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: exit status from QUIT when called under --script was lost
28 (reported by Hubert Kauker)
29 * bug fix: MAKE-ARRAY for non-zero :INITIAL-ELEMENT always used the
30 same implementation of FILL to initialize the array, even if a more
31 efficient one was available (reported by Stas Boukarev, thanks to
33 * bug fix: potential miscompilation of array stack allocation on x86 and
34 x86-64. (reported by Time Tossavainen)
35 * bug fix: some forms of AND, OR, and COND resulted in expansions that could
36 result in their subforms being treated as top level forms. (reported by
38 * bug fix: On x86/x86-64 alien functions declared to return integers shorter
39 than a machine register could leave garbage in the high bits of the
40 result register (bug 316325).
41 * bug fix: disable address space randomization Linux/x86-64 as well,
42 not just x86-64. (reported by Ken Olum)
43 * bug fix: Attempting to DEREF an (ALIEN (* T)) would produce a WARNING and
44 generate incorrect code.
45 * bug fix: #201; type inference for CONS and ARRAY types could derive
46 wrong results in the presence of eg. RPLACA or ADJUST-ARRAY.
47 * bug fix: special variables with a proclaimed specific subtype of FUNCTION
48 could not be assigned to or bound with PROGV. (reported by Lorenz
50 * bug fix: the value of CL:- in the inspector was the previous expression
51 evaluated rather than the expression being evaluated.
52 * bug fix: constants can no longer be locally declared special.
53 * bug fix: signals delivered to threads started from foreign land (read:
54 directly by pthread_create, not by MAKE-THREAD) are redirected to a Lisp
55 thread by blocking all signals and resignalling.
57 changes in sbcl-1.0.28 relative to 1.0.27:
58 * a number of bugs in cross-compilation have been fixed, with the ultimate
59 result that building under (at least) clisp should be much more reliable.
60 * minor incompatible changes: echo-streams now propagate unread-char to the
61 underlying input stream, and no longer permit unreading more than one
63 * improvement: on x86/x86-64 Lisp call frames now have the same layout as C
64 frames, allowing for instance more reliable backtraces.
65 * improvement: the debugger REPL can now reference lexical variables
66 by name directly for code compiled with (DEBUG 3).
67 * improvement: errors from malformed declarations now have better source
68 paths associated with them. (thanks to Tobias Rittweiler)
69 * optimization: faster local calls on x86/x86-64
70 * bug fix: some error messages for out-of-bound array indexes confused the
71 index and the bound. (thanks to Stas Boukarev)
72 * bug fix: pretty printing malformed DEFPACKAGE forms (thanks to Sidney
74 * bug fix: running regressions tests in shells without OSTYPE set now works.
75 (reported by Harald Hanche-Olsen)
76 * bug fix: more robust static space exhaustion signalling from
77 MAKE-STATIC-VECTOR (thanks to Daniel Lowe)
78 * bug fix: (SETF DOCUMENTATION) for anonymous function now throws the
79 docstring away instead of storing it under names such as (LAMBDA (X)).
80 (reported by Leslie Polzer)
81 * bug fix: timers could go off in the wrong order, be delayed indefinitely
82 (thanks to Ole Arndt for the patch)
83 * bug fix: RESTART-FRAME and RETURN-FROM-FRAME stack corruption
84 * bug fix: the discriminating function for PRINT-OBJECT no longer preserves
85 potentially-invalid effective methods in its cache.
86 * bug fix: SB-INTROSPECT:FIND-DEFINITION-SOURCE now works with funcallable
87 instances as well (thanks to Paul Khuong)
88 * bug fix: using RUN-PROGRAM does not interfere with SB-POSIX:WAIT,
89 SB-POSIX:WAITPID and their C equivalents.
90 * bug fix: RUN-PROGRAM does not crash on Darwin when stressed.
92 changes in sbcl-1.0.27 relative to 1.0.26:
93 * new port: support added for x86-64 OpenBSD. (thanks to Josh Elsasser)
94 * new port: support added for x86-64 Solaris. (thanks to Alex Viskovatoff)
95 * improvement: the system either recovers from stack exhaustion or dies
96 properly as opposed to leaving the user uncertain of whether the handler
97 trampled on some random memory next to the stack or having to rely on
98 --lose-on-corruption (which is still a good idea to use in production
99 because stack exhaustion can happen in signal handlers which will likely
101 * bug fix: fix gc related interrupt handling bug on ppc (regression from
102 1.0.25.37, reported by Harald Hanche-Olsen)
103 * bug fix: work around signal delivery bug in darwin (regression from
104 1.0.25.44, reported by Sidney Markowitz)
105 * bug fix: fix ERROR leaking memory (reported by David Thompson)
107 changes in sbcl-1.0.26 relative to 1.0.25:
108 * incompatible change: an interruption (be it a function passed to
109 INTERRUPT-THREAD or a timer function) runs in an environment where
110 interrupts can be enabled. The interruption can use
111 WITH-INTERRUPTS or WITHOUT-INTERRUPTS as it sees fit. Use
112 WITHOUT-INTERRUPTS to avoid nesting of interruptions and
113 potentially running out of stack. Keep in mind that in the absence
114 of WITHOUT-INTERRUPTS some potentially blocking operation such as
115 acquiring a lock can enable interrupts.
116 * incompatible change: GC-OFF and GC-ON are removed, as they were
117 always unsafe. Use WITHOUT-GCING instead.
118 * new feature: runtime option --disable-ldb
119 * new feature: runtime option --lose-on-corruption to die at the
120 slightest hint of possibly non-recoverable errors: running out of
121 memory, stack, alien stack, binding stack, encountering a memory
122 fault, etc. In the absence of --lose-on-corruption a warning is
124 * enhancement: detect binding stack exhaustion
125 * enhancement: detect alien stack exhaustion on x86/x86-64
126 * improvement: generally more stable and reliable interrupt handling
127 * improvement: there is a per thread interruption queue,
128 interruptions are executed in order of arrival
129 * improvement: a repeating timer reschedules itself when the it has
130 finished, but expiration times are spaced equally. If an
131 expiration time is in the past it will trigger after a short grace
132 period that may give a chance to other things to run.
133 * optimization: slightly faster gc on multithreaded builds
134 * optimization: faster WITHOUT-GCING
135 * bug fix: when JOIN-THREAD signals an error, do it when not holding
136 important locks so that the debugger/handler doesn't produce
137 recursive errors or deadlock.
138 * bug fix: real-time signals are not used anymore, so no more
139 hanging when the system wide real-time signal queue gets full.
140 * bug fix: INTERRUPT-THREAD on a dying thread could produce memory
142 * bug fix: finalizers, gc hooks never run in a WITHOUT-INTERRUPTS
143 * bug fix: fix random memory faults related to interrupts on alpha
144 * bug fix: fix random memory faults related to interrupts on sparc
145 * bug fix: fix deadlocks related to starting threads
146 * bug fix: fix deadlines on locks on futex platforms
147 * bug fix: restore errno in signal handlers
148 * bug fix: fix deadlocks related to hash tables
149 * bug fix: fix deadlocks in pcl
151 changes in sbcl-1.0.25 relative to 1.0.24:
152 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
153 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
154 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
155 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
156 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
157 better name and does not return values so stale on multiprocessor systems.
158 Also, HOLDING-MUTEX-P was added for about the only sane usage of
160 * improvement: unithread builds keep track of MUTEX-VALUE.
161 * improvement: reading from a TWO-WAY-STREAM does not touch the output
162 stream anymore making it thread safe to have a concurrent reader and
163 a writer, for instance, in a pipe.
164 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
165 arguments, making compiler-macros for SETF-functions able to inspect
166 their constant arguments.
167 * improvement: COMPILE-FILE reports times with millisecond accuracy
168 (thanks to Luis Oliveira)
169 * optimization: CHAR-CODE type derivation has been improved, making
170 TYPEP elimination on subtypes of CHARACTER work better. (reported
171 by Tobias Rittweiler, patch by Paul Khuong)
172 * bug fix: setting alien structure fields of type struct by value now
173 computes the right offset for the memory copy.
174 * bug fix: compilation problem involving inlined calls to aliens with
175 result type VOID. (reported by Ken Olum)
176 * bug fix: #235a; sequential inline expasion in different policies no
177 longer reuses the functional from the previous expansion site.
178 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
179 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
180 * bug fix: named ENUMs in multiply-referenced alien record types no longer
181 cause an implied type redefinition. Regression from 1.0.21.29.
182 * bug fix: sign-extension of small signed return values now works with C
183 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
184 * improvements to the Windows port:
185 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
186 Westfold, thanks to Rudi Schlatte)
188 changes in sbcl-1.0.24 relative to 1.0.23:
189 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
190 vector of a multidimensional SIMPLE-ARRAY.
191 * new feature: the system now signals a continuable error if standard
192 readtable modification is attempted. (thanks to Tobias Rittweiler)
193 * new feature: DIRECTORY has been extended with a non-standard keyword
194 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
195 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
196 has been robustified and documented better. (thanks to Robert Goldman)
197 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
198 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
199 * tradeoff: constant FORMAT control strings are now compiled unless
200 SPACE > SPEED (previously only when SPEED > SPACE.)
201 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
203 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
205 * bug fix: lexical type declarations are now correctly reported by
206 SB-CLTL2. (reported by Larry D'Anna)
207 * bug fix: STRING-TO-OCTETS did not handle :START properly when
208 using UTF-8 as external format. (thanks to Luis Oliveira)
209 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
210 are signalled correctly. (thanks to Stas Boukarev)
211 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
212 designator. (thanks to Tobias Rittweiler)
213 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
214 designator, and returns T instead of the function. (thanks to
216 * bug fix: direct superclasses of STANDARD-CLASS and
217 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
218 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
219 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
220 when no class with the named slot yet exists no longer causes a
221 compile-time style-warning.
222 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
223 in safe code. (reported by Didier Verna)
224 * bug fix: #430; nested structure constructors can stack allocate.
225 * bug fix: on some 64-bit platforms dynamic space size was truncated
226 to #xffff0000 bytes. (reported by Benjamin Lambert)
227 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
228 REPL unusable. (reported by Daniel Herring)
229 * bug fix: return values of READ-SEQUENCE did not take :START into
230 account on file streams, regressions since 1.0.12.22. (reported by
231 Thomas Russ, patch by Paul Khuong)
232 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
233 method specializer used to confuse permuation vector optimization.
234 * bug fix: system inserted bogus implicit type declarations for local
235 special variables in DEFMETHOD bodies.
236 * bug fix: #354; duplicated frames in backtraces due to
237 non-tail-call-optimized XEPs to functions with return type NIL
239 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
240 initializes structure object slots according to DEFSTRUCT initforms,
241 and DEFSTRUCT forms :INCLUDEind structure classes defined using
242 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
243 (reported by Bruno Haible and Stephen Wilson)
244 * bug fix: #395; fill-pointer output streams used now support
245 element-type BASE-CHAR as well.
246 * bug fix: compiler error when attempting to derive return value of
247 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
249 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
251 * bug fix: Mac OS X binaries should now be portable between Leopard
254 changes in sbcl-1.0.23 relative to 1.0.22:
255 * enhancement: when disassembling method functions, disassembly
256 for the associated fast function is also produced.
257 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
259 * optimization: printing with *PRINT-PRETTY* true is now more
260 efficient as long as the object being printed doesn't require
261 special handling by the pretty printer.
262 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
263 now interact correctly with type declarations.
264 * partial bug fix: PCL detects infinite recursion during wrapper
265 validation. (thanks to Attila Lendvai)
266 * bug fix: #426; nested function calls are inlined properly.
267 Previously if FOO was an inline function, in calls of the form
268 (FOO (FOO ...)) the outer call was not inlined.
269 * bug fix: long long arguments passed in registers to alien
270 functions on PowerPC Linux and NetBSD are now handled correctly
271 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
273 changes in sbcl-1.0.22 relative to 1.0.21:
274 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
275 for the shared object in the current directory, but passes the native
276 namestring of the designated pathname to the operation system's shared
277 object loading function as-it.
