECASE once and for all based on the type of the
4675 single sequence argument). (The old non-inline implementation
4676 of the general M-argument sequence-of-length-N case required
4677 O(M*N*N) time when any of the sequence arguments were LISTs.)
4678 * The QUIT :UNIX-CODE keyword argument has been renamed to
4679 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4680 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4682 * Raymond Wiker's patches to port dynamic loading from Linux to
4683 FreeBSD have been added.
4684 * The BUGS file is now more nearly up to date, thanks in large part
4685 to Martin Atzmueller's review of it.
4686 * The debugger now flushes standard output streams before it begins
4687 its output ("debugger invoked" and so forth).
4688 * The core version number and fasl file version number have both
4689 been incremented, because of incompatible changes in the layout
4691 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4692 used to not be called for a saved Lisp image.)
4693 * Martin Atzmueller's version of a patch to fix a compiler crash,
4694 as posted on sbcl-devel 13 September 2000, has been installed.
4695 * Instead of installing Martin Atzmueller's patch for the
4696 compiler transform for SUBSEQ, I deleted the compiler transform,
4697 and transforms for some similar consing operations.
4698 * A bug in signal handling which kept TRACE from working on OpenBSD
4700 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4701 away by constant folding
4702 * The system now defines its address space constants in one place
4703 (in the Lisp sources), and propagates them automatically elsewhere
4704 (through GENESIS and the sbcl.h file). Therefore, patching the
4705 address map is less unnecessarily tedious and error-prone. The
4706 Lisp names of address space constants have also been systematized.
4707 * CVS tags like dollar-Header-dollar have been removed from
4708 the sources, because they have never saved me trouble and
4709 they've been source of trouble working with patches and other
4710 diff-related operations.
4711 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4712 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4714 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4716 * The system has been ported to OpenBSD.
4717 * The system now compiles with a simple "sh make.sh" on the systems
4718 that it's supported on. I.e., now you no longer need to tweak
4719 text in the target-features.lisp-expr and symlinks in src/runtime/
4720 by hand, the make.sh takes care of it for you.
4721 * The system is no longer so grossly inefficient when compiling code
4722 involving vectors implemented as general (not simple) vectors (VECTOR T),
4723 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4724 which dares to use the various sequence functions on non-simple
4725 vectors, takes less of a performance hit.
4726 * There is now a primitive type predicate VECTOR-T-P
4727 to test for the (VECTOR T) type, so that e.g.
4728 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4729 can now be compiled with some semblance of efficiency. (The old code
4730 turned the type declaration into a full call to %TYPEP at runtime!)
4731 * AREF on (VECTOR T) is still not fast, since it's still compiled
4732 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4733 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4734 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4735 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4736 without making full calls to SUBTYPEP at runtime.
4737 (Some analogous efficiency issues for non-simple vectors specialized to
4738 element types other than T, or for non-simple multidimensional arrays,
4739 have not been addressed. They could almost certainly be handled the
4740 same way if anyone is motivated to do so.)
4741 * The changes in array handling break binary compatibility, so
4742 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4743 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4744 instead of (VALUES T T).
4745 * By following the instructions that Dan Barlow posted to sbcl-devel
4746 on 2 July 2000, I was able to enable primitive dynamic object
4747 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4748 functionality is not implemented (since it calls ld to resolve
4749 library references automatically, requiring RUN-PROGRAM for its
4750 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4751 try to resolve library references) is now supported.
4752 * The system now flushes the standard output streams when it terminates,
4753 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4754 them at several other probably-convenient times, e.g. in each pass of
4755 the toplevel read-eval-print loop, and after evaluating a form given
4756 as an "--eval" command-line option. (These changes were motivated by a
4757 discussion of stream flushing issues on cmucl-imp in August 2000.)
4758 * The source transform for TYPEP of array types no longer assumes
4759 that an array whose element type is a not-yet-defined type
4760 is implemented as an array of T, but instead punts, so that the
4761 type will be interpreted at runtime.
4762 * There is now some support for cross-compiling in make.sh: each of
4763 the phases of make.sh has its own script. (This should be transparent
4764 to people doing ordinary, non-cross-compile builds.)
4765 * Since my laptop doesn't have hundreds of megabytes of memory like
4766 my desktop machine, I became more motivated to do some items on
4767 my to-do list in order to reduce the size of the system a little:
4768 ** Arrange for various needed-only-at-cold-init things to be
4769 uninterned after cold init. To support this, those things have
4770 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4771 symbols with such names are now uninterned after cold init).
4772 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4773 fasl files for cold load.
4774 ** Remove the old compiler structure pooling code (which used to
4775 be conditional on the target feature :SB-ALLOC) completely.
4776 ** Redo the representation of some data in cold init to be more compact.
4777 (I also looked into supporting byte compiled code at bootstrap time,
4778 which would probably reduce the size of the system a lot, but that
4779 looked too complicated, so I punted for now.)
4780 * The maximum signal nesting depth in the src/runtime/ support code has
4781 been reduced from 4096 to 256. (I don't know any reason for the very
4782 large old value. If the new smaller value turns out to break something,
4783 I'll probably just bump it back up.)
4784 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4786 * Many, many bugs reported by Peter Van Eynde have been added to
4787 the BUGS list; some have even been fixed.
4788 * While enabling dynamic object file loading, I tried to make the
4789 code easier to understand, renaming various functions and variables
4790 with less ambiguous names, and changing some function calling
4791 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4792 * While trying to figure out how to do the OpenBSD port, I tried to
4793 clean up some of the code in src/runtime/. In particular, I dropped
4794 support for non-POSIX signal handling, added various comments,
4795 tweaked the code to reduce the number of compilation warnings, and
4796 renamed some files to increase consistency.
4797 * To support the new automatic configuration functionality in make.sh,
4798 the source file target-features.lisp-expr has been replaced with the
4799 source file base-target-features.lisp-expr and the machine-generated
4800 file local-target-features.lisp-expr.
4801 * fixed a stupid quoting error in make.sh so that using CMU CL
4802 "lisp -batch" as cross-compilation host works again
4804 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4806 * DESCRIBE no longer tries to call itself recursively to describe
4807 bound/fbound values, so that it no longer fails on symbols which are
4808 bound to themselves (like keywords, T, and NIL).
4809 * DESCRIBE now works on generic functions.
4810 * The printer now prints less-screwed-up representations of closures
4811 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4812 * A private symbol is used instead of the :EMPTY keyword previously
4813 used to mark empty slots in hash tables. Thus
4814 (DEFVAR *HT* (MAKE-HASH-TABLE))
4815 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4816 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4817 now does what ANSI says that it should. (You can still get
4818 similar noncompliant behavior if bang on the hash table
4819 implementation with all the symbols you get back from
4820 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4821 This breaks binary compatibility, since tests for equality to
4822 :EMPTY are wired into things like the macroexpansion of
4823 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4825 * There's now a minimal placeholder implementation for CL:STEP,
4826 as required by ANSI.
4827 * An obscure bug in the interaction of the normal compiler, the byte
4828 compiler, inlining, and structure predicates has been patched
4829 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4830 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4832 * Missing ordinary arguments in a macro call are now detected even
4833 when the macro lambda list contains &KEY or &REST.
4834 * The debugger no longer complains about encountering the top of the
4835 stack when you type "FRAME 0" to explicitly instruct it to go to
4836 the top of the stack. And it now prints the frame you request even
4837 if it's the current frame (instead of saying "You are here.").
4838 * As specified by ANSI, the system now always prints keywords
4839 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4841 * The default initial SIZE of HASH-TABLEs is now smaller.
4842 * Type information from CLOS class dispatch is now propagated
4843 into DEFMETHOD bodies, so that e.g.
4844 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4846 is now basically equivalent to
4847 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4848 (DECLARE (TYPE SINGLE-FLOAT X))
4850 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4851 operation, without having to do run-time type dispatch.
4852 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4853 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4854 * Since I don't seem to be making big file reorganizations very often
4855 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4856 is overflowing my ability to conveniently back them up), I've finally
4857 checked the system into CVS. (The CVS repository is on my home system,
4858 not at SourceForge -- putting it on SourceForge might come later.)
4859 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4860 high-level GC-NOTIFY-FOO functions send their output. (There's
4861 still very little control of where low-level verbose GC functions
4862 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4863 less than it used to -- the GC-NOTIFY-FOO functions are now under
4864 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4865 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4866 in only one place in the source code, and propagates it automatically
4867 everywhere that it's needed. Thus e.g. when I bump the version from
4868 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4869 * The C source files now include boilerplate legalese and documentation
4870 at the head of each file (just as the Lisp source files already did).
4871 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4872 to his page will be replaced with a link to his new CLiki service.
4874 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4876 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4877 * The build process now looks for GNU make under the default name "gmake",
4878 instead of "make" as it used to. If GNU make is not available as "gmake"
4879 on your system, you can change this default behavior by setting the
4880 GNUMAKE environment variable.
4881 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4882 can build without error under CMU CL.
4884 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4886 * There is now a partial SBCL user manual (with some new text and some
4887 text cribbed from the CMU CL manual).
4888 * The beginnings of a profiler have been added (starting with the
4889 CMU CL profiler and simplifying and cleaning up). Eventually the
4890 main interface should be through the TRACE macro, but for now,
4891 it's still accessed through vaguely CMU-CL-style functions and macros
4892 exported from the package SB-PROFILE.
4893 * Some problems left over from porting CMU CL to the new
4894 cross-compilation bootstrap process have been cleaned up:
4895 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4896 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4897 part when getting rid of the compiler *BACKEND* stuff.)
4898 ** The value of *NULL-TYPE* was screwed up, because it was
4899 being initialized before the type system knew the final
4900 definition of the 'NULL type. This screwed up several key
4901 optimizations in the compiler, causing inefficiency in all sorts
4902 of places. (I found it because I wanted to understand why
4903 GET-INTERNAL-RUN-TIME was consing.)
4904 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4905 PROCLAIM FTYPE information. Unfortunately this broke binary
4906 compatibility again, since now the forms output by DEFGENERIC
4907 to refer to functions which didn't exist in 0.6.3.
4908 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4909 can use the new (as of 0.6.3) transform for SOME into MAP into
4911 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4912 values to (INTEGER 0 1000000), so that the time code will no longer
4913 occasionally get blown up by Linux returning 1000000 microseconds
4914 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4915 its output conform to the ANSI spec. (Alas, this makes its output
4916 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4917 * A full call to MAP NIL with a single sequence argument no longer conses.
4918 * fixes to problems pointed out by Martin Atzmueller:
4919 * The manual page no longer talks about multiprocessing as though
4920 it were currently supported.
4921 * The ILISP support patches have been removed from the distribution,
4922 because as of version 5.10.1, ILISP now supports SBCL without us
4923 having to maintain patches.
4924 * added a modified version of Raymond Toy's recent CMU CL patch for
4925 EQUALP comparison of HASH-TABLE
4927 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4929 * The system still can't cross-compile itself with
4930 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4931 entails), but at least it can compile more of itself that way
4932 than it used to be able to, and various buglets which were uncovered
4933 by trying to cross-compile itself that way have now been fixed.
4934 * This release breaks binary compatibility again. This time
4935 at least I've incremented the FASL file format version to 2, so that the
4936 problem can be detected reliably instead of just causing weird errors.
4937 * various new style warnings:
4938 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4939 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4940 ** using the lexical binding of a variable named in the *FOO* style
4941 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4942 as specified by ANSI.
4943 * *RANDOM-STATE* is no longer automatically initialized from
4944 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4945 default behavior of the system is to repeat its behavior every time
4946 it's run. If you'd like to change this behavior, you can always
4947 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4948 old convention there was no comparably easy way to get the system to
4949 repeat its behavior every time it was run.
4950 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4951 FTYPE declarations has been removed, in favor of their ANSI
4952 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4953 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4954 the types of their sequence arguments can be determined at compile time.
4955 This is done through a new open code expansion for MAP which eliminates
4956 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4957 argument types can be determined at compile time.
4958 * The optimizer now transforms COERCE into an identity operation when it
4959 can prove that the coerced object is already of the correct type. (This
4960 can be a win for machine generated code, including the output of other
4961 optimization transforms, such as the MAP transform above.)
4962 * Credit information has been moved from source file headers into CREDITS.
4963 * Source file headers have been made more standard.
4964 * The CASE macro now compiles without complaining even when it has
4967 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4969 * (Note that the way that the PCL macroexpansions were rewritten
4970 to accommodate the change in DEFGENERIC below breaks binary
4971 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4972 not run under sbcl-0.6.2. Once we get out of alpha releases,
4973 i.e. hit release 1.0.0, we'll probably try to maintain binary
4974 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4975 and sbcl-1.4.4. Until then, however, it might be fairly common
4976 for maintenance releases to break binary compatibility.)
4977 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4979 * The reporting of unhandled signals has been changed to print some
4980 explanatory text as well as the report form. (Previously only
4981 the report form was printed.)
4982 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4983 it defines, so that the compiler no longer issues undefined function
4984 warnings for compiled-but-not-yet-loaded generic functions.
4985 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4986 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4987 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4988 section of the ANSI spec are supported.
4989 * The "" nickname for the "KEYWORD" package has been removed.
4990 The reader still handles symbol tokens which begin with a package marker
4991 as keywords, but it doesn't expose its mechanism for doing so in the
4992 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4993 * The system now issues STYLE-WARNINGs for contradictory TYPE
4994 proclamations. (Warnings for contradictory FTYPE proclamations would
4995 be nice too, but those can't be done usefully unless the type system
4996 is made smarter about FUNCTION types.)
4997 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4998 systematized, so that "*target-*.lisp is supposed to exist only on the
4999 target and imply that there's a related file which exists on the
5000 host, and *host-*.lisp is supposed to exist only on the host and imply
5001 that there's a related file which exists on the target. This involves a
5002 lot of renaming. Hopefully the acute confusion caused by the renaming
5003 will be justified by the reduction in chronic confusion..
5004 ** runtime-type.lisp -> early-target-type.lisp
5005 ** target-type.lisp -> late-target-type.lisp
5006 ** early-host-format.lisp -> early-format.lisp
5007 ** late-host-format.lisp -> late-format.lisp
5008 ** host-error.lisp -> misc-error.lisp
5009 ** early-error.lisp -> early-target-error.lisp
5010 ** late-error.lisp -> late-target-error.lisp
5011 ** host-defboot.lisp -> early-defboot.lisp
5012 ** code/misc.lisp -> code/target-misc.lisp
5013 ** code/host-misc.lisp -> code/misc.lisp
5014 ** code/numbers.lisp -> code/target-numbers.lisp
5015 ** code/early-numbers.lisp -> numbers.lisp
5016 ** early-host-type.lisp -> early-type.lisp
5017 ** late-host-type.lisp -> late-type.lisp
5018 ** host-typep.lisp -> typep.lisp
5019 ** load.lisp -> target-load.lisp
5020 ** host-load.lisp -> load.lisp
5021 ** host-disassem.lisp -> disassem.lisp
5022 ** host-insts.lisp -> insts.lisp
5023 ** byte-comp.lisp -> target-byte-comp.lisp
5024 ** host-byte-comp.lisp -> byte-comp.lisp
5025 ** host-signal.lisp -> signal.lisp
5026 ** host-defstruct.lisp -> defstruct.lisp
5027 ** late-target-type.lisp -> deftypes-for-target.lisp
5028 Furthermore, several other previously target-only files foo.lisp (e.g.
5029 hash-table.lisp and random.lisp) have been split into a target-and-host
5030 foo.lisp file and a target-only target-foo.lisp file, with their key type
5031 definitions in the target-and-host part, so that the cross-compiler will
5032 know more about target types.
5033 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5034 gone away. In their place are various *BACKEND-FOO* variables
5035 corresponding to the slots of the old structure.
5036 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5037 into the target SBCL has been fixed.
5038 * The chill.lisp system for loading cold code into a running SBCL
5040 * Support for the CMU CL "scavenger hook" extension has been removed.
5041 (It was undocumented and unused in the CMU CL sources that SBCL was
5042 derived from, and stale in sbcl-0.6.1.)
5043 * Various errors in the cross-compiler type system were detected
5044 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5045 (enabling various consistency checks). Many of them were fixed,
5046 but some hard problems remain, so the compiler is back to
5047 running without *TYPE-SYSTEM-INITIALIZED* for now.
5048 * As part of the cross-compiler type system cleanup, I implemented
5049 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5050 * I have started adding UNCROSS calls throughout the type system
5051 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5052 on cross-compiler input files will be able to go away, and various
5054 * CONSTANTP now returns true for quoted forms (as explicitly required
5057 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5059 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5060 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5061 (MAKE-LIST -1). In the longer term, it ought to become true
5062 that declarations are assertions even at SAFETY 1. For now, it's not
5063 quite true even at SAFETY 3, but it's at least more nearly true..
5064 (Note that this change seems to increases the size of the system by
5065 O(5%) and to decrease the speed of the compiler by 20% or more.)
5066 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5067 for the problem of printing dozens of lines of distracting information
5068 about low-level system machinery as part of the top stack frame
5069 on entry to the debugger when an undefined function was called.
5070 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5071 is not reset to COMMON-LISP-USER.
5072 * Compilation of stuff related to dyncount.lisp has been made conditional
5073 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5074 smaller. The various dyncount-related symbols have been moved into
5075 a new "SB-DYNCOUNT" package.
5076 * tty-inspect.lisp has been renamed to inspect.lisp.
5077 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5078 feature has gone away. (When we eventually port to other flavors of
5079 libc and/or Unix, we'll try to make the differences between flavors
5080 invisible at the user level.)
5081 * Various other *FEATURES* tags, and/or their associated conditionals,
5082 have been removed if obsolescent, or given better documentation, or
5083 sometimes given more-mnemonic names.
5085 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5087 * tidied up "make.sh" script
5088 * tidied up system directory structure
5089 * better "clean.sh" behavior
5090 * added doc/FOR-CMUCL-DEVELOPERS
5091 * many many small tweaks to output format, e.g. removing possibly-confusing
5092 trailing #\. character in DESCRIBE-INSTANCE
5093 * (EQUALP #\A 'A) no longer signals an error.
5094 * new hashing code, including EQUALP hashing
5095 * tidied up Lisp initialization and toplevel
5096 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5097 * command line argument processing
5098 * added POSIX-GETENV function to deal with Unix-ish environment variables
5099 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5100 terminating SBCL on EOF
5101 * non-verbose GC by default
5102 * There is no more "sbcl" shell script; the sbcl file is now the C
5103 runtime executable (just like CMU CL).
5104 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5106 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5107 symbol and package information as Lisp native symbol and package objects
5108 instead of strings naming symbols and strings naming packages. This way,
5109 whenever packages are renamed (as in warm init), debug information is
5110 transformed along with everything else.
5111 * tweaked the optimization policy declarations which control the building
5112 of SBCL itself. Now, among other things, the system no longer saves
5113 source location debugging information. (This helps two problems at once
5114 by reducing SBCL size and by keeping SBCL from trying to look for its
5115 sources -- which may not exist -- when reporting errors.)
5116 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5117 debugging and testing purposes
5118 * cleaned up printing, making the printer call PRINT-OBJECT for
5119 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5120 methods, giving nearly-ANSI behavior
5121 * converted almost all special variables to use *FOO* naming convention
5122 * deleted PARSE-TIME functionality, since it can be done portably
5123 * moved some files out of cold init into warm init
5124 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5126 * regularized formatting of source files
5127 * added an install.sh script
5128 * fixed ridiculous memory usage of cross-compiler by making
5129 compiler/alloc.lisp not try to do pooling unless it can hook
5130 itself into the GC of the cross-compilation host. Now the system
5131 builds nicely on my old laptop.
5132 * added :SB-ALLOC in target-features.lisp-expr
5133 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5134 was not implemented)
5135 * re-did condition handling and note reporting in the compiler. Notes
5136 are no longer handled by signalling conditions. Style warnings
5137 and warnings are handled more correctly and reported in such a way
5138 that it's easy to find one or the other in your output (so that you
5139 can e.g. figure out which of many problems caused COMPILE-FILE to
5141 * changed the severity of several compiler warnings from full WARNING
5142 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5143 compiler note reporting so that it doesn't use the condition system
5144 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5145 COMPILE-FILE command)
5146 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5147 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5148 no longer supported, since their implementation was deeply intertwingled
5149 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5150 * removed bogus "support" for compiler macros named (SETF FOO), and
5151 removed the compiler macro for SETF INFO (but only after making a fool
5152 of myself on the cmucl-imp mailing list by posting a bogus patch for
5153 DEFINE-COMPILER-MACRO..)
5154 * Compiled files containing forms which have side effects on the Lisp
5155 reader (such as DEFPACKAGE forms) are now handled more correctly.
5156 (Compiler queuing of top level lambdas has been suppressed by setting
5157 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5158 * deleted various currently-unused source files, e.g. gengc.lisp. They
5159 may be added back at some point e.g. when porting to other architectures,
5160 but until they are it's distracting to distribute them and to try to
5162 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5163 there were so many of them they're just distractions, and UNCROSS is
5164 known to be able to handle the current sources
5165 * moved PROFILE functionality into TRACE, so that it will be clear
5166 how the wrapping and unwrapping of functions when you profile them
5167 interacts with the wrapping and unwrapping of functions when you
5168 trace them. (Actually, the functionality isn't there yet, but at least
5169 the interface specification is there. Hopefully, the functionality will
5170 arrive with some future maintenance release.)
5171 * removed host-oops.lisp
5172 * changed signature of QUIT function to allow UNIX-CODE argument
5173 * fixed READ-SEQUENCE bug
5174 * tweaked verbose GC output so that it looks more like the progress
5175 output that ANSI specifies for functions like LOAD
5176 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5177 * added <http://sbcl.sourceforge.com> to the banner information printed by