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: compiler now generates faster array typechecking code.
16 * optimization: ARRAY-DIMENSION is now faster for multidimensional and
18 * optimization: multidimensional array accesses in the absence of type
19 information regarding array rank are approximately 10% faster due to
20 open coding of ARRAY-RANK.
21 * improvement: SBCL now emits a compiler note where stack allocation was
22 requested but could not be provided.
23 * improvement: better MACHINE-VERSION responses. (thanks to Josh Elsasser)
24 * improvement: pretty-printing loop has been implemented properly. (thanks
26 * documentation: CLOS slot typechecing policy has been documented.
27 * bug fix: better floating point exception handling on x86/OpenBSD.
28 (thanks to Josh Elsasser)
29 * bug fix: exit status from QUIT when called under --script was lost
30 (reported by Hubert Kauker)
31 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
32 same implementation of FILL to initialize the array, even if a more
33 efficient one was available (reported by Stas Boukarev, thanks to
35 * bug fix: potential miscompilation of array stack allocation on x86 and
36 x86-64. (reported by Time Tossavainen)
37 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
38 result in their subforms being treated as top level forms. (reported by
40 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
41 than a machine register could leave garbage in the high bits of the
42 result register (bug 316325).
43 * bug fix: disable address space randomization Linux/x86-64 as well,
44 not just x86-64. (reported by Ken Olum)
45 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
46 generate incorrect code.
47 * bug fix: #201; type inference for CONS and ARRAY types could derive
48 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
49 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
50 could not be assigned to or bound with PROGV. (reported by Lorenz
52 * bug fix: the value of CL:- in the inspector was the previous expression
53 evaluated rather than the expression being evaluated.
54 * bug fix: constants can no longer be locally declared special.
55 * bug fix: signals delivered to threads started from foreign land (read:
56 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
57 thread by blocking all signals and resignalling.
59 changes in sbcl-1.0.28 relative to 1.0.27:
60 * a number of bugs in cross-compilation have been fixed, with the ultimate
61 result that building under (at least) clisp should be much more reliable.
62 * minor incompatible changes: echo-streams now propagate unread-char to the
63 underlying input stream, and no longer permit unreading more than one
65 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
66 frames, allowing for instance more reliable backtraces.
67 * improvement: the debugger REPL can now reference lexical variables
68 by name directly for code compiled with (DEBUG 3).
69 * improvement: errors from malformed declarations now have better source
70 paths associated with them. (thanks to Tobias Rittweiler)
71 * optimization: faster local calls on x86/x86-64
72 * bug fix: some error messages for out-of-bound array indexes confused the
73 index and the bound. (thanks to Stas Boukarev)
74 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
76 * bug fix: running regressions tests in shells without OSTYPE set now works.
77 (reported by Harald Hanche-Olsen)
78 * bug fix: more robust static space exhaustion signalling from
79 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
80 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
81 docstring away instead of storing it under names such as (LAMBDA (X)).
82 (reported by Leslie Polzer)
83 * bug fix: timers could go off in the wrong order, be delayed indefinitely
84 (thanks to Ole Arndt for the patch)
85 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
86 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
87 potentially-invalid effective methods in its cache.
88 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
89 instances as well (thanks to Paul Khuong)
90 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
91 SB-POSIX:WAITPID and their C equivalents.
92 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
94 changes in sbcl-1.0.27 relative to 1.0.26:
95 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
96 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
97 * improvement: the system either recovers from stack exhaustion or dies
98 properly as opposed to leaving the user uncertain of whether the handler
99 trampled on some random memory next to the stack or having to rely on
100 --lose-on-corruption (which is still a good idea to use in production
101 because stack exhaustion can happen in signal handlers which will likely
103 * bug fix: fix gc related interrupt handling bug on ppc (regression from
104 1.0.25.37, reported by Harald Hanche-Olsen)
105 * bug fix: work around signal delivery bug in darwin (regression from
106 1.0.25.44, reported by Sidney Markowitz)
107 * bug fix: fix ERROR leaking memory (reported by David Thompson)
109 changes in sbcl-1.0.26 relative to 1.0.25:
110 * incompatible change: an interruption (be it a function passed to
111 INTERRUPT-THREAD or a timer function) runs in an environment where
112 interrupts can be enabled. The interruption can use
113 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
114 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
115 potentially running out of stack. Keep in mind that in the absence
116 of WITHOUT-INTERRUPTS some potentially blocking operation such as
117 acquiring a lock can enable interrupts.
118 * incompatible change: GC-OFF and GC-ON are removed, as they were
119 always unsafe. Use WITHOUT-GCING instead.
120 * new feature: runtime option --disable-ldb
121 * new feature: runtime option --lose-on-corruption to die at the
122 slightest hint of possibly non-recoverable errors: running out of
123 memory, stack, alien stack, binding stack, encountering a memory
124 fault, etc. In the absence of --lose-on-corruption a warning is
126 * enhancement: detect binding stack exhaustion
127 * enhancement: detect alien stack exhaustion on x86/x86-64
128 * improvement: generally more stable and reliable interrupt handling
129 * improvement: there is a per thread interruption queue,
130 interruptions are executed in order of arrival
131 * improvement: a repeating timer reschedules itself when the it has
132 finished, but expiration times are spaced equally. If an
133 expiration time is in the past it will trigger after a short grace
134 period that may give a chance to other things to run.
135 * optimization: slightly faster gc on multithreaded builds
136 * optimization: faster WITHOUT-GCING
137 * bug fix: when JOIN-THREAD signals an error, do it when not holding
138 important locks so that the debugger/handler doesn't produce
139 recursive errors or deadlock.
140 * bug fix: real-time signals are not used anymore, so no more
141 hanging when the system wide real-time signal queue gets full.
142 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
144 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
145 * bug fix: fix random memory faults related to interrupts on alpha
146 * bug fix: fix random memory faults related to interrupts on sparc
147 * bug fix: fix deadlocks related to starting threads
148 * bug fix: fix deadlines on locks on futex platforms
149 * bug fix: restore errno in signal handlers
150 * bug fix: fix deadlocks related to hash tables
151 * bug fix: fix deadlocks in pcl
153 changes in sbcl-1.0.25 relative to 1.0.24:
154 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
155 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
156 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
157 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
158 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
159 better name and does not return values so stale on multiprocessor systems.
160 Also, HOLDING-MUTEX-P was added for about the only sane usage of
162 * improvement: unithread builds keep track of MUTEX-VALUE.
163 * improvement: reading from a TWO-WAY-STREAM does not touch the output
164 stream anymore making it thread safe to have a concurrent reader and
165 a writer, for instance, in a pipe.
166 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
167 arguments, making compiler-macros for SETF-functions able to inspect
168 their constant arguments.
169 * improvement: COMPILE-FILE reports times with millisecond accuracy
170 (thanks to Luis Oliveira)
171 * optimization: CHAR-CODE type derivation has been improved, making
172 TYPEP elimination on subtypes of CHARACTER work better. (reported
173 by Tobias Rittweiler, patch by Paul Khuong)
174 * bug fix: setting alien structure fields of type struct by value now
175 computes the right offset for the memory copy.
176 * bug fix: compilation problem involving inlined calls to aliens with
177 result type VOID. (reported by Ken Olum)
178 * bug fix: #235a; sequential inline expasion in different policies no
179 longer reuses the functional from the previous expansion site.
180 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
181 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
182 * bug fix: named ENUMs in multiply-referenced alien record types no longer
183 cause an implied type redefinition. Regression from 1.0.21.29.
184 * bug fix: sign-extension of small signed return values now works with C
185 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
186 * improvements to the Windows port:
187 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
188 Westfold, thanks to Rudi Schlatte)
190 changes in sbcl-1.0.24 relative to 1.0.23:
191 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
192 vector of a multidimensional SIMPLE-ARRAY.
193 * new feature: the system now signals a continuable error if standard
194 readtable modification is attempted. (thanks to Tobias Rittweiler)
195 * new feature: DIRECTORY has been extended with a non-standard keyword
196 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
197 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
198 has been robustified and documented better. (thanks to Robert Goldman)
199 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
200 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
201 * tradeoff: constant FORMAT control strings are now compiled unless
202 SPACE > SPEED (previously only when SPEED > SPACE.)
203 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
205 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
207 * bug fix: lexical type declarations are now correctly reported by
208 SB-CLTL2. (reported by Larry D'Anna)
209 * bug fix: STRING-TO-OCTETS did not handle :START properly when
210 using UTF-8 as external format. (thanks to Luis Oliveira)
211 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
212 are signalled correctly. (thanks to Stas Boukarev)
213 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
214 designator. (thanks to Tobias Rittweiler)
215 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
216 designator, and returns T instead of the function. (thanks to
218 * bug fix: direct superclasses of STANDARD-CLASS and
219 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
220 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
221 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
222 when no class with the named slot yet exists no longer causes a
223 compile-time style-warning.
224 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
225 in safe code. (reported by Didier Verna)
226 * bug fix: #430; nested structure constructors can stack allocate.
227 * bug fix: on some 64-bit platforms dynamic space size was truncated
228 to #xffff0000 bytes. (reported by Benjamin Lambert)
229 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
230 REPL unusable. (reported by Daniel Herring)
231 * bug fix: return values of READ-SEQUENCE did not take :START into
232 account on file streams, regressions since 1.0.12.22. (reported by
233 Thomas Russ, patch by Paul Khuong)
234 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
235 method specializer used to confuse permuation vector optimization.
236 * bug fix: system inserted bogus implicit type declarations for local
237 special variables in DEFMETHOD bodies.
238 * bug fix: #354; duplicated frames in backtraces due to
239 non-tail-call-optimized XEPs to functions with return type NIL
241 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
242 initializes structure object slots according to DEFSTRUCT initforms,
243 and DEFSTRUCT forms :INCLUDEind structure classes defined using
244 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
245 (reported by Bruno Haible and Stephen Wilson)
246 * bug fix: #395; fill-pointer output streams used now support
247 element-type BASE-CHAR as well.
248 * bug fix: compiler error when attempting to derive return value of
249 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
251 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
253 * bug fix: Mac OS X binaries should now be portable between Leopard
256 changes in sbcl-1.0.23 relative to 1.0.22:
257 * enhancement: when disassembling method functions, disassembly
258 for the associated fast function is also produced.
259 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
261 * optimization: printing with *PRINT-PRETTY* true is now more
262 efficient as long as the object being printed doesn't require
263 special handling by the pretty printer.
264 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
265 now interact correctly with type declarations.
266 * partial bug fix: PCL detects infinite recursion during wrapper
267 validation. (thanks to Attila Lendvai)
268 * bug fix: #426; nested function calls are inlined properly.
269 Previously if FOO was an inline function, in calls of the form
270 (FOO (FOO ...)) the outer call was not inlined.
271 * bug fix: long long arguments passed in registers to alien
272 functions on PowerPC Linux and NetBSD are now handled correctly
273 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
275 changes in sbcl-1.0.22 relative to 1.0.21:
276 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
277 for the shared object in the current directory, but passes the native
278 namestring of the designated pathname to the operation system's shared
279 object loading function as-it.
280 * minor incompatible change: --disable-debugger toplevel option now takes
281 effect before processing of initialization files and --eval or --load
283 * new feature: new commandline argument: --script, which supports
284 shebang lines. See documentation for details. (based on work by
286 * new feature: SAVE-LISP-AND-DIE can save current values of
287 --dynamic-space-size and --control-stack-size in the executable core,
288 causing it to skip normal runtime option processing. See documentation
289 for details. (thanks to Zach Beane)
290 * enhancement: inoccous calls to EVAL or generic functions dispatching
291 on subclasses of eg. STREAM no longer cause compiler notes to appear.
292 * enhancement: the system no longer resignals errors from --load and
293 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
294 associated with the original error to be lost. (thanks to Ariel
296 * enhancement: :DONT-SAVE keyword argument has been added to
297 LOAD-SHARED-OBJECT for controlling interaction with
298 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
299 the effects of an earlier LOAD-SHARED-OBJECT call.
300 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
301 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
302 provided. (reported by Cedric St-Jean)
303 * bug fix: circularity handling in the reader did not treat raw
304 structure slots correctly. (reported by Cedric St-Jean)
305 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
306 file descriptors when there were none.
307 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
308 was broken. (reporteed by Willem Broekema)
309 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
310 pathnames without a directory.
311 * bug fix: COMPILE-FILE did not warn about undefined variable
312 references at toplevel, and LOAD of the resulting fasl did
314 * bug fix: functions with non-required arguments used to end up with
315 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
316 * bug fix: redefining a function with non-required arguments didn't
317 update the system's knowledge about its call signature properly.
318 * bug fix: fixed #431; incompatible alien record type redefinitions
319 are detected and handled. (reported by Neil Haven)
320 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
321 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
323 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
324 after alien stack frames.
325 * bug fix: strings passed to mk{s,d}temp lacked null termination
327 changes in sbcl-1.0.21 relative to 1.0.20:
328 * new feature: the compiler is able to track the effective type of a
329 generic function across method addition and removal.
330 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
331 appropriately typed structure slots without locking.
332 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
333 information like those gathered by TIME using a programming-friendly
335 * new feature: TIME reports time taken even if the form performs a
336 non-local transfer of control.
337 * enhancement: reduced conservativism on GENCGC platforms: on
338 average 45% less pages pinned (measured from SBCL self build).
339 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
340 mutate constant symbols or violate declaimed type of the symbol.
341 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
342 sleeping on the semaphore in heavily contested semaphores.
343 * bug fix: semaphores and condition variables were not interrupt
345 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
346 owned by other threads anymore.
347 * bug fix: FIND on lists called KEY outside the specified
348 subsequence. (reported by budden)
349 * bug fix: LOG doesn't use single-float intermediate results when
350 given mixed integer and double-float arguments, leading to better
351 precision. (reported by Bob Felts)
352 * bug fix: LOG with base zero returned values of inconsistent type.
354 changes in sbcl-1.0.20 relative to 1.0.19:
355 * minor incompatible change: OPTIMIZE qualities
356 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
357 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
358 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
359 * documentation: some slot access efficiency guidelines have been
360 added to the user manual.
361 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
362 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
363 as ASSOC and MEMEBER.
364 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
365 to more efficient EQ-comparison versions more often.
366 * optimization: enhanced derivation of DOLIST iteration variable type
368 * optimization: constant folding of simple (LIST ...) forms as DOLIST
370 * optimization: runtime lookup of function definitions can be
371 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
372 * optimization: compiler is able to derive the return type of
373 (AREF (THE STRING X) Y) as being CHARACTER.
374 * optimization: CLRHASH on empty hash-tables no longer does pointless
375 work. (thanks to Alec Berryman)
376 * optimization: REPLACE deftransforms don't punt when :START1 or
378 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
379 no longer emits a code deletion note. (reported by Andrew
381 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
382 functions with an odd number of &OPTIONAL arguments, a &REST
383 argument, and one or more &KEY arguments at the call site.
384 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
385 arguments appeared at call sites as well.
386 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
387 READ-LINE &co can no longer read from them afterwards. (reported
389 * bug fix: fixed #427: unused local aliens no longer cause compiler
390 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
392 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
393 alien-stack leakage. (reported by Andy Hefner)
394 * bug fix: PROGV signals an error when an attempt to violate declared
395 type of a variable or bind a constant is made.
396 * bug fix: SET signals an error when an attempt to violate declared
397 type of a variable is made.
398 * bug fix: restart computation during the execution of a restart
399 test function no longer causes infinite recursion. (reported by
401 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
402 signals an error. (thanks to Pierre Mai)
403 * bug fix: EXPORT left symbol unexported in conflict situations.
404 (thanks to Michael Weber)
405 * bug fix: correctly handle name conflicts involving CL:NIL.
406 (thanks to Michael Weber)
407 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
408 conflicts arising from USEing package with conflicting symbols
409 correctly. (thanks to Michael Weber)
411 changes in sbcl-1.0.19 relative to 1.0.18:
412 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
413 warnings that go otherwise unhandled will be muffled if they are
414 of the type that's the value of this variable.
415 * optimization: stack allocation is slightly more efficient on x86
417 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
418 where a raw slot always is initialized using the initform whose
419 type is not know sufficiently well a compile-time are now compiled
420 correctly. (reported by John Morrison)
421 * bug fix: compiler no longer makes erronous assumptions in the
422 presence of non-foldable SATISFIES types.
423 * bug fix: stack analysis missed cleanups of dynamic-extent
424 arguments in non-let-converted calls to local functions.
425 * improvements to the Windows port:
426 ** adjusted address spaces for building on both Win32 and
427 Win64. (thanks for John Connors)
428 * fixed some bugs revealed by Paul Dietz' test suite:
429 ** interval arithmetic during type derivation used inexact integer
430 to single-float coercions.
431 ** arithmetic operations involving large integers and single
432 floats give the same results in compiled and interpreted code.
433 ** deriving the result type of COERCE no longer signals an error
434 if the derived type of the second argument is a MEMBER type
435 containing invalid type specifiers.
436 ** ADJOIN with constant NIL as second argument works correctly.
437 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
439 changes in sbcl-1.0.18 relative to 1.0.17:
440 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
441 profiles only the current thread.
442 * minor incompatible change: changes to SYMBOL-VALUE of constants
443 defined with DEFCONSTANT now signal an error.
444 * enhancement: SB-SPROF now has support for wallclock profiling,
445 and is also able to profile specific threads. REPORT output
446 has also additional sorting options.
447 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
449 * optimization: structure allocation has been improved
450 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
451 ** out of line constructors are ~10% faster.
452 ** inline constructors are ~15% faster.
453 ** inline constructors are capable of dynamic extent allocation
454 (generally on x86 and x86-64, in some cases on other platforms
456 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
458 * optimization: file compiler is now able to coalesce non-circular
459 lists, non-base strings, and bit-vectors. Additionally, constants
460 are never referenced through SYMBOL-VALUE at runtime.
461 * optimization: code defining methods on PRINT-OBJECT (and other
462 generic functions in the COMMON-LISP package) now loads faster.
463 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
465 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
466 fasl is now deleted. (reported by Attila Lendvai)
467 * bug fix: READ-LINE always returned NIL for the last line in files.
468 (reported by Yoshinori Tahara)
469 * bug fix: more accurate disassembly annotations of foreign function
470 calls. (thanks to Andy Hefner)
471 * bug fix: trimming non-simple strings and non-string string
472 designators when the there is nothing to trim works properly.
473 (thanks to James Knight)
474 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
475 (thanks to Travis Cross)
476 * fixed some bugs revealed by Paul Dietz' test suite:
477 ** NIL is a valid function name (regression at 1.0.13.38)
478 ** FILL on lists was missing its return value (regression at 1.0.12.27)
479 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
480 fill pointers properly (regression at 1.0.12.23)
481 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
482 displacement indices properly (regression at 1.0.12.23)
484 changes in sbcl-1.0.17 relative to 1.0.16:
485 * temporary regression: user code can no longer allocate closure
486 variable storage on stack, due to bug 419 without explicitly
487 requesting it. Please consult sbcl-devel for advice if you need to
488 use this feature in the meanwhile.
489 * new feature: runtime argument --control-stack-size can be used to
490 adjust thread default control stack size.
491 * enhancement: improved TIME output
492 ** all times are reported using the measured accuracy (milliseconds
493 for real and GC times, microseconds for everything else.)
494 ** processor cycle counts on x86 and x86-64.
495 ** interpreted forms are counted for both evaluator modes.
496 ** number of lambdas converted by the compiler is reported.
497 ** CPU percentage report (computed from real and total run time.)
498 ** more comprehensive run time reporting, using a condenced format
499 ** interperted form, lambda, and page fault counts are omitted
501 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
503 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
504 * optimization: two argument forms of LAST are upto ~10% faster
505 in normal SPEED policies.
506 * optimization: NCONC no longer needs to heap cons its &REST list
507 in normal SPEED policies.
508 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
510 * bug fix: UNION and NUNION work with :TEST-NOT once more,
511 regression since 1.0.9.1. (thanks to Eric Marsden)
512 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
513 since 1.0.15.36. (thanks to Paul Khuong)
514 * bug fix: LAST when always returned the whole list when given a bignum
515 as the second argument.
516 * bug fix: dynamic extent allocation of nested lists and vectors
517 could leak to otherwise accessible parts.
518 * bug fix: invalid optimization of heap-allocated alien variable
520 * bug fix: fasl header checking is less vulnerable to different
521 platform word lengths.
522 * bug fix: more correct assembler syntax for GNU binutils
523 2.18.50.0.4 support. (thanks to Marijn Schouten)
524 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
526 * bug fix: the fopcompiler can handle LOCALLY forms (with no
527 declarations) successfully. (reported by Attila Lendvai)
529 changes in sbcl-1.0.16 relative to 1.0.15:
530 * minor incompatible change: revert the changes to sb-posix's error
531 signaling added in 1.0.14.
532 * minor incompatible change: change PROBE-FILE back to returning
533 NIL whenever we can't get a truename, as was the case before 1.0.14.
534 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
535 inherits from ERROR instead of just CONDITION.
536 * new feature: SB-INTROSPECT can provide source locations for instances
537 as well. (thanks to Tobias Rittweiler)
538 * optimization: binding special variables now generates smaller code
539 on threaded platforms.
540 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
541 and cases where no :TEST is given but the compiler can infer that
542 the element to search is of type (OR FIXNUM (NOT NUMBER)).
543 * optimization: better LOGNOT on fixnums.
544 * optimization: modular arithmetic for a particular requested width
545 is implemented using a tagged representation unless a better
546 representation is available.
547 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
548 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
549 for wellformedness and returns a specialized vector. (reported by
550 Francois-Rene Rideau)
551 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
552 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
553 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
554 unscheduling timer at the same time as another timer fires.
555 * bug fix: don't reschedule timers for dead threads.
556 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
557 * bug fix: copying output from RUN-PROGRAM to a stream signalled
558 bogus errors if select() was interrupted.
559 * enhancement: add support for fcntl's struct flock to SB-POSIX.
561 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
562 * enhancement: cleaner backtraces for interactive interrupts, as
563 well as other cases where the interesting frames used to be
564 obscured by interrupt handling frames.
565 * enhancement: untracing a whole package using (UNTRACE "FOO") is
566 now supported, and tracing a whole package using (TRACE "FOO") now
567 traces SETF-functions as well.
568 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
569 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
570 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
571 * unreadably printed representation of hash-tables now includes
573 * bug fix: partially fixed #188: type propagation from assignments
574 is now more efficient.
575 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
576 'DOUBLE-FLOAT) are not flushable.
577 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
578 full address of the object, and none of the tag bits.
579 * bug fix: readably printing hash-tables now respects other printer
580 control variables. (reported by Cedric St-Jean)
581 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
582 keyword with MAKE-HASH-TABLE.
583 * bug fix: export SB-POSIX:MKSTEMP.
584 * bug fix: SORT was not interrupt safe.
585 * bug fix: XREF accounts for the last node of each basic-block as
587 * bug fix: MAKE-INSTANCE optimizations interacted badly with
588 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
589 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
590 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
591 create obsolete instances in the case of redefinition or
592 obsoletion of a superclass. (thanks to Andy Hefner)
593 * bug fix: Support for the Alpha architecture has been revived; it had
594 suffered somewhat from lack of maintenance since sbcl-1.0.
595 * improvements to the Windows port:
596 ** The system detects the codepage to use at startup, instead of
597 using the value from the saved core. (thanks to Kei Suzuki)
599 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
600 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
601 (see documentation for details.)
602 * revived support for OpenBSD (contributed by Josh Elsasser)
603 * partially fixed bug #108: ROOM no longer suffers from occasional
604 (AVER (SAP= CURRENT END)) failures.
605 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
606 bodies no longer cause a WARNING to be signalled. (reported by
608 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
609 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
610 the bug in TRUNCATE.) (reported by Michael Weber)
611 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
612 no samples. (reported by Andy Hefner)
613 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
614 no longer appear as (NIL ...) frames in backtraces.
615 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
616 is now actually exported from SB-EXT as documented. (reported by
618 * bug fix: sb-aclrepl now correctly understands how to inspect
619 single-floats on 64-bit platforms where single-floats are not boxed.
620 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
621 yet been finalized. (reported by Levente Meszaros)
622 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
623 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
624 SBCL-specific optimize qualities.
626 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
627 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
628 an executable in the search path, and does so in the child
629 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
630 been removed; it can be found in the CVS history, for somebody who
631 needs that search behavior (see the manual).
632 * minor incompatible change: compiler policy re. weakening type
633 checks has changed: now type checks are weakened only if SAFETY < 2
635 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
636 unparsing of directory pathnames as files. Analogously,
637 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
638 filename to parse into a directory pathname.
639 * enhancement: implicit generic function creation now signals a
640 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
641 users can bind handlers for and muffle around calls to LOAD.
642 * enhancement: RUN-PROGRAM allows unicode arguments and environments
643 to be used (using the default stream external format), and allows
644 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
645 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
646 strings and vectors whose element-type or simplicity is not fully
647 known at compile-time.
648 * optimization: STRING-TRIM and related functions no longer allocate
649 a new string when no trimming needs to be performed. These functions
650 are also faster than before when the input string has been declared
652 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
653 * optimization: READ-LINE is significantly faster for files containing
655 * optimization: non-open coded uses of character comparison operators
656 (e.g. char=) no longer cons when called with more than one parameter
657 on platforms supporting dynamic-extent allocation.
658 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
659 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
660 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
662 * bug fix: some sequence functions elided bounds checking when
664 * bug fix: too liberal weakening of union-type checks when SPEED >
666 * bug fix: more bogus fixnum declarations in ROOM implementation
669 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
670 * new feature: MAKE-HASH-TABLE now experimentally accepts a
671 :SYNCHRONIZED argument, which makes the hash-table safe for
672 concurrent accesses (but not iteration.) See also:
673 SB-EXT:WITH-LOCKED-HASH-TABLE, and
674 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
675 * optimization: CONCATENATE on strings is an order of magnitude faster
676 in code compiled with (> SPEED SPACE).
677 * optimization: SUBSEQ is ~50% faster on lists.
678 * bug fix: bug 417 fixed -- source location reporting is now more robust.
679 * bug fix: SUBSEQ on a list will now correctly signal an error if if
680 END is smaller then START.
681 * bug fix: SB-PROFILE will no longer report extra consing for nested
682 calls to profiled functions.
683 * bug fix: ROOM implementation had bogus fixnum declarations which
684 could cause type-errors when calling ROOM on large images.
685 * bug fix: if file compilation is aborted, the partial fasl is now
686 deleted, and COMPILE-FILE returns NIL as the primary value.
687 * bug fix: number of thread safety issues relating to SBCL's internal
688 hash-table usage have been fixed.
689 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
690 be returned from its body when the values were being returned
691 using unknown-values return convection and the W-P-O was wrapped
692 inside an UNWIND-PROTECT.
693 * bug fix: sb-posix should now compile again under Windows, enabling
696 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
697 * incompatible change: hash-table accessor functions are no longer
698 automatically protected by locks. Concurrent accesses on the same hash-table
699 from multiple threads can give inconsistent results or even corrupt the
700 hash-table completely. Multi-threaded applications should do their own
701 locking at the correct granularity. In the current implementation it is
702 still safe to have multiple readers access the same table, but it's not
703 guaranteed that this property will be maintained in future releases.
704 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
705 and will signal an error at runtime.
706 * enhancement: SB-THREAD package now exports a semaphore interface.
707 * enhancement: CONS can now stack-allocate on x86 and
708 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
710 * enhancement: nested lists can now be stack allocated on
711 platforms providing stack allocation support.
712 * enhancement: dynamic-extent support has been extended to support
713 cases where there are multiple possible sources for the stack
715 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
716 if the mutex is uncontested on Linux.
717 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
718 as specified by the CLHS. (thanks to Tobias Rittweiler)
719 * bug fix: NaN comparison now works on x86-64.
720 * bug fix: CLOSE :ABORT T on a stream with pending output now
722 * bug fix: instances of non-standard metaclasses using standard
723 instance structure protocol sometimes missed the slot type checks
725 * bug fix: known functions can be inlined in high-debug code.
727 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
728 * minor incompatible change: the MSI installer on Windows no longer
729 associates .lisp and .fasl files with the installed SBCL.
730 * minor incompatible change: :UNIX is no longer present in *FEATURES*
731 on Windows. (thanks to Luis Oliviera)
732 * new feature: SB-CLTL2 contrib module now implements
733 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
734 * optimization: scavenging weak pointers is now more efficient,
735 requiring O(1) instead of O(N) per weak pointer to identify
736 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
737 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
738 method bodies using constant slot-names when the first argument is
739 a specializer parameter for the method.
740 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
741 names, when the class of the instance is a direct instance
742 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
743 as slow as the constant slot-name case.)
744 * optimization: member type construction is now O(N) instead
746 * optimization: UNION and NUNION are now O(N+M) for large
747 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
749 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
750 now more readable in environments like Slime which display it.
751 (thanks to Tobias C. Rittweiler)
752 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
753 was able to infer the correct argument type for the object on which
754 the CAS operation was being performed.
755 * bug fix: copy propagation interfered with parallel assignment
756 semantics in local calls. (reported by Paul Khuong)
757 * bug fix: the signed modular fixnum shift compiled to wrong code on
758 x86 and x86-64. (spotted by a slight modification to some of
761 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
762 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
763 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
764 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
766 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
767 and generic function now print the object identity for anonymous
768 instances. (thanks to Attila Lendvai)
769 * enhancement: as an extension to MOP, SBCL now supports SETF of
770 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
771 (thanks to Attila Lendvai)
772 * enhancement: sb-cover has an alternate annotation mode which puts
773 the form annotations at the car of the form, instead of annotating
774 it completely. The benefit of this mode is that it shows better
775 which forms were instrumented by the compiler.
776 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
777 by printer control variables. (reported by Dan Corkill)
778 * bug fix: system leaked memory when delayed output was performed by
779 the OS in smaller chunks then expected. (thanks to David Smith)
780 * bug fix: system leaked memory when file streams were not closed
782 * bug fix: large objects written to slow streams that were modified
783 after the write could end up with the modified state written to
784 the underlying file descriptor.
785 * bug fix: multiple threads operating in parallel on the same stream
786 could cause buffer-overflows.
787 * bug fix: source location information is stored correctly
788 (broken since 1.0.6). This bug would generally show up as the
789 Slime debugger highlighting the wrong form when the sldb-show-source
791 * bug fix: Branch forms are again annotated as branches in the sb-cover
793 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
794 on x86-64. (reported by Gregory Vanuxem)
795 * bug fix: EQUALP could return wrong results for structures with raw
796 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
797 word). (reported by Vjacheslav Fyodorov)
798 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
801 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
802 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
803 atomic compare-and-swap operations on threaded platforms.
804 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
805 allows assigning a global minimum value to optimization qualities
806 (overriding proclamations and declarations).
807 * enhancement: closed over variables can be stack-allocated on x86
809 * performance bug fix: GETHASH and (SETF GETHASH) are once again
811 * optimization: slot definition lookup is now O(1). This speeds up
812 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
813 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
814 * optimization: ASSOC and MEMBER can now be open-coded for all
815 combinations of keyword arguments when second argument is constant
816 and SPEED >= SPACE. In other cases a specialized version is
818 * bug fix: using obsoleted structure instances with TYPEP and
819 generic functions now signals a sensible error.
820 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
821 (reported by Kristoffer Kvello)
822 * bug fix: backtrace construction is now more careful when making
823 lisp-objects from pointers on the stack, to avoid creating bogus
824 objects that can be seen by the GC.
825 * bug fix: defaulting of values in contexts expecting more than 7
826 variables now works on x86-64. (reported by Christopher Laux)
827 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
829 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
830 as the property-list of a symbol.
831 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
832 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
833 situations requested, are once again file-compileable. (reported
836 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
837 * MOP improvement: support for user-defined subclasses of
838 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
839 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
840 the expansion of DEFMETHOD, is responsible for generating a form
841 which creates a list of specializers when evaluated. Additional
842 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
843 debugging and introspective support.
844 * minor incompatible change: the (unsupported) spinlock interface
845 has changed: free spinlock now has the value NIL, and a held spinlock
846 has the owning thread as its value.
847 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
848 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
850 * enhancement: name of a socket-stream is now "a socket" instead of
852 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
853 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
854 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
855 for generic functions. (Thanks to Tobias C. Rittweiler)
856 * optimization: bignum printing speed has been improved by 20-40%
857 (depending on the bignum size.)
858 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
860 * bug fix: the cache used by the CLOS to store precomputed effective
861 methods, slot offsets, and constant return values is now thread and
863 * bug fix: generic function dispatch function updating is now thread
864 and interrupt safe (in the sense that the known issues have been
866 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
867 * bug fix: interning EQL-specializers is now thread and interrupt safe.
868 * bug fix: asdf systems with dependencies to the SB-POSIX or
869 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
870 * bug fix: interrupt safety of applicable method computation has been
873 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
874 * new contrib: sb-cover, an experimental code coverage tool, is included
876 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
877 significantly faster.
878 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
879 produce more efficient code on x86-64 if the compiler can determine
880 that (+ INDEX <constant>) does not require a bounds check and FOO
881 has an element type at least 8 bits wide.
882 * enhancement: a new, experimental synchronous timeout facility is
883 provided. Refer to SB-SYS:WITH-DEADLINE for details.
884 * enhancement: when a symbol name conflict error arises, the
885 conflicting symbols are always printed with a package prefix.
886 (thanks to Kevin Reid)
887 * enhancement: stepping is now once again supported on the SPARC and
888 MIPS platforms. (It is also now more likely to work on CheneyGC
890 * enhancement: sb-sprof can now also track and report accurate call
892 * bug fixes: the treatment of non-standard subclasses of
893 SB-MOP:SPECIALIZER is more correct.
894 * incompatible change: PURIFY no longer copies the data from the
895 dynamic space into the static and read-only spaces on platforms
896 that use the generational garbage collector
897 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
899 * bug fix: GC race condition occasionally resulting in crashes with
900 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
902 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
904 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
905 a STYLE-WARNING to be signalled (regression from 1.0.4.)
906 * bug fix: an asynchronous interrupt could previously leave the
907 system running with GC inhibited.
908 * bug fix: a DECLARE form evaluated at top-level now causes an error
909 rather than silently (or verbosely) returning NIL.
910 * bug fix: trying to dissassemble functions compiled at high DEBUG could
911 result in a error being signalled due to source form lookup errors.
912 (reported by Peter Graves)
914 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
915 * incompatible change: removed writer methods for host-ent-name,
916 host-ent-addresses -- changing the values did not update the DNS
918 * minor incompatible change: changed experimental JOIN-THREAD interface
919 * documentation: the manual now lists reader and writer methods
920 in class slot documentation sections. (thanks to Richard M Kreuter)
921 * documentation: unwinding from asyncronous events has been
922 documented as unsafe.
923 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
924 in multithreaded application code.
925 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
926 platforms. (thanks to James Anderson for the optimization hint)
927 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
929 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
930 variants no longer cons.
931 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
932 their NOT- variants no longer cons.
933 * optimization: EQUAL hash tables no longer use SXHASH for objects
934 of all data types, but instead use an EQL hash for types for which
935 EQUAL is the same as EQL.
936 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
937 are significantly faster.
938 * optimization: new STRING-OUTPUT-STREAM implementation is much
939 faster for large outputs and conses 30% less on average.
940 * enhancement: XREF information is now collected for references made
941 to global variables using SYMBOL-VALUE and a constant argument.
942 * enhancement: SIGINT now causes a specific condition
943 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
944 * bug fix: FILE-POSITION used to signal an error in some cases where
945 ANSI requires it to return NIL.
946 * bug fix: ADJUST-ARRAY is now interrupt-safe.
947 * bug fix: adding and removing fd-handlers is now interrupt-safe.
948 * bug fix: inlined calls to C now ensure 16byte stack alignment on
950 * bug fix: bad type declaration in the CLOS implementation has
951 been fixed. (reported by James Anderson)
952 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
953 (thanks to Lutz Euler)
954 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
955 required by the CLHS. (reported by Samium Gromoff)
956 * bug fix: dead unbound variable references now signal an error.
957 (reported by Marco Monteiro)
958 * bug fix: / with an unused value was being deleted in safe code.
959 (thanks to Marco Monteiro and Kevin Reid)
960 * bug fix: number of characters that can be written onto a single
961 line in a file is unlimited.
962 * bug fix: some GC deadlocks caused by asynchronous interrupts have
963 been fixed by inhibiting interrupts for when GC is disabled.
964 * bug fix: some interrupt safety issues with GETHASH, (SETF
965 GETHASH), CLRHASH and REMHASH have been fixed.
966 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
967 specifier no longer causes infinite recursion.
968 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
969 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
970 * bug fix: an &environment argument with the correct variable information
971 is passed to macros that are expanded during byte compilation.
972 (reported by Samium Gromoff)
973 * bug fix: a logic error was causing the hash function for bignums to
974 have excessive amounts of collisions. (reported by Faré Rideau)
975 * bug fix: modifying the contents of an array could change the return
976 value of SXHASH on that array, which is only allowed for strings
977 and bit vectors. (bug introduced in 0.9.16)
978 * improvement: the x86-64/darwin port now passes all tests (except
979 for the debugger tests) but should still be considered
980 experimental until this is fixed.
981 * improvement: a style-warning is signaled for CASE (etc) clauses with
982 duplicate keys. (thanks to Kevin Reid)
983 * improvement: macos/x86 and macos/x86-64 now use mach exception
984 handlers for dealing with illegal instructions (for trapping and
985 error handling) and memory protection violations (for GC).
987 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
988 * new platform: experimental support for x86-64/darwin (MacOS).
989 * incompatible change: the thread-safe (on most platforms) getaddrinfo
990 and getnameinfo sockets functions are used instead of gethostbyaddr
991 and gethostbyname, on platforms where the newer functions are available.
992 As a result, the ALIASES field of HOST-ENT will always be NIL on these
994 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
995 don't print the startup banner, but behave as if --noinform was passed
996 as a command line argument. (thanks to Kevin Reid)
997 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
998 * optimization: code using alien values with undeclared types is much faster.
999 * optimization: the compiler is now able to open code SEARCH in more cases.
1000 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
1001 * bug fix: using standardized COMMON-LISP special variables as loop
1002 variables no longer signals bogus package lock violations. (reported
1004 * bug fix: declaring local loop variables to be of a range-limited type
1005 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1006 (reported by Andras Simon)
1007 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1008 bugs remain on x86-64.)
1009 * bug fix: the #= and ## reader macros now interact reasonably with
1010 funcallable instances.
1011 * bug fix: type-checks for function arguments were compiled using the
1012 compiler policy settings of the wrong lexical environment (in 1.0.2
1014 * bug fix: SHADOW accepts characters as string designators, as required
1015 by the spec (thanks to Eric Marsden)
1016 * bug fix: fixed GC safety issues when foreign functions are called with
1017 non-base strings as arguments
1018 * bug fix: more consistent error output for fatal-compiler-warnings, like
1020 * bug fix: functions with proper names were showing up as NIL in
1021 backtraces (reported by Edi Weitz, regression in 1.0.2)
1023 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1024 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1025 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1026 produce more efficient code on the x86 if the compiler can determine
1027 that (+ INDEX <constant>) does not require a bounds check and FOO
1028 has an element type at least 8 bits wide.
1029 * bug fix: references to undefined variables are handled the same way
1030 in toplevel forms as in the normal compiler
1031 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1033 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1034 (thanks to Jon Buller)
1035 * improvement: faster compilation times for complex functions
1036 * improvement: added readlink support to SB-POSIX (thanks to Richard
1039 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1040 * improvement: experimental support for mach exception handling on
1041 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1043 * improvement: support for GBK external format.
1044 (thanks to Chun Tian (binghe))
1045 * improvement: the debugger now displays variables that have been closed
1046 over, in code compiled with (DEBUG 2) or higher
1047 * improvement: support for executable cores on NetBSD (thanks to
1049 * new feature: added a RESTART-FRAME debugger command
1050 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1051 be used to provide an implementation for FILE-POSITION on Gray streams
1052 (thanks to Eric Marsden)
1053 * new feature: add syslog, openlog and closelog support to SB-POSIX
1054 (thanks to Richard Kreuter)
1055 * optimization: the function call overhead in code compiled with
1056 a high DEBUG optimization setting is significantly
1057 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1058 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1059 (thanks to Tony Martinez)
1060 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1061 evaluated once (reported by Kevin Reid)
1062 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1063 works again (thanks to Kilian Sprotte)
1064 * bug fix: an error is signaled for tagbodies with duplicate tags
1065 (thanks to Stephen Wilson)
1066 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1067 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1068 stack frames from alien callbacks.
1069 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1070 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1071 65k pages (thanks to David Woodhouse)
1072 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1074 changes in sbcl-1.0.1 relative to sbcl-1.0:
1075 * new platform: FreeBSD/x86-64, including support for threading.
1076 * new feature: the compiler stores cross-referencing information
1077 about function calls (who-calls), macroexpansion (who-macroexpands)
1078 and special variables (who-binds, who-sets, who-references) for code
1079 compiled with (< SPACE 3). This information is available through the
1080 sb-introspect contrib.
1081 * new feature: users may subclass SEQUENCE, and have instances of
1082 these classes interoperate with standard Common Lisp functions if
1083 a number of methods are defined. (This feature is experimental
1084 and the interface subject to change based on feedback from SBCL
1085 users and the general community)
1086 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1087 x86/x86-64, rather than the previous fixed depth of 8
1088 * improvement: another pthread back-end of mutex "pthread-futex".
1089 * bug fix: non-ascii command-line arguments are processed correctly
1090 (thanks to Yaroslav Kavenchuk)
1091 * bug fix: non-required arguments were not passed correctly when a method
1092 defined using DEFMETHOD was called from a mop-generated method using
1093 CALL-NEXT-METHOD (reported by Pascal Costanza)
1094 * bug fix: recursion is now permitted in accessors through
1095 SLOT-UNBOUND. (reported by Pascal Costanza)
1096 * bug fix: an error was signaled at startup if the HOME environment
1097 variable was defined, but had an empty value (reported by Peter Van Eynde)
1098 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1099 signal a PROGRAM-ERROR, not a BUG.
1100 * bug fix: SB-POSIX:READDIR works when built with large file support.
1101 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1103 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1104 for better automatic indentation support. (thanks to Matt Pillsbury)
1105 * optimization: loading generic functions no longer takes O(n^2) time,
1106 proportional to the amount of methods in the generic function
1107 (reported by Todd Sabin and Jeremy Brown)
1108 * optimization: the FIND and POSITION family of sequence functions
1109 are significantly faster on arrays whose element types have been
1111 * improvements to the Windows port:
1112 ** Intermittent heap corruption problems have been fixed. (thanks
1113 to Alastair Bridgewater)
1114 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1116 ** Lisp is able to unwind foreign exception frames from alien
1117 callbacks. (thanks to Alastair Bridgewater)
1119 changes in sbcl-1.0 relative to sbcl-0.9.18:
1120 * improvement: experimental support for threading on FreeBSD/x86.
1121 (thanks to NIIMI Satoshi)
1122 * improvement: runtime option --dynamic-space-size can be used
1123 to set the size of the dynamic space reserved on startup.
1124 * improvement: floating point modes in effect are now saved in
1125 core, and restored on startup.
1126 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1127 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1128 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1129 threads are not running after *SAVE-HOOKS* have run.
1130 * improvement: writes to CLOS instance slots are type-checked in code
1131 compiled with (SAFETY 3)
1132 * improvement: floating-point exception handling on FreeBSD (thanks to
1134 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1135 (thanks to Zach Beane)
1136 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1138 * improvement: added support for the Shift-JIS external format.
1139 (contributed by NIIMI Satoshi)
1140 * improvement: callbacks are supported on Linux/PPC. (thanks to
1142 * bug fix: compiler bug triggered by a (non-standard) VALUES
1143 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1144 * bug fix: file compiler no longer confuses validated and already
1145 dumped structurres. (reported by Kaersten Poeck)
1146 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1147 fill-pointer signals a type-error as required. (thanks to
1149 * bug fix: disassemly of funcallable instances works.
1150 * bug fix: single stepping on PPC.
1151 * bug fix: fix thread-safety problems in the type system (generally
1152 manifesting as nonsensical errors like "STRING is a bad type specifier
1153 for sequences" or "The value 1 is not of type FIXNUM")
1154 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1155 (reported by Josip Gracin)
1156 * bug fix: an error is signaled for attempts to displace arrays with
1157 incompatible element types (thanks to Mario Mommer)
1158 * bug fix: more correct handling of wide characters in debug info
1159 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1160 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1161 and don't cause extra consing
1162 * optimization: MAP and MAP-INTO are significantly faster on vectors
1163 whose elements types have been declared.
1164 * Improvements to SB-SPROF:
1165 ** Support for allocation profiling
1166 ** Reduced profiling overhead, especially for long profiling runs
1167 * Improvements to the Windows port:
1168 ** floating point exceptions are now reported correctly.
1169 ** stack exhaustion detection works partially.
1170 ** more accurate GET-INTERNAL-REAL-TIME.
1171 ** better RUN-PROGRAM behaviour in case of errors in the child
1173 ** PROBE-FILE now simplifies pathnames correctly.
1174 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1175 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1176 1st 1970 (although time timezone data will be unreliable).
1177 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1179 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1180 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1181 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1183 * bug fix: two potential GC deadlocks affecting threaded builds.
1184 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1185 returns T (reported by Anton Kazennikov)
1186 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1187 with non-variable places
1188 * bug fix: remove a race condition in the setting of
1189 funcallable-instance functions, this should make threaded CLOS
1190 code more stable against memory faults.
1191 * bug fix: corruption of specials when unbinding is interrupted by an
1192 asynchronous unwind (reported by Hannu Koivisto)
1193 * improvement: the debugger will now also display local variables that
1194 are only used once, for code compiled with a DEBUG optimization quality
1197 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1198 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1199 * incompatible change: External-format support for FFI calls. The
1200 SB-ALIEN:C-STRING no longer implies an ASCII
1201 external-format. Instead the string is subject to external-format
1202 conversion. Additionally return values of type SB-ALIEN:C-STRING
1203 are by default of type (SIMPLE-ARRAY CHARACTER), not
1204 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1205 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1206 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1207 * incompatible change: SB-EXT package no longer contains the
1208 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1209 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1211 * incompatible change: the single-stepper is no longer available
1212 on Alpha, Mips and Sparc platforms.
1213 * minor incompatible change: the direct superclasses of
1214 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1215 not (STANDARD-OBJECT FUNCTION). This makes the
1216 class-precedence-lists of GENERIC-FUNCTION and
1217 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1219 * minor incompatible change: the default stream external format on
1220 non-unicode SBCL is no longer detected from the locale, but is always
1222 * new feature: Add a version of evaluator that uses an interpreter instead
1223 of the compiler. EVAL still uses the compiler by default, to switch it
1224 to use the interpreter, set the value of the variable
1225 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1226 * minor incompatible change: the single-stepper REPL has been merged
1227 with the normal debugger (see the "Stepping" heading of the debugger help
1228 for more details). The debugger command STEP will no longer switch
1229 to the single-stepper REPL.
1230 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1231 for a type now works.
1232 * bug fix: loading of default sysinit file works. (thanks to Leonid
1234 * bug fix: better detection of circularities in the file-compiler.
1235 (reported by Marco Monteiro)
1236 * bug fix: the CL pathname functions now work with files that have
1237 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1238 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1239 systems with Unix98 pty semantics.
1240 * bug fix: ASDF-INSTALL will now work with bsd tar.
1241 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1243 * bug fix: timers expiring in dead threads no longer cause a
1244 type-error (reported by Paul "Nonny Mouse").
1245 * bug fix: thanks to more lightweight single-stepper instrumentation,
1246 code compiled with (DEBUG 3) will compile and execute significantly faster,
1247 and will have more accurate type-inferencing than before
1248 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1249 whose bindings are modified
1250 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1251 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1252 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1253 CL:SOFTWARE-VERSION with proper version
1255 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1256 * feature: implemented the READER-METHOD-CLASS and
1257 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1258 as specified by AMOP.
1259 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1261 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1262 (thanks to Lutz Euler)
1263 * optimization: hashing of general arrays and vectors has been
1264 improved. (reported by Any Fingerhut)
1265 * enhancement: SB-INTROSPECT is now able to find definitions of
1266 profiled functions. (thanks to Troels Henriksen)
1267 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1268 as well, allowing compiler-macros for SETF-functions to expand.
1269 * enhancement: step-instrumentation no longer wraps known
1270 single-value functions with multiple-value context, allowing
1271 better type inference.
1272 * fixed bug #337: use of MAKE-METHOD in method combination now works
1273 even in the presence of user-defined method classes. (reported by
1274 Bruno Haible and Pascal Costanza)
1275 * fixed bug #339(c): if there are applicable methods not part of any
1276 long-form method-combination group, call INVALID-METHOD-ERROR.
1277 (reported by Bruno Haible)
1278 * fixed bug #361: the :FUNCTION initarg in the protocol for
1279 initialization of methods can now be used to override
1280 internally-produced optimized functions. (reported by Bruno
1282 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1283 system-provided lambda expression no longer cause warnings about
1284 unbound #:|pv-table| symbols.
1285 * bug fix: improved the handling of type declarations and the
1286 detection of violations for keyword arguments with non-constant
1288 * bug fix: potentially erronous calls to PATHNAME and
1289 MERGE-PATHNAMES were being flushed in some cases.
1290 (reported by Richard Kreuter)
1291 * bug fix: compiled calls to TYPEP were mishandling obsolete
1292 instances. (reported by James Bielman and Attila Lendvai)
1293 * bug fix: format strings with ~> without matching ~< no longer
1294 trigger an AVER, but signal an understandable error instead.
1295 (reported by Antonio Martinez)
1296 * bug fix: specifying an output-file without a pathname-type for
1297 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1298 of input-file instead of "fasl". (reported by Robert Dodier)
1299 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1300 (reported by James Y Knight).
1301 * bug fix: compiler-macros-function did not consider the environment
1302 argument for shadowing by local functions.
1303 * bug fix: compiler-macros expansion was inhibited by local INLINE
1305 * bug fix: inline expansions of known functions were subject to
1306 step-instrumentation in high debug policies, leading to problems
1307 with type-inference.
1308 * bug fix: compiler failed to differentiate between different CONS
1309 types in some cases.
1310 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1311 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1312 * bug fix: #368: incorrect use of expressed vs. upgraded array
1314 * bug fix: #306a: more precise unions of array types.
1315 * thread-safety improvements:
1316 ** CONDITION-WAIT could return early on Linux, if the thread was
1317 interrupted and subsequently continued with SIGCONT.
1318 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1320 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1321 * added support for the ucs-2 external format. (contributed by Ivan
1323 * minor incompatible change: pretty printing of objects of type
1324 (cons symbol) is, in the default pprint-dispatch-table, now
1325 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1327 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1328 called later in a class's lifetime, possibly as late as when the
1329 first instance of the class is created. Previously,
1330 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1331 class became finalizeable.
1332 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1333 for multibyte external-format streams. (thanks to "vbzoli")
1334 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1335 with :ALLOCATION :CLASS inherited from superclasses of the
1337 * fixed bug: anonymous classes can now be created using the :NAME
1338 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1339 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1340 * fixed bug: core-files saved with :EXECUTABLE T can again be
1341 executed when SBCL_HOME isn't set. (reported by James Knight)
1342 * fixed bug: toplevel LOCALLY forms with declarations could
1343 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1344 * fixed bug: printing from several different threads using different
1345 values of *print-case* could cause invalid output, due to
1346 some internal special variables of the printer not being bound
1347 thread-locally (reported by Max Mikhanosha)
1348 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1349 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1350 (reported by Pascal Costanza)
1351 * fixed bug: FINALIZE-INHERITANCE is called from
1352 REINITIALIZE-INSTANCE on classes when the class has previously
1353 been finalized, as required by AMOP.
1354 * minor code generation optimizations:
1355 ** better register allocation in CLOS dispatching functions
1356 ** overflow detection when coercing signed bytes to fixnums on x86-64
1357 ** is now implemented with one IMUL instruction instead of three shifts
1358 ** more efficient bit-vector access on x86 and x86-64
1359 ** more efficient access to raw structure slots on x86 and x86-64
1360 * fixed some bugs revealed by Paul Dietz' test suite:
1361 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1362 return its argument.
1364 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1365 * feature: thread support on Solaris/x86, and experimental thread support
1367 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1369 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1370 default method for SLOT-UNBOUND.
1371 * minor incompatible change: prevent the user from specializing the
1372 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1373 somewhat counter to the intent of the protocol, I (CSR) think, and
1374 additionally it just doesn't work in SBCL as currently
1375 implemented, thanks to optimizations (that are always valid for
1376 the other three SLOT-VALUEish functions, but not for the setter).
1377 * minor incompatible change: the :SB-LDB feature is now enabled by
1378 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1379 the low-level debugger.
1380 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1381 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1382 down to an SB-SYS:FD-STREAM, allowing direct communication in
1383 more cases. Temporary files and pipes are still used as a fallback
1385 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1386 on the non-reentrant C functions localtime(3) and gmtime(3).
1387 * bug fix: global optimization policy was not visible in LOCALLY and
1389 * bug fix: class objects can be used as specializers in methods.
1390 (reported by Pascal Costanza)
1391 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1392 with a type error. (reported by blitz_ on #lisp)
1393 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1394 failed with a type error. (reported by Pascal Bourguignon)
1395 * bug fix: merging pathnames against defaults with :DIRECTORY
1396 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1397 (reported by James Y Knight)
1398 * bug fix: saving large (>2GB) cores on x86-64 now works
1399 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1400 constant index and a (simple-array (signed-byte 32)) array
1401 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1402 error (patch by Robert J. Macomber)
1403 * feature: TIME now displays the amount of run-time spent in GC
1404 * bug fix: The debugger now does a better job of respecting
1405 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1406 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1407 workaround for bug 403.)
1408 * bug fix: floating point exception handling now works on FreeBSD
1409 (thanks to NIIMI Satoshi)
1410 * fixed some bugs revealed by Paul Dietz' test suite:
1411 ** MISC.641: LET-conversion were not supposed to work in late
1413 * improvements to the Win32/x86 port:
1414 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1415 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1417 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1418 * new feature: source path information is generated for macro-expansion
1419 errors for use in IDE's like Slime (thanks to Helmut Eller)
1420 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1421 * bug fix: compiler does not loop forever on an invalid type in
1423 * improvement: compilation of most CLOS applications is significantly
1425 * optimization: added a limited bytecode compiler for simple toplevel
1426 forms, speeding up compilation and FASL loading
1427 * bug fix: the statistical profiler now properly distinguishes anonymous
1430 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1431 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1432 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1433 into subclasses of the finalized class. Now user calls to
1434 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1435 system return before any subclasses are finalized.
1436 * minor incompatible change: The reader no longer ignores errors
1437 regarding non-existent packages in #+ and #- feature tests.
1438 * new feature: command line options --no-sysinit, --no-userinit to
1439 inhibit loading the corresponding init files
1440 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1441 for sending data through UDP sockets (thanks to François-René Rideau)
1442 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1443 error is signalled instead (thanks to François-René Rideau)
1444 * minor incompatible change: Error signalling behaviour of lexical
1445 operations violating package locks has changed slightly. Refer to
1446 documentation on package locks for details.
1447 * enhancement: EVAL can process IF-expressions without resorting to the
1449 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1450 constant-folded properly due to missing out-of-line functions.
1451 (reported by Mika Pihlajamäki)
1452 * bug fix: LISTEN sometimes returned T even in cases where no data was
1453 immediately available from the stream
1454 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1455 were reversed. (reported by Levente Mészáros)
1456 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1457 when given a vector displaced to an adjustable array. (reported
1459 * bug fix: derivation of float boundaries from numbers outside the
1460 appropriate float range (reported by John Wiseman)
1461 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1463 * bug fix: precomputing cache entries for generic functions with
1464 some subclasses of specializers as yet invalid does not attempt to
1465 fill a cache line with a negative offset. (reported by Levente
1467 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1468 allow condition class objects as arguments to DOCUMENTATION and
1469 (SETF DOCUMENTATION); only find and set documentation for
1470 structure names for the STRUCTURE doc-type. (suggested by Gary
1472 * improvements to the Win32/x86 port:
1473 ** (user-homedir-pathname) and default initialization file
1474 locations now know about the user's "Documents and Settings"
1475 directory (thanks to Yaroslav Kavenchuk)
1476 ** run-program is implemented (thanks to Mike Thomas)
1477 ** sockets support (thanks to Timothy Ritchey)
1478 ** better backtrace support (thanks to Alastair Bridgewater)
1479 ** sb-grovel supported
1480 ** asdf-install and sb-posix work somewhat
1481 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1482 * improvements to the Solaris/x86 port:
1483 ** works on Solaris 11/Solaris Express
1484 ** floating-point exception handling support
1485 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1486 * fixed some bugs revealed by Paul Dietz' test suite:
1487 ** REMOVE-METHOD returns its generic function argument even when
1488 the method is not one of the generic functions' methods.
1489 ** objects declared as MEMBER types can be admissible arguments to
1490 structure accessors.
1491 ** printing characters should simply be printed by the FORMAT ~:C
1493 ** compiler failure when compiling functions with hairy constant
1494 defaults for optional parameters.
1495 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1496 function, which is already optimized.
1498 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1499 * new platform: experimental support for SBCL x86/Darwin, including
1500 MacOS X 10.4.5 on Intel.
1501 * new feature: Unicode character names are now known to the system
1502 (through CHAR-NAME and NAME-CHAR).
1503 * new feature: the filesystem location of SBCL's core file is
1504 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1505 * minor incompatible change: the contrib modules SB-POSIX and
1506 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1507 this change is to make it easier to distribute
1508 location-independent binaries.
1509 * bug fix: as implied by AMOP, standardized classes no longer have
1510 slots named by external symbols of public packages. (reported by
1512 * optimization: calling functions via constant symbols -- as in
1513 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1514 via the function object as in (FUNCALL #'FOO).
1515 * optimization: on x86 and x86-64, the calling convention for the
1516 case of dealing with an unknown number of values has been altered
1517 to be friendlier to the prediction heuristics implemented,
1518 particularly with reference to CALL and RET pairing. (thanks to
1519 Alastair Bridgewater)
1520 * optimization: on x86, the code for access to thread-local symbol
1521 values and binding/unbinding of thread-local symbols is smaller.
1522 (thanks to Alastair Bridgewater)
1523 * enhancement: CONSTANTP is now able to determine constantness of
1524 more complex forms, including calls to constant-foldable standardized
1525 functions and some special forms beyond QUOTE.
1526 * fixed bug: occasional GC crashes on Solaris/x86
1527 * optimization: x86-64 supports stack allocation of results of simple
1528 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1529 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1530 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1531 (thanks to James Knight)
1532 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1533 by the standard (thanks to Stephen Compall)
1535 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1536 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1537 be used for bundling the runtime and the core file into one
1538 executable binary. This feature is not currently supported on all SBCL
1539 platforms. (thanks to James Bielman and NIIMI Satoshi)
1540 * new feature: a generational or ephemeral garbage collector is now
1541 the default on the PowerPC platform (both Linux and Darwin). The
1542 old Cheney (stop and copy) collector is a build-time option.
1543 (thanks to Cyrus Harmon, after Raymond Toy)
1544 * minor incompatible change: the method by which SBCL finds its
1545 contributed modules has changed; it no longer relies on symbolic
1546 links from an $SBCL_HOME/systems directory, but searches directly
1547 in the subdirectories of $SBCL_HOME.
1548 * enhancement: the dynamic heap size on the Linux/PPC platform is
1549 markedly larger, even using the older Cheney garbage collector.
1550 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1551 certain complicated string types. (reported by Gary King)
1552 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1553 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1554 * fixed bug: DOCUMENTATION on structures no longer signals an
1555 error if no structure type of the right name exists. (reported by
1557 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1558 applicable non-standard primary or :AROUND method on
1559 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1560 no longer cause unbound variable errors. (reported by Kevin Reid)
1561 * fixed bug: metacircle resolution in cases where methods have slots
1562 added before the slots from STANDARD-METHOD. (reported by Jean
1564 * fixed bug: the Power PC assembly code for calling into lisp
1565 sometimes computed its return address wrongly (depending on the
1566 15th bit of the address, fixed at link time). This used to
1567 manifest itself in a segmentation violation while building PCL.
1568 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1569 many others over the years)
1570 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1571 the stack pointer (treated as a signed integer) is negative.
1572 (thanks to Peter van Eynde)
1574 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1575 * new platform: experimental support for the Windows operating
1576 system has been added. (thanks to Alastair Bridgewater)
1577 * fixed several bugs in and robustified the PPC FFI (including
1578 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1579 * bug fix: allow non-simple string symbol names (reported by Paul
1581 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1583 * bug fix: saving a core corrupted callbacks on x86/x86-64
1584 * bug fix: closed a loophole in metacircularity detection and
1585 grounding in the PCL implementation of CLOS.
1586 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1587 (reported by tomppa on #lisp)
1588 * bug fix: clear the direction flag on Lisp -> C transitions, as
1589 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1590 (reported by Andrej Grozin and Hendrik Maryns)
1591 * optimization: major improvements to GC efficiency on GENCGC platforms
1592 * optimization: faster implementation of EQUAL
1593 * optimization: emit more efficient opcodes for some common
1594 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1595 * optimization: several other minor code-generation improvements on x86-64
1596 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1598 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1599 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1600 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1601 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1602 by ANSI as a generic function, and for consistency (SETF
1603 GENERIC-FUNCTION-NAME) is treated likewise.
1604 * fixed bug #233.b: make constraint propagation notice when a variable
1605 value is changed after it is referenced but before it is used
1606 * fixed bug #296: no more arbitrary behaviour from filenames with
1607 odd characters as --load arguments. (reported by Adam Warner)
1608 * fixed bug #390: :CHARACTER-SET pathname components now work as
1609 expected. (reported by Tim Daly Jr)
1610 * fixed bug #391: complicated :TYPE intersections in slot
1611 definitions no longer cause an error in PCL internals.
1612 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1613 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1614 returning the number of octets which would be written to the
1615 file-stream. (thanks to Robert J. Macomber)
1616 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1617 arguments to the internal stream functions. (thanks to David
1619 * bug fix: stack exhaustion is now handled on the x86/SunOS
1620 platform. (thanks to Daisuke Homma)
1621 * bug fix: more accurate ROOM results on GENCGC platforms
1622 * optimization: improved type inference for arithmetic-for
1623 index variables in LOOP
1624 * optimization: faster floating-point SQRT on x86-64
1625 * fixed some bugs revealed by Paul Dietz' test suite:
1626 ** DOCUMENTATION returns NIL instead of "" for method combinations
1627 that don't have a docstring
1629 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1630 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1631 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1632 therefore are not customizeable by user code (as seems to be at
1633 least permitted and maybe required by AMOP). As a consolation,
1634 however, the SBCL implementation of these functions now calls
1635 REINITIALIZE-INSTANCE as specified by AMOP.
1636 * bug fix: slot-definition documentation is propagated as per ANSI
1637 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1638 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1639 argument, as do eslotd initialization methods. (from Pascal
1640 Costanza's "Closer" project)
1641 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1642 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1644 * bug fix: it is now possible to have more than one subclass of
1645 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1646 (reported by Bruno Haible, Pascal Costanza and others)
1647 * bug fix: the dependent update protocol now works for generic
1648 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1649 and Pascal Costanza)
1650 * bug fix: condition-class instances corresponding to
1651 DEFINE-CONDITION forms are now created eagerly. (reported by
1652 Kalle Olavi Niemitalo on comp.lang.lisp)
1653 * bug fix: floating point printing is more accurate in some
1654 circumstances. (thanks to Simon Alexander)
1655 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1656 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1657 * bug fix: callbacks on OS X now preserve stack-alignment.
1658 (thanks to Cyrus Harmon)
1659 * enhancement: the x86-64 disassembler is much better at
1660 disassembling SSE instructions. (thanks to Lutz Euler)
1661 * enhancement: CHECK-TYPE on a variable now allows type-inference
1662 (thanks to Vincent Arkesteijn)
1663 * optimization: improved performance of EUC-JP external format.
1664 (thanks to NIIMI Satoshi)
1665 * optimization: performance improvements to IO on file streams of
1666 :ELEMENT-TYPE CHARACTER
1667 * optimization: much faster memory allocation on x86-64
1668 * optimization: faster garbage collections (latency and throughput)
1670 * optimization: faster bignum subtraction and division on x86 and x86-64
1671 (thanks to Lutz Euler)
1672 * optimization: more accurate type-inference for loops that use a
1673 floating point index variable or a negative step.
1675 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1676 * bug fix: add a workaround to SBCL looping infinitely at startup on
1677 Linux kernels with apparently buggy implementations of personality().
1678 (thanks to Svein Ove Aas)
1679 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1680 (thanks to David Lichteblau)
1681 * bug fix: local bindings shadow global symbol macros for
1682 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1683 * optimization: non-open coded uses of numeric comparison operators
1684 (e.g. >) no longer cons when called with more than one parameter
1685 on platforms supporting dynamic-extent allocation.
1686 * enhancement: saving cores with foreign code loaded is now supported
1687 on MIPS/Linux in addition to the previously supported platforms.
1688 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1689 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1690 * bug fix: division by zero in sb-sprof when no samples were collected
1691 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1692 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1693 (thanks to Alexander Kjeldaas)
1694 * new feature: ignoring the return values of destructive functions
1695 like NREVERSE signals a compile-time style-warning.
1696 (thanks to Kevin Reid)
1697 * new feature: the alignment of alien structure fields can be
1698 explicitly specified. (thanks to Cyrus Harmon)
1699 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1700 and dump core on SIGQUIT
1702 ** incompatible change: threads do not inherit values of specials
1703 from their parents (see manual)
1704 ** bug fix: threads stacks belonging to dead threads are freed by the
1705 next exiting thread, no need to gc to collect thread stacks anymore
1706 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1707 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1708 ** bug fix: dynamic variable and thread start related gc lossage
1709 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1711 * fixed some bugs revealed by Paul Dietz' test suite:
1712 ** SUBTYPEP is slightly more accurate on heinously complicated
1713 CONS types where some of the members have uncertain (in the
1714 NIL, NIL sense) type relationships to each other.
1716 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1717 no longer purified unless :PURIFY T is explicitly specified.
1718 ** Non-purified cores are significantly smaller than before
1720 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1721 * new feature: timers based on Zach Beane's excellent timer package
1722 * added support for the following external formats: koi8-u,
1723 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1724 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1725 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1726 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1727 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1728 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1729 * incompatible change: a threaded SBCL will no longer revert to
1730 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1731 * bug fix: interrupts are disabled until startup is complete; no
1732 more sigsegvs when receiving a signal to soon
1733 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1735 * bug fix: add a workaround for the memory randomization features in
1736 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1737 workaround will only be in effect on systems with the proc filesystem
1739 * bug fix: printing objects of type HASH-TABLE signals a
1740 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1742 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1743 have been running for over 50 days. (reported by Gilbert Baumann)
1744 * bug fix: the logic for getting names of functions gets less
1745 confused when confronded with alternate-metaclass
1746 funcallable-instances. (reported by Cyrus Harmon)
1747 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1748 consistent, even on internal alternate-metaclass objects.
1749 * bug fix: flush closure information collected by physical
1750 environment analysis prepass before the main pass. (bug reported
1752 * bug fix: compiler pack phase does not modify a hashtable, which is
1753 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1754 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1755 may be expanded. (reported by Alexander <malishev> on c.l.l)
1756 * bug fix: no more style-warnings when compiling code using LAST
1757 under high-SPEED low-SPACE settings. (reported by David Wallin
1759 * fixed some incompatibilities between SBCL's MOP and the MOP
1761 ** the METAOBJECT class is now implemented;
1762 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1763 STANDARD-OBJECT, as required;
1764 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1765 now compatible (as required by VALIDATE-SUPERCLASS); there
1766 remains a consistency requirement over the presence or absence
1767 of the FUNCTION class in the superclass list of finalized
1768 classes; see the manual for more details;
1769 ** the :AROUND method for COMPUTE-SLOTS on
1770 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1771 requested slot ordering.
1773 ** bug fix: parent thread now can be gc'ed even with a live
1775 ** bug fix: binding a special with PROGV to no value is not
1776 the same as that symbol not having been bound (thanks to
1778 ** bug fix: binding specials is thread safe (thanks to
1780 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1781 and SB-SPROF do not die with 'no handler for signal XX in
1782 interrupt_handle_now(..)' anymore
1783 ** bug fix: WITH-TIMEOUT works with multiple running threads
1784 * fixed some bugs revealed by Paul Dietz' test suite:
1785 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1786 the :method-class keyword argument.
1788 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1789 * new port: the Solaris operating system on x86 processors is now
1790 mostly supported, though some rough edges in the environment
1791 remain. (thanks to Daisuke Homma)
1792 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1793 space for its heap. (thanks to Thiemo Seufer)
1794 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1795 closures. (thanks to Thiemo Seufer)
1796 * minor incompatible change: eof selects abort in the debugger.
1797 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1798 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1800 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1801 deprecated, and will go away in a future revision of SBCL.
1802 * minor incompatible change: GC-ON and GC-OFF are no longer
1803 implemented with a counter, it does not matter how many times gc
1804 is switched on or off
1805 * bug fix: discriminating functions for generic function classes
1806 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1807 make invalid assumptions about method precedence order. (reported
1809 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1810 platforms. (thanks to Thiemo Seufer)
1811 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1812 * bug fix: correct pathname printing with printer escaping is on.
1813 (thanks to Kevin Reid)
1814 * bug fix: complex VOP definitions in "user-space" no longer trigger
1815 package locks. (reported by Zach Beane)
1816 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1817 for funcallable-instance functions, and is no different from
1819 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1820 characters in the current readtable. (reported by Nicholas Neuss)
1821 * bug fix: SUBTYPEP on various CONS types returns more a more
1822 accurate acknowledgment of its certainty.
1823 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1824 lists in some cases. This partially fixes bug 384.
1825 * flush all standard streams before prompting in the REPL and the
1827 * bug fix: signal handling and triggering gc do not conflict
1828 directly anymore, in particular a high frequency sb-sprof does
1829 not prevent gc from running
1830 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1831 approximation for timezone and DST information between the
1832 universal time epoch and the smallest negative 32-bit time_t.
1833 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1834 year 1899 so long as the encoded time is non-negative.
1835 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1836 the correct answer for non-constant rotate amounts when performing
1837 an inline 32-bit rotation.
1839 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1840 there is only one thread in the session
1841 ** bug fix: memory leak for streams created in one thread and
1842 written to in another
1843 ** bug fix: lockup when compiled with gcc4
1844 ** bug fix: race that allows the gc to be triggered when gc is
1846 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1847 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1849 ** bug fix: finished threads are gc'ed properly
1850 * fixed some bugs revealed by Paul Dietz' test suite:
1851 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1852 has arguments with a different set of applicable methods from
1853 the orignal arguments.
1854 ** The type error thrown by MAP now has a correct expected-type
1856 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1857 name a compiled function.
1858 ** Three MISC tests where a large bignum was improperly coerced to
1859 a float that couldn't represent that bignum during type
1860 derivation were fixed.
1861 ** SUBTYPEP can now handle types involving the negation of a
1862 list-form FUNCTION type.
1863 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1864 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1865 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1867 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1868 * New feature: Experimental support for bivalent streams: streams
1869 opened with :element-type :default now allow character and binary
1870 (unsigned-byte 8) I/O
1871 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1872 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1874 * bug fix: no more highly sporadic "couldn't check whether ~S is
1875 readable" when reading a stream and an interrupt hits in the middle
1876 of a select system call
1877 * compiler better recognizes complex arrays (reported by Eduardo
1879 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1880 for variables on SBCL built with linkage-tables. (reported by Luis
1882 * various error reporting improvements.
1883 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1884 (Thanks to Thiemo Seufer)
1885 * bug fix: nested reader invokations work correctly
1886 * bug fix: it is possible to have simultaneous references to foreign
1887 code and foreign data with the same name.
1889 ** added x86-64 support
1890 ** incompatible change: the threading api now works with thread
1891 objects instead of thread ids
1892 ** bug fix: threads are protected from signals and interruption when
1893 starting up or going down
1894 ** bug fix: a race where an exiting thread could lose its stack to gc
1895 ** bug fix: don't halt on infinite error in threads if possible
1896 ** fixed numerous gc deadlocks introduced in the pthread merge
1897 ** bug fix: fixed thread safety issues in read and print
1898 ** bug fix: debugger doesn't hang on session lock if interrupted at
1899 an inappropriate moment
1900 ** bug fix: run-program is now thread safe(r)
1901 ** bug fix: inner with-recursive-lock no longer releases the mutex
1902 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1903 * fixed some bugs revealed by Paul Dietz' test suite:
1904 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1905 EXPECTED-TYPE slots filled.
1906 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1908 ** MAKE-CONDITION accepts classes as type-designators.
1909 ** COMPILE may never return NIL.
1910 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1911 range before calling Unix time functions
1913 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1914 * numerous signal handling fixes to increase stability
1915 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1916 * minor incompatible change: we now correctly canonize default
1917 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1918 as per the MOP, rather than the historical (INITARG INITFUNCTION
1919 INITFORM). (reported by Bruno Haible)
1920 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1921 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1922 for more information.
1923 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1924 pathname is a directory pathname.
1925 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1926 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1928 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1929 :METACLASS argument in addition to a class name. (reported by
1930 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1931 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1932 (reported by Bruno Haible; patch from Gerd Moellmann)
1933 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1935 * bug fix: more cleanups to the floating point exception handling on
1936 x86-64 (thanks to James Knight)
1937 * bug fix: the compiler does not try to constant fold calls of
1938 COERCE and BIT-* functions when they return freshly constructed
1939 objects. (reported by Nikodemus Siivola and Paul Dietz)
1940 * optimization: improved the MIPS versions of generic (in the
1941 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1942 * optimization: direct conversion of (unsigned-byte 32) to floats on
1943 the PowerPC platform.
1944 * optimization: structure instances with raw slots now use less
1945 memory, and probably show better memory locality. (thanks to
1947 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1948 are treated as requests for stack allocation on the x86-64,
1949 Alpha-32, PPC and SPARC platforms.
1950 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1951 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1952 (thanks to Tony Martinez)
1954 ** gcing a dead thread can no longer lead to lockups
1955 ** threads block signals until they are set up properly
1956 ** errno is no longer shared by threads
1957 ** interrupt-thread restores the eflags register on x86
1958 ** fixed some lockups due to gc/thread interaction
1959 * fixed some bugs revealed by Paul Dietz' test suite:
1960 ** invalid dotted lists no longer raise a read error when
1961 *READ-SUPPRESS* is T
1962 ** signal an error if a symbol that names a declaration is used
1963 as the name of a type, or vice versa
1964 ** allow using the (declare (typespec var*)) abbreviation for
1965 (declare (type typespec var*)) with all type specifiers
1966 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1967 ** always return NIL from PPRINT-POP when OBJECT is NIL
1968 ** don't signal errors when pretty-printing malformed LABELS,
1969 FLET or MACROLET forms
1970 ** declarations in a DOLIST body are also in scope for the
1972 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1974 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1977 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1978 * fixed cross-compiler leakages that prevented building a 32-bit
1979 target with a 64-bit host compiler.
1980 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1981 opened with :IF-EXISTS :OVERWRITE.
1982 * fixed bug 281: error for an invalid qualifier in a short-form method
1983 combination method is not signalled until the faulty method is called.
1984 * bug fix: iteration variable type inferrer failed to deal with open
1985 intervals. (reported by Alan Shields)
1986 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1987 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1988 or more runtime options were provided to the sbcl binary.
1989 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1990 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1991 * optimization: there's now a fast-path for fixnum arguments in the
1992 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1993 * optimization: the code generated on x86-64 is more compact thanks
1994 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1995 * bug fix: floating-point exception handling now partly works on
1996 x86-64. (Thanks to James Knight)
1997 * improvement to the MIPS backend from Thiemo Seufer: C-style
1998 64-bit long long arguments and return values to alien functions
2000 * fixed some bugs revealed by Paul Dietz' test suite:
2001 ** the type-error signalled from WARN has a filled-in DATUM slot.
2002 ** the type-error required when a stream is not associated with
2003 a file has the stream as its datum.
2004 ** type-errors on single-floats on x86-64 no longer have
2005 :INVALID-OBJECT as the datum
2006 ** the type-errors signalled for invalid function names now have
2007 a correct expected type
2008 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2009 for structures defined with a :TYPE. Documentation strings for
2010 typed structures are no longer immediately discarded
2011 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2012 broadcast streams and synonym streams. FILE-LENGTH now also works
2013 on broadcast streams.
2015 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2016 * incompatible change: the --noprogrammer option, deprecated since
2017 version 0.7.5, has been removed. Please use the equivalent
2018 --disable-debugger option instead.
2019 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2021 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2022 inoperational for a while now) has been completely removed.
2023 * null lexical environments are now printed as #<NULL-LEXENV>,
2024 significantly reducing the amount of clutter in typical backtraces.
2025 * documentation on weak pointers, finalization, and after GC hooks
2026 has been added to the manual.
2027 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2028 as other specialized array types, is much faster. SUBSEQ and
2029 COPY-SEQ on such arrays have also been sped up.
2030 * optimization: EQL is now more efficient when at least other argument
2031 is known to be of type (OR FIXNUM (NOT NUMBER)).
2032 * fixed bug: the runtime is now less vulnerable to changes in the
2033 size of the SBCL object on OS X, and virtual memory is reserved for
2034 all spaces that need to be at a fixed address.
2035 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2036 * fixed bug: finalizers and after GC hooks that cause consing are now
2038 * fixed bug: compiler error messages and summaries are now printed to
2039 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2040 * fixed inference of the upper bound of an iteration variable.
2041 (reported by Rajat Datta).
2042 * fixed bug 211e: calling local functions with duplicated constant
2043 keyword argument no longer causes a bogus style warning about an
2045 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2046 ftype declaration to be disregarded. (reported by Dave Roberts)
2047 * fixed bug 373: caused by erronous compilation of references to alien
2048 variables in the runtime on ppc/darwin.
2049 * fixed bug 376: CONJUGATE type deriver.
2050 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2051 (reported by Baughn on #lisp)
2052 * a cross-compiler bug on non-x86 platforms has been identified and
2053 fixed. (thanks to Bruno Haible)
2054 * improvements to the MIPS runtime code for increased stability.
2055 (thanks to Thiemo Seufer)
2056 * increased the maximimum compact environment size to allow
2057 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2058 * partial workaround for bug 135: don't name the function we're
2059 calling for hairy cases of EVAL, so as not to accumulate one environment
2060 entry per call to EVAL. (reported by Kevin Reid)
2061 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2062 * optimization: type testing for non-vector arrays should be faster.
2063 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2065 * bug fix: setting 31st element of a bit vector to zero did not work
2067 * bug fix: redefining a class definition which failed due to a
2068 previous accessor / function clash now works (but see BUGS entry
2069 #380 for more problems in this area). (thanks to Zach Beane)
2070 * the long form of DEFINE-METHOD-COMBINATION disables method group
2071 checking when given a single method group with pattern *. (thanks
2073 * on x86 compiler supports stack allocation of results of simple
2074 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2075 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2076 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2077 * fixed some bugs related to Unicode integration:
2078 ** the restarts for recovering from input and output encoding
2079 errors only appear when there is in fact such an error to
2081 * fixed some bugs revealed by Paul Dietz' test suite:
2082 ** MISC.549 and similar: late transformation of unsafe type
2083 assertions into derived types caused unexpected code
2085 ** SCALE-FLOAT type deriver is less wrong.
2086 ** type derivers for EXP, LOG and similar functions compute result
2087 types for complex arguments better.
2088 ** (MISC.563) CONJUGATE type deriver works for very restricted
2090 ** out-of-line type testers for character strings are available.
2091 ** EQUAL compiler transform understands specialness of objects
2093 ** accessing double-floats stored on the stack now works on x86-64.
2094 ** debugger internals could sometimes create invalid lispobjs,
2095 resulting in GC crashes.
2096 ** MISC.548: type check weakening can convert required type into
2098 ** initialization forms for bindings are not in scope of free special
2101 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2102 * incompatible change: thread support for non-NPTL systems has
2103 been removed - locking is buggy and unreliable. A threaded
2104 SBCL build will now warn at startup and refuse to create
2105 new threads, unless futex support is detected in the kernel
2106 * incompatible change: the top level REPL now has only an ABORT
2107 restart associated with it, not TOPLEVEL and ABORT as it used to.
2108 TOP and TOPLEVEL are now available as debugger commands for
2109 returning to the top level.
2110 * incompatible change: forms evaluated in the REPL now use the
2111 global optimization policy.
2112 * incompatible change: user- and system-initialization files are
2113 no longer processed with LOAD, but by READ and EVAL; hence the
2114 global optimization policy, startup package, readtable, etc,
2116 * The .fasl file format number has been incremented because of
2117 various incompatible changes.
2118 * internal entry point details and argument counts no longer appear
2119 in backtraces unless explicitly requested by setting
2120 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2121 * built-in and standard functions no longer have names like "top
2122 level local call to FOO".
2123 * fixed bug 32: functions defined in non-null lexical environments
2124 now have more legible printed representation
2125 * fixed bug 33: functions defined in non-null lexical environemnts
2126 are now more amenable to inspection by INSPECT.
2127 * workaround for bug 354: XEPs no longer appear in backtraces unless
2128 explicitly requested.
2129 * fixed bug: receiving the signal which results from attempting to
2130 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2131 SIGBUS on other *BSDs) on architectures where the C stack is also
2132 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2133 to debugger confusion or wild instability. (reported by Cheuksan
2135 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2136 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2137 (reported by Lutz Euler)
2138 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2139 compiled code. (reported by Paul Dietz)
2140 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2141 specialized parameter to a method to something that is not TYPEP
2142 the specializer is now possible.
2143 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2144 face of package deletion.
2145 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2146 pathname. (reported by David Tolpin on c.l.l)
2147 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2148 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2149 STRING-STREAM and FILE-STREAM, respectively.
2150 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2151 than previously; in addition, its test suite is now run on build.
2152 (thanks to Luke Gorrie)
2153 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2154 * optimization: added a immediate representation for single-floats
2156 * fixed some bugs revealed by Paul Dietz' test suite:
2157 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2158 ** attempting to create a package with a colliding nickname causes
2159 correctable errors to be signalled.
2160 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2161 ** several x86-64 backend bugs related to sign-extension of immediate
2164 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2165 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2166 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2168 * fixed loading of multiply forward-referenced layouts.
2169 (thanks to Cheuksan Wang)
2170 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2171 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2172 (see COMPILE-FILE documentation for details of :PRINT option)
2173 * fixed bugs 19 and 317: fixed-format floating point printing is
2174 more accurate. This also fixes a bug reported by Adam Warner
2175 related to the ~@F format directive.
2176 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2178 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2179 dispatch table if the from-char is a dispatch macro character.
2180 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2181 by the wordsize no longer ignore the last word. (reported by Lutz
2183 * fixed bug in type checking of dynamic-extent variables. (reported
2185 * optimization: sequence traversal functions use their freedom to
2186 coerce function designators to functions.
2187 * optimization: code with many calls to CLOS methods specialized on
2188 CLOS classes has had redundant type checks removed.
2189 * optimization: added declarations to speed up operations that access
2190 the internal character database (for example STRING-UPCASE)
2191 * optimization: comparison operations between floats and sufficiently small
2192 fixnums no longer create extra rationals
2193 * fixed some bugs related to Unicode integration:
2194 ** portions of multibyte characters at the end of buffers for
2195 character-based file input are correctly transferred to the
2196 start of the buffer at the next read.
2197 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2198 passing it through to OPEN.
2199 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2200 argument given to internal calls to OPEN.
2201 ** the built-in comment readers (introduced by character sequences
2202 ";" and "#|") are more forgiving to encoding errors; they will
2203 STYLE-WARN and then attempt to resync the stream at a character
2204 boundary. (thanks to Teemu Kalvas)
2205 * fixed some bugs in the x86-64 port:
2206 ** Negative short int return values from c-calls are sign-extended
2208 ** The stack is aligned to 16-bytes for c-calls, as required by
2209 the ABI. (thanks to Cheuksan Wang)
2210 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2211 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2212 for enums. (thanks to Vincent Arkesteijn)
2213 ** Multiple small optimizations and bugfixes for floating point
2215 * fixed some bugs revealed by Paul Dietz' test suite:
2216 ** Space, Tab, Linefeed, Return and Page have the invalid
2217 secondary constituent character trait.
2218 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2220 ** WITH-INPUT-FROM-STRING should only update the index place on
2222 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2223 works more reliably.
2224 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2225 with one-dimensional arrays with fill pointers.
2226 ** TYPE-OF failed on a complex with an integer realpart and a
2228 ** compiler failure during type inference for the code of form
2229 (IF (EQL X (THE ...)) ...) (MISC.535).
2231 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2232 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2233 platform. The port supports 61-bit fixnums, large memory spaces
2234 and reloading shared object files.
2235 * enhancement: saving cores with foreign code loaded is now
2236 supported on ppc/Darwin in addition to the previously supported
2238 * enhancement: the statistical profiler now walks deeper into the
2239 call stack for more meaningful call-graphs and accrued time
2240 reports (x86/x86-64 only). It also now reports time spent in
2242 * enhancement: it is now possible to trace most individual methods
2243 of a generic function in addition to tracing the generic function
2245 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2247 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2248 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2249 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2250 produces an error. (thanks to Vincent Arkesteijn)
2251 * bug fix: NAMESTRING on pathnames with :WILD components in their
2252 directories works correctly. (thanks to Artem V. Andreev)
2253 * fixed bug 125: compiler preserves identity of closures. (reported
2255 * bug fixed: functions with &REST arguments sometimes failed with
2256 "Undefined function" when compiled with (DEBUG 3). (reported by
2258 * bug fix: overflow during compiling of setting element of a bit
2259 vector with constant index and value. (reported by Timmy Douglas)
2260 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2261 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2262 * on x86 compiler partially supports stack allocation of dynamic-extent
2264 * GO and RETURN-FROM do not check the extent of their exit points
2265 when compiled with SAFETY 0.
2266 * fixed some bugs related to Unicode integration:
2267 ** encoding and decoding errors are now much more robustly
2268 handled; it should now be possible to recover even from invalid
2269 input or output to the terminal. (thanks to Teemu Kalvas)
2270 ** provided a first cut at implementing STRING-TO-OCTETS and
2271 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2272 ** altered the SB-MD5 contributed module slightly, changing the
2273 interface just enough for it to be supportable for builds where
2274 lisp characters are not eight bits.
2275 * fixed some bugs revealed by Paul Dietz' test suite:
2276 ** the FORMATTER-generated functions for ~V[ conditionals require
2277 the correct number of arguments.
2278 ** READ-FROM-STRING returns the mandated second value when applied
2279 to displaced strings.
2280 ** the #\Rubout and #\Backspace characters are treated as invalid
2281 constituent characters by the tokenizer.
2283 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2284 * new feature: reloading changed shared object files with
2285 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2286 * new feature: references to foreign variables and functions
2287 can now be compiled and loaded before the corresponding shared
2288 object file is loaded, as long as the foreign definitions are
2289 available at runtime.
2290 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2291 supported. (thanks to Dan Debertin)
2292 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2293 just keywords. (thanks to Vincent Arkesteijn)
2294 * fixed bug #331: structure-class instances corresponding to
2295 DEFSTRUCT forms are now created eagerly.
2296 * fixed bug #345: backtraces from calls to undefined functions work
2297 on x86 as well. Related bug #61 is now also partially fixed on x86
2298 (backtraces from throws to unknown catch tags.)
2299 * bug fix: lambda-list parsing is now stricter vrt. order and number
2300 of lambda-list keywords.
2301 * bug fix: as specified by AMOP, an error is signalled if a
2302 class-option appears multiple times in a DEFCLASS form. (reported
2304 * bug fix: RANDOM can be compiled when the compiler derives the type
2305 of its numeric argument as a disjoint set of small integers.
2306 (reported by Paul Dietz)
2307 * bug fix: starting a core saved with shared objects loaded when
2308 those objects are not available no longer causes threaded SBCL to
2309 hang. (reported by Sean Ross)
2310 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2312 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2313 argument, which must always be NIL. (reported by Kalle Niemitalo)
2314 * bug fix: printing 1.0d+23 no longer results in an error.
2315 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2316 * bug fix: structure slot setters preserve evaluation order. (thanks
2318 * fixed some bugs related to Unicode integration:
2319 ** RUN-PROGRAM can allow its child to take input from a Lisp
2320 stream. (reported by Stefan Scholl)
2321 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2322 locales. (reported by Ken Causey)
2323 * fixed some bugs revealed by Paul Dietz' test suite:
2324 ** INCF, DECF and REMF evaluate their place form as specified in
2326 ** FORMATTER expands ~{ ~} iteration directives with V or #
2327 parameters correctly.
2328 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2329 consequent uses no arguments correctly.
2330 ** the system has a partial understanding of the (COMPLEX RATIO)
2332 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2335 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2336 * new feature: a build-time option (controlled by the :SB-UNICODE
2337 keyword feature, enabled by default) for building the system with
2338 support for the entire 21-bit character space defined by the
2340 * new feature: the system now has rudimentary external-format
2341 support; the primary user-visible change at this time is that
2342 characters with the high bit set (such as lower-case-e-acute) will
2343 print correctly to a terminal in a UTF-8 environment.
2344 * minor incompatible change: BASE-CHAR no longer names a class;
2345 however, CHARACTER continues to do so, as required by ANSI.
2346 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2347 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2348 has been moved to the SB-EXT package (temporarily re-exported from
2350 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2351 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2353 * fixed bug #351: better error-handlind and reporting for malformed
2355 * fixed bug #350: bignum-printing is now more memory-efficient,
2356 allowing printing of very large bignums, eg. (expt 2 10000000).
2357 (reported by Bruno Haible)
2358 * fixed bug #302: better primitive-type selection for intersection
2360 * fixed bug #308: non-graphic characters now all have names, as
2361 required. (reported by Bruno Haible)
2362 * bug fix: redefining a class with different superclasses now correctly
2363 removes it from the direct-subclasses of its previous superclasses.
2364 (reported by David Morse)
2365 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2366 new value now works. (reported by Bruno Haible)
2367 * bug fix: correct canonicalization of multiple non-standard slot
2368 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2369 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2370 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2371 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2372 now exists, an signals an error.
2373 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2374 during macroexpansion for non-iteration variables as well. (reported
2375 by Bruno Haible for CMUCL)
2376 * bug fix: Cyclic structures and unprintable objects in compiler
2377 messages no longer cause errors. (reported by Bruno Haible)
2378 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2379 and READ-FROM-STRING all now return a primary value of NIL if
2380 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2381 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2382 (reported by Bruno Haible for CMUCL)
2383 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2384 do not designate a condition. (reported by Bruno Haible for
2386 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2387 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2388 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2389 permitting resolution in favour of any of the conflicting symbols.
2390 (reported by Bruno Haible for CMUCL)
2391 * bug fix: EQUAL compiler optimizations is less aggressive on
2392 strings which can potentially compare true despite having distinct
2393 specialized array element types.
2394 * bug fix: unit enumerations can be defined without dividing by
2395 zero. (thanks to Vincent Arkesteijn)
2396 * FORMAT compile-time argument count checking has been enhanced.
2397 (report from Bruno Haible for CMUCL)
2398 * a partial workaround for the bug 262: the compiler does not try to
2399 inline-expand a local function doing RETURN-FROM from a deleted
2400 BLOCK. (thanks to Peter Denno for the bug report and to David
2401 Wragg for the simple test case)
2402 * fixed some bugs revealed by Paul Dietz' test suite:
2403 ** RENAME-PACKAGE allows all package designators as new package
2405 ** constraint propagation and lambda variable substitution are
2406 more cautious in dealing with partially deleted code.
2407 ** compiler sometimes lost reoptimization passes.
2408 ** CERROR, given a condition as condition designator, passes its
2409 remaining arguments to the continue format control without
2411 ** Case-altering FORMAT directives work correctly on non-ASCII
2413 ** The REAL type specifier handles bounds outside the floating
2414 point ranges without signalling FLOATING-POINT-OVERFLOW.
2415 ** Functions with IR1-transformations can create intercomponent
2416 references to global functions.
2417 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2419 ** FORMAT ~R treats a nil value for its first parameter correctly.
2421 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2422 * enhancement: saving cores with foreign code loaded is now
2423 supported on x86/NetBSD and sparc/Linux in addition to the previously
2424 supported platforms.
2425 * bug fix: on some platforms repeated installations caused multiple
2426 copies of HTML documentation to be installed -- should not happen
2427 any more. (reported by Stefan Scholl)
2428 * bug fix: parsing self-recursive alien record types multiple times
2429 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2430 original patch by Helmut Eller for CMUCL)
2431 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2432 (thanks to Richard Kreuter)
2433 * bug fix: defining classes whose accessors are methods on existing
2434 generic functions in other (locked) packages no longer signals
2435 bogus package lock violations. (reported by François-René Rideau)
2436 * bug fix: special variables as DEFMETHOD parameters no longer have
2437 associated bogus type declarations. (reported by David Wragg and
2439 * bug fix: read-write consistency on streams of element-type
2440 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2441 * bug fix: redefiniton of the only method of a generic function with
2442 no DEFGENERIC no longer emits a full WARNING. In addition,
2443 redefinition of generic functions with no DEFGENERIC to an
2444 incompatible lambda list now signals an error. (thanks to Zach
2446 * bug fix: DEFGENERIC now works even when there's a function of the
2447 same name in an enclosing lexical environment. (thanks to Zach
2449 * fixed compiler failure, caused by instrumenting code during
2450 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2451 * optimization: added loop analysis and improved register allocation
2452 * fixed some bugs revealed by Paul Dietz' test suite:
2453 ** POSITION on displaced vectors with non-zero displacement
2454 returns the right answer.
2455 ** (SIMPLE-STRING) is a valid type specifier for sequence
2457 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2459 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2460 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2462 ** The justification version of the FORMAT ~< directive treats
2463 non-zero minpad parameter correctly.
2465 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2466 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2467 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2468 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2469 the supported interface.
2470 * new feature: Single-stepping of code compiled with DEBUG 2 or
2471 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2472 * new feature: saving cores with foreign code loaded is now
2473 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2474 Timothy Moore's work for CMUCL)
2475 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2476 parameters to * instead of NIL if no initform is supplied.
2477 (reported by Johan Bockgaard)
2478 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2479 correctly when FUNCALL appears as the car of the form. Note:
2480 despite this FUNCALL forms are not currently subject to
2481 compiler-macro expansion. (port of Raymond Toy's fix for the
2482 same from CMUCL, reported by Johan Bockgaard)
2483 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2484 (thanks for Teemu Kalvas)
2485 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2486 case instead of terminating the iteration (thanks for Julian
2487 Squires, Sean Champ and Raymond Toy)
2488 * bug fix: incorrect expansion of defgeneric that caused a style
2489 warning. (thanks for Zach Beane)
2490 * bug fix: slot accessor effective method computation works properly
2491 for classes with multiple non-standard applicable methods on
2492 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2493 * on x86 compiler supports stack allocation of results of LIST and
2494 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2495 CMUCL implementation by Gerd Moellmann)
2496 * fixed some bugs revealed by Paul Dietz' test suite:
2497 ** FORMAT strings with both the ~<~:;~> form of the justification
2498 directive and pretty-printing directives cause an error.
2500 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2501 * incompatible change: the internal functions
2502 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2503 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2504 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2505 instead of the old functions.
2506 * new feature: on platforms where "dladdr" is available foreign
2507 function names now appear in backtraces. (based on Helmut Eller's
2509 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2510 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2512 * documentation: ASDF manual is now distributed with SBCL.
2513 * bug fix: SBCL can now load its contributed modules with REQUIRE
2514 even if the system-provided entries have been removed from
2515 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2517 * bug fix: backtraces involving undefined functions or assembly
2518 routines are more informative. (thanks to Brian Downing)
2519 * bug fix: mutually referent alien structures now work correctly.
2520 (reported by Rick Taube)
2521 * bug fix: structures defined by WITH-ALIEN can be referred to
2522 within other definitions in the same WITH-ALIEN.
2523 * bug fix: division operators (MOD, TRUNCATE and the like) with
2524 constant zero divisors and integer dividends no longer generate
2526 * bug fix: provide default methods for INPUT-STREAM-P and
2527 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2528 * bug fix: improve the detection and resolution of MOP metacycles
2529 (where effective method computation depends on the generic function
2530 having its effective method computed). (reported by Bruno Haible)
2531 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2532 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2533 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2534 * bug fix: pathnames with relative directory components can be
2535 represented relative to default pathnames.
2536 * optimization: in taking the GCD of bignums, reduce the two bignums
2537 to approximately the same size (using Euclid's algorithm) before
2538 applying the more sophisticated binary GCD. (thanks to Juho
2540 * optimization: COUNT on bitvectors now operates word-at-a-time.
2541 * optimization: ASH with a positive, but not necessarily constant,
2542 (leftwards) shift, when in a modular context, is compiled to a
2544 * fixed some bugs revealed by Paul Dietz' test suite:
2545 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2546 if the corresponding argument is NIL.
2547 ** FORMAT directives accepting a minpad parameter treat negative
2549 ** literal commas as character parameters to FORMAT directives are
2551 ** literal spaces directly after ~<Newline> directives within a
2552 format-logical-block (~:< ~@:>) do not induce :FILL-style
2553 conditional newlines.
2554 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2555 as specified: it no longer includes conditional newlines).
2556 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2558 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2559 :PER-LINE-PREFIX argument does not evaluate to a string.
2561 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2562 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2563 the manual for details; for now, package locks can be disabled by
2564 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2565 if no major problems are found then it is likely that they will be
2566 compiled in unconditionally.
2567 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2568 unsupported operators on all platforms. To load a shared library
2569 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2570 object file, link it into a shared library outside of SBCL and
2571 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2573 * bug workaround: ROOM T and the GC were not working together
2574 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2575 reporting facility aren't preserved. That reporting has been
2576 disabled, so now until and unless someone figures out how to make it
2577 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2578 * minor incompatible change: as threatened around sbcl-0.8.0, the
2579 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2580 The symbols are also exported from SB-PCL for backwards
2581 compatibility, but more so than before SB-PCL should be treated as
2582 an implementation-internal package.
2583 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2585 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2586 It is known as of this release not to work on the Alpha, however.
2587 * fixed bug #167: errors signalled due to illegal syntax in method
2588 bodies are now more legible.
2589 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2590 designators and can hence be used with TYPEP.
2591 * fixed bug #333: CHECK-TYPE now ensures that the type error
2592 signalled, if any, has the right object to be accessed by
2593 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2594 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2595 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2597 * fixed bug #334: programmatic addition of slots using specialized
2598 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2599 and :ALLOCATION :CLASS slots.
2600 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2601 float exponents. (rereported by Peter Seibel)
2602 * fixed a bug: #\Space (and other whitespace characters) are no
2603 longer considered to be macro characters in standard syntax by
2604 GET-MACRO-CHARACTER.
2605 * fixed bug: initialization of condition class metaobjects no longer
2606 causes an instance of the condition to be created. (reported by
2608 * fixed bug: it is now possible to have slots such that
2609 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2610 description is neither :INSTANCE nor :CLASS.
2611 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2612 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2613 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2614 non-standard SLOT-DEFINITION-ALLOCATION values.
2615 * fixed bugs: various race conditions handling exiting threads.
2616 CL-PPCRE's thread torture test now passes.
2617 * fixed arguably-a-bug: GC time proportional to number of threads in
2618 system even when most of them are idle
2619 * optimization: improved performance of BIT and SBIT on bit-vectors.
2620 * .fasl file incompatibility: The fasl file version number has
2621 been incremented because of changes associated with package locks.
2623 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2624 * minor incompatible change: the system no longer provides
2625 optimization hints (conditions which are TYPEP
2626 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2627 mode; these hints are still emitted when higher SPEED optimization
2629 * new contrib module: a sampling profiler (profiling by statistical
2630 sampling, rather than by instrumenting functions) is available as
2631 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2632 * the behaviour of the standard function ED is now customizeable by
2633 third parties through a hook variable: see ED's documentation
2634 string for information on the protocol.
2635 * the compiler no longer emits efficiency notes for (FUNCALL X)
2636 when the type of X is uncertain under default optimization
2638 * fixed bug 276: mutating a binding of a specialized parameter to a
2639 method to something that is not TYPEP the specializer is now
2641 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2642 exhibit the required behaviour on the x86 platform. (thanks to
2643 Peter van Eynde, Eric Marsden and Bruno Haible)
2644 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2645 even for difficult arguments. (reported by Peter Graves)
2646 * fixed bug 141a: the backquote printer now descends quoted
2648 * fixed another bug in backquote printing: no more destructive
2649 modification of the form's list structure. (reported by Brian
2651 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2652 move between its address being taken and the call to
2653 interrupt_thread, fixing a crashing race condition.
2654 * the SB-POSIX contrib implementation has been adjusted so that it
2655 no longer exhibits ridiculously poor performance when constructing
2656 instances corresponding to C structs.
2658 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2659 * minor incompatible change: the sb-grovel contrib now treats C
2660 structures as alien (in the sense of SB-ALIEN) objects rather than
2661 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2662 has implications for memory management of client code
2663 (sb-grovel-returned objects must now be manually managed) and for
2664 type safety (alien objects now have full types).
2665 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2666 used to control emission of compiler diagnostics, rather than the
2667 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2668 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2669 quality should be considered deprecated.
2670 * install.sh now installs the user manual as well
2671 * (not quite a new documentable feature, but worth considering in
2672 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2673 beginnings of a semantically meaningful condition hierarchy is
2674 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2676 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2677 designator as the defaults argument.
2678 * fixed bug: Displaced arrays whose displaced-to array has become
2679 too small now cause ARRAY-DIMENSION to signal an error, providing
2680 for safer bounds-checking. (reported by Bruno Haible)
2681 * fixed bug: DEFCLASS slot definitions with identical :READER and
2682 :WRITER names now signal a reasonable error. (reported by Thomas
2684 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2686 * fixed bug: Invalid slot specification errors now print correctly.
2687 (thanks to Zach Beane)
2688 * fixed bug 320: Shared to local slot value transfers in class
2689 redefinitions now happen corrently from superclasses as
2690 well. (reported by Bruno Haible)
2691 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2693 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2694 improper lists correctly. (reported by Bruno Haible)
2695 * fixed bug 313: source-transform for <fun-name> was erroneously
2696 applied to a call of a value of a variable with name <fun-name>.
2697 (reported by Antonio Menezes Leitao)
2698 * fixed bug 307: The obsolete instance protocol ensures that
2699 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2700 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2701 remove dead unknown-values globs from the middle of the stack.
2702 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2703 easier to resume long computations after using *BREAK-ON-SIGNALS*
2704 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2705 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2706 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2707 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2708 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2709 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2711 * fixed a (fairly theoretical) bug in string printing: if
2712 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2713 does not have array-element-type equal to the most general string
2715 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2716 function name. (thanks to Nikodemus Siivola)
2717 * fixed bug: compile-time format string checker failed on
2718 non-closed ~{. (reported by Thomas F Burdick)
2719 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2720 don't warn on legal loop code involving more than one aggregate
2721 boolean. (thanks to Nikodemus Siivola)
2722 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2723 in keyword position to MAKE-PACKAGE was not being reported as
2725 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2726 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2728 * fixed bugs 280 and 312: the checking for multiple definitions in a
2729 file is less likely to become confused by uses of inline
2731 * fixed bug: the #S reader macro performs the keyword coercion
2732 specified for slot names. (reported by Kalle Niemitalo)
2733 * fixed bug: lambda lists may contain symbols whose names start with
2734 & but are not lambda-list-keywords; their occurrence triggers a
2736 * fixed bug 321: define-method-combination argument lambda lists do
2737 not cause invalid code to be generated when &optional or &aux
2738 variables have default values. (reported by Bruno Haible)
2739 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2740 have CLOS classes; this appears also to have fixed reported
2741 instability in stack exhaustion detection.
2742 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2743 MUFFLE-WARNING when no associated restart is present is now
2745 * optimization: rearranged the expansion of various defining macros
2746 so that each expands into only one top-level form in a
2747 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2749 * optimization: used a previously little-used slot in symbols to
2750 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2752 * fixed some bugs revealed by Paul Dietz' test suite:
2753 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2755 ** Fixed an optimization bug related to inheritance of initforms
2756 from local to shared slots.
2757 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2758 ** CAST optimizer forgot to flush argument derived type.
2759 ** print/read consistency on floats is now orders of magnitude
2760 more likely. (thanks also to Bruno Haible for a similar report
2762 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2763 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2764 for objects of type REAL. Make it so.
2765 ** SXHASH is specified (infelicitously) to respect similarity,
2766 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2767 it so. (thanks to Markus Ziegler)
2768 ** on the Alpha, the compiler succeeds in compiling functions
2769 returning a known number of arguments greater than 63.
2770 ** fixed handling of invalid NIL arguments in keyword position
2772 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2774 ** provided a readably-printable representation for RANDOM-STATE
2776 ** ensured that pathnames, where they have a namestring, always
2777 print using #P"..." syntax.
2779 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2780 * Support for the forthcoming 2.0 version of the NetBSD kernel
2781 running on x86 hardware has been added. (thanks to Perry
2782 E. Metzger most immediately, and others for their past work)
2783 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2784 current release version; SBCL's OpenBSD support had been broken
2785 since about the time of OpenBSD's switch to ELF binary format.)
2786 * [placeholder for DX summary]
2787 ** user code with &REST lists declared dynamic-extent, under high
2788 speed or space and low safety and debug optimization policy.
2789 * The manual has been converted to Texinfo format and the debugger
2790 chapter from the cmucl manual has been added.
2791 * A facility has been added to extract documentation strings from
2792 sbcl and store them as Texinfo-formatted snippets for inclusion in
2793 the manual (via Texinfo's @include directive)
2794 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2795 the test case to Dave Roberts)
2796 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2797 pointers. (reported by Sean Ross)
2798 * bug fix: PROFILE output is printed nicely even for large numerical
2799 values. (thanks to Zach Beane)
2800 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2801 greater than 32 handle EOF correctly.
2802 * bug fix: on X86 an immediate argument of the IMUL instruction is
2803 correctly printed in disassembly. (thanks to Lutz Euler)
2804 * bug fix: class slots in redefined classes preserve their old
2805 values. (thanks to Bruno Haible and Nikodemus Siivola)
2806 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2807 succeeds. (reported by Marco Baringer)
2808 * fixed some bugs revealed by Paul Dietz' test suite:
2809 ** READ-SEQUENCE now works on ECHO-STREAMs.
2810 ** RATIONALIZE works more according to its specification. (thanks
2813 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2814 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2815 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2816 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2817 should matter to you only if you rebind the printer control
2818 variables and then find you want different bindings in the
2819 debugger than in the ordinary execution of your program.)
2820 * The runtime build system has been tweaked to support building
2821 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2822 assembler and linker. (thanks to Nikodemus Siivola)
2823 * Unbound, undefined, undeclared variables now trigger full
2824 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2825 more often programmer error than deliberate exploitation of undefined
2827 * optimization: the hash algorithm for strings has changed to one
2828 that is less vulnerable to spurious collisions. (thanks to Juho
2830 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2831 needless bounds checking. (thanks to Juho Snellman)
2832 * optimization: implemented multiplication as a modular
2833 (UNSIGNED-BYTE 32) operation on the PPC backend.
2834 * fixed some bugs revealed by Paul Dietz' test suite:
2835 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2837 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2838 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2839 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2841 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2843 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2845 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2846 their output stream on EOF from read.
2847 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2848 have been read to end-of-file.
2849 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2851 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2852 description of determination of which consecutive characters
2854 ** Printing the "Space" character with escaping on now yields "#\\ ",
2855 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2856 ** Reading floating-point numbers with *READ-BASE* set to a number
2857 less than 10 works correctly.
2858 ** Reading floating-point numbers with *READ-BASE* set to a number
2859 more than 10 works correctly.
2860 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2861 the readtable currently in effect.
2863 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2864 * minor incompatible change: parsing of namestrings on a physical
2865 (Unix) host has changed; numbers after the final #\. in a
2866 namestring are no longer interpreted as a version field. This is
2867 intented to be largely invisible to the user, except that the
2868 meaning of the namestring "*.*.*" has changed: it now refers to a
2869 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2870 should usually be replaced by
2871 (make-pathname :name :wild :type :wild :version :wild)
2872 with the added benefit that this is more likely to be portable.
2873 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2874 option to OPEN now signals an error if the file being opened
2875 exists; this may have an impact on existing code.
2876 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2877 signals received, on the PowerPC platforms (both Linux and
2878 Darwin). (thanks to Pierre Mai for pointing out the location of
2880 * several fixes on OS X: The system now builds and runs cleanly on
2881 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2882 (thanks to Brian Mastenbrook)
2883 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2884 second-resolution: integer multiples of 1/3600 between -24 and 24.
2885 (thanks to Vincent Arkesteijn)
2886 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2887 type when called with 1 argument; PEEK-CHAR checked type of
2888 PEEK-TYPE only after having read first character from a
2889 stream. (reported by Peter Graves)
2890 * bug fix: the garbage collector now has much better locality
2891 behaviour, and in particular no longer treats all memory as being
2892 exhausted when many small objects point to each other in a deeply
2894 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2895 recognized as being TYPEP their class.
2896 * bug fix: the PUSHNEW documentation string has been corrected.
2897 (thanks to Vincent Arkesteijn)
2898 * bug fix: defaulting of the value for the last of an atypically
2899 large number of multiple values being bound was not being
2900 performed correctly on the Alpha or PPC platforms
2901 * optimization: implemented multiplication as a modular
2902 (UNSIGNED-BYTE 32) operation on the x86 backend.
2903 * optimization: SEARCH on simple-base-strings can now be open-coded.
2904 (see also contrib/compiler-extras.lisp for inspiration for
2905 teaching the compiler about the Boyer-Moore algorithm).
2906 * value, returned by MAX (and MIN) called with several EQUALP, but
2907 not EQL, arguments now does not depend on compiler settings.
2908 * fixed some bugs revealed by Paul Dietz' test suite:
2909 ** in stack analysis liveness information is propagated from
2910 non-local entry points.
2911 ** pathwise CAST removing failed when the CAST node did not start
2913 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2914 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2916 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2917 host is already defined.
2918 ** RENAME-FILE works on streams instead of signalling an internal
2920 ** PEEK-CHAR uses the current readtable when determining whether
2921 or not a character is whitespace.
2922 ** MERGE-PATHNAMES handles the case when the pathname does not
2923 specify a name while the default-pathname specifies a version.
2924 ** Pathnames now stand a better chance of respecting print/read
2926 ** Attempting to use standardized file system operators with a
2927 pathname with invalid :DIRECTORY components signals a
2929 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2931 ** DIRECTORY on logical pathnames is more correct.
2932 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2933 signal a TYPE-ERROR if their argument is not a stream.
2934 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2935 designator argument does not designate a stream.
2936 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2937 examining the synonym.
2938 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2940 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2941 element-type larger than ([UN]SIGNED-BYTE 32).
2943 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2944 * When built with the :SB-FUTEX feature, threaded builds now take
2945 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2946 for faster/more reliable mutex and condition variable support.
2947 * Incompatible change (but one you probably shouldn't have been using
2948 anyway): the interface and code for arbitrating between multiple
2949 threads in the same user session has been redesigned.
2950 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2951 error when called without an explicit environment argument.
2952 (thanks to Nikodemus Siivola)
2953 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2954 confused about their position. (thanks to Adam Warner and Gerd
2956 * bug fix: Pretty printing backquoted forms with unquotations in the
2957 argument list position of various code constructs such as LAMBDA
2958 now works correctly. (reported by Paul Dietz)
2959 * bug fix: Pretty printing unquotations no longer loses all
2960 stream position information.
2961 * optimization: performance of string output streams is now less
2962 poor for multiple small sequence writes.
2963 * optimization: performance of CSUBTYPEP in the presence of complex
2964 expressions involving CONS and NOT many times has been improved.
2965 (reported by Paul Dietz)
2966 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2968 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2969 streams for the REPL, for the debugger, and so on. (thanks to
2971 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2972 * fixed some bugs revealed by Paul Dietz' test suite:
2973 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2974 is ignored if it is too small, rather than propagating through
2975 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2976 ** extremely complex negations of CONS types were not being
2977 sufficiently canonicalized, leading to inconsistencies in
2979 ** VALUES tranformer lost derived type.
2981 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2982 * fixed a bootstrapping bug: the build process no longer assumes
2983 that the various BOOLE-related constants have the same value in
2984 host and target lisps. (noted by Paul Dietz' test suite on an
2985 SBCL binary built from CLISP)
2986 * The system can now be dynamically linked on the MIPS platform,
2987 which enables dynamic loading of foreign code from Lisp. (thanks
2988 to Ralf Baechle for discussions on the MIPS ABI)
2989 * The system now records debugging information for its own source
2990 files in a filesystem-position-independent manner, relative to
2991 the "SYS" logical host.
2992 * fixed a compiler bug: MV-LET convertion did not check references
2993 to the "max args" entry point. (reported by Brian Downing)
2994 * tweaked disassembly notes to be less confident about proclaiming
2995 some instruction as an LRA. (thanks to Brian Downing)
2996 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2997 now each have their own history, command character, and other
2998 characteristics. (thanks to David Lichteblau)
2999 * fixed some bugs revealed by Paul Dietz' test suite:
3000 ** compiler failure in compiling LOGAND expressions including a
3002 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3003 shift greater than 32.
3004 ** FLUSH-DEST did not mark blocks for type check regeneration.
3005 ** HANDLER-CASE failed to accept declarations in handler clauses
3006 in some circumstances.
3008 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3009 * New code in contrib/sb-introspect (still probably not entirely
3010 stable yet) provides some support for smart Lisp development
3011 environments like SLIME.
3012 * The conditions signalled for errors occurring when loading .fasl
3013 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3014 in a way which should help ASDF recover gracefully.
3015 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3016 stuff has been cleaned up. If you code contrib/ stuff, this might
3017 affect you, and you can look at contrib/README, contrib/STANDARDS,
3018 and/or the 0.8.4.27 diff to check.
3019 * In full calls the compiler now does not generate checks for declared
3020 argument types for all arguments.
3021 * various threading fixes
3022 ** and some experimental patches which didn't make it into
3023 the main tree for this release, but which are shipped in
3024 contrib/experimental-thread.patch as a possible fix for some
3025 failures (deadlock, spinning...) in GC-intensive multithreaded
3027 * fixed PPC build problem (source code incompatibility of different
3028 library versions): added offsetof() hackery which attempts to divine
3029 where glibc maintainers put uc_mcontext today
3030 * fixed bug 282: compiler does not trust type assertions while passing
3031 arguments to a full call.
3032 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3033 &OPTIONAL VALUES type parameter.
3034 * fix bug 214: algorithm for noting rejected templates is now more
3035 similar to that of template seletion. (also reported by rydis on
3037 * fixed bug 141b: printing backquoted information readably and prettily
3038 inserts a space where necessary.
3039 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3040 (CHARACTER 10) are now reported as errors, rather than propagated
3041 as unknown types. (reported by piso on #lisp)
3042 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3043 with values NIL and :ERROR. (thanks to Milan Zamazal)
3044 * fixed bug 191c: CLOS now does proper keyword argument checking as
3045 described in CLHS 7.6.5 and 7.6.5.1.
3046 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3047 longer raise an error; further, using a list as a for-as-arithmetic
3048 counter now raises a meaningful error.
3049 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3050 now understood by sequence creation functions such as MAKE-SEQUENCE
3052 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3053 a STRING-INPUT-STREAM.
3054 * compiler enhancement: SIGNUM is now better able to derive the type
3056 * type declarations inside WITH-SLOTS are checked. (reported by
3058 * fixed some bugs revealed by Paul Dietz' test suite:
3059 ** incorrect optimization of TRUNCATE for a positive first
3060 argument and negative second.
3061 ** compiler failure in let-convertion during flushing dead code.
3062 ** compiler failure while deriving type of TRUNCATE on an
3063 interval, containing 0.
3064 ** ASH of a negative bignum by a negative bignum count now returns
3066 ** intersection of CONS types now canonicalizes properly, fixing
3067 inconsistencies in SUBTYPEP.
3069 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3070 * incompatible change: The --disable-debugger command line
3071 option now clobbers the debugger at a more fundamental
3072 level, by redefining #'INVOKE-DEBUGGER instead of by
3073 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3074 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3075 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3076 enter the debugger REPL and then suffer recursive errors
3077 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3078 while under the new system, BREAK in --disable-debugger mode
3079 terminates the system just as an unhandled error would.
3080 * fixed compiler performance when processing loops with a step >1;
3081 * bug fix: DOCUMENTATION now retrieves generic function
3082 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3083 support has been systematized, and now supports the methods
3084 specified by ANSI, along with a default method and a method for
3085 slot documentation. (reported by Nathan Froyd)
3086 * bug fix: effective methods associated with a generic function are
3087 no longer cached over a change of that generic function's method
3088 combination. (reported by Andreas Fuchs)
3089 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3090 names a non-existent directory. (thanks to Andreas Fuchs)
3091 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3092 the PPC platform, lead to stack corruption; this has been fixed.
3093 (reported by Rainer Joswig)
3094 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3095 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3096 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3097 arithmetic in general, is now much more reliable on the PPC
3099 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3100 platform now returns the right answer.
3101 * optimization: restored some effective method precomputation in
3102 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3103 precomputation is now tunable.
3104 * optimization: compiler-internal data structure use has been
3105 reviewed, and changes have been made that should improve the
3106 performance of the compiler by about 20%.
3107 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3108 simple-base-strings and simple-bit-vectors is improved.
3109 * optimization: the optimization of 32-bit logical and arithmetic
3110 functions introduced in version 0.8.3 on the x86 has been
3111 implemented on the mips, ppc and sparc platforms; an
3112 implementation of the same facility, but for 64-bit arithmetic,
3113 has been added for the alpha.
3114 * microoptimization: the compiler is better able to make use of the
3115 x86 LEA instruction for multiplication by constants.
3116 * bug fix: in some situations compiler did not report usage of
3117 generic arithmetic in (SPEED 3) policy.
3118 * bug 145b fix: compiler used wrong type specifier while converting
3119 MEMBER-types to numeric.
3120 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3122 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3123 index argument now works properly on the Alpha platform.
3124 * bug fix: floating point exception treatment on the Alpha platform
3126 * bug fix: FILE-POSITION works much better on string input and
3127 output streams. (thanks to Nikodemus Siivola)
3128 * bug fix: many threading/garbage collection symptoms sorted.
3129 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3130 might be pseudo-atomic.
3131 * internal change: Stopping for GC is now done with signals not
3132 ptrace. GC is now done in whichever thread wanted it, instead of
3134 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3136 * bug fix: result form in DO is not contained in the implicit
3138 * incompatible change: ICR structure is changed; the value part of
3139 CONTINUATION is now called LVAR; corresponding functions are
3140 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3141 * added type deriver for ISQRT (thanks to Robert E. Brown).
3142 * bug fix: better support for loading from the command line when an
3143 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3145 * fixed some bugs revealed by Paul Dietz' test suite:
3146 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3147 ** ROUND and FROUND now give the right answer when given very
3148 small float arguments.
3149 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3151 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3152 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3153 ** LCM should return a non-negative integer.
3154 ** PARSE-INTEGER returned the index of a terminator instead of the
3155 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3156 ** PARSE-INTEGER returned an incorrect index being applied to a
3158 ** LCM with two arguments of 0 returns 0 rather than signalling
3160 ** unsigned addition of a 32-bit constant with the high bit set no
3161 longer causes an internal compiler error on the x86.
3162 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3163 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3164 with negative last argument.
3165 ** byte specifiers with zero size and position no longer cause
3166 an error during type derivation.
3167 ** bignum multiplication on the Alpha platform now returns the
3169 * porting: The system now builds on SuSE AMD64, although it still
3170 generates a 32-bit binary.
3171 * .fasl file incompatibility: The fasl file version number has
3172 been incremented (because of the changes to internal compiler
3173 data structures referred to above).
3175 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3176 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3177 more accurately, on the Darwin kernel running on PowerPC hardware.
3178 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3179 * Compiler code deletion notes now signal a condition of type
3180 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3181 an associated MUFFLE-WARNING restart.
3182 * The compiler now performs limited argument count validation of
3183 constant format strings in FORMAT, and where appropriate in ERROR,
3184 CERROR and WARN. (thanks to Gerd Moellmann)
3185 * New ASDF-INSTALL contrib can be used for automatic download and
3186 installation of third-party Lisp code from CCLAN or other sites
3188 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3189 which forces another thread to execute a function supplied by the
3191 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3192 now accept and act upon their :ELEMENT-TYPE keyword argument.
3193 (reported by Martin Atzmueller, Edi Weitz)
3194 * bug fix: FILE-POSITION now accepts position designators up to
3195 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3196 is the greater. (thanks to Patrik Nordebo)
3197 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3198 Kalle Olavi Niemitalo)
3199 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3200 printable. (reported by Eric Marsden)
3201 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3202 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3203 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3204 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3205 (thanks to Stig Sandoe)
3206 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3207 UNREAD-CHAR now correctly includes the unread character in the
3208 target sequence. (thanks to Gerd Moellmann)
3209 * bug fix in threaded builds: the system can now be suspended and
3210 resumed by shell job control with minimal disruption.
3211 * bug fixes in times and timezones >2038AD
3212 * better handling of "where is GNU make?" problem in build scripts
3213 (thanks to Nikodemus Siivola)
3214 * new optimization: inside a named function any reference to a
3215 function with the same name is considered to be a self-reference;
3216 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3217 optimization quality.
3218 * new optimization on x86: logical functions and + now have
3219 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3220 used when the result is truncated to 32 bits.
3221 * VALUES declaration is partially enabled.
3222 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3223 * fixed some bugs revealed by Paul Dietz' test suite:
3224 ** The system now obeys the constraint imposed by
3225 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3226 types form a lattice under type intersection.
3227 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3228 ** ASSOC now ignores NIL elements in an alist.
3229 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3230 and (1+ MOST-POSITIVE-FIXNUM) answers.
3231 ** The addition of a method with invalid qualifiers to a generic
3232 function does not cause an error to be signalled immediately;
3233 a warning is signalled, and the error is generated only on
3234 calling the generic function.
3235 * changed .fasl file version number, in part to add type codes for
3236 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3237 obscure ANSI requirements
3239 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3240 * fixed bug 148: failure to inline-expand a local function left
3241 garbage, confusing the compiler.
3242 * fixed bugs 3cd: structure slot readers perform type check if the
3243 slot can have an invalid value (i.e. it is either not initialized
3244 or can be written with a less specific slot writer).
3245 * bug fix: the compiler now traps array references to elements off
3246 the end of an array; previously, the bounds checking in some
3247 circumstances could go off-by-one.
3248 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3250 * type declarations for array element types now obey the description
3251 on the CLHS page "Declaration TYPE", as per discussions on
3252 sbcl-help around 2003-05-08. This means that a declaration
3253 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3254 declaration, all references to BAR will be asserted or assumed
3255 (with THE, so dependent on compiler policy) to involve objects of
3256 type FOO. Note that no such declaration is implied in
3257 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3258 * declared types of functions from the "Conditions"
3259 chapter. (reported by Paul Dietz)
3260 * bug fix: CERROR accepts a function as its first argument.
3261 * bug fix: NTH an NTHCDR accept a bignum as index
3262 arguments. (reported by Adam Warner)
3263 * optimization: character compare routines now optimize comparing
3264 against a constant character. (reported by Gilbert Baumann)
3265 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3266 argument now works properly on the MIPS platform.
3267 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3268 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3269 declared to have a functional type, does not cause an error
3271 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3272 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3273 * fixed bug: if last continuation of a deleted block has a
3274 destination, this destination should be deleted too. (reported by
3276 * fixed a bug in the bootstrap process: the host compiler's values
3277 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3278 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3279 test case from Patrik Nordebo)
3280 * improved the ability of the disassembler on the PPC platform to
3281 provide helpful disassembly notes.
3282 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3283 the class in more cases than previously.
3284 * bug fix: FILE-POSITION now understands :START and :END for
3285 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3286 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3287 8) data. (thanks to David Lichteblau)
3288 * bug fix: it is possible to add a method to a generic function
3289 without lambda list.
3290 * bug fix: reader failed to signal END-OF-FILE inside an
3291 object representation. (reported by Nikodemus Siivola)
3292 * fixed some bugs revealed by Paul Dietz' test suite:
3293 ** LAST and [N]BUTLAST should accept a bignum.
3294 ** condition slot accessors are methods.
3295 ** (VECTOR NIL) is a subtype of STRING.
3297 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3298 * minor incompatible change: some nonsensical specialized lambda
3299 lists (used in DEFMETHOD) which were previously ignored now signal
3301 * minor incompatible change: the system is now aware of the types of
3302 variables in the COMMON-LISP package, and will signal errors for
3303 most violations of these type constraints (where previously they
3304 were silently accepted).
3305 * minor incompatible change: COMPILE-FILE now uses the freedom
3306 afforded (ANSI 3.2.2.3) to use derived function types for
3307 functions defined in the same file. This also permits the system
3308 to warn on static type mismatches and function
3309 redefinition. (Currently it does not work with high DEBUG level.)
3310 * minor incompatible change: VALUES declaration is disabled.
3311 * When issuing notes, the compiler now signals a condition of type
3312 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3313 restart for use in user handlers. It is expected that the
3314 COMPILER-NOTE condition will eventually become a condition
3315 supertype to a hierarchy of note types, which will then be
3316 handleable in a similar fashion. However, at the moment, no such
3317 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3318 but it's an implementation detail, not a classification for the
3320 * Changes in type checking closed the following bugs:
3321 ** type checking of unused values (192b, 194d, 203);
3322 ** template selection based on unsafe type assertions (192c, 236);
3323 ** type checking in branches (194bc).
3324 * A short form of VALUES type specifier has ANSI meaning (it has
3325 increased the number of situations when SBCL cannot perform type
3327 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3328 keywords or constants is permissible.
3329 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3330 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3331 classes. (thanks to Antonio Martinez)
3332 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3333 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3334 argument to a pretty-print stream. Instead, it leaves any such
3335 operations to DESCRIBE-OBJECT methods.
3336 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3338 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3339 lambda lists. (thanks to David Lichteblau)
3340 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3341 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3342 (thanks to Gerd Moellmann)
3343 * fixed bug 246: increased compilation speed of long
3344 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3346 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3348 * DEFCONSTANT now throws a condition of type
3349 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3350 constant to a non-EQL value; CONTINUE and ABORT restarts
3351 respectively change and preserve the value.
3352 * fixed bug 63: The code walker, part of the implementation of CLOS,
3353 is now better at handling symbol macros.
3354 * bug fix: There is no longer an internal implementation type named
3355 CL:LENGTH. (reported by Raymond Toy)
3356 * bug fix: In macro-like defining macros/special operators the
3357 implicit block does not enclose the lambda list.
3358 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3359 enclosed in parentheses) are not suitable as type specifiers, and
3360 their use properly signals an error now.
3361 * bug fix: An argument count mismatch for a type specifier in code
3362 being compiled no longer causes an unhandled error at compile
3363 time, but signals a compile-time warning.
3364 * fixed simple vector readable printing
3365 * bug fix: DESCRIBE takes more care over whether the class
3366 precedence list slot of a class is bound before accessing it.
3367 (reported by Markus Krummenacker)
3368 * bug fix: FORMATTER can successfully compile pretty-printer format
3369 strings which use variants of the ~* directive inside.
3370 * bug fix: SEARCH now applies its TEST predicate to the elements of
3371 the arguments in the correct order. (thanks to Wolfhard Buss)
3372 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3373 in type inference. (thanks to Robert E. Brown)
3374 * bug fix: Reading in symbols with an explicit package name of ""
3375 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3376 * fixed some bugs revealed by Paul Dietz' test suite:
3377 ** NIL is now allowed as a structure slot name.
3378 ** Arbitrary numbers, not just REALs, are allowed in certain
3379 circumstances in LOOP for-as-arithmetic clauses.
3380 ** Multiple class redefinitions before slot access no longer
3381 causes a type error.
3382 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3383 association between the name and a class.
3384 ** Generic functions with non-standard method-combination and over
3385 five methods all of which return constants no longer return NIL
3386 after the first few invocations. (thanks to Gerd Moellmann)
3387 ** CALL-NEXT-METHOD with no arguments now passes the original
3388 values of the arguments, even in the presence of assignment.
3389 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3391 ** DISASSEMBLE works with closures and funcallable instances.
3392 ** ADD-METHOD now returns the generic function, not the new method.
3393 ** FIND-METHOD signals an error if the lengths of the specializers
3394 is incompatible with the generic function, even if the ERRORP
3396 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3397 which its argument is a member.
3398 ** DEFCLASS only redefines the class named by its class-name
3399 argument if that name is the proper name of the class;
3400 otherwise, it creates a new class.
3401 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3402 of the UNBOUND-SLOT condition to the name of the slot.
3403 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3404 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3405 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3406 SLOT-MAKUNBOUND in the specified fashion.
3408 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3409 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3410 cross-compilation host. As a consequence, we can now bootstrap our
3411 way up to SBCL starting with a bare gcc toolchain and human-readable
3412 source code (first the source to CLISP, then the source to SBCL).
3413 * A contributed module containing a partial implementation of the
3414 simple-streams interface has been included. (thanks to Rudi
3416 * A contributed module implementing the RFC1321 Message Digest
3417 Algorithm, known as MD5, has been included.
3418 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3419 no longer has any effect, as the code controlled by this feature
3420 has been deleted. (As far as we know, no-one has ever built using
3421 this feature, and its semantics were confused in any case).
3422 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3423 work on conditions (as required by the ANSI specification),
3424 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3425 expected behaviour on conditions. Users should note, however,
3426 that such behaviour is not required by the ANSI specification,
3427 and so use of this behaviour may render their code unportable.
3428 * fixed some bugs revealed by Paul Dietz' test suite:
3429 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3431 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3432 ** SXHASH on condition objects no longer returns NIL.
3433 ** :ALLOCATION :CLASS slots are better treated; their values are
3434 updated on class redefinition, and initforms inherited from
3435 superclasses are applied.
3436 ** REMOVE-METHOD returns its generic function argument even when
3437 no method was removed.
3438 ** SHARED-INITIALIZE now initializes the values of the requested
3439 slots, including those with :ALLOCATION :CLASS.
3440 ** ALLOCATE-INSTANCE now works on structure classes defined via
3441 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3443 ** SLOT-EXISTS-P now works on conditions, as well as structures
3445 ** MAKE-LOAD-FORM now has the required methods on
3446 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3447 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3448 keyword, and now implements the SLOT-NAMES argument.
3449 ** methods with &OPTIONAL arguments no longer allow too many
3450 arguments to be passed in the call without error.
3451 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3452 option is consistent with the required arguments of the generic
3453 function lambda list.
3454 * bug fix: REQUIRE accepts a string designator. (Thanks to
3456 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3457 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3458 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3459 * bug fix: The compiler now checks for duplicated variables in macro
3461 * bug fix: SETQ on globals returns the correct value.
3462 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3463 not a COMPILER-ERROR (followed by some other strange error on
3464 choosing the CONTINUE restart).
3465 * bug fix: make.sh and friends are now more consistent in the way that
3466 they look for GNU "make".
3468 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3469 * experimental native threads support (on x86 Linux >=2.4 only).
3470 This is not compiled in by default: you need to add :SB-THREAD to
3471 the target features. See the "Beyond ANSI" chapter of the manual
3473 * fix for longstanding nonANSIism: The old distinction between
3474 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3475 The return value from CL:FIND-CLASS is now a CLOS class, and
3476 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3477 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3478 * An interface to the MetaObject Protocol, as described in Kiczales,
3479 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3480 MIT Press, 1991, is available from the SB-MOP package.
3481 * incompatible change: the SB-PCL package should now be considered
3482 a private implementation detail, and no longer a semi-private MOP
3484 * minor incompatible change: due to rearrangement for threads, the
3485 control stack and binding stack are now allocated at arbitrary
3486 addresses instead of being hardcoded per-port. Users affected by
3487 this probably have to be doing advanced things with shared
3488 libraries, and will know who they are.
3489 * minor incompatible change: Previously, all --eval forms used were
3490 processed with READ before any of them were processed with EVAL.
3491 Now each --eval form is processed with both READ and EVAL before
3492 the next --eval form is processed. (Thus package operations like
3493 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3494 work as the user might reasonably expect.)
3495 * minor incompatible change: *STANDARD-INPUT* is now only an
3496 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3498 * minor incompatible change: Y-OR-N-P is now character-oriented,
3499 not line oriented. Also, YES-OR-NO-P now works without errors.
3500 (thanks to Antonio Martinez)
3501 * sb-aclrepl module improvements: an integrated inspector, added
3502 repl features, and a bug fix to :trace command.
3503 * Known functions, which cannot be open coded by the backend, are
3504 considered to be able to check types of their arguments. (fixing
3505 a bug report by Nathan J. Froyd)
3506 * fixed a bug in computing method discriminating functions: It is
3507 now possible to define methods specialized on classes which have
3508 forward-referenced superclasses. (thanks to Gerd Moellmann)
3509 * fixed evaluation order in optional entries (reported by Gilbert
3511 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3512 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3513 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3514 required-by-AMOP NAME argument, as well as CLASS and
3515 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3516 * fixed bug 20: DEFMETHOD can define methods using names that are
3517 not the proper names of classes to designate class specializers.
3518 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3519 Unix file descriptors, instead of blowing up. (thanks to Antonio
3521 * Garbage collection refactoring: user-visible change is that a
3522 call to the GC function during WITHOUT-GCING will not do garbage
3523 collection until the end of the WITHOUT-GCING. If you were doing
3524 this you were probably losing anyway.
3525 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3526 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3527 TYPEP the latter but not the former.
3528 * The compiler issues a full WARNING for calls to undefined functions
3529 with names from the CL package.
3530 * MAP-INTO for a vector destination is open coded. (reported by
3531 Brian Downing on c.l.l)
3532 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3533 documentation string.
3534 * fixed some bugs revealed by Paul Dietz' test suite:
3535 ** COPY-ALIST now signals an error if its argument is a dotted
3537 ** Condition slots are now accessed more correctly in the presence
3538 of multiple initargs for a given slot.
3539 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3540 exclude restarts of the same name associated with a different
3542 ** DEFCLASS of forward-referenced classes with another
3543 forward-referenced class in the superclasses list no longer
3545 ** Condition slots are now initialized once each, not multiple
3546 times. (thanks to Gerd Moellmann)
3547 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3549 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3550 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3551 arguments contain duplicated elements.
3552 ** RESTART-CASE understands local macros.
3553 ** RESTART-CASE associates exactly its own restarts with a condition.
3554 ** ENDP in safe mode checks its argument to be of type LIST.
3555 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3556 in question is unbound.
3557 ** Optimized MAKE-INSTANCE functions no longer cause internal
3558 assertion failures in the presence of duplicate initargs.
3559 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3560 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3561 propagated even in the ctor optimized implementation of
3563 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3565 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3566 * a better implementation of SXHASH on (simple) bit vectors,
3567 measured both in execution speed and in distribution of results
3568 over the positive fixnums, has been installed. Likewise, a better
3569 implementation of EQUAL for simple bit vectors is now available.
3570 * fixed CEILING optimization for a divisor of form 2^k.
3571 * fixed bug 240 (emitting extra style warnings "using the lexical
3572 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3573 by Antonio Martinez)
3574 * fixed SXHASH, giving different results for NIL depending on type
3575 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3576 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3577 variable. (found by Rolf Wester)
3578 * SBCL does not ignore type declarations for special
3579 variables. (reported by rif on c.l.l 2003-03-05)
3580 * some bug fixes in contrib/sb-aclrepl/
3581 * fixed some bugs revealed by Paul Dietz' test suite:
3582 ** a bug in the CONS type specifier, whereby the CAR and CDR
3583 types got intertwined, has been fixed;
3584 ** the type system is now able to reason about the interaction
3585 between INTEGER and RATIO types more completely;
3586 ** APPEND, [N]REVERSE and NRECONC check that those their
3587 arguments, which must be proper lists, are really so;
3588 ** An array specialized to be unable to hold elements has been
3589 implemented, as required -- yes, really -- by ANSI;
3590 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3591 on malformed property lists;
3593 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3594 * incompatible packaging change: in line with Unix convention,
3595 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3596 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3597 * REQUIRE and PROVIDE are now optionally capable of doing something
3598 useful. See the documentation string for REQUIRE.
3599 * infrastructure for a managed SBCL contrib system: contributed
3600 modules in this release include:
3601 ** the ASDF system definition facility;
3602 ** an interface to the BSD Sockets API;
3603 ** an ACL-like convenience interface to the repl;
3604 (thanks to Kevin Rosenberg)
3605 ** an implementation of ROTATE-BYTE, with efficient implementation
3607 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3608 gives a closer approximation to the right answer than previously.
3609 (thanks to Raymond Toy)
3610 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3611 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3612 optional environment arguments, as required by ANSI.
3613 * fixed bugs in other functions taking environment objects, allowing
3614 calls with an explicit NIL environment argument to be compiled
3616 * fixed bug 228: primary return values from
3617 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3618 COMPILE or FUNCTION.
3619 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3620 :TYPE will no longer signal errors on innocuous objects.
3621 * fixed bug 231b: SETQ is better at respecting type declarations in
3622 the lexical environment.
3623 * fixed a bug in DEFCLASS: classes named by symbols with no or
3624 unprintable packages can now be defined.
3625 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3626 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3627 * fixed compiler failure related to checking types of functions.
3628 (reported by Robert E. Brown)
3629 * the compiler is now much more consistent in its error-checking
3630 treatment of bounding index arguments to sequence functions: in
3631 (SAFETY 3) code, errors will be signalled in almost all cases if
3632 invalid sequence bounding indices are passed to functions defined
3633 by ANSI to operate on sequences.
3634 * fixed a bug in the build procedure: documentation of SBCL-specific
3635 packages is now preserved and available in the final Lisp image.
3636 * lifted FDEFINITION lookup out of loops in the implementation of
3637 many list operations. (thanks to Robert E. Brown)
3638 * fixed a bug in the reader: the #n# reader macro now works for
3639 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3640 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3641 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3642 E. Brown and Tony Martinez respectively)
3643 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3644 SLOT-BOUNDP: in particular, it is now able to optimize them much
3645 better, and is now not vulnerable to having packages renamed.
3646 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3647 always called when a slot is not present in an instance. (thanks
3649 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3650 longer signal an error on unreading a character following EOF on
3651 the previous constituent stream. (thanks to Tony Martinez)
3652 * fixed some bugs revealed by Paul Dietz' test suite:
3653 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3654 not just nonnegative fixnums;
3655 ** the logical bit-array operators such as BIT-AND now accept an
3656 explicit NIL for their "opt-arg" argument (to indicate a
3657 freshly-consed result bit-array);
3658 ** ELT now signals an error on an invalid sequence index in safe
3660 ** the type system is now cleverer about negations of numeric
3661 types, and consequently understands the BIGNUM and RATIO types
3663 ** the type system is now cleverer about the interaction between
3664 INTEGER and RATIO types: while bugs still remain, many more
3665 cases are accurately computed;
3666 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3667 if it is in the last clause;
3668 ** CONSTANTLY now correctly returns a side-effect-free function in
3670 ** DECLARE is no longer treated as a special-operator; in
3671 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3672 * incremented fasl file version number due to the change in the
3673 DEFSTRUCT-SLOT-DESCRIPTION structure.
3675 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3676 * minor incompatible change: code processed by the "interpreter" or
3677 EVAL now has a compilation optimization policy of (DEBUG 2)
3678 (changed from (DEBUG 1)) to improve debuggability of interactive
3679 development, and to allow the use of the debug RETURN command in
3681 * an experimental implementation of the RETURN command for the
3682 debugger has been included. (thanks to Frederik Kuivinen)
3683 * fixed bug 62: constraints were not propagated into a loop.
3684 * fixed bug in embedded calls of SORT (reported and investigated by
3686 * fixed some bugs revealed by Paul Dietz' test suite:
3687 ** printing and reading of arrays with some dimensions having
3688 length 0 (thanks to Gerd Moellmann);
3689 ** BOA constructor with &AUX argument without a default value does
3690 not cause a type error;
3691 ** CONSTANTP now returns true for all self-evaluating objects.
3693 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3694 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3695 accessors that are related by inheritance, as specified in the
3696 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3697 Valtteri Vuorikoski)
3698 * The compiler is now able to inline functions that were defined in
3699 a complex lexical environment (e.g. inside a MACROLET).
3700 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3702 * Support for the upcoming FreeBSD-5.0 release has been included.
3703 (thanks to Dag-Erling Smorgrav)
3704 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3705 effect when it is not in a toplevel context.
3706 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3707 stand a better chance of being correct. (thanks to Gerd
3709 * fixed bug in COERCE, which now signals an error on coercing a
3710 rational to a bounded real type which excludes the expected
3712 * The compiler is now able to derive types more accurately from the
3713 COERCE and COMPILE functions.
3714 * fixed bug 223: functional binding is considered to be constant
3715 only for symbols in the CL package.
3716 * fixed bug 231: SETQ did not check the type of a variable being set
3717 (reported by Robert E. Brown)
3718 * A new optimization for MAKE-INSTANCE has been included, fixing
3719 various bugs (including relating to :ALLOCATION :CLASS slots and
3720 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3722 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3723 ** As required by ANSI, LOOP now disallows anonymous collection
3724 clauses such as COLLECT I in conjunction with aggregate boolean
3725 clauses such as THEREIS (= I 1);
3726 ** LOOP now signals an error when any variable is reused in the
3727 same loop (including the potentially useful construct analogous
3728 to WITH A = 1 WITH A = (1+ A);
3729 ** IT is only a special loop symbol within the first clause of a
3730 conditional loop clause;
3731 ** LOOP with a typed iteration variable over a hashtable now
3732 signals a type error iff it should.
3733 * fixed some other bugs revealed by Paul Dietz' test suite:
3734 ** FILE-STREAM now names the class previously known as FD-STREAM;
3735 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3736 argument) no longer signals an error;
3737 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3738 of the slot symbol, rather than using the current package
3739 ((:CONC-NAME "") continues to intern the slot's name in the
3741 * incremented fasl file version number, because of the incompatible
3742 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3743 of the new implementation of DEFINE-COMPILER-MACRO.
3745 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3746 * Support for building SBCL for MIPS platforms running in
3747 little-endian mode has now been checked in, and basic
3748 functionality on said platforms verified.
3749 * minor incompatible change: PCL now records the pathname of a file
3750 in which methods and the like are defined, rather than its
3752 * minor incompatible change: TRUENAME now considers the truename of
3753 a file naming a directory to be the pathname with :DIRECTORY
3754 component indicating that directory.
3755 * minor incompatible change: a NAMED clause in the extended form of
3756 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3757 reason for the previous behaviour is unclear.
3758 * more systematization and improvement of CLOS and MOP conformance
3759 in PCL (thanks to Gerd Moellman and Pierre Mai):
3760 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3762 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3763 primary methods with no specializers;
3764 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3766 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3767 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3768 FORWARD-REFERENCED-CLASSes; error reporting on
3769 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3771 ** SXHASH on CLOS instances now uses a slot internal to the
3772 instance to return different numbers on distinct instances,
3773 while preserving the same return value through invocations of
3775 ** DEFMETHOD signals errors when methods with longer incongruent
3776 lambda lists are added to generic functions;
3777 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3778 CLASS, as specified in AMOP;
3779 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3780 on the order returned by the primary method for classes of
3781 class STANDARD-CLASS;
3782 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3783 * fixed some bugs shown by Paul Dietz' test suite:
3784 ** DOLIST puts its body in TAGBODY;
3785 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3787 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3788 value producing form;
3789 ** if more variables are given to PROGV than values, extra
3790 variables are bound and made to have no value;
3791 ** NSUBSTITUTE on list arguments gets the right answer with
3793 ** ELT signals an error of type TYPE-ERROR when the index argument
3794 is not a valid sequence index;
3795 ** LOOP signals (at macroexpansion time) an error of type
3796 PROGRAM-ERROR when duplicate variable names are found;
3797 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3798 ** FUNCALL of special-operators now cause an error of type
3800 ** PSETQ now works as required in the presence of side-effecting
3801 symbol-macro places;
3802 ** NCONC accepts any object as its last argument;
3803 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3805 ** loop-for-as-package does not require a package to be explicitely
3807 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3809 * fixed bug 166: compiler preserves "there is a way to go"
3810 invariant when deleting code.
3811 * fixed bug 172: macro lambda lists with required arguments after
3812 &REST arguments now cause an error to be signalled. (thanks to
3814 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3815 bug. (thanks to Matthew Danish)
3816 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3818 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3819 arguments are lexically rebound. (thanks to Gerd Moellmann and
3821 * fixed bug 194: error messages are now more informative when there
3822 is no primary method applicable in a call to a generic
3823 function. (thanks to Gerd Moellmann)
3824 * fixed bug in command line argument checking (thanks to Julian
3826 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3827 (thanks to Matthew Danish)
3828 * incremented fasl file version number, because of the
3829 SXHASH-related changes in the layout of CLOS data structures
3831 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3832 * minor incompatible change: The runtime (the Unix executable named
3833 "sbcl") is now much pickier about the .core files it will load.
3834 Essentially it now requires .core files to descend from the same
3835 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3836 as the runtime does. (The intent is to prevent the crashes which
3837 can occur, and which can even be reported as mysterious failures,
3838 when people patch the sources or change the build parameters
3839 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3840 sbcl and .core files.)
3841 * fixed bug: VALUES-LIST is no longer optimized away.
3842 * fixed bug 142: The FFI conversion of C string values to Lisp
3843 string values no longer conses excessively. (thanks to Nathan
3844 Froyd porting Raymond Toy's fix to CMU CL)
3845 * began to systematize and improve MOP conformance in PCL (thanks to
3846 Nathan Froyd, Gerd Moellman and Pierre Mai):
3847 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3849 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3850 ** FINALIZE-INHERITANCE is now called on class finalization;
3851 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3852 argument precedence order.
3853 * fixed bug 202: The compiler no longer fails on functions whose
3854 derived types contradict their declared type.
3855 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3856 so it can be non-toplevel.
3857 * The fasl file version number has changed (because of the new
3858 implementation of DEFMACRO).
3859 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3860 safe code, that any length requirement by their type-specifier
3861 argument is valid. The exceptions to this are described in bug
3863 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3864 functions now check the types of their inputs as required by ANSI.
3865 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3866 introduced symbol is DECLAREd to be SPECIAL.
3867 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3868 * fixed bug: :COUNT argument to sequence functions may be negative
3869 * fixed bug: body of DO-SYMBOLS may contain declarations
3870 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3871 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3872 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3873 symbol macro only once
3874 * fixed printing of call frame when argument list is unavailable
3875 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3876 * compiler no longer signals WARNING on unknown keyword
3879 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3880 * A beta-quality port to the mips architecture running Linux,
3881 based on the old CMUCL backend, has been made. It has been tested
3882 on a big-endian kernel, and works sufficiently well to be able to
3883 rebuild itself; it has not been tested in little-endian mode.
3884 * fixed an inconsistency between gencgc.c and purify.c which made
3885 dumping/loading .core files unreliable
3886 * fixed bug 120a: The compiler now deals correctly with IFs where
3887 the consequent is the same as the alternative, instead of
3888 misderiving the return type. (thanks to Alexey Dejneka)
3889 * fixed bug 113: Logical pathnames are now dumpable (the logical
3890 host is resolved at load-time, throwing an error if it is not
3892 * fixed bug 174: FORMAT's error message is slightly clearer when a
3893 non-printing character is used in a format directive.
3894 * fixed several bugs in compiler checking of type declarations, i.e.
3895 violations of the Python "declarations are assertions" principle
3896 (thanks to Alexey Dejneka)
3897 * fixed several bugs in PCL's error checking (thanks to Gerd
3899 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3901 * fixed bug in compilation of functions as first class values
3902 (thanks to Antonio Martinez-Shotton)
3903 * The compiler's handling TYPE-ERRORs which it can prove will
3904 inevitably happen at runtime has been cleaned up and corrected
3905 in several ways. (thanks to Alexey Dejneka)
3906 * improved argument type checking for various basic arithmetic
3907 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3908 lavished on them in the past that they can be compiled in many
3909 ways in different special cases
3910 * fixed bug 181: compiler checks validity of user supplied type
3912 * cleaned up code flushing in optimization: Function calls which
3913 should signal errors for safety purposes (e.g. which ANSI says
3914 should signal errors when their arguments are of incorrect type)
3915 are no longer optimized away.
3916 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3917 * incremented fasl file version number, because changes in the
3918 implementation of sequence functions like COERCE caused
3919 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3920 in old inline expansions) to become undefined. (Actually these
3921 changes were later undone, so we might very well be binary
3922 compatible with 0.7.7 after all, but leaving the version number
3923 incremented seemed like the simplest and most conservative
3926 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3927 * An alpha-quality port to the parisc architecture running Linux,
3928 based on the old CMUCL backend, has been made. This, even more so
3929 than the other backends, should be considered still a work in
3930 progress; known problems include that the Linux kernel in 64-bit
3931 mode does not propagate the correct sigcontext structure to
3932 userspace, and consequently SBCL on a parisc64 kernel will not
3934 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3935 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3936 suppressed.) Also now that the compiler is looking at declarations
3937 in the environment, it checks optimization declarations as well,
3938 and suppresses inlining when (> DEBUG SPEED).
3939 * More fixes have been made to treatment of floating point exception
3940 treatment and other Unix signals. In particular, floating point
3941 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3942 * The detection and handling of control stack exhaustion (infinite
3943 or very deeply nested recursion) has changed. Stack exhaustion
3944 detection is now done by write-protecting pages at the OS level
3945 and applies at all optimization settings; when found, a
3946 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3947 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3948 be caught using IGNORE-ERRORS.
3949 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3950 correctly when transforming recursive LABELS functions to LETs.
3951 (thanks to Alexey Dejneka)
3952 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3953 that are names of constants or global variables.
3954 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3955 alien routines with docstrings.
3956 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3957 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3959 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3960 (thanks to Eric Marsden)
3961 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3962 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3963 * Bug fix: The SPARC backend can now compile functions involving
3964 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3965 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3966 to a functions expecting a PATHNAME-DESIGNATOR.
3967 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3968 lambda lists. (thanks to Alexey Dejneka)
3969 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3970 functions so that EXPORTing the name of the class doesn't cause
3971 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3972 to Antonio Martinez for reporting this)
3973 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3974 constant-folded without failing an assertion. (thanks to Einar
3975 Floystad Dorum for reporting this)
3976 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3977 types are now dealt with more correctly by the compiler.
3978 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3979 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3980 behaviour analogous to RENAME-FILE. This puts its behaviour more
3981 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3983 * The fasl file version number has changed again. (because of the
3984 bug fix involving the names of PCL MAKE-INSTANCE functions)
3986 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3987 * bug fix: Floating point exceptions are treated much more
3988 consistently on the x86/Linux and PPC/Linux platforms.
3989 * Array initialization with :INITIAL-ELEMENT is now much faster for
3990 cases when the compiler cannot open code the array creation, but
3991 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3992 array accesses have also seen a speed increase.
3993 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3994 specified. (This was at the root of some bad interactions between
3995 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3997 * bug fix: Internal error arguments for undefined functions are now
3998 computed correctly on the PPC/Linux platform.
3999 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
4000 Raymond Toy's patch for CMU CL)
4001 * Support for the Solaris 9 operating environment has been included
4002 (thanks to Daniel Merritt)
4003 * A very ugly but hopefully complete draft of the missing FFI chapter
4004 of the manual has been created by reformatting the corresponding
4005 CMU CL manual chapter into (currently very ugly and incoherent)
4006 DocBook and bringing it up to date for SBCL behavior. Thus, the
4007 manual is now essentially complete, at least by my extreme
4008 once-and-only-once standards, whereby it's acceptable to refer to
4009 the doc strings of SB-EXT functions as the primary documentation.
4010 * The fasl file version number has changed again, due to cleanup of
4011 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4012 is no longer a static symbol.)
4014 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4015 * SBCL now builds with OpenMCL (version 0.12) as the
4016 cross-compilation host; also, more progress has been made toward
4017 bootstrapping under CLISP.
4018 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4020 * bug 158 fixed: The compiler can now deal with integer loop
4021 increments different from 1; fixing this turned out also to fix
4023 * bug 169 fixed: no more bogus warnings about using lexical bindings
4024 despite the presence of perfectly good SPECIAL declarations (thanks
4025 to David Lichteblau)
4026 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4027 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4028 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4030 * bug 180 fixed: Method combination specifications no longer ignore
4031 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4032 * bug fix: Structure type predicate functions now check their argument
4033 count as they should.
4034 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4035 correctly. (thanks to Pierre Mai)
4036 * minor incompatible change: The --noprogrammer option is deprecated
4037 in favor of the new --disable-debugger option, which is very similar.
4038 (The major difference is that it takes effect at a slightly different
4039 time at startup, causing handling of errors in --sysinit and
4040 --userinit files will be affected differently.) The
4041 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4042 been added to allow this functionality to be controlled from ordinary
4043 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4044 maintainers, who might want to run non-interactive scripts to
4045 build SBCL cores which will later be used interactively.)
4046 * minor incompatible change: The LOAD function no longer, when given
4047 a wild pathname to load, loads all files matching that pathname.
4048 Instead, an error of type FILE-ERROR is signalled.
4050 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4051 * bug 147 fixed: The compiler preserves its block link/count
4052 invariants more correctly now so that it doesn't crash. (thanks
4054 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4056 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4057 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4058 no-value-for-this-character as NIL (as specified by ANSI).
4059 * HOST-NAMESTRING on physical pathnames now returns a string that is
4060 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4061 (thanks to Christophe Rhodes)
4062 * The Alpha port handles icache flushing more correctly. (thanks to
4064 * More progress has been made toward bootstrapping under CLISP. (thanks
4065 to Christophe Rhodes)
4066 * The fasl file format has changed again, because dynamic loading
4067 on OpenBSD (which has non-ELF object files) motivated some cleanups
4068 in the way that foreign symbols are transformed and passed around.
4069 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4070 is no longer treated as whitespace by the reader, but instead as
4071 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4072 |A
\7fB|, instead of A as it used to.
4074 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4075 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4076 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4077 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4078 this as of 0.7.1.45, but was left out of the previous news section
4079 (thanks to Dan Barlow)
4080 * SBCL now runs on the Solaris operating system on SPARC architectures
4081 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4082 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4083 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4084 * SPARC backend cleanups, allowing builds of cores optimized for V8
4085 and V9 SPARCS, and also emission of code targeted to a particular
4086 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4088 * SBCL is closer to bootstrapping under CLISP, thanks to various
4089 fixes by Christophe Rhodes.
4090 * The fasl file format has changed again, to allow the compiler's
4091 INFO database to support symbol macros.
4092 * The user manual (in doc/) is formatted into HTML more nicely.
4093 (thanks to coreythomas)
4094 * The system is smarter about SUBTYPEP relationships, especially
4095 those involving NOT types (including types such as ATOM which are
4096 represented internally using NOT types). Thus SUBTYPEP is less
4097 likely to return (VALUES NIL NIL) in general, and in particular
4098 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4100 * The fasl file format has changed again, because the internal
4101 representation of types now includes a new slot to support the new
4102 SUBTYPEP-of-NOT-types logic.
4103 * (not a change in the main branch of SBCL, but a related prototype
4104 which can hopefully be merged into the main branch of SBCL in the
4105 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4106 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4107 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4108 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4109 dispatch macro character combinations. (thanks to Alexey Dejneka)
4110 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4111 that superficially look like logical namestrings correctly.
4112 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4114 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4115 default supplied-p) for &optional and &key arguments. (thanks to
4118 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4119 * incompatible change: The compiler is now less aggressive about
4120 tail call optimization, doing it only when (> SPACE DEBUG) or
4121 (> SPEED DEBUG). (This is an incompatible change because there are
4122 programs which relied on the old CMU-CL-style behavior to optimize
4123 away their unbounded recursion which will now die of stack overflow.)
4124 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4125 for non-GENCGC systems has been increased to 20M (since that
4126 seems much closer to the likely performance optimum for modern
4127 systems than the old 4M value was)
4128 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4129 and *DEBUG-PRINT-LEVEL*
4130 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4131 of CMU CL's support for SPARC, and various endianness and other
4132 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4133 * new syntactic sugar for the Unix command line: --load foo.bar is now
4134 an alternate notation for --eval '(load "foo.bar")'.
4136 ** The system now detects stack overflow and handles it gracefully,
4137 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4138 optimization settings. (This is a good thing in general, and
4139 its introduction in this version should be particularly timely
4140 for anyone whose code fails because of suppression of tail
4142 ** The system now hunts for the C variable "environ" in a more
4143 devious way, to avoid segfaults when the C library version differs
4144 between compile time and run time. (thanks to Christophe Rhodes)
4145 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4146 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4147 ** The compiler no longer issues bogus style warnings for undefined
4148 classes in the same source file as the DEFCLASSes which defined
4149 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4151 ** fixes in CONDITION class precedence list for undefined function
4152 errors (thanks to Alexei Dejneka)
4153 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4154 correctly. (thanks to Dan Barlow)
4155 ** portability fixes aiming at bootstrapping under CLISP (thanks
4156 to Dave McDonald and Christophe Rhodes)
4157 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4158 ** fixes in type translation and and type inference (thanks to
4160 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4161 ** various fixes in the new ports (thanks to Dan Barlow)
4162 * several changes related to debugging:
4163 ** suppression of tail recursion, as noted above
4164 ** stack overflow detection, as noted above
4165 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4166 is now the default. (For some time encapsulation has been more
4167 reliable than the breakpoint-based :ENCAPSULATE NIL
4168 implementation, at least on X86 systems; and I just noticed that
4169 encapsulation also seems closer to the spirit of the ANSI
4172 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4174 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4175 up properly again. (There was a packaging bug in 0.7.0 which
4176 left their definitions in SB-SYS::LOAD-FOREIGN and
4177 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4178 vital for most things which interface to C-level interfaces,
4179 like extensions working with sockets or databases or
4180 Perl-compatible regexes or whatever, and the need to fix
4181 this bug is the main reason that 0.7.1 was released so
4183 ** DEFGENERIC is now choosier about the methods it redefines, so that
4184 reLOADing a previously-LOADed file containing DEFGENERICs does
4185 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4186 cycle now works as it should. (thanks to Alexey Dejneka)
4187 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4188 fixed by Christophe Rhodes. (He actually submitted this patch
4189 months ago, and I delayed until after 0.7.0.)
4190 ** Bug 111 (internal compiler confusion about runtime checks on
4191 FUNCTION types) was fixed by Alexey Dejneka.
4192 * Some internal cleanups (getting rid of variables which aren't
4193 needed now that the byte interpreter is gone) caused the fasl
4194 file format number to change again.
4196 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4197 * major incompatible change: The default fasl file extension, i.e. the
4198 default extension for files produced by COMPILE-FILE, has changed
4199 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4201 ** There are many changes in the implementation of the compiler.
4202 SBCL is now essentially a compiler-only implementation of ANSI
4203 Common Lisp. EVAL still "interprets" a few special cases, but
4204 almost all the interesting cases are handled by creating
4205 a LAMBDA expression, calling COMPILE on it, then calling
4206 FUNCALL on the result.
4207 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4208 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4209 Since the code is newer, there might still be some new bugs
4210 (though not as many as before Martin Atzmueller's fixes:-). But
4211 the new code is substantially simpler and clearer, and hopefully
4212 any remaining bugs will be simpler, less fundamental, and more
4213 fixable then the bugs in the old code.
4214 ** The revised compiler is still a little unsteady on its feet.
4216 *** The debugging information it produces (particularly the names
4217 of FUNCTION objects) is sometimes much less useful than what
4218 the old compiler produced.
4219 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4220 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4221 within a MACROLET) has been temporarily weakened.
4222 ** There are new compiler optimizations for various functions:
4223 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4224 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4225 *** the math functions TRUNCATE, FLOOR, and CEILING
4226 *** the function-of-all-trades COERCE
4227 Mostly these should be transparent, but there's one
4228 potentially-annoying problem (bug 117): when the compiler
4229 inline-expands a function and does type analysis on the result,
4230 it can create control paths which have type mismatches, and
4231 when it can't prove that those control paths aren't taken,
4232 it will issue WARNINGs about the type mismatches. This is
4233 a particular problem in practice for the new sequence functions.
4234 It's not clear how this should be fixed, and for now, a
4235 workaround is given in the entry for 117 in the BUGS file.
4236 ** (Because of the interaction between the two previous items --
4237 occasional inlining problems and new inline expansions -- some
4238 of the new sequence function optimizations won't really kick in
4239 completely until debugging information, and then inlining, are
4240 straightened out in some future version.)
4241 * minor incompatible changes:
4242 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4243 DIRECTORY no longer implicitly promotes NIL slots of its
4244 pathname argument to :WILD. In particular, when you ask for the
4245 contents of a directory (which you used to be able to do without
4246 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4247 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4248 ** changes in behavior that ANSI explicitly defines to be
4249 implementation dependent:
4250 *** The new compiler-only implementation still conforms with ANSI,
4251 but acts a little different than before. Besides the obvious
4252 changes in performance tradeoffs (that the cost per form passed
4253 to EVAL has gone up, and the cost per form executed by EVAL
4254 has gone down), the behavior of the system changes a little
4255 because there are no longer any interpreted function objects.
4256 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4257 e.g. doing COMPILE on the output of interactive DEFUN is
4259 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4261 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4262 has changed from (:CL) to NIL.
4263 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4264 e.g. #\Newline and #\Tab, don't have names specified in the
4265 ANSI Common Lisp standard, is now based on their ASCII symbolic
4266 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4267 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4268 are no longer used for output.
4269 ** changes in internal implementation constants:
4270 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4271 4 million. (If your application spends a lot of time GCing and
4272 you have a lot of RAM, you might want to experiment with
4273 increasing it even more.)
4274 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4275 However, almost all old code should still continue to work without
4276 immediate update, as SB-C-CALL is now a (deprecated) nickname
4278 ** Old operator names in the style DEF-FOO are now deprecated in
4279 favor of new corresponding names DEFINE-FOO, for consistency with
4280 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4281 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4282 internal symbols, but a few supported extensions like
4283 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4284 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4285 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4286 etc. as you get deeper into recursive calls to the debugger
4287 command loop, instead of the old "5]", "5]]", "5]]]"
4288 sequence. (I was motivated to do this when squabbles between
4289 ILISP and SBCL left me very deeply nested in the debugger. In the
4290 short term, this change will probably provoke more ILISP/SBCL
4291 squabbles, but hopefully it will be an improvement in the long run.)
4292 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4293 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4294 function, since declaiming return types involving aliens is
4295 (1) annoyingly messy to do by hand and (2) vital to efficient
4296 compilation of code which calls such functions.
4297 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4298 longer reexported by the SB-EXT package. They're solely useful
4299 for alien code, so it seems more logical that you should get
4300 them from the SB-ALIEN package, not in SB-EXT.
4301 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4302 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4303 features. Instead, the code that they used to control is always
4304 built into the system.
4305 * many other bug fixes
4306 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4307 advantage of the new EVAL-WHEN stuff and to clean them up in
4308 general, and they are now more ANSI-compliant in a number of
4309 ways. Martin Atzmueller is responsible for a lot of this.
4310 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4312 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4313 *** correct ERROR type for various file operations
4314 *** some fixes for Lisp streams
4315 *** DEFMETHOD syntax checking
4316 *** changing old weird representation of debug information as
4317 strings (which, among their other deficiencies, don't transform
4318 correctly when you rename packages, and don't change their
4319 print representation when you change things like *PACKAGE*
4320 and *PRINT-LENGTH*) to symbols and lists of symbols
4321 He also made several improvements and fixed several bugs in DESCRIBE.
4322 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4324 *** misbehavior of WRITE-STRING/WRITE-LINE
4325 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4326 and several other LOOP problems as well
4327 *** DIRECTORY when similar filenames are present
4328 *** DEFGENERIC with :METHOD options
4329 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4330 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4331 *** argument ordering in FIND with :TEST option
4332 *** mishandled package designator argument in APROPOS-LIST
4333 *** various problems in the backquote readmacro
4334 *** a bug in APROPOS
4335 *** probably some others that I'm not describing very well here,
4336 since the CVS log documents them by reference to sbcl-devel
4337 messages, and the SourceForge archives aren't working well.:-(
4338 ** Dan Barlow improved the Alpha port (and is making progress on the
4339 PPC port, for those of you who think different).
4340 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4341 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4342 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4343 continues to work on the SPARC port (for those of you in a position
4344 to look down upon our little PC-compatible boxes from a great height).
4345 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4346 and uses that copy, rather than the current dynamic value, when
4347 it's trying to decide whether to truncate output. Thus e.g.
4348 (let ((*print-lines* 50))
4349 (pprint-logical-block (stream nil)
4351 (let ((*print-lines* 8))
4352 (print (aref possiblybigthings i) stream)))))
4353 should now truncate the logical block only at 50 lines, instead of
4354 often truncating it at 8 lines, as it did before.
4355 * The doc/cmucl/ directory, containing old CMU CL documentation
4356 from the time of the fork, is no longer part of the base system.
4357 SourceForge has shut down its anonymous FTP service, and with it
4358 my original plan for distributing the old CMU CL documentation
4359 there. For now, if you need these files you can download an old
4360 SBCL source release and extract them from it.
4361 * The fasl file version number changed again, for dozens of reasons,
4362 some of which are apparent above.
4364 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4365 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4366 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4367 patch, so that DISASSEMBLE output is much nicer.
4368 * The code in the SB-PROFILE package now seems reasonably stable.
4369 I still haven't decided what the final interface should look like
4370 (I'd like PROFILE to interact cleanly with TRACE, since both
4371 facilities use function encapsulation) but if you have a need
4372 for profiling now, you can probably use it successfully with
4373 the current CMU-CL-style interface.
4374 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4375 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4376 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4377 link now returns the dangling link itself, and for similar
4378 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4379 link itself. (In these cases the old code signalled an error and
4380 looped endlessly, respectively.) Thus, DIRECTORY now works even
4381 in the presence of dangling and cyclic symbolic links.
4382 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4383 is now a supported extension again, since the consensus on
4384 sbcl-devel was that it can be useful for ordinary development
4385 work, not just for debugging SBCL itself.
4386 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4387 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4388 that currently-defined functions might be redefined later with
4389 different return types.
4390 * Hash tables can be printed readably, as inspired by CMU CL code
4391 of Eric Marsden and SBCL code of Martin Atzmueller.
4392 * better error handling in CLOS method combination, thanks to
4393 Martin Atzmueller porting Pierre Mai's CMU CL patches
4394 * more overflow fixes for >16Mbyte I/O buffers
4395 * A bug in READ has been fixed, so that now a single Ctrl-D
4396 character suffices to cause end-of-file on character streams.
4397 In particular, now you only need one Ctrl-D at the command
4398 line (not two) to exit SBCL.
4399 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4401 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4402 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4403 does the right thing.
4404 * The implementation of some type tests, especially for CONDITION
4405 types, is now tidier and maybe faster, due to CMU CL code
4406 originally by Douglas Crosher, ported by Martin Atzmueller.
4407 * Some math functions have been fixed, and there are new
4408 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4409 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4410 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4411 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4412 likely to be built into sbcl-0.7.0.)
4413 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4414 CL, ported to SBCL by Martin Atzmueller)
4415 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4416 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4417 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4418 * There's a new slam.sh hack to shorten the edit/compile/debug
4419 cycle for low-level changes to SBCL itself, and a new
4420 :SB-AFTER-XC-CORE target feature to control the generation of
4421 the after-xc.core file needed by slam.sh.
4422 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4423 COMPILE-FILE is no longer supported, so that now every function
4424 gets an entry point, so that block compilation looks a little
4425 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4426 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4427 returns the number of bytes consed since the system started,
4428 rather than the number consed since the first time the function
4429 was called. (The new definition parallels ANSI functions like
4430 CL:GET-INTERNAL-RUN-TIME.)
4431 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4432 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4433 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4434 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4435 * Fasl file version numbers are now independent of the target CPU,
4436 since historically most system changes which required version
4437 number changes have affected all CPUs equally. Similarly,
4438 the byte fasl file version is now equal to the ordinary
4441 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4442 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4443 is no longer recognized. I apologize for this, because it was
4444 listed in SB-EXT as a supported extension, but I found that
4445 its existing behavior was poorly specified, as well as incorrectly
4446 specified, and it looked like too much of a mess to straighten it
4447 out. I have enough on my hands trying to get ANSI stuff to work..
4448 * many patches ported from CMU CL by Martin Atzmueller, with
4449 half a dozen bug fixes in pretty-printing and the debugger, and
4450 half a dozen others elsewhere
4451 * fixed bug 13: Floating point infinities are now supported again.
4452 They might still be a little bit flaky, but thanks to bug reports
4453 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4454 as flaky as they were.
4455 * The --noprogrammer command line option is now supported. (Its
4456 behavior is slightly different in detail from what the old man
4457 page claimed it would do, but it's still appropriate under the
4458 same circumstances that the man page talks about.)
4459 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4460 are now supported, and enabled by default. Thus, the compiler can
4461 handle many floating point and complex operations much less
4462 inefficiently. (Thus e.g. you can implement a complex FFT
4464 * The compiler now detects type mismatches between DECLAIM FTYPE
4465 and DEFUN better, and implements CHECK-TYPE more correctly, and
4466 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4467 thanks to patches from Martin Atzmueller.
4468 * various fixes to make the cross-compiler more portable to
4469 ANSI-conforming-but-different cross-compilation hosts (notably
4470 Lispworks for Windows, following bug reports from Arthur Lemmens)
4471 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4472 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4473 fixed thanks to Pierre Mai's CMU CL patches.
4474 * fixes to keep the system from overflowing internal counters when
4475 it tries to use i/o buffers larger than 16M bytes
4476 * fixed bug 45a: Various internal functions required to support
4477 complex special functions have been merged from CMU CL sources.
4478 (When I was first setting up SBCL, I misunderstood a compile-time
4479 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4480 * improved support for type intersection and union, fixing bug 12
4481 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4482 more obscure bugs as well
4483 * some steps toward byte-compiling non-performance-critical
4484 parts of the system, courtesy of patches from Martin Atzmueller
4485 * Christophe Rhodes has made some debian packages of sbcl at
4486 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4487 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4488 stable, but are nonetheless usable. When he's ready, I'd be happy
4489 to add them to the SourceForge "File Releases" section. (And if
4490 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4491 * new fasl file format version number (because of changes in
4492 internal representation of (OR ..) types to accommodate the new
4493 support for (AND ..) types, among other things)
4495 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4496 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4498 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4500 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4501 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4502 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4503 * DESCRIBE now works on structure objects again.
4504 * Most function call argument type mismatches are now handled as
4505 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4506 know whether the function will be redefined before the call is
4507 executed. (The compiler could flag local calls with full WARNINGs,
4508 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4509 it doesn't keep track of enough information to know whether calls
4510 are local in this sense.)
4511 * Compiler output is now more verbose, with messages truncated
4512 later than before. (There should be some supported way for users
4513 to override the default verbosity, but I haven't decided how to
4514 provide it yet, so this behavior is still controlled by the internal
4515 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4516 src/compiler/ir1util.lisp.)
4517 * Fasl file format version numbers have increased again, because
4518 support for the Gray streams extension changes the layout of the
4519 system's STREAM objects.
4520 * The Gray subclassable streams extension now works, thanks to a
4521 patch from Martin Atzmueller.
4522 * The full LOAD-FOREIGN extension (not just the primitive
4523 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4524 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4525 but like most other programs, it defaults to copying the Unix
4526 environment from the original process instead of starting the
4527 new process in an empty environment.
4528 * Extensions which manipulate the Unix environment now support
4529 an :ENVIRONMENT keyword option which doesn't smash case or
4530 do other bad things. The CMU-CL-style :ENV option is retained
4531 for porting convenience.
4532 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4533 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4535 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4537 * A patch from Martin Atzmueller seems to have solved the SIGINT
4538 problem, and as far as we know, signal-handling now works cleanly.
4539 (If you find any new bugs, please report them!)
4540 * The system no longer defaults Lisp source file names to types
4541 ".l", ".cl", or ".lsp", but only to ".lisp".
4542 * The compiler no longer uses special default file extensions for
4543 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4544 seems to expect a single default extension for all compiled code,
4545 and there's no compelling reason to try to stretch the standard
4546 to allow two different extensions.) Instead, byte-compiled files
4547 default to the same extension as native-compiled files.
4548 * Fasl file format version numbers have increased again, because
4549 a rearrangement of internal implementation packages made some
4550 dumped symbols in old fasl files unreadable in new cores.
4551 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4552 many fewer weird special cases.
4553 * Bug #17 (differing COMPILE-FILE behavior between logical and
4554 physical pathnames) has been fixed, and some related misbehavior too,
4555 thanks to a patch from Martin Atzmueller.
4556 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4557 by Tim Moore, ported to SBCL by Martin Atzmueller.
4558 * Martin Atzmueller fixed several filesystem-related problems,
4559 including bug #36, in part by porting CMU CL patches, which were
4560 written in part by Paul Werkowski.
4561 * More compiler warnings in src/runtime/ are gone, thanks to
4562 more patches from Martin Atzmueller.
4563 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4566 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4568 * DESCRIBE now works on CONDITION objects.
4569 * The debugger now handles errors which arise when trying to print
4570 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4572 * The build system now uses an additional file, customize-target-features.lisp,
4573 to allow local modifications to the target *FEATURES* list. (The point of
4574 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4575 debugging features enabled, without having to worry about propagating it
4576 into everyone's system when I do a "cvs update".) When no
4577 customize-target-features.lisp file exists, the target *FEATURES* list
4578 should be constructed the same way as before.
4579 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4580 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4581 warn about behavior which is undefined under the ANSI spec. Note: This
4582 is specified by ANSI, but it's not very popular with programmers.
4583 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4584 macro in the SBCL sources for an example of a workaround which you
4585 might use to make portable ANSI-standard code which does what you want.
4586 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4587 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4588 when it's used as a non-toplevel form. (This is required in order
4589 to implement the DEFCONSTANT-EQX macro.)
4590 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4591 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4592 funny things, due to bugs in the implementation of EVAL-WHEN
4593 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4594 non-toplevel DEFCONSTANTs are uncommon.)
4595 * The core file version number and fasl file version number have been
4596 incremented, because the old noncompliant DEFCONSTANT behavior involved
4597 calling functions which no longer exist, and because I also took the
4598 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4599 * fixed bug 1 (error handling before read-eval-print loop starts), and
4600 redid debugger restarts and related debugger commands somewhat while
4602 ** The QUIT debugger command is gone, since it did something
4603 rather different than the SB-EXT:QUIT command, and since it never
4604 worked properly outside the main toplevel read/eval/print loop.
4605 Invoking the new TOPLEVEL restart provides the same functionality.
4606 ** The GO debugger command is also gone, since you can just invoke
4607 the CONTINUE restart directly instead.
4608 ** The TOP debugger command is also gone, since it's redundant with the
4609 FRAME 0 command, and since it interfered with abbreviations for the
4611 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4613 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4614 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4615 some porting work by Martin Atzmueller)
4616 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4617 with a boatload of patches which he ported to SBCL. Now that those
4619 ** The system tries to make sure that its low-priority messages
4620 are prefixed by semicolons, to help people who like to use
4621 syntax highlighting in their ILISP buffer. (This patch
4622 was originally due to Raymond Toy.)
4623 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4624 patches originally written by Raymond Toy.
4625 ** The compiler understands coercion between single-value and
4626 multiple-VALUES type expressions better, getting rid of some very
4627 weird behavior, thanks to patches originally by Robert MacLachlan
4628 and Douglas Crosher.
4629 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4630 lambda lists, thanks to a patch originally by Pierre Mai.
4631 ** The system no longer bogusly warns about "abbreviated type
4633 ** The compiler gets less confused by inlining and RETURN-FROM,
4634 thanks to some patches originally by Tim Moore.
4635 ** The system no longer hangs when dumping circular lists to fasl
4636 files, thanks to a patch originally from Douglas Crosher.
4637 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4638 undefined function error.
4639 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4640 in structure slot initforms) for now, documented workaround instead:-|
4641 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4642 * fixed bug 5: added stubs for various Gray stream functions called
4643 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4644 installed, at least appropriate type errors are generated
4645 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4646 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4647 consistently in DEFMETHOD forms.
4648 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4649 it doesn't seem to affect SBCL after all
4650 * The C runtime system now builds with better optimization and many
4651 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4653 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4655 * The system is now under CVS at SourceForge (instead of the
4656 CVS repository on my home machine).
4657 * The new signal handling code has been tweaked to treat register
4658 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4659 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4660 downstream, e.g. in debug-int.lisp, has implicit dependencies
4661 on the unsignedness of integer representation of machine words,
4662 and that caused the system to bomb out with infinite regress
4663 when trying to recover from type errors involving signed values,
4664 e.g. (BUTLAST '(1 2 3) -1).)
4665 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4666 (This was one of the bugs Peter Van Eynde reported back in July.)
4667 * The system now uses code inspired by Colin Walters' O(N)
4668 implementation of MAP (from the cmucl-imp@cons.org mailing
4669 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4670 inline the MAP operation, and there is more than one
4671 sequence argument to the MAP call (so that it can't just
4672 do ETYPECASE once and for all based on the type of the
4673 single sequence argument). (The old non-inline implementation
4674 of the general M-argument sequence-of-length-N case required
4675 O(M*N*N) time when any of the sequence arguments were LISTs.)
4676 * The QUIT :UNIX-CODE keyword argument has been renamed to
4677 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4678 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4680 * Raymond Wiker's patches to port dynamic loading from Linux to
4681 FreeBSD have been added.
4682 * The BUGS file is now more nearly up to date, thanks in large part
4683 to Martin Atzmueller's review of it.
4684 * The debugger now flushes standard output streams before it begins
4685 its output ("debugger invoked" and so forth).
4686 * The core version number and fasl file version number have both
4687 been incremented, because of incompatible changes in the layout
4689 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4690 used to not be called for a saved Lisp image.)
4691 * Martin Atzmueller's version of a patch to fix a compiler crash,
4692 as posted on sbcl-devel 13 September 2000, has been installed.
4693 * Instead of installing Martin Atzmueller's patch for the
4694 compiler transform for SUBSEQ, I deleted the compiler transform,
4695 and transforms for some similar consing operations.
4696 * A bug in signal handling which kept TRACE from working on OpenBSD
4698 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4699 away by constant folding
4700 * The system now defines its address space constants in one place
4701 (in the Lisp sources), and propagates them automatically elsewhere
4702 (through GENESIS and the sbcl.h file). Therefore, patching the
4703 address map is less unnecessarily tedious and error-prone. The
4704 Lisp names of address space constants have also been systematized.
4705 * CVS tags like dollar-Header-dollar have been removed from
4706 the sources, because they have never saved me trouble and
4707 they've been source of trouble working with patches and other
4708 diff-related operations.
4709 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4710 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4712 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4714 * The system has been ported to OpenBSD.
4715 * The system now compiles with a simple "sh make.sh" on the systems
4716 that it's supported on. I.e., now you no longer need to tweak
4717 text in the target-features.lisp-expr and symlinks in src/runtime/
4718 by hand, the make.sh takes care of it for you.
4719 * The system is no longer so grossly inefficient when compiling code
4720 involving vectors implemented as general (not simple) vectors (VECTOR T),
4721 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4722 which dares to use the various sequence functions on non-simple
4723 vectors, takes less of a performance hit.
4724 * There is now a primitive type predicate VECTOR-T-P
4725 to test for the (VECTOR T) type, so that e.g.
4726 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4727 can now be compiled with some semblance of efficiency. (The old code
4728 turned the type declaration into a full call to %TYPEP at runtime!)
4729 * AREF on (VECTOR T) is still not fast, since it's still compiled
4730 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4731 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4732 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4733 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4734 without making full calls to SUBTYPEP at runtime.
4735 (Some analogous efficiency issues for non-simple vectors specialized to
4736 element types other than T, or for non-simple multidimensional arrays,
4737 have not been addressed. They could almost certainly be handled the
4738 same way if anyone is motivated to do so.)
4739 * The changes in array handling break binary compatibility, so
4740 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4741 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4742 instead of (VALUES T T).
4743 * By following the instructions that Dan Barlow posted to sbcl-devel
4744 on 2 July 2000, I was able to enable primitive dynamic object
4745 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4746 functionality is not implemented (since it calls ld to resolve
4747 library references automatically, requiring RUN-PROGRAM for its
4748 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4749 try to resolve library references) is now supported.
4750 * The system now flushes the standard output streams when it terminates,
4751 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4752 them at several other probably-convenient times, e.g. in each pass of
4753 the toplevel read-eval-print loop, and after evaluating a form given
4754 as an "--eval" command-line option. (These changes were motivated by a
4755 discussion of stream flushing issues on cmucl-imp in August 2000.)
4756 * The source transform for TYPEP of array types no longer assumes
4757 that an array whose element type is a not-yet-defined type
4758 is implemented as an array of T, but instead punts, so that the
4759 type will be interpreted at runtime.
4760 * There is now some support for cross-compiling in make.sh: each of
4761 the phases of make.sh has its own script. (This should be transparent
4762 to people doing ordinary, non-cross-compile builds.)
4763 * Since my laptop doesn't have hundreds of megabytes of memory like
4764 my desktop machine, I became more motivated to do some items on
4765 my to-do list in order to reduce the size of the system a little:
4766 ** Arrange for various needed-only-at-cold-init things to be
4767 uninterned after cold init. To support this, those things have
4768 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4769 symbols with such names are now uninterned after cold init).
4770 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4771 fasl files for cold load.
4772 ** Remove the old compiler structure pooling code (which used to
4773 be conditional on the target feature :SB-ALLOC) completely.
4774 ** Redo the representation of some data in cold init to be more compact.
4775 (I also looked into supporting byte compiled code at bootstrap time,
4776 which would probably reduce the size of the system a lot, but that
4777 looked too complicated, so I punted for now.)
4778 * The maximum signal nesting depth in the src/runtime/ support code has
4779 been reduced from 4096 to 256. (I don't know any reason for the very
4780 large old value. If the new smaller value turns out to break something,
4781 I'll probably just bump it back up.)
4782 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4784 * Many, many bugs reported by Peter Van Eynde have been added to
4785 the BUGS list; some have even been fixed.
4786 * While enabling dynamic object file loading, I tried to make the
4787 code easier to understand, renaming various functions and variables
4788 with less ambiguous names, and changing some function calling
4789 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4790 * While trying to figure out how to do the OpenBSD port, I tried to
4791 clean up some of the code in src/runtime/. In particular, I dropped
4792 support for non-POSIX signal handling, added various comments,
4793 tweaked the code to reduce the number of compilation warnings, and
4794 renamed some files to increase consistency.
4795 * To support the new automatic configuration functionality in make.sh,
4796 the source file target-features.lisp-expr has been replaced with the
4797 source file base-target-features.lisp-expr and the machine-generated
4798 file local-target-features.lisp-expr.
4799 * fixed a stupid quoting error in make.sh so that using CMU CL
4800 "lisp -batch" as cross-compilation host works again
4802 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4804 * DESCRIBE no longer tries to call itself recursively to describe
4805 bound/fbound values, so that it no longer fails on symbols which are
4806 bound to themselves (like keywords, T, and NIL).
4807 * DESCRIBE now works on generic functions.
4808 * The printer now prints less-screwed-up representations of closures
4809 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4810 * A private symbol is used instead of the :EMPTY keyword previously
4811 used to mark empty slots in hash tables. Thus
4812 (DEFVAR *HT* (MAKE-HASH-TABLE))
4813 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4814 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4815 now does what ANSI says that it should. (You can still get
4816 similar noncompliant behavior if bang on the hash table
4817 implementation with all the symbols you get back from
4818 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4819 This breaks binary compatibility, since tests for equality to
4820 :EMPTY are wired into things like the macroexpansion of
4821 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4823 * There's now a minimal placeholder implementation for CL:STEP,
4824 as required by ANSI.
4825 * An obscure bug in the interaction of the normal compiler, the byte
4826 compiler, inlining, and structure predicates has been patched
4827 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4828 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4830 * Missing ordinary arguments in a macro call are now detected even
4831 when the macro lambda list contains &KEY or &REST.
4832 * The debugger no longer complains about encountering the top of the
4833 stack when you type "FRAME 0" to explicitly instruct it to go to
4834 the top of the stack. And it now prints the frame you request even
4835 if it's the current frame (instead of saying "You are here.").
4836 * As specified by ANSI, the system now always prints keywords
4837 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4839 * The default initial SIZE of HASH-TABLEs is now smaller.
4840 * Type information from CLOS class dispatch is now propagated
4841 into DEFMETHOD bodies, so that e.g.
4842 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4844 is now basically equivalent to
4845 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4846 (DECLARE (TYPE SINGLE-FLOAT X))
4848 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4849 operation, without having to do run-time type dispatch.
4850 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4851 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4852 * Since I don't seem to be making big file reorganizations very often
4853 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4854 is overflowing my ability to conveniently back them up), I've finally
4855 checked the system into CVS. (The CVS repository is on my home system,
4856 not at SourceForge -- putting it on SourceForge might come later.)
4857 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4858 high-level GC-NOTIFY-FOO functions send their output. (There's
4859 still very little control of where low-level verbose GC functions
4860 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4861 less than it used to -- the GC-NOTIFY-FOO functions are now under
4862 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4863 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4864 in only one place in the source code, and propagates it automatically
4865 everywhere that it's needed. Thus e.g. when I bump the version from
4866 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4867 * The C source files now include boilerplate legalese and documentation
4868 at the head of each file (just as the Lisp source files already did).
4869 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4870 to his page will be replaced with a link to his new CLiki service.
4872 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4874 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4875 * The build process now looks for GNU make under the default name "gmake",
4876 instead of "make" as it used to. If GNU make is not available as "gmake"
4877 on your system, you can change this default behavior by setting the
4878 GNUMAKE environment variable.
4879 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4880 can build without error under CMU CL.
4882 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4884 * There is now a partial SBCL user manual (with some new text and some
4885 text cribbed from the CMU CL manual).
4886 * The beginnings of a profiler have been added (starting with the
4887 CMU CL profiler and simplifying and cleaning up). Eventually the
4888 main interface should be through the TRACE macro, but for now,
4889 it's still accessed through vaguely CMU-CL-style functions and macros
4890 exported from the package SB-PROFILE.
4891 * Some problems left over from porting CMU CL to the new
4892 cross-compilation bootstrap process have been cleaned up:
4893 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4894 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4895 part when getting rid of the compiler *BACKEND* stuff.)
4896 ** The value of *NULL-TYPE* was screwed up, because it was
4897 being initialized before the type system knew the final
4898 definition of the 'NULL type. This screwed up several key
4899 optimizations in the compiler, causing inefficiency in all sorts
4900 of places. (I found it because I wanted to understand why
4901 GET-INTERNAL-RUN-TIME was consing.)
4902 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4903 PROCLAIM FTYPE information. Unfortunately this broke binary
4904 compatibility again, since now the forms output by DEFGENERIC
4905 to refer to functions which didn't exist in 0.6.3.
4906 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4907 can use the new (as of 0.6.3) transform for SOME into MAP into
4909 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4910 values to (INTEGER 0 1000000), so that the time code will no longer
4911 occasionally get blown up by Linux returning 1000000 microseconds
4912 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4913 its output conform to the ANSI spec. (Alas, this makes its output
4914 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4915 * A full call to MAP NIL with a single sequence argument no longer conses.
4916 * fixes to problems pointed out by Martin Atzmueller:
4917 * The manual page no longer talks about multiprocessing as though
4918 it were currently supported.
4919 * The ILISP support patches have been removed from the distribution,
4920 because as of version 5.10.1, ILISP now supports SBCL without us
4921 having to maintain patches.
4922 * added a modified version of Raymond Toy's recent CMU CL patch for
4923 EQUALP comparison of HASH-TABLE
4925 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4927 * The system still can't cross-compile itself with
4928 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4929 entails), but at least it can compile more of itself that way
4930 than it used to be able to, and various buglets which were uncovered
4931 by trying to cross-compile itself that way have now been fixed.
4932 * This release breaks binary compatibility again. This time
4933 at least I've incremented the FASL file format version to 2, so that the
4934 problem can be detected reliably instead of just causing weird errors.
4935 * various new style warnings:
4936 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4937 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4938 ** using the lexical binding of a variable named in the *FOO* style
4939 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4940 as specified by ANSI.
4941 * *RANDOM-STATE* is no longer automatically initialized from
4942 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4943 default behavior of the system is to repeat its behavior every time
4944 it's run. If you'd like to change this behavior, you can always
4945 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4946 old convention there was no comparably easy way to get the system to
4947 repeat its behavior every time it was run.
4948 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4949 FTYPE declarations has been removed, in favor of their ANSI
4950 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4951 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4952 the types of their sequence arguments can be determined at compile time.
4953 This is done through a new open code expansion for MAP which eliminates
4954 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4955 argument types can be determined at compile time.
4956 * The optimizer now transforms COERCE into an identity operation when it
4957 can prove that the coerced object is already of the correct type. (This
4958 can be a win for machine generated code, including the output of other
4959 optimization transforms, such as the MAP transform above.)
4960 * Credit information has been moved from source file headers into CREDITS.
4961 * Source file headers have been made more standard.
4962 * The CASE macro now compiles without complaining even when it has
4965 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4967 * (Note that the way that the PCL macroexpansions were rewritten
4968 to accommodate the change in DEFGENERIC below breaks binary
4969 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4970 not run under sbcl-0.6.2. Once we get out of alpha releases,
4971 i.e. hit release 1.0.0, we'll probably try to maintain binary
4972 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4973 and sbcl-1.4.4. Until then, however, it might be fairly common
4974 for maintenance releases to break binary compatibility.)
4975 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4977 * The reporting of unhandled signals has been changed to print some
4978 explanatory text as well as the report form. (Previously only
4979 the report form was printed.)
4980 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4981 it defines, so that the compiler no longer issues undefined function
4982 warnings for compiled-but-not-yet-loaded generic functions.
4983 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4984 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4985 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4986 section of the ANSI spec are supported.
4987 * The "" nickname for the "KEYWORD" package has been removed.
4988 The reader still handles symbol tokens which begin with a package marker
4989 as keywords, but it doesn't expose its mechanism for doing so in the
4990 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4991 * The system now issues STYLE-WARNINGs for contradictory TYPE
4992 proclamations. (Warnings for contradictory FTYPE proclamations would
4993 be nice too, but those can't be done usefully unless the type system
4994 is made smarter about FUNCTION types.)
4995 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4996 systematized, so that "*target-*.lisp is supposed to exist only on the
4997 target and imply that there's a related file which exists on the
4998 host, and *host-*.lisp is supposed to exist only on the host and imply
4999 that there's a related file which exists on the target. This involves a
5000 lot of renaming. Hopefully the acute confusion caused by the renaming
5001 will be justified by the reduction in chronic confusion..
5002 ** runtime-type.lisp -> early-target-type.lisp
5003 ** target-type.lisp -> late-target-type.lisp
5004 ** early-host-format.lisp -> early-format.lisp
5005 ** late-host-format.lisp -> late-format.lisp
5006 ** host-error.lisp -> misc-error.lisp
5007 ** early-error.lisp -> early-target-error.lisp
5008 ** late-error.lisp -> late-target-error.lisp
5009 ** host-defboot.lisp -> early-defboot.lisp
5010 ** code/misc.lisp -> code/target-misc.lisp
5011 ** code/host-misc.lisp -> code/misc.lisp
5012 ** code/numbers.lisp -> code/target-numbers.lisp
5013 ** code/early-numbers.lisp -> numbers.lisp
5014 ** early-host-type.lisp -> early-type.lisp
5015 ** late-host-type.lisp -> late-type.lisp
5016 ** host-typep.lisp -> typep.lisp
5017 ** load.lisp -> target-load.lisp
5018 ** host-load.lisp -> load.lisp
5019 ** host-disassem.lisp -> disassem.lisp
5020 ** host-insts.lisp -> insts.lisp
5021 ** byte-comp.lisp -> target-byte-comp.lisp
5022 ** host-byte-comp.lisp -> byte-comp.lisp
5023 ** host-signal.lisp -> signal.lisp
5024 ** host-defstruct.lisp -> defstruct.lisp
5025 ** late-target-type.lisp -> deftypes-for-target.lisp
5026 Furthermore, several other previously target-only files foo.lisp (e.g.
5027 hash-table.lisp and random.lisp) have been split into a target-and-host
5028 foo.lisp file and a target-only target-foo.lisp file, with their key type
5029 definitions in the target-and-host part, so that the cross-compiler will
5030 know more about target types.
5031 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5032 gone away. In their place are various *BACKEND-FOO* variables
5033 corresponding to the slots of the old structure.
5034 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5035 into the target SBCL has been fixed.
5036 * The chill.lisp system for loading cold code into a running SBCL
5038 * Support for the CMU CL "scavenger hook" extension has been removed.
5039 (It was undocumented and unused in the CMU CL sources that SBCL was
5040 derived from, and stale in sbcl-0.6.1.)
5041 * Various errors in the cross-compiler type system were detected
5042 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5043 (enabling various consistency checks). Many of them were fixed,
5044 but some hard problems remain, so the compiler is back to
5045 running without *TYPE-SYSTEM-INITIALIZED* for now.
5046 * As part of the cross-compiler type system cleanup, I implemented
5047 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5048 * I have started adding UNCROSS calls throughout the type system
5049 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5050 on cross-compiler input files will be able to go away, and various
5052 * CONSTANTP now returns true for quoted forms (as explicitly required
5055 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5057 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5058 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5059 (MAKE-LIST -1). In the longer term, it ought to become true
5060 that declarations are assertions even at SAFETY 1. For now, it's not
5061 quite true even at SAFETY 3, but it's at least more nearly true..
5062 (Note that this change seems to increases the size of the system by
5063 O(5%) and to decrease the speed of the compiler by 20% or more.)
5064 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5065 for the problem of printing dozens of lines of distracting information
5066 about low-level system machinery as part of the top stack frame
5067 on entry to the debugger when an undefined function was called.
5068 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5069 is not reset to COMMON-LISP-USER.
5070 * Compilation of stuff related to dyncount.lisp has been made conditional
5071 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5072 smaller. The various dyncount-related symbols have been moved into
5073 a new "SB-DYNCOUNT" package.
5074 * tty-inspect.lisp has been renamed to inspect.lisp.
5075 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5076 feature has gone away. (When we eventually port to other flavors of
5077 libc and/or Unix, we'll try to make the differences between flavors
5078 invisible at the user level.)
5079 * Various other *FEATURES* tags, and/or their associated conditionals,
5080 have been removed if obsolescent, or given better documentation, or
5081 sometimes given more-mnemonic names.
5083 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5085 * tidied up "make.sh" script
5086 * tidied up system directory structure
5087 * better "clean.sh" behavior
5088 * added doc/FOR-CMUCL-DEVELOPERS
5089 * many many small tweaks to output format, e.g. removing possibly-confusing
5090 trailing #\. character in DESCRIBE-INSTANCE
5091 * (EQUALP #\A 'A) no longer signals an error.
5092 * new hashing code, including EQUALP hashing
5093 * tidied up Lisp initialization and toplevel
5094 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5095 * command line argument processing
5096 * added POSIX-GETENV function to deal with Unix-ish environment variables
5097 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5098 terminating SBCL on EOF
5099 * non-verbose GC by default
5100 * There is no more "sbcl" shell script; the sbcl file is now the C
5101 runtime executable (just like CMU CL).
5102 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5104 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5105 symbol and package information as Lisp native symbol and package objects
5106 instead of strings naming symbols and strings naming packages. This way,
5107 whenever packages are renamed (as in warm init), debug information is
5108 transformed along with everything else.
5109 * tweaked the optimization policy declarations which control the building
5110 of SBCL itself. Now, among other things, the system no longer saves
5111 source location debugging information. (This helps two problems at once
5112 by reducing SBCL size and by keeping SBCL from trying to look for its
5113 sources -- which may not exist -- when reporting errors.)
5114 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5115 debugging and testing purposes
5116 * cleaned up printing, making the printer call PRINT-OBJECT for
5117 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5118 methods, giving nearly-ANSI behavior
5119 * converted almost all special variables to use *FOO* naming convention
5120 * deleted PARSE-TIME functionality, since it can be done portably
5121 * moved some files out of cold init into warm init
5122 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5124 * regularized formatting of source files
5125 * added an install.sh script
5126 * fixed ridiculous memory usage of cross-compiler by making
5127 compiler/alloc.lisp not try to do pooling unless it can hook
5128 itself into the GC of the cross-compilation host. Now the system
5129 builds nicely on my old laptop.
5130 * added :SB-ALLOC in target-features.lisp-expr
5131 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5132 was not implemented)
5133 * re-did condition handling and note reporting in the compiler. Notes
5134 are no longer handled by signalling conditions. Style warnings
5135 and warnings are handled more correctly and reported in such a way
5136 that it's easy to find one or the other in your output (so that you
5137 can e.g. figure out which of many problems caused COMPILE-FILE to
5139 * changed the severity of several compiler warnings from full WARNING
5140 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5141 compiler note reporting so that it doesn't use the condition system
5142 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5143 COMPILE-FILE command)
5144 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5145 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5146 no longer supported, since their implementation was deeply intertwingled
5147 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5148 * removed bogus "support" for compiler macros named (SETF FOO), and
5149 removed the compiler macro for SETF INFO (but only after making a fool
5150 of myself on the cmucl-imp mailing list by posting a bogus patch for
5151 DEFINE-COMPILER-MACRO..)
5152 * Compiled files containing forms which have side effects on the Lisp
5153 reader (such as DEFPACKAGE forms) are now handled more correctly.
5154 (Compiler queuing of top level lambdas has been suppressed by setting
5155 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5156 * deleted various currently-unused source files, e.g. gengc.lisp. They
5157 may be added back at some point e.g. when porting to other architectures,
5158 but until they are it's distracting to distribute them and to try to
5160 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5161 there were so many of them they're just distractions, and UNCROSS is
5162 known to be able to handle the current sources
5163 * moved PROFILE functionality into TRACE, so that it will be clear
5164 how the wrapping and unwrapping of functions when you profile them
5165 interacts with the wrapping and unwrapping of functions when you
5166 trace them. (Actually, the functionality isn't there yet, but at least
5167 the interface specification is there. Hopefully, the functionality will
5168 arrive with some future maintenance release.)
5169 * removed host-oops.lisp
5170 * changed signature of QUIT function to allow UNIX-CODE argument
5171 * fixed READ-SEQUENCE bug
5172 * tweaked verbose GC output so that it looks more like the progress
5173 output that ANSI specifies for functions like LOAD
5174 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5175 * added <http://sbcl.sourceforge.com> to the banner information printed by