278 * minor incompatible change: --disable-debugger toplevel option now takes
279 effect before processing of initialization files and --eval or --load
281 * new feature: new commandline argument: --script, which supports
282 shebang lines. See documentation for details. (based on work by
284 * new feature: SAVE-LISP-AND-DIE can save current values of
285 --dynamic-space-size and --control-stack-size in the executable core,
286 causing it to skip normal runtime option processing. See documentation
287 for details. (thanks to Zach Beane)
288 * enhancement: inoccous calls to EVAL or generic functions dispatching
289 on subclasses of eg. STREAM no longer cause compiler notes to appear.
290 * enhancement: the system no longer resignals errors from --load and
291 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
292 associated with the original error to be lost. (thanks to Ariel
294 * enhancement: :DONT-SAVE keyword argument has been added to
295 LOAD-SHARED-OBJECT for controlling interaction with
296 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
297 the effects of an earlier LOAD-SHARED-OBJECT call.
298 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
299 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
300 provided. (reported by Cedric St-Jean)
301 * bug fix: circularity handling in the reader did not treat raw
302 structure slots correctly. (reported by Cedric St-Jean)
303 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
304 file descriptors when there were none.
305 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
306 was broken. (reporteed by Willem Broekema)
307 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
308 pathnames without a directory.
309 * bug fix: COMPILE-FILE did not warn about undefined variable
310 references at toplevel, and LOAD of the resulting fasl did
312 * bug fix: functions with non-required arguments used to end up with
313 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
314 * bug fix: redefining a function with non-required arguments didn't
315 update the system's knowledge about its call signature properly.
316 * bug fix: fixed #431; incompatible alien record type redefinitions
317 are detected and handled. (reported by Neil Haven)
318 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
319 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
321 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
322 after alien stack frames.
323 * bug fix: strings passed to mk{s,d}temp lacked null termination
325 changes in sbcl-1.0.21 relative to 1.0.20:
326 * new feature: the compiler is able to track the effective type of a
327 generic function across method addition and removal.
328 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
329 appropriately typed structure slots without locking.
330 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
331 information like those gathered by TIME using a programming-friendly
333 * new feature: TIME reports time taken even if the form performs a
334 non-local transfer of control.
335 * enhancement: reduced conservativism on GENCGC platforms: on
336 average 45% less pages pinned (measured from SBCL self build).
337 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
338 mutate constant symbols or violate declaimed type of the symbol.
339 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
340 sleeping on the semaphore in heavily contested semaphores.
341 * bug fix: semaphores and condition variables were not interrupt
343 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
344 owned by other threads anymore.
345 * bug fix: FIND on lists called KEY outside the specified
346 subsequence. (reported by budden)
347 * bug fix: LOG doesn't use single-float intermediate results when
348 given mixed integer and double-float arguments, leading to better
349 precision. (reported by Bob Felts)
350 * bug fix: LOG with base zero returned values of inconsistent type.
352 changes in sbcl-1.0.20 relative to 1.0.19:
353 * minor incompatible change: OPTIMIZE qualities
354 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
355 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
356 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
357 * documentation: some slot access efficiency guidelines have been
358 added to the user manual.
359 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
360 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
361 as ASSOC and MEMEBER.
362 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
363 to more efficient EQ-comparison versions more often.
364 * optimization: enhanced derivation of DOLIST iteration variable type
366 * optimization: constant folding of simple (LIST ...) forms as DOLIST
368 * optimization: runtime lookup of function definitions can be
369 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
370 * optimization: compiler is able to derive the return type of
371 (AREF (THE STRING X) Y) as being CHARACTER.
372 * optimization: CLRHASH on empty hash-tables no longer does pointless
373 work. (thanks to Alec Berryman)
374 * optimization: REPLACE deftransforms don't punt when :START1 or
376 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
377 no longer emits a code deletion note. (reported by Andrew
379 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
380 functions with an odd number of &OPTIONAL arguments, a &REST
381 argument, and one or more &KEY arguments at the call site.
382 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
383 arguments appeared at call sites as well.
384 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
385 READ-LINE &co can no longer read from them afterwards. (reported
387 * bug fix: fixed #427: unused local aliens no longer cause compiler
388 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
390 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
391 alien-stack leakage. (reported by Andy Hefner)
392 * bug fix: PROGV signals an error when an attempt to violate declared
393 type of a variable or bind a constant is made.
394 * bug fix: SET signals an error when an attempt to violate declared
395 type of a variable is made.
396 * bug fix: restart computation during the execution of a restart
397 test function no longer causes infinite recursion. (reported by
399 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
400 signals an error. (thanks to Pierre Mai)
401 * bug fix: EXPORT left symbol unexported in conflict situations.
402 (thanks to Michael Weber)
403 * bug fix: correctly handle name conflicts involving CL:NIL.
404 (thanks to Michael Weber)
405 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
406 conflicts arising from USEing package with conflicting symbols
407 correctly. (thanks to Michael Weber)
409 changes in sbcl-1.0.19 relative to 1.0.18:
410 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
411 warnings that go otherwise unhandled will be muffled if they are
412 of the type that's the value of this variable.
413 * optimization: stack allocation is slightly more efficient on x86
415 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
416 where a raw slot always is initialized using the initform whose
417 type is not know sufficiently well a compile-time are now compiled
418 correctly. (reported by John Morrison)
419 * bug fix: compiler no longer makes erronous assumptions in the
420 presence of non-foldable SATISFIES types.
421 * bug fix: stack analysis missed cleanups of dynamic-extent
422 arguments in non-let-converted calls to local functions.
423 * improvements to the Windows port:
424 ** adjusted address spaces for building on both Win32 and
425 Win64. (thanks for John Connors)
426 * fixed some bugs revealed by Paul Dietz' test suite:
427 ** interval arithmetic during type derivation used inexact integer
428 to single-float coercions.
429 ** arithmetic operations involving large integers and single
430 floats give the same results in compiled and interpreted code.
431 ** deriving the result type of COERCE no longer signals an error
432 if the derived type of the second argument is a MEMBER type
433 containing invalid type specifiers.
434 ** ADJOIN with constant NIL as second argument works correctly.
435 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
437 changes in sbcl-1.0.18 relative to 1.0.17:
438 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
439 profiles only the current thread.
440 * minor incompatible change: changes to SYMBOL-VALUE of constants
441 defined with DEFCONSTANT now signal an error.
442 * enhancement: SB-SPROF now has support for wallclock profiling,
443 and is also able to profile specific threads. REPORT output
444 has also additional sorting options.
445 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
447 * optimization: structure allocation has been improved
448 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
449 ** out of line constructors are ~10% faster.
450 ** inline constructors are ~15% faster.
451 ** inline constructors are capable of dynamic extent allocation
452 (generally on x86 and x86-64, in some cases on other platforms
454 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
456 * optimization: file compiler is now able to coalesce non-circular
457 lists, non-base strings, and bit-vectors. Additionally, constants
458 are never referenced through SYMBOL-VALUE at runtime.
459 * optimization: code defining methods on PRINT-OBJECT (and other
460 generic functions in the COMMON-LISP package) now loads faster.
461 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
463 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
464 fasl is now deleted. (reported by Attila Lendvai)
465 * bug fix: READ-LINE always returned NIL for the last line in files.
466 (reported by Yoshinori Tahara)
467 * bug fix: more accurate disassembly annotations of foreign function
468 calls. (thanks to Andy Hefner)
469 * bug fix: trimming non-simple strings and non-string string
470 designators when the there is nothing to trim works properly.
471 (thanks to James Knight)
472 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
473 (thanks to Travis Cross)
474 * fixed some bugs revealed by Paul Dietz' test suite:
475 ** NIL is a valid function name (regression at 1.0.13.38)
476 ** FILL on lists was missing its return value (regression at 1.0.12.27)
477 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
478 fill pointers properly (regression at 1.0.12.23)
479 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
480 displacement indices properly (regression at 1.0.12.23)
482 changes in sbcl-1.0.17 relative to 1.0.16:
483 * temporary regression: user code can no longer allocate closure
484 variable storage on stack, due to bug 419 without explicitly
485 requesting it. Please consult sbcl-devel for advice if you need to
486 use this feature in the meanwhile.
487 * new feature: runtime argument --control-stack-size can be used to
488 adjust thread default control stack size.
489 * enhancement: improved TIME output
490 ** all times are reported using the measured accuracy (milliseconds
491 for real and GC times, microseconds for everything else.)
492 ** processor cycle counts on x86 and x86-64.
493 ** interpreted forms are counted for both evaluator modes.
494 ** number of lambdas converted by the compiler is reported.
495 ** CPU percentage report (computed from real and total run time.)
496 ** more comprehensive run time reporting, using a condenced format
497 ** interperted form, lambda, and page fault counts are omitted
499 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
501 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
502 * optimization: two argument forms of LAST are upto ~10% faster
503 in normal SPEED policies.
504 * optimization: NCONC no longer needs to heap cons its &REST list
505 in normal SPEED policies.
506 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
508 * bug fix: UNION and NUNION work with :TEST-NOT once more,
509 regression since 1.0.9.1. (thanks to Eric Marsden)
510 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
511 since 1.0.15.36. (thanks to Paul Khuong)
512 * bug fix: LAST when always returned the whole list when given a bignum
513 as the second argument.
514 * bug fix: dynamic extent allocation of nested lists and vectors
515 could leak to otherwise accessible parts.
516 * bug fix: invalid optimization of heap-allocated alien variable
518 * bug fix: fasl header checking is less vulnerable to different
519 platform word lengths.
520 * bug fix: more correct assembler syntax for GNU binutils
521 2.18.50.0.4 support. (thanks to Marijn Schouten)
522 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
524 * bug fix: the fopcompiler can handle LOCALLY forms (with no
525 declarations) successfully. (reported by Attila Lendvai)
527 changes in sbcl-1.0.16 relative to 1.0.15:
528 * minor incompatible change: revert the changes to sb-posix's error
529 signaling added in 1.0.14.
530 * minor incompatible change: change PROBE-FILE back to returning
531 NIL whenever we can't get a truename, as was the case before 1.0.14.
532 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
533 inherits from ERROR instead of just CONDITION.
534 * new feature: SB-INTROSPECT can provide source locations for instances
535 as well. (thanks to Tobias Rittweiler)
536 * optimization: binding special variables now generates smaller code
537 on threaded platforms.
538 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
539 and cases where no :TEST is given but the compiler can infer that
540 the element to search is of type (OR FIXNUM (NOT NUMBER)).
541 * optimization: better LOGNOT on fixnums.
542 * optimization: modular arithmetic for a particular requested width
543 is implemented using a tagged representation unless a better
544 representation is available.
545 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
546 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
547 for wellformedness and returns a specialized vector. (reported by
548 Francois-Rene Rideau)
549 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
550 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
551 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
552 unscheduling timer at the same time as another timer fires.
553 * bug fix: don't reschedule timers for dead threads.
554 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
555 * bug fix: copying output from RUN-PROGRAM to a stream signalled
556 bogus errors if select() was interrupted.
557 * enhancement: add support for fcntl's struct flock to SB-POSIX.
559 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
560 * enhancement: cleaner backtraces for interactive interrupts, as
561 well as other cases where the interesting frames used to be
562 obscured by interrupt handling frames.
563 * enhancement: untracing a whole package using (UNTRACE "FOO") is
564 now supported, and tracing a whole package using (TRACE "FOO") now
565 traces SETF-functions as well.
566 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
567 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
568 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
569 * unreadably printed representation of hash-tables now includes
571 * bug fix: partially fixed #188: type propagation from assignments
572 is now more efficient.
573 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
574 'DOUBLE-FLOAT) are not flushable.
575 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
576 full address of the object, and none of the tag bits.
577 * bug fix: readably printing hash-tables now respects other printer
578 control variables. (reported by Cedric St-Jean)
579 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
580 keyword with MAKE-HASH-TABLE.
581 * bug fix: export SB-POSIX:MKSTEMP.
582 * bug fix: SORT was not interrupt safe.
583 * bug fix: XREF accounts for the last node of each basic-block as
585 * bug fix: MAKE-INSTANCE optimizations interacted badly with
586 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
587 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
588 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
589 create obsolete instances in the case of redefinition or
590 obsoletion of a superclass. (thanks to Andy Hefner)
591 * bug fix: Support for the Alpha architecture has been revived; it had
592 suffered somewhat from lack of maintenance since sbcl-1.0.
593 * improvements to the Windows port:
594 ** The system detects the codepage to use at startup, instead of
595 using the value from the saved core. (thanks to Kei Suzuki)
597 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
598 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
599 (see documentation for details.)
600 * revived support for OpenBSD (contributed by Josh Elsasser)
601 * partially fixed bug #108: ROOM no longer suffers from occasional
602 (AVER (SAP= CURRENT END)) failures.
603 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
604 bodies no longer cause a WARNING to be signalled. (reported by
606 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
607 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
608 the bug in TRUNCATE.) (reported by Michael Weber)
609 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
610 no samples. (reported by Andy Hefner)
611 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
612 no longer appear as (NIL ...) frames in backtraces.
613 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
614 is now actually exported from SB-EXT as documented. (reported by
616 * bug fix: sb-aclrepl now correctly understands how to inspect
617 single-floats on 64-bit platforms where single-floats are not boxed.
618 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
619 yet been finalized. (reported by Levente Meszaros)
620 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
621 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
622 SBCL-specific optimize qualities.
624 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
625 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
626 an executable in the search path, and does so in the child
627 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
628 been removed; it can be found in the CVS history, for somebody who
629 needs that search behavior (see the manual).
630 * minor incompatible change: compiler policy re. weakening type
631 checks has changed: now type checks are weakened only if SAFETY < 2
633 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
634 unparsing of directory pathnames as files. Analogously,
635 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
636 filename to parse into a directory pathname.
637 * enhancement: implicit generic function creation now signals a
638 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
639 users can bind handlers for and muffle around calls to LOAD.
640 * enhancement: RUN-PROGRAM allows unicode arguments and environments
641 to be used (using the default stream external format), and allows
642 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
643 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
644 strings and vectors whose element-type or simplicity is not fully
645 known at compile-time.
646 * optimization: STRING-TRIM and related functions no longer allocate
647 a new string when no trimming needs to be performed. These functions
648 are also faster than before when the input string has been declared
650 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
651 * optimization: READ-LINE is significantly faster for files containing
653 * optimization: non-open coded uses of character comparison operators
654 (e.g. char=) no longer cons when called with more than one parameter
655 on platforms supporting dynamic-extent allocation.
656 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
657 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
658 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
660 * bug fix: some sequence functions elided bounds checking when
662 * bug fix: too liberal weakening of union-type checks when SPEED >
664 * bug fix: more bogus fixnum declarations in ROOM implementation
667 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
668 * new feature: MAKE-HASH-TABLE now experimentally accepts a
669 :SYNCHRONIZED argument, which makes the hash-table safe for
670 concurrent accesses (but not iteration.) See also:
671 SB-EXT:WITH-LOCKED-HASH-TABLE, and
672 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
673 * optimization: CONCATENATE on strings is an order of magnitude faster
674 in code compiled with (> SPEED SPACE).
675 * optimization: SUBSEQ is ~50% faster on lists.
676 * bug fix: bug 417 fixed -- source location reporting is now more robust.
677 * bug fix: SUBSEQ on a list will now correctly signal an error if if
678 END is smaller then START.
679 * bug fix: SB-PROFILE will no longer report extra consing for nested
680 calls to profiled functions.
681 * bug fix: ROOM implementation had bogus fixnum declarations which
682 could cause type-errors when calling ROOM on large images.
683 * bug fix: if file compilation is aborted, the partial fasl is now
684 deleted, and COMPILE-FILE returns NIL as the primary value.
685 * bug fix: number of thread safety issues relating to SBCL's internal
686 hash-table usage have been fixed.
687 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
688 be returned from its body when the values were being returned
689 using unknown-values return convection and the W-P-O was wrapped
690 inside an UNWIND-PROTECT.
691 * bug fix: sb-posix should now compile again under Windows, enabling
694 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
695 * incompatible change: hash-table accessor functions are no longer
696 automatically protected by locks. Concurrent accesses on the same hash-table
697 from multiple threads can give inconsistent results or even corrupt the
698 hash-table completely. Multi-threaded applications should do their own
699 locking at the correct granularity. In the current implementation it is
700 still safe to have multiple readers access the same table, but it's not
701 guaranteed that this property will be maintained in future releases.
702 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
703 and will signal an error at runtime.
704 * enhancement: SB-THREAD package now exports a semaphore interface.
705 * enhancement: CONS can now stack-allocate on x86 and
706 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
708 * enhancement: nested lists can now be stack allocated on
709 platforms providing stack allocation support.
710 * enhancement: dynamic-extent support has been extended to support
711 cases where there are multiple possible sources for the stack
713 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
714 if the mutex is uncontested on Linux.
715 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
716 as specified by the CLHS. (thanks to Tobias Rittweiler)
717 * bug fix: NaN comparison now works on x86-64.
718 * bug fix: CLOSE :ABORT T on a stream with pending output now
720 * bug fix: instances of non-standard metaclasses using standard
721 instance structure protocol sometimes missed the slot type checks
723 * bug fix: known functions can be inlined in high-debug code.
725 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
726 * minor incompatible change: the MSI installer on Windows no longer
727 associates .lisp and .fasl files with the installed SBCL.
728 * minor incompatible change: :UNIX is no longer present in *FEATURES*
729 on Windows. (thanks to Luis Oliviera)
730 * new feature: SB-CLTL2 contrib module now implements
731 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
732 * optimization: scavenging weak pointers is now more efficient,
733 requiring O(1) instead of O(N) per weak pointer to identify
734 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
735 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
736 method bodies using constant slot-names when the first argument is
737 a specializer parameter for the method.
738 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
739 names, when the class of the instance is a direct instance
740 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
741 as slow as the constant slot-name case.)
742 * optimization: member type construction is now O(N) instead
744 * optimization: UNION and NUNION are now O(N+M) for large
745 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
747 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
748 now more readable in environments like Slime which display it.
749 (thanks to Tobias C. Rittweiler)
750 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
751 was able to infer the correct argument type for the object on which
752 the CAS operation was being performed.
753 * bug fix: copy propagation interfered with parallel assignment
754 semantics in local calls. (reported by Paul Khuong)
755 * bug fix: the signed modular fixnum shift compiled to wrong code on
756 x86 and x86-64. (spotted by a slight modification to some of
759 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
760 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
761 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
762 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
764 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
765 and generic function now print the object identity for anonymous
766 instances. (thanks to Attila Lendvai)
767 * enhancement: as an extension to MOP, SBCL now supports SETF of
768 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
769 (thanks to Attila Lendvai)
770 * enhancement: sb-cover has an alternate annotation mode which puts
771 the form annotations at the car of the form, instead of annotating
772 it completely. The benefit of this mode is that it shows better
773 which forms were instrumented by the compiler.
774 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
775 by printer control variables. (reported by Dan Corkill)
776 * bug fix: system leaked memory when delayed output was performed by
777 the OS in smaller chunks then expected. (thanks to David Smith)
778 * bug fix: system leaked memory when file streams were not closed
780 * bug fix: large objects written to slow streams that were modified
781 after the write could end up with the modified state written to
782 the underlying file descriptor.
783 * bug fix: multiple threads operating in parallel on the same stream
784 could cause buffer-overflows.
785 * bug fix: source location information is stored correctly
786 (broken since 1.0.6). This bug would generally show up as the
787 Slime debugger highlighting the wrong form when the sldb-show-source
789 * bug fix: Branch forms are again annotated as branches in the sb-cover
791 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
792 on x86-64. (reported by Gregory Vanuxem)
793 * bug fix: EQUALP could return wrong results for structures with raw
794 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
795 word). (reported by Vjacheslav Fyodorov)
796 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
799 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
800 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
801 atomic compare-and-swap operations on threaded platforms.
802 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
803 allows assigning a global minimum value to optimization qualities
804 (overriding proclamations and declarations).
805 * enhancement: closed over variables can be stack-allocated on x86
807 * performance bug fix: GETHASH and (SETF GETHASH) are once again
809 * optimization: slot definition lookup is now O(1). This speeds up
810 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
811 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
812 * optimization: ASSOC and MEMBER can now be open-coded for all
813 combinations of keyword arguments when second argument is constant
814 and SPEED >= SPACE. In other cases a specialized version is
816 * bug fix: using obsoleted structure instances with TYPEP and
817 generic functions now signals a sensible error.
818 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
819 (reported by Kristoffer Kvello)
820 * bug fix: backtrace construction is now more careful when making
821 lisp-objects from pointers on the stack, to avoid creating bogus
822 objects that can be seen by the GC.
823 * bug fix: defaulting of values in contexts expecting more than 7
824 variables now works on x86-64. (reported by Christopher Laux)
825 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
827 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
828 as the property-list of a symbol.
829 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
830 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
831 situations requested, are once again file-compileable. (reported
834 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
835 * MOP improvement: support for user-defined subclasses of
836 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
837 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
838 the expansion of DEFMETHOD, is responsible for generating a form
839 which creates a list of specializers when evaluated. Additional
840 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
841 debugging and introspective support.
842 * minor incompatible change: the (unsupported) spinlock interface
843 has changed: free spinlock now has the value NIL, and a held spinlock
844 has the owning thread as its value.
845 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
846 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
848 * enhancement: name of a socket-stream is now "a socket" instead of
850 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
851 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
852 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
853 for generic functions. (Thanks to Tobias C. Rittweiler)
854 * optimization: bignum printing speed has been improved by 20-40%
855 (depending on the bignum size.)
856 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
858 * bug fix: the cache used by the CLOS to store precomputed effective
859 methods, slot offsets, and constant return values is now thread and
861 * bug fix: generic function dispatch function updating is now thread
862 and interrupt safe (in the sense that the known issues have been
864 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
865 * bug fix: interning EQL-specializers is now thread and interrupt safe.
866 * bug fix: asdf systems with dependencies to the SB-POSIX or
867 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
868 * bug fix: interrupt safety of applicable method computation has been
871 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
872 * new contrib: sb-cover, an experimental code coverage tool, is included
874 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
875 significantly faster.
876 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
877 produce more efficient code on x86-64 if the compiler can determine
878 that (+ INDEX <constant>) does not require a bounds check and FOO
879 has an element type at least 8 bits wide.
880 * enhancement: a new, experimental synchronous timeout facility is
881 provided. Refer to SB-SYS:WITH-DEADLINE for details.
882 * enhancement: when a symbol name conflict error arises, the
883 conflicting symbols are always printed with a package prefix.
884 (thanks to Kevin Reid)
885 * enhancement: stepping is now once again supported on the SPARC and
886 MIPS platforms. (It is also now more likely to work on CheneyGC
888 * enhancement: sb-sprof can now also track and report accurate call
890 * bug fixes: the treatment of non-standard subclasses of
891 SB-MOP:SPECIALIZER is more correct.
892 * incompatible change: PURIFY no longer copies the data from the
893 dynamic space into the static and read-only spaces on platforms
894 that use the generational garbage collector
895 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
897 * bug fix: GC race condition occasionally resulting in crashes with
898 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
900 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
902 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
903 a STYLE-WARNING to be signalled (regression from 1.0.4.)
904 * bug fix: an asynchronous interrupt could previously leave the
905 system running with GC inhibited.
906 * bug fix: a DECLARE form evaluated at top-level now causes an error
907 rather than silently (or verbosely) returning NIL.
908 * bug fix: trying to dissassemble functions compiled at high DEBUG could
909 result in a error being signalled due to source form lookup errors.
910 (reported by Peter Graves)
912 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
913 * incompatible change: removed writer methods for host-ent-name,
914 host-ent-addresses -- changing the values did not update the DNS
916 * minor incompatible change: changed experimental JOIN-THREAD interface
917 * documentation: the manual now lists reader and writer methods
918 in class slot documentation sections. (thanks to Richard M Kreuter)
919 * documentation: unwinding from asyncronous events has been
920 documented as unsafe.
921 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
922 in multithreaded application code.
923 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
924 platforms. (thanks to James Anderson for the optimization hint)
925 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
927 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
928 variants no longer cons.
929 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
930 their NOT- variants no longer cons.
931 * optimization: EQUAL hash tables no longer use SXHASH for objects
932 of all data types, but instead use an EQL hash for types for which
933 EQUAL is the same as EQL.
934 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
935 are significantly faster.
936 * optimization: new STRING-OUTPUT-STREAM implementation is much
937 faster for large outputs and conses 30% less on average.
938 * enhancement: XREF information is now collected for references made
939 to global variables using SYMBOL-VALUE and a constant argument.
940 * enhancement: SIGINT now causes a specific condition
941 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
942 * bug fix: FILE-POSITION used to signal an error in some cases where
943 ANSI requires it to return NIL.
944 * bug fix: ADJUST-ARRAY is now interrupt-safe.
945 * bug fix: adding and removing fd-handlers is now interrupt-safe.
946 * bug fix: inlined calls to C now ensure 16byte stack alignment on
948 * bug fix: bad type declaration in the CLOS implementation has
949 been fixed. (reported by James Anderson)
950 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
951 (thanks to Lutz Euler)
952 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
953 required by the CLHS. (reported by Samium Gromoff)
954 * bug fix: dead unbound variable references now signal an error.
955 (reported by Marco Monteiro)
956 * bug fix: / with an unused value was being deleted in safe code.
957 (thanks to Marco Monteiro and Kevin Reid)
958 * bug fix: number of characters that can be written onto a single
959 line in a file is unlimited.
960 * bug fix: some GC deadlocks caused by asynchronous interrupts have
961 been fixed by inhibiting interrupts for when GC is disabled.
962 * bug fix: some interrupt safety issues with GETHASH, (SETF
963 GETHASH), CLRHASH and REMHASH have been fixed.
964 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
965 specifier no longer causes infinite recursion.
966 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
967 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
968 * bug fix: an &environment argument with the correct variable information
969 is passed to macros that are expanded during byte compilation.
970 (reported by Samium Gromoff)
971 * bug fix: a logic error was causing the hash function for bignums to
972 have excessive amounts of collisions. (reported by Faré Rideau)
973 * bug fix: modifying the contents of an array could change the return
974 value of SXHASH on that array, which is only allowed for strings
975 and bit vectors. (bug introduced in 0.9.16)
976 * improvement: the x86-64/darwin port now passes all tests (except
977 for the debugger tests) but should still be considered
978 experimental until this is fixed.
979 * improvement: a style-warning is signaled for CASE (etc) clauses with
980 duplicate keys. (thanks to Kevin Reid)
981 * improvement: macos/x86 and macos/x86-64 now use mach exception
982 handlers for dealing with illegal instructions (for trapping and
983 error handling) and memory protection violations (for GC).
985 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
986 * new platform: experimental support for x86-64/darwin (MacOS).
987 * incompatible change: the thread-safe (on most platforms) getaddrinfo
988 and getnameinfo sockets functions are used instead of gethostbyaddr
989 and gethostbyname, on platforms where the newer functions are available.
990 As a result, the ALIASES field of HOST-ENT will always be NIL on these
992 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
993 don't print the startup banner, but behave as if --noinform was passed
994 as a command line argument. (thanks to Kevin Reid)
995 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
996 * optimization: code using alien values with undeclared types is much faster.
997 * optimization: the compiler is now able to open code SEARCH in more cases.
998 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
999 * bug fix: using standardized COMMON-LISP special variables as loop
1000 variables no longer signals bogus package lock violations. (reported
1002 * bug fix: declaring local loop variables to be of a range-limited type
1003 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
1004 (reported by Andras Simon)
1005 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
1006 bugs remain on x86-64.)
1007 * bug fix: the #= and ## reader macros now interact reasonably with
1008 funcallable instances.
1009 * bug fix: type-checks for function arguments were compiled using the
1010 compiler policy settings of the wrong lexical environment (in 1.0.2
1012 * bug fix: SHADOW accepts characters as string designators, as required
1013 by the spec (thanks to Eric Marsden)
1014 * bug fix: fixed GC safety issues when foreign functions are called with
1015 non-base strings as arguments
1016 * bug fix: more consistent error output for fatal-compiler-warnings, like
1018 * bug fix: functions with proper names were showing up as NIL in
1019 backtraces (reported by Edi Weitz, regression in 1.0.2)
1021 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
1022 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
1023 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
1024 produce more efficient code on the x86 if the compiler can determine
1025 that (+ INDEX <constant>) does not require a bounds check and FOO
1026 has an element type at least 8 bits wide.
1027 * bug fix: references to undefined variables are handled the same way
1028 in toplevel forms as in the normal compiler
1029 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
1031 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
1032 (thanks to Jon Buller)
1033 * improvement: faster compilation times for complex functions
1034 * improvement: added readlink support to SB-POSIX (thanks to Richard
1037 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
1038 * improvement: experimental support for mach exception handling on
1039 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
1041 * improvement: support for GBK external format.
1042 (thanks to Chun Tian (binghe))
1043 * improvement: the debugger now displays variables that have been closed
1044 over, in code compiled with (DEBUG 2) or higher
1045 * improvement: support for executable cores on NetBSD (thanks to
1047 * new feature: added a RESTART-FRAME debugger command
1048 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
1049 be used to provide an implementation for FILE-POSITION on Gray streams
1050 (thanks to Eric Marsden)
1051 * new feature: add syslog, openlog and closelog support to SB-POSIX
1052 (thanks to Richard Kreuter)
1053 * optimization: the function call overhead in code compiled with
1054 a high DEBUG optimization setting is significantly
1055 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
1056 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
1057 (thanks to Tony Martinez)
1058 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
1059 evaluated once (reported by Kevin Reid)
1060 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
1061 works again (thanks to Kilian Sprotte)
1062 * bug fix: an error is signaled for tagbodies with duplicate tags
1063 (thanks to Stephen Wilson)
1064 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
1065 * bug fix: Win32 port can now handle foreign code unwinding Lisp
1066 stack frames from alien callbacks.
1067 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
1068 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
1069 65k pages (thanks to David Woodhouse)
1070 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
1072 changes in sbcl-1.0.1 relative to sbcl-1.0:
1073 * new platform: FreeBSD/x86-64, including support for threading.
1074 * new feature: the compiler stores cross-referencing information
1075 about function calls (who-calls), macroexpansion (who-macroexpands)
1076 and special variables (who-binds, who-sets, who-references) for code
1077 compiled with (< SPACE 3). This information is available through the
1078 sb-introspect contrib.
1079 * new feature: users may subclass SEQUENCE, and have instances of
1080 these classes interoperate with standard Common Lisp functions if
1081 a number of methods are defined. (This feature is experimental
1082 and the interface subject to change based on feedback from SBCL
1083 users and the general community)
1084 * improvement: sb-sprof traces call stacks to an arbitrary depth on
1085 x86/x86-64, rather than the previous fixed depth of 8
1086 * improvement: another pthread back-end of mutex "pthread-futex".
1087 * bug fix: non-ascii command-line arguments are processed correctly
1088 (thanks to Yaroslav Kavenchuk)
1089 * bug fix: non-required arguments were not passed correctly when a method
1090 defined using DEFMETHOD was called from a mop-generated method using
1091 CALL-NEXT-METHOD (reported by Pascal Costanza)
1092 * bug fix: recursion is now permitted in accessors through
1093 SLOT-UNBOUND. (reported by Pascal Costanza)
1094 * bug fix: an error was signaled at startup if the HOME environment
1095 variable was defined, but had an empty value (reported by Peter Van Eynde)
1096 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
1097 signal a PROGRAM-ERROR, not a BUG.
1098 * bug fix: SB-POSIX:READDIR works when built with large file support.
1099 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
1101 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
1102 for better automatic indentation support. (thanks to Matt Pillsbury)
1103 * optimization: loading generic functions no longer takes O(n^2) time,
1104 proportional to the amount of methods in the generic function
1105 (reported by Todd Sabin and Jeremy Brown)
1106 * optimization: the FIND and POSITION family of sequence functions
1107 are significantly faster on arrays whose element types have been
1109 * improvements to the Windows port:
1110 ** Intermittent heap corruption problems have been fixed. (thanks
1111 to Alastair Bridgewater)
1112 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
1114 ** Lisp is able to unwind foreign exception frames from alien
1115 callbacks. (thanks to Alastair Bridgewater)
1117 changes in sbcl-1.0 relative to sbcl-0.9.18:
1118 * improvement: experimental support for threading on FreeBSD/x86.
1119 (thanks to NIIMI Satoshi)
1120 * improvement: runtime option --dynamic-space-size can be used
1121 to set the size of the dynamic space reserved on startup.
1122 * improvement: floating point modes in effect are now saved in
1123 core, and restored on startup.
1124 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
1125 startup, not time since first call to GET-INTERNAL-REAL-TIME.
1126 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
1127 threads are not running after *SAVE-HOOKS* have run.
1128 * improvement: writes to CLOS instance slots are type-checked in code
1129 compiled with (SAFETY 3)
1130 * improvement: floating-point exception handling on FreeBSD (thanks to
1132 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
1133 (thanks to Zach Beane)
1134 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
1136 * improvement: added support for the Shift-JIS external format.
1137 (contributed by NIIMI Satoshi)
1138 * improvement: callbacks are supported on Linux/PPC. (thanks to
1140 * bug fix: compiler bug triggered by a (non-standard) VALUES
1141 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1142 * bug fix: file compiler no longer confuses validated and already
1143 dumped structurres. (reported by Kaersten Poeck)
1144 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1145 fill-pointer signals a type-error as required. (thanks to
1147 * bug fix: disassemly of funcallable instances works.
1148 * bug fix: single stepping on PPC.
1149 * bug fix: fix thread-safety problems in the type system (generally
1150 manifesting as nonsensical errors like "STRING is a bad type specifier
1151 for sequences" or "The value 1 is not of type FIXNUM")
1152 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1153 (reported by Josip Gracin)
1154 * bug fix: an error is signaled for attempts to displace arrays with
1155 incompatible element types (thanks to Mario Mommer)
1156 * bug fix: more correct handling of wide characters in debug info
1157 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1158 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1159 and don't cause extra consing
1160 * optimization: MAP and MAP-INTO are significantly faster on vectors
1161 whose elements types have been declared.
1162 * Improvements to SB-SPROF:
1163 ** Support for allocation profiling
1164 ** Reduced profiling overhead, especially for long profiling runs
1165 * Improvements to the Windows port:
1166 ** floating point exceptions are now reported correctly.
1167 ** stack exhaustion detection works partially.
1168 ** more accurate GET-INTERNAL-REAL-TIME.
1169 ** better RUN-PROGRAM behaviour in case of errors in the child
1171 ** PROBE-FILE now simplifies pathnames correctly.
1172 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1173 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1174 1st 1970 (although time timezone data will be unreliable).
1175 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1177 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1178 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1179 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1181 * bug fix: two potential GC deadlocks affecting threaded builds.
1182 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1183 returns T (reported by Anton Kazennikov)
1184 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1185 with non-variable places
1186 * bug fix: remove a race condition in the setting of
1187 funcallable-instance functions, this should make threaded CLOS
1188 code more stable against memory faults.
1189 * bug fix: corruption of specials when unbinding is interrupted by an
1190 asynchronous unwind (reported by Hannu Koivisto)
1191 * improvement: the debugger will now also display local variables that
1192 are only used once, for code compiled with a DEBUG optimization quality
1195 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1196 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1197 * incompatible change: External-format support for FFI calls. The
1198 SB-ALIEN:C-STRING no longer implies an ASCII
1199 external-format. Instead the string is subject to external-format
1200 conversion. Additionally return values of type SB-ALIEN:C-STRING
1201 are by default of type (SIMPLE-ARRAY CHARACTER), not
1202 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1203 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1204 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1205 * incompatible change: SB-EXT package no longer contains the
1206 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1207 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1209 * incompatible change: the single-stepper is no longer available
1210 on Alpha, Mips and Sparc platforms.
1211 * minor incompatible change: the direct superclasses of
1212 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1213 not (STANDARD-OBJECT FUNCTION). This makes the
1214 class-precedence-lists of GENERIC-FUNCTION and
1215 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1217 * minor incompatible change: the default stream external format on
1218 non-unicode SBCL is no longer detected from the locale, but is always
1220 * new feature: Add a version of evaluator that uses an interpreter instead
1221 of the compiler. EVAL still uses the compiler by default, to switch it
1222 to use the interpreter, set the value of the variable
1223 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1224 * minor incompatible change: the single-stepper REPL has been merged
1225 with the normal debugger (see the "Stepping" heading of the debugger help
1226 for more details). The debugger command STEP will no longer switch
1227 to the single-stepper REPL.
1228 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1229 for a type now works.
1230 * bug fix: loading of default sysinit file works. (thanks to Leonid
1232 * bug fix: better detection of circularities in the file-compiler.
1233 (reported by Marco Monteiro)
1234 * bug fix: the CL pathname functions now work with files that have
1235 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1236 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1237 systems with Unix98 pty semantics.
1238 * bug fix: ASDF-INSTALL will now work with bsd tar.
1239 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1241 * bug fix: timers expiring in dead threads no longer cause a
1242 type-error (reported by Paul "Nonny Mouse").
1243 * bug fix: thanks to more lightweight single-stepper instrumentation,
1244 code compiled with (DEBUG 3) will compile and execute significantly faster,
1245 and will have more accurate type-inferencing than before
1246 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1247 whose bindings are modified
1248 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1249 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1250 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1251 CL:SOFTWARE-VERSION with proper version
1253 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1254 * feature: implemented the READER-METHOD-CLASS and
1255 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1256 as specified by AMOP.
1257 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1259 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1260 (thanks to Lutz Euler)
1261 * optimization: hashing of general arrays and vectors has been
1262 improved. (reported by Any Fingerhut)
1263 * enhancement: SB-INTROSPECT is now able to find definitions of
1264 profiled functions. (thanks to Troels Henriksen)
1265 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1266 as well, allowing compiler-macros for SETF-functions to expand.
1267 * enhancement: step-instrumentation no longer wraps known
1268 single-value functions with multiple-value context, allowing
1269 better type inference.
1270 * fixed bug #337: use of MAKE-METHOD in method combination now works
1271 even in the presence of user-defined method classes. (reported by
1272 Bruno Haible and Pascal Costanza)
1273 * fixed bug #339(c): if there are applicable methods not part of any
1274 long-form method-combination group, call INVALID-METHOD-ERROR.
1275 (reported by Bruno Haible)
1276 * fixed bug #361: the :FUNCTION initarg in the protocol for
1277 initialization of methods can now be used to override
1278 internally-produced optimized functions. (reported by Bruno
1280 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1281 system-provided lambda expression no longer cause warnings about
1282 unbound #:|pv-table| symbols.
1283 * bug fix: improved the handling of type declarations and the
1284 detection of violations for keyword arguments with non-constant
1286 * bug fix: potentially erronous calls to PATHNAME and
1287 MERGE-PATHNAMES were being flushed in some cases.
1288 (reported by Richard Kreuter)
1289 * bug fix: compiled calls to TYPEP were mishandling obsolete
1290 instances. (reported by James Bielman and Attila Lendvai)
1291 * bug fix: format strings with ~> without matching ~< no longer
1292 trigger an AVER, but signal an understandable error instead.
1293 (reported by Antonio Martinez)
1294 * bug fix: specifying an output-file without a pathname-type for
1295 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1296 of input-file instead of "fasl". (reported by Robert Dodier)
1297 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1298 (reported by James Y Knight).
1299 * bug fix: compiler-macros-function did not consider the environment
1300 argument for shadowing by local functions.
1301 * bug fix: compiler-macros expansion was inhibited by local INLINE
1303 * bug fix: inline expansions of known functions were subject to
1304 step-instrumentation in high debug policies, leading to problems
1305 with type-inference.
1306 * bug fix: compiler failed to differentiate between different CONS
1307 types in some cases.
1308 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1309 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1310 * bug fix: #368: incorrect use of expressed vs. upgraded array
1312 * bug fix: #306a: more precise unions of array types.
1313 * thread-safety improvements:
1314 ** CONDITION-WAIT could return early on Linux, if the thread was
1315 interrupted and subsequently continued with SIGCONT.
1316 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1318 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1319 * added support for the ucs-2 external format. (contributed by Ivan
1321 * minor incompatible change: pretty printing of objects of type
1322 (cons symbol) is, in the default pprint-dispatch-table, now
1323 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1325 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1326 called later in a class's lifetime, possibly as late as when the
1327 first instance of the class is created. Previously,
1328 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1329 class became finalizeable.
1330 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1331 for multibyte external-format streams. (thanks to "vbzoli")
1332 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1333 with :ALLOCATION :CLASS inherited from superclasses of the
1335 * fixed bug: anonymous classes can now be created using the :NAME
1336 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1337 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1338 * fixed bug: core-files saved with :EXECUTABLE T can again be
1339 executed when SBCL_HOME isn't set. (reported by James Knight)
1340 * fixed bug: toplevel LOCALLY forms with declarations could
1341 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1342 * fixed bug: printing from several different threads using different
1343 values of *print-case* could cause invalid output, due to
1344 some internal special variables of the printer not being bound
1345 thread-locally (reported by Max Mikhanosha)
1346 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1347 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1348 (reported by Pascal Costanza)
1349 * fixed bug: FINALIZE-INHERITANCE is called from
1350 REINITIALIZE-INSTANCE on classes when the class has previously
1351 been finalized, as required by AMOP.
1352 * minor code generation optimizations:
1353 ** better register allocation in CLOS dispatching functions
1354 ** overflow detection when coercing signed bytes to fixnums on x86-64
1355 ** is now implemented with one IMUL instruction instead of three shifts
1356 ** more efficient bit-vector access on x86 and x86-64
1357 ** more efficient access to raw structure slots on x86 and x86-64
1358 * fixed some bugs revealed by Paul Dietz' test suite:
1359 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1360 return its argument.
1362 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1363 * feature: thread support on Solaris/x86, and experimental thread support
1365 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1367 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1368 default method for SLOT-UNBOUND.
1369 * minor incompatible change: prevent the user from specializing the
1370 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1371 somewhat counter to the intent of the protocol, I (CSR) think, and
1372 additionally it just doesn't work in SBCL as currently
1373 implemented, thanks to optimizations (that are always valid for
1374 the other three SLOT-VALUEish functions, but not for the setter).
1375 * minor incompatible change: the :SB-LDB feature is now enabled by
1376 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1377 the low-level debugger.
1378 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1379 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1380 down to an SB-SYS:FD-STREAM, allowing direct communication in
1381 more cases. Temporary files and pipes are still used as a fallback
1383 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1384 on the non-reentrant C functions localtime(3) and gmtime(3).
1385 * bug fix: global optimization policy was not visible in LOCALLY and
1387 * bug fix: class objects can be used as specializers in methods.
1388 (reported by Pascal Costanza)
1389 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1390 with a type error. (reported by blitz_ on #lisp)
1391 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1392 failed with a type error. (reported by Pascal Bourguignon)
1393 * bug fix: merging pathnames against defaults with :DIRECTORY
1394 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1395 (reported by James Y Knight)
1396 * bug fix: saving large (>2GB) cores on x86-64 now works
1397 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1398 constant index and a (simple-array (signed-byte 32)) array
1399 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1400 error (patch by Robert J. Macomber)
1401 * feature: TIME now displays the amount of run-time spent in GC
1402 * bug fix: The debugger now does a better job of respecting
1403 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1404 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1405 workaround for bug 403.)
1406 * bug fix: floating point exception handling now works on FreeBSD
1407 (thanks to NIIMI Satoshi)
1408 * fixed some bugs revealed by Paul Dietz' test suite:
1409 ** MISC.641: LET-conversion were not supposed to work in late
1411 * improvements to the Win32/x86 port:
1412 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1413 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1415 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1416 * new feature: source path information is generated for macro-expansion
1417 errors for use in IDE's like Slime (thanks to Helmut Eller)
1418 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1419 * bug fix: compiler does not loop forever on an invalid type in
1421 * improvement: compilation of most CLOS applications is significantly
1423 * optimization: added a limited bytecode compiler for simple toplevel
1424 forms, speeding up compilation and FASL loading
1425 * bug fix: the statistical profiler now properly distinguishes anonymous
1428 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1429 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1430 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1431 into subclasses of the finalized class. Now user calls to
1432 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1433 system return before any subclasses are finalized.
1434 * minor incompatible change: The reader no longer ignores errors
1435 regarding non-existent packages in #+ and #- feature tests.
1436 * new feature: command line options --no-sysinit, --no-userinit to
1437 inhibit loading the corresponding init files
1438 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1439 for sending data through UDP sockets (thanks to François-René Rideau)
1440 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1441 error is signalled instead (thanks to François-René Rideau)
1442 * minor incompatible change: Error signalling behaviour of lexical
1443 operations violating package locks has changed slightly. Refer to
1444 documentation on package locks for details.
1445 * enhancement: EVAL can process IF-expressions without resorting to the
1447 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1448 constant-folded properly due to missing out-of-line functions.
1449 (reported by Mika Pihlajamäki)
1450 * bug fix: LISTEN sometimes returned T even in cases where no data was
1451 immediately available from the stream
1452 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1453 were reversed. (reported by Levente Mészáros)
1454 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1455 when given a vector displaced to an adjustable array. (reported
1457 * bug fix: derivation of float boundaries from numbers outside the
1458 appropriate float range (reported by John Wiseman)
1459 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1461 * bug fix: precomputing cache entries for generic functions with
1462 some subclasses of specializers as yet invalid does not attempt to
1463 fill a cache line with a negative offset. (reported by Levente
1465 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1466 allow condition class objects as arguments to DOCUMENTATION and
1467 (SETF DOCUMENTATION); only find and set documentation for
1468 structure names for the STRUCTURE doc-type. (suggested by Gary
1470 * improvements to the Win32/x86 port:
1471 ** (user-homedir-pathname) and default initialization file
1472 locations now know about the user's "Documents and Settings"
1473 directory (thanks to Yaroslav Kavenchuk)
1474 ** run-program is implemented (thanks to Mike Thomas)
1475 ** sockets support (thanks to Timothy Ritchey)
1476 ** better backtrace support (thanks to Alastair Bridgewater)
1477 ** sb-grovel supported
1478 ** asdf-install and sb-posix work somewhat
1479 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1480 * improvements to the Solaris/x86 port:
1481 ** works on Solaris 11/Solaris Express
1482 ** floating-point exception handling support
1483 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1484 * fixed some bugs revealed by Paul Dietz' test suite:
1485 ** REMOVE-METHOD returns its generic function argument even when
1486 the method is not one of the generic functions' methods.
1487 ** objects declared as MEMBER types can be admissible arguments to
1488 structure accessors.
1489 ** printing characters should simply be printed by the FORMAT ~:C
1491 ** compiler failure when compiling functions with hairy constant
1492 defaults for optional parameters.
1493 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1494 function, which is already optimized.
1496 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1497 * new platform: experimental support for SBCL x86/Darwin, including
1498 MacOS X 10.4.5 on Intel.
1499 * new feature: Unicode character names are now known to the system
1500 (through CHAR-NAME and NAME-CHAR).
1501 * new feature: the filesystem location of SBCL's core file is
1502 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1503 * minor incompatible change: the contrib modules SB-POSIX and
1504 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1505 this change is to make it easier to distribute
1506 location-independent binaries.
1507 * bug fix: as implied by AMOP, standardized classes no longer have
1508 slots named by external symbols of public packages. (reported by
1510 * optimization: calling functions via constant symbols -- as in
1511 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1512 via the function object as in (FUNCALL #'FOO).
1513 * optimization: on x86 and x86-64, the calling convention for the
1514 case of dealing with an unknown number of values has been altered
1515 to be friendlier to the prediction heuristics implemented,
1516 particularly with reference to CALL and RET pairing. (thanks to
1517 Alastair Bridgewater)
1518 * optimization: on x86, the code for access to thread-local symbol
1519 values and binding/unbinding of thread-local symbols is smaller.
1520 (thanks to Alastair Bridgewater)
1521 * enhancement: CONSTANTP is now able to determine constantness of
1522 more complex forms, including calls to constant-foldable standardized
1523 functions and some special forms beyond QUOTE.
1524 * fixed bug: occasional GC crashes on Solaris/x86
1525 * optimization: x86-64 supports stack allocation of results of simple
1526 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1527 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1528 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1529 (thanks to James Knight)
1530 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1531 by the standard (thanks to Stephen Compall)
1533 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1534 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1535 be used for bundling the runtime and the core file into one
1536 executable binary. This feature is not currently supported on all SBCL
1537 platforms. (thanks to James Bielman and NIIMI Satoshi)
1538 * new feature: a generational or ephemeral garbage collector is now
1539 the default on the PowerPC platform (both Linux and Darwin). The
1540 old Cheney (stop and copy) collector is a build-time option.
1541 (thanks to Cyrus Harmon, after Raymond Toy)
1542 * minor incompatible change: the method by which SBCL finds its
1543 contributed modules has changed; it no longer relies on symbolic
1544 links from an $SBCL_HOME/systems directory, but searches directly
1545 in the subdirectories of $SBCL_HOME.
1546 * enhancement: the dynamic heap size on the Linux/PPC platform is
1547 markedly larger, even using the older Cheney garbage collector.
1548 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1549 certain complicated string types. (reported by Gary King)
1550 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1551 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1552 * fixed bug: DOCUMENTATION on structures no longer signals an
1553 error if no structure type of the right name exists. (reported by
1555 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1556 applicable non-standard primary or :AROUND method on
1557 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1558 no longer cause unbound variable errors. (reported by Kevin Reid)
1559 * fixed bug: metacircle resolution in cases where methods have slots
1560 added before the slots from STANDARD-METHOD. (reported by Jean
1562 * fixed bug: the Power PC assembly code for calling into lisp
1563 sometimes computed its return address wrongly (depending on the
1564 15th bit of the address, fixed at link time). This used to
1565 manifest itself in a segmentation violation while building PCL.
1566 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1567 many others over the years)
1568 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1569 the stack pointer (treated as a signed integer) is negative.
1570 (thanks to Peter van Eynde)
1572 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1573 * new platform: experimental support for the Windows operating
1574 system has been added. (thanks to Alastair Bridgewater)
1575 * fixed several bugs in and robustified the PPC FFI (including
1576 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1577 * bug fix: allow non-simple string symbol names (reported by Paul
1579 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1581 * bug fix: saving a core corrupted callbacks on x86/x86-64
1582 * bug fix: closed a loophole in metacircularity detection and
1583 grounding in the PCL implementation of CLOS.
1584 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1585 (reported by tomppa on #lisp)
1586 * bug fix: clear the direction flag on Lisp -> C transitions, as
1587 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1588 (reported by Andrej Grozin and Hendrik Maryns)
1589 * optimization: major improvements to GC efficiency on GENCGC platforms
1590 * optimization: faster implementation of EQUAL
1591 * optimization: emit more efficient opcodes for some common
1592 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1593 * optimization: several other minor code-generation improvements on x86-64
1594 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1596 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1597 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1598 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1599 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1600 by ANSI as a generic function, and for consistency (SETF
1601 GENERIC-FUNCTION-NAME) is treated likewise.
1602 * fixed bug #233.b: make constraint propagation notice when a variable
1603 value is changed after it is referenced but before it is used
1604 * fixed bug #296: no more arbitrary behaviour from filenames with
1605 odd characters as --load arguments. (reported by Adam Warner)
1606 * fixed bug #390: :CHARACTER-SET pathname components now work as
1607 expected. (reported by Tim Daly Jr)
1608 * fixed bug #391: complicated :TYPE intersections in slot
1609 definitions no longer cause an error in PCL internals.
1610 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1611 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1612 returning the number of octets which would be written to the
1613 file-stream. (thanks to Robert J. Macomber)
1614 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1615 arguments to the internal stream functions. (thanks to David
1617 * bug fix: stack exhaustion is now handled on the x86/SunOS
1618 platform. (thanks to Daisuke Homma)
1619 * bug fix: more accurate ROOM results on GENCGC platforms
1620 * optimization: improved type inference for arithmetic-for
1621 index variables in LOOP
1622 * optimization: faster floating-point SQRT on x86-64
1623 * fixed some bugs revealed by Paul Dietz' test suite:
1624 ** DOCUMENTATION returns NIL instead of "" for method combinations
1625 that don't have a docstring
1627 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1628 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1629 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1630 therefore are not customizeable by user code (as seems to be at
1631 least permitted and maybe required by AMOP). As a consolation,
1632 however, the SBCL implementation of these functions now calls
1633 REINITIALIZE-INSTANCE as specified by AMOP.
1634 * bug fix: slot-definition documentation is propagated as per ANSI
1635 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1636 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1637 argument, as do eslotd initialization methods. (from Pascal
1638 Costanza's "Closer" project)
1639 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1640 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1642 * bug fix: it is now possible to have more than one subclass of
1643 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1644 (reported by Bruno Haible, Pascal Costanza and others)
1645 * bug fix: the dependent update protocol now works for generic
1646 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1647 and Pascal Costanza)
1648 * bug fix: condition-class instances corresponding to
1649 DEFINE-CONDITION forms are now created eagerly. (reported by
1650 Kalle Olavi Niemitalo on comp.lang.lisp)
1651 * bug fix: floating point printing is more accurate in some
1652 circumstances. (thanks to Simon Alexander)
1653 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1654 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1655 * bug fix: callbacks on OS X now preserve stack-alignment.
1656 (thanks to Cyrus Harmon)
1657 * enhancement: the x86-64 disassembler is much better at
1658 disassembling SSE instructions. (thanks to Lutz Euler)
1659 * enhancement: CHECK-TYPE on a variable now allows type-inference
1660 (thanks to Vincent Arkesteijn)
1661 * optimization: improved performance of EUC-JP external format.
1662 (thanks to NIIMI Satoshi)
1663 * optimization: performance improvements to IO on file streams of
1664 :ELEMENT-TYPE CHARACTER
1665 * optimization: much faster memory allocation on x86-64
1666 * optimization: faster garbage collections (latency and throughput)
1668 * optimization: faster bignum subtraction and division on x86 and x86-64
1669 (thanks to Lutz Euler)
1670 * optimization: more accurate type-inference for loops that use a
1671 floating point index variable or a negative step.
1673 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1674 * bug fix: add a workaround to SBCL looping infinitely at startup on
1675 Linux kernels with apparently buggy implementations of personality().
1676 (thanks to Svein Ove Aas)
1677 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1678 (thanks to David Lichteblau)
1679 * bug fix: local bindings shadow global symbol macros for
1680 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1681 * optimization: non-open coded uses of numeric comparison operators
1682 (e.g. >) no longer cons when called with more than one parameter
1683 on platforms supporting dynamic-extent allocation.
1684 * enhancement: saving cores with foreign code loaded is now supported
1685 on MIPS/Linux in addition to the previously supported platforms.
1686 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1687 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1688 * bug fix: division by zero in sb-sprof when no samples were collected
1689 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1690 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1691 (thanks to Alexander Kjeldaas)
1692 * new feature: ignoring the return values of destructive functions
1693 like NREVERSE signals a compile-time style-warning.
1694 (thanks to Kevin Reid)
1695 * new feature: the alignment of alien structure fields can be
1696 explicitly specified. (thanks to Cyrus Harmon)
1697 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1698 and dump core on SIGQUIT
1700 ** incompatible change: threads do not inherit values of specials
1701 from their parents (see manual)
1702 ** bug fix: threads stacks belonging to dead threads are freed by the
1703 next exiting thread, no need to gc to collect thread stacks anymore
1704 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1705 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1706 ** bug fix: dynamic variable and thread start related gc lossage
1707 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1709 * fixed some bugs revealed by Paul Dietz' test suite:
1710 ** SUBTYPEP is slightly more accurate on heinously complicated
1711 CONS types where some of the members have uncertain (in the
1712 NIL, NIL sense) type relationships to each other.
1714 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1715 no longer purified unless :PURIFY T is explicitly specified.
1716 ** Non-purified cores are significantly smaller than before
1718 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1719 * new feature: timers based on Zach Beane's excellent timer package
1720 * added support for the following external formats: koi8-u,
1721 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1722 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1723 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1724 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1725 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1726 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1727 * incompatible change: a threaded SBCL will no longer revert to
1728 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1729 * bug fix: interrupts are disabled until startup is complete; no
1730 more sigsegvs when receiving a signal to soon
1731 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1733 * bug fix: add a workaround for the memory randomization features in
1734 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1735 workaround will only be in effect on systems with the proc filesystem
1737 * bug fix: printing objects of type HASH-TABLE signals a
1738 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1740 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1741 have been running for over 50 days. (reported by Gilbert Baumann)
1742 * bug fix: the logic for getting names of functions gets less
1743 confused when confronded with alternate-metaclass
1744 funcallable-instances. (reported by Cyrus Harmon)
1745 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1746 consistent, even on internal alternate-metaclass objects.
1747 * bug fix: flush closure information collected by physical
1748 environment analysis prepass before the main pass. (bug reported
1750 * bug fix: compiler pack phase does not modify a hashtable, which is
1751 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1752 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1753 may be expanded. (reported by Alexander <malishev> on c.l.l)
1754 * bug fix: no more style-warnings when compiling code using LAST
1755 under high-SPEED low-SPACE settings. (reported by David Wallin
1757 * fixed some incompatibilities between SBCL's MOP and the MOP
1759 ** the METAOBJECT class is now implemented;
1760 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1761 STANDARD-OBJECT, as required;
1762 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1763 now compatible (as required by VALIDATE-SUPERCLASS); there
1764 remains a consistency requirement over the presence or absence
1765 of the FUNCTION class in the superclass list of finalized
1766 classes; see the manual for more details;
1767 ** the :AROUND method for COMPUTE-SLOTS on
1768 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1769 requested slot ordering.
1771 ** bug fix: parent thread now can be gc'ed even with a live
1773 ** bug fix: binding a special with PROGV to no value is not
1774 the same as that symbol not having been bound (thanks to
1776 ** bug fix: binding specials is thread safe (thanks to
1778 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1779 and SB-SPROF do not die with 'no handler for signal XX in
1780 interrupt_handle_now(..)' anymore
1781 ** bug fix: WITH-TIMEOUT works with multiple running threads
1782 * fixed some bugs revealed by Paul Dietz' test suite:
1783 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1784 the :method-class keyword argument.
1786 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1787 * new port: the Solaris operating system on x86 processors is now
1788 mostly supported, though some rough edges in the environment
1789 remain. (thanks to Daisuke Homma)
1790 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1791 space for its heap. (thanks to Thiemo Seufer)
1792 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1793 closures. (thanks to Thiemo Seufer)
1794 * minor incompatible change: eof selects abort in the debugger.
1795 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1796 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1798 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1799 deprecated, and will go away in a future revision of SBCL.
1800 * minor incompatible change: GC-ON and GC-OFF are no longer
1801 implemented with a counter, it does not matter how many times gc
1802 is switched on or off
1803 * bug fix: discriminating functions for generic function classes
1804 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1805 make invalid assumptions about method precedence order. (reported
1807 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1808 platforms. (thanks to Thiemo Seufer)
1809 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1810 * bug fix: correct pathname printing with printer escaping is on.
1811 (thanks to Kevin Reid)
1812 * bug fix: complex VOP definitions in "user-space" no longer trigger
1813 package locks. (reported by Zach Beane)
1814 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1815 for funcallable-instance functions, and is no different from
1817 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1818 characters in the current readtable. (reported by Nicholas Neuss)
1819 * bug fix: SUBTYPEP on various CONS types returns more a more
1820 accurate acknowledgment of its certainty.
1821 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1822 lists in some cases. This partially fixes bug 384.
1823 * flush all standard streams before prompting in the REPL and the
1825 * bug fix: signal handling and triggering gc do not conflict
1826 directly anymore, in particular a high frequency sb-sprof does
1827 not prevent gc from running
1828 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1829 approximation for timezone and DST information between the
1830 universal time epoch and the smallest negative 32-bit time_t.
1831 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1832 year 1899 so long as the encoded time is non-negative.
1833 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1834 the correct answer for non-constant rotate amounts when performing
1835 an inline 32-bit rotation.
1837 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1838 there is only one thread in the session
1839 ** bug fix: memory leak for streams created in one thread and
1840 written to in another
1841 ** bug fix: lockup when compiled with gcc4
1842 ** bug fix: race that allows the gc to be triggered when gc is
1844 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1845 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1847 ** bug fix: finished threads are gc'ed properly
1848 * fixed some bugs revealed by Paul Dietz' test suite:
1849 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1850 has arguments with a different set of applicable methods from
1851 the orignal arguments.
1852 ** The type error thrown by MAP now has a correct expected-type
1854 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1855 name a compiled function.
1856 ** Three MISC tests where a large bignum was improperly coerced to
1857 a float that couldn't represent that bignum during type
1858 derivation were fixed.
1859 ** SUBTYPEP can now handle types involving the negation of a
1860 list-form FUNCTION type.
1861 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1862 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1863 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1865 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1866 * New feature: Experimental support for bivalent streams: streams
1867 opened with :element-type :default now allow character and binary
1868 (unsigned-byte 8) I/O
1869 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1870 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1872 * bug fix: no more highly sporadic "couldn't check whether ~S is
1873 readable" when reading a stream and an interrupt hits in the middle
1874 of a select system call
1875 * compiler better recognizes complex arrays (reported by Eduardo
1877 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1878 for variables on SBCL built with linkage-tables. (reported by Luis
1880 * various error reporting improvements.
1881 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1882 (Thanks to Thiemo Seufer)
1883 * bug fix: nested reader invokations work correctly
1884 * bug fix: it is possible to have simultaneous references to foreign
1885 code and foreign data with the same name.
1887 ** added x86-64 support
1888 ** incompatible change: the threading api now works with thread
1889 objects instead of thread ids
1890 ** bug fix: threads are protected from signals and interruption when
1891 starting up or going down
1892 ** bug fix: a race where an exiting thread could lose its stack to gc
1893 ** bug fix: don't halt on infinite error in threads if possible
1894 ** fixed numerous gc deadlocks introduced in the pthread merge
1895 ** bug fix: fixed thread safety issues in read and print
1896 ** bug fix: debugger doesn't hang on session lock if interrupted at
1897 an inappropriate moment
1898 ** bug fix: run-program is now thread safe(r)
1899 ** bug fix: inner with-recursive-lock no longer releases the mutex
1900 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1901 * fixed some bugs revealed by Paul Dietz' test suite:
1902 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1903 EXPECTED-TYPE slots filled.
1904 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1906 ** MAKE-CONDITION accepts classes as type-designators.
1907 ** COMPILE may never return NIL.
1908 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1909 range before calling Unix time functions
1911 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1912 * numerous signal handling fixes to increase stability
1913 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1914 * minor incompatible change: we now correctly canonize default
1915 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1916 as per the MOP, rather than the historical (INITARG INITFUNCTION
1917 INITFORM). (reported by Bruno Haible)
1918 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1919 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1920 for more information.
1921 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1922 pathname is a directory pathname.
1923 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1924 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1926 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1927 :METACLASS argument in addition to a class name. (reported by
1928 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1929 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1930 (reported by Bruno Haible; patch from Gerd Moellmann)
1931 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1933 * bug fix: more cleanups to the floating point exception handling on
1934 x86-64 (thanks to James Knight)
1935 * bug fix: the compiler does not try to constant fold calls of
1936 COERCE and BIT-* functions when they return freshly constructed
1937 objects. (reported by Nikodemus Siivola and Paul Dietz)
1938 * optimization: improved the MIPS versions of generic (in the
1939 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1940 * optimization: direct conversion of (unsigned-byte 32) to floats on
1941 the PowerPC platform.
1942 * optimization: structure instances with raw slots now use less
1943 memory, and probably show better memory locality. (thanks to
1945 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1946 are treated as requests for stack allocation on the x86-64,
1947 Alpha-32, PPC and SPARC platforms.
1948 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1949 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1950 (thanks to Tony Martinez)
1952 ** gcing a dead thread can no longer lead to lockups
1953 ** threads block signals until they are set up properly
1954 ** errno is no longer shared by threads
1955 ** interrupt-thread restores the eflags register on x86
1956 ** fixed some lockups due to gc/thread interaction
1957 * fixed some bugs revealed by Paul Dietz' test suite:
1958 ** invalid dotted lists no longer raise a read error when
1959 *READ-SUPPRESS* is T
1960 ** signal an error if a symbol that names a declaration is used
1961 as the name of a type, or vice versa
1962 ** allow using the (declare (typespec var*)) abbreviation for
1963 (declare (type typespec var*)) with all type specifiers
1964 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1965 ** always return NIL from PPRINT-POP when OBJECT is NIL
1966 ** don't signal errors when pretty-printing malformed LABELS,
1967 FLET or MACROLET forms
1968 ** declarations in a DOLIST body are also in scope for the
1970 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1972 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1975 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1976 * fixed cross-compiler leakages that prevented building a 32-bit
1977 target with a 64-bit host compiler.
1978 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1979 opened with :IF-EXISTS :OVERWRITE.
1980 * fixed bug 281: error for an invalid qualifier in a short-form method
1981 combination method is not signalled until the faulty method is called.
1982 * bug fix: iteration variable type inferrer failed to deal with open
1983 intervals. (reported by Alan Shields)
1984 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1985 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1986 or more runtime options were provided to the sbcl binary.
1987 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1988 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1989 * optimization: there's now a fast-path for fixnum arguments in the
1990 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1991 * optimization: the code generated on x86-64 is more compact thanks
1992 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1993 * bug fix: floating-point exception handling now partly works on
1994 x86-64. (Thanks to James Knight)
1995 * improvement to the MIPS backend from Thiemo Seufer: C-style
1996 64-bit long long arguments and return values to alien functions
1998 * fixed some bugs revealed by Paul Dietz' test suite:
1999 ** the type-error signalled from WARN has a filled-in DATUM slot.
2000 ** the type-error required when a stream is not associated with
2001 a file has the stream as its datum.
2002 ** type-errors on single-floats on x86-64 no longer have
2003 :INVALID-OBJECT as the datum
2004 ** the type-errors signalled for invalid function names now have
2005 a correct expected type
2006 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
2007 for structures defined with a :TYPE. Documentation strings for
2008 typed structures are no longer immediately discarded
2009 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
2010 broadcast streams and synonym streams. FILE-LENGTH now also works
2011 on broadcast streams.
2013 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
2014 * incompatible change: the --noprogrammer option, deprecated since
2015 version 0.7.5, has been removed. Please use the equivalent
2016 --disable-debugger option instead.
2017 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
2019 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
2020 inoperational for a while now) has been completely removed.
2021 * null lexical environments are now printed as #<NULL-LEXENV>,
2022 significantly reducing the amount of clutter in typical backtraces.
2023 * documentation on weak pointers, finalization, and after GC hooks
2024 has been added to the manual.
2025 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
2026 as other specialized array types, is much faster. SUBSEQ and
2027 COPY-SEQ on such arrays have also been sped up.
2028 * optimization: EQL is now more efficient when at least other argument
2029 is known to be of type (OR FIXNUM (NOT NUMBER)).
2030 * fixed bug: the runtime is now less vulnerable to changes in the
2031 size of the SBCL object on OS X, and virtual memory is reserved for
2032 all spaces that need to be at a fixed address.
2033 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
2034 * fixed bug: finalizers and after GC hooks that cause consing are now
2036 * fixed bug: compiler error messages and summaries are now printed to
2037 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
2038 * fixed inference of the upper bound of an iteration variable.
2039 (reported by Rajat Datta).
2040 * fixed bug 211e: calling local functions with duplicated constant
2041 keyword argument no longer causes a bogus style warning about an
2043 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
2044 ftype declaration to be disregarded. (reported by Dave Roberts)
2045 * fixed bug 373: caused by erronous compilation of references to alien
2046 variables in the runtime on ppc/darwin.
2047 * fixed bug 376: CONJUGATE type deriver.
2048 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
2049 (reported by Baughn on #lisp)
2050 * a cross-compiler bug on non-x86 platforms has been identified and
2051 fixed. (thanks to Bruno Haible)
2052 * improvements to the MIPS runtime code for increased stability.
2053 (thanks to Thiemo Seufer)
2054 * increased the maximimum compact environment size to allow
2055 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
2056 * partial workaround for bug 135: don't name the function we're
2057 calling for hairy cases of EVAL, so as not to accumulate one environment
2058 entry per call to EVAL. (reported by Kevin Reid)
2059 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
2060 * optimization: type testing for non-vector arrays should be faster.
2061 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
2063 * bug fix: setting 31st element of a bit vector to zero did not work
2065 * bug fix: redefining a class definition which failed due to a
2066 previous accessor / function clash now works (but see BUGS entry
2067 #380 for more problems in this area). (thanks to Zach Beane)
2068 * the long form of DEFINE-METHOD-COMBINATION disables method group
2069 checking when given a single method group with pattern *. (thanks
2071 * on x86 compiler supports stack allocation of results of simple
2072 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
2073 * support for building and running on Mac OS X 10.4 "Tiger" has been added
2074 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
2075 * fixed some bugs related to Unicode integration:
2076 ** the restarts for recovering from input and output encoding
2077 errors only appear when there is in fact such an error to
2079 * fixed some bugs revealed by Paul Dietz' test suite:
2080 ** MISC.549 and similar: late transformation of unsafe type
2081 assertions into derived types caused unexpected code
2083 ** SCALE-FLOAT type deriver is less wrong.
2084 ** type derivers for EXP, LOG and similar functions compute result
2085 types for complex arguments better.
2086 ** (MISC.563) CONJUGATE type deriver works for very restricted
2088 ** out-of-line type testers for character strings are available.
2089 ** EQUAL compiler transform understands specialness of objects
2091 ** accessing double-floats stored on the stack now works on x86-64.
2092 ** debugger internals could sometimes create invalid lispobjs,
2093 resulting in GC crashes.
2094 ** MISC.548: type check weakening can convert required type into
2096 ** initialization forms for bindings are not in scope of free special
2099 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
2100 * incompatible change: thread support for non-NPTL systems has
2101 been removed - locking is buggy and unreliable. A threaded
2102 SBCL build will now warn at startup and refuse to create
2103 new threads, unless futex support is detected in the kernel
2104 * incompatible change: the top level REPL now has only an ABORT
2105 restart associated with it, not TOPLEVEL and ABORT as it used to.
2106 TOP and TOPLEVEL are now available as debugger commands for
2107 returning to the top level.
2108 * incompatible change: forms evaluated in the REPL now use the
2109 global optimization policy.
2110 * incompatible change: user- and system-initialization files are
2111 no longer processed with LOAD, but by READ and EVAL; hence the
2112 global optimization policy, startup package, readtable, etc,
2114 * The .fasl file format number has been incremented because of
2115 various incompatible changes.
2116 * internal entry point details and argument counts no longer appear
2117 in backtraces unless explicitly requested by setting
2118 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
2119 * built-in and standard functions no longer have names like "top
2120 level local call to FOO".
2121 * fixed bug 32: functions defined in non-null lexical environments
2122 now have more legible printed representation
2123 * fixed bug 33: functions defined in non-null lexical environemnts
2124 are now more amenable to inspection by INSPECT.
2125 * workaround for bug 354: XEPs no longer appear in backtraces unless
2126 explicitly requested.
2127 * fixed bug: receiving the signal which results from attempting to
2128 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
2129 SIGBUS on other *BSDs) on architectures where the C stack is also
2130 the Lisp stack (x86 and x86-64) from foreign code no longer leads
2131 to debugger confusion or wild instability. (reported by Cheuksan
2133 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
2134 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
2135 (reported by Lutz Euler)
2136 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
2137 compiled code. (reported by Paul Dietz)
2138 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
2139 specialized parameter to a method to something that is not TYPEP
2140 the specializer is now possible.
2141 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2142 face of package deletion.
2143 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2144 pathname. (reported by David Tolpin on c.l.l)
2145 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2146 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2147 STRING-STREAM and FILE-STREAM, respectively.
2148 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2149 than previously; in addition, its test suite is now run on build.
2150 (thanks to Luke Gorrie)
2151 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2152 * optimization: added a immediate representation for single-floats
2154 * fixed some bugs revealed by Paul Dietz' test suite:
2155 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2156 ** attempting to create a package with a colliding nickname causes
2157 correctable errors to be signalled.
2158 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2159 ** several x86-64 backend bugs related to sign-extension of immediate
2162 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2163 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2164 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2166 * fixed loading of multiply forward-referenced layouts.
2167 (thanks to Cheuksan Wang)
2168 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2169 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2170 (see COMPILE-FILE documentation for details of :PRINT option)
2171 * fixed bugs 19 and 317: fixed-format floating point printing is
2172 more accurate. This also fixes a bug reported by Adam Warner
2173 related to the ~@F format directive.
2174 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2176 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2177 dispatch table if the from-char is a dispatch macro character.
2178 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2179 by the wordsize no longer ignore the last word. (reported by Lutz
2181 * fixed bug in type checking of dynamic-extent variables. (reported
2183 * optimization: sequence traversal functions use their freedom to
2184 coerce function designators to functions.
2185 * optimization: code with many calls to CLOS methods specialized on
2186 CLOS classes has had redundant type checks removed.
2187 * optimization: added declarations to speed up operations that access
2188 the internal character database (for example STRING-UPCASE)
2189 * optimization: comparison operations between floats and sufficiently small
2190 fixnums no longer create extra rationals
2191 * fixed some bugs related to Unicode integration:
2192 ** portions of multibyte characters at the end of buffers for
2193 character-based file input are correctly transferred to the
2194 start of the buffer at the next read.
2195 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2196 passing it through to OPEN.
2197 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2198 argument given to internal calls to OPEN.
2199 ** the built-in comment readers (introduced by character sequences
2200 ";" and "#|") are more forgiving to encoding errors; they will
2201 STYLE-WARN and then attempt to resync the stream at a character
2202 boundary. (thanks to Teemu Kalvas)
2203 * fixed some bugs in the x86-64 port:
2204 ** Negative short int return values from c-calls are sign-extended
2206 ** The stack is aligned to 16-bytes for c-calls, as required by
2207 the ABI. (thanks to Cheuksan Wang)
2208 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2209 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2210 for enums. (thanks to Vincent Arkesteijn)
2211 ** Multiple small optimizations and bugfixes for floating point
2213 * fixed some bugs revealed by Paul Dietz' test suite:
2214 ** Space, Tab, Linefeed, Return and Page have the invalid
2215 secondary constituent character trait.
2216 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2218 ** WITH-INPUT-FROM-STRING should only update the index place on
2220 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2221 works more reliably.
2222 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2223 with one-dimensional arrays with fill pointers.
2224 ** TYPE-OF failed on a complex with an integer realpart and a
2226 ** compiler failure during type inference for the code of form
2227 (IF (EQL X (THE ...)) ...) (MISC.535).
2229 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2230 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2231 platform. The port supports 61-bit fixnums, large memory spaces
2232 and reloading shared object files.
2233 * enhancement: saving cores with foreign code loaded is now
2234 supported on ppc/Darwin in addition to the previously supported
2236 * enhancement: the statistical profiler now walks deeper into the
2237 call stack for more meaningful call-graphs and accrued time
2238 reports (x86/x86-64 only). It also now reports time spent in
2240 * enhancement: it is now possible to trace most individual methods
2241 of a generic function in addition to tracing the generic function
2243 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2245 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2246 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2247 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2248 produces an error. (thanks to Vincent Arkesteijn)
2249 * bug fix: NAMESTRING on pathnames with :WILD components in their
2250 directories works correctly. (thanks to Artem V. Andreev)
2251 * fixed bug 125: compiler preserves identity of closures. (reported
2253 * bug fixed: functions with &REST arguments sometimes failed with
2254 "Undefined function" when compiled with (DEBUG 3). (reported by
2256 * bug fix: overflow during compiling of setting element of a bit
2257 vector with constant index and value. (reported by Timmy Douglas)
2258 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2259 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2260 * on x86 compiler partially supports stack allocation of dynamic-extent
2262 * GO and RETURN-FROM do not check the extent of their exit points
2263 when compiled with SAFETY 0.
2264 * fixed some bugs related to Unicode integration:
2265 ** encoding and decoding errors are now much more robustly
2266 handled; it should now be possible to recover even from invalid
2267 input or output to the terminal. (thanks to Teemu Kalvas)
2268 ** provided a first cut at implementing STRING-TO-OCTETS and
2269 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2270 ** altered the SB-MD5 contributed module slightly, changing the
2271 interface just enough for it to be supportable for builds where
2272 lisp characters are not eight bits.
2273 * fixed some bugs revealed by Paul Dietz' test suite:
2274 ** the FORMATTER-generated functions for ~V[ conditionals require
2275 the correct number of arguments.
2276 ** READ-FROM-STRING returns the mandated second value when applied
2277 to displaced strings.
2278 ** the #\Rubout and #\Backspace characters are treated as invalid
2279 constituent characters by the tokenizer.
2281 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2282 * new feature: reloading changed shared object files with
2283 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2284 * new feature: references to foreign variables and functions
2285 can now be compiled and loaded before the corresponding shared
2286 object file is loaded, as long as the foreign definitions are
2287 available at runtime.
2288 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2289 supported. (thanks to Dan Debertin)
2290 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2291 just keywords. (thanks to Vincent Arkesteijn)
2292 * fixed bug #331: structure-class instances corresponding to
2293 DEFSTRUCT forms are now created eagerly.
2294 * fixed bug #345: backtraces from calls to undefined functions work
2295 on x86 as well. Related bug #61 is now also partially fixed on x86
2296 (backtraces from throws to unknown catch tags.)
2297 * bug fix: lambda-list parsing is now stricter vrt. order and number
2298 of lambda-list keywords.
2299 * bug fix: as specified by AMOP, an error is signalled if a
2300 class-option appears multiple times in a DEFCLASS form. (reported
2302 * bug fix: RANDOM can be compiled when the compiler derives the type
2303 of its numeric argument as a disjoint set of small integers.
2304 (reported by Paul Dietz)
2305 * bug fix: starting a core saved with shared objects loaded when
2306 those objects are not available no longer causes threaded SBCL to
2307 hang. (reported by Sean Ross)
2308 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2310 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2311 argument, which must always be NIL. (reported by Kalle Niemitalo)
2312 * bug fix: printing 1.0d+23 no longer results in an error.
2313 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2314 * bug fix: structure slot setters preserve evaluation order. (thanks
2316 * fixed some bugs related to Unicode integration:
2317 ** RUN-PROGRAM can allow its child to take input from a Lisp
2318 stream. (reported by Stefan Scholl)
2319 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2320 locales. (reported by Ken Causey)
2321 * fixed some bugs revealed by Paul Dietz' test suite:
2322 ** INCF, DECF and REMF evaluate their place form as specified in
2324 ** FORMATTER expands ~{ ~} iteration directives with V or #
2325 parameters correctly.
2326 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2327 consequent uses no arguments correctly.
2328 ** the system has a partial understanding of the (COMPLEX RATIO)
2330 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2333 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2334 * new feature: a build-time option (controlled by the :SB-UNICODE
2335 keyword feature, enabled by default) for building the system with
2336 support for the entire 21-bit character space defined by the
2338 * new feature: the system now has rudimentary external-format
2339 support; the primary user-visible change at this time is that
2340 characters with the high bit set (such as lower-case-e-acute) will
2341 print correctly to a terminal in a UTF-8 environment.
2342 * minor incompatible change: BASE-CHAR no longer names a class;
2343 however, CHARACTER continues to do so, as required by ANSI.
2344 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2345 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2346 has been moved to the SB-EXT package (temporarily re-exported from
2348 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2349 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2351 * fixed bug #351: better error-handlind and reporting for malformed
2353 * fixed bug #350: bignum-printing is now more memory-efficient,
2354 allowing printing of very large bignums, eg. (expt 2 10000000).
2355 (reported by Bruno Haible)
2356 * fixed bug #302: better primitive-type selection for intersection
2358 * fixed bug #308: non-graphic characters now all have names, as
2359 required. (reported by Bruno Haible)
2360 * bug fix: redefining a class with different superclasses now correctly
2361 removes it from the direct-subclasses of its previous superclasses.
2362 (reported by David Morse)
2363 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2364 new value now works. (reported by Bruno Haible)
2365 * bug fix: correct canonicalization of multiple non-standard slot
2366 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2367 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2368 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2369 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2370 now exists, an signals an error.
2371 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2372 during macroexpansion for non-iteration variables as well. (reported
2373 by Bruno Haible for CMUCL)
2374 * bug fix: Cyclic structures and unprintable objects in compiler
2375 messages no longer cause errors. (reported by Bruno Haible)
2376 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2377 and READ-FROM-STRING all now return a primary value of NIL if
2378 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2379 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2380 (reported by Bruno Haible for CMUCL)
2381 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2382 do not designate a condition. (reported by Bruno Haible for
2384 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2385 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2386 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2387 permitting resolution in favour of any of the conflicting symbols.
2388 (reported by Bruno Haible for CMUCL)
2389 * bug fix: EQUAL compiler optimizations is less aggressive on
2390 strings which can potentially compare true despite having distinct
2391 specialized array element types.
2392 * bug fix: unit enumerations can be defined without dividing by
2393 zero. (thanks to Vincent Arkesteijn)
2394 * FORMAT compile-time argument count checking has been enhanced.
2395 (report from Bruno Haible for CMUCL)
2396 * a partial workaround for the bug 262: the compiler does not try to
2397 inline-expand a local function doing RETURN-FROM from a deleted
2398 BLOCK. (thanks to Peter Denno for the bug report and to David
2399 Wragg for the simple test case)
2400 * fixed some bugs revealed by Paul Dietz' test suite:
2401 ** RENAME-PACKAGE allows all package designators as new package
2403 ** constraint propagation and lambda variable substitution are
2404 more cautious in dealing with partially deleted code.
2405 ** compiler sometimes lost reoptimization passes.
2406 ** CERROR, given a condition as condition designator, passes its
2407 remaining arguments to the continue format control without
2409 ** Case-altering FORMAT directives work correctly on non-ASCII
2411 ** The REAL type specifier handles bounds outside the floating
2412 point ranges without signalling FLOATING-POINT-OVERFLOW.
2413 ** Functions with IR1-transformations can create intercomponent
2414 references to global functions.
2415 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2417 ** FORMAT ~R treats a nil value for its first parameter correctly.
2419 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2420 * enhancement: saving cores with foreign code loaded is now
2421 supported on x86/NetBSD and sparc/Linux in addition to the previously
2422 supported platforms.
2423 * bug fix: on some platforms repeated installations caused multiple
2424 copies of HTML documentation to be installed -- should not happen
2425 any more. (reported by Stefan Scholl)
2426 * bug fix: parsing self-recursive alien record types multiple times
2427 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2428 original patch by Helmut Eller for CMUCL)
2429 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2430 (thanks to Richard Kreuter)
2431 * bug fix: defining classes whose accessors are methods on existing
2432 generic functions in other (locked) packages no longer signals
2433 bogus package lock violations. (reported by François-René Rideau)
2434 * bug fix: special variables as DEFMETHOD parameters no longer have
2435 associated bogus type declarations. (reported by David Wragg and
2437 * bug fix: read-write consistency on streams of element-type
2438 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2439 * bug fix: redefiniton of the only method of a generic function with
2440 no DEFGENERIC no longer emits a full WARNING. In addition,
2441 redefinition of generic functions with no DEFGENERIC to an
2442 incompatible lambda list now signals an error. (thanks to Zach
2444 * bug fix: DEFGENERIC now works even when there's a function of the
2445 same name in an enclosing lexical environment. (thanks to Zach
2447 * fixed compiler failure, caused by instrumenting code during
2448 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2449 * optimization: added loop analysis and improved register allocation
2450 * fixed some bugs revealed by Paul Dietz' test suite:
2451 ** POSITION on displaced vectors with non-zero displacement
2452 returns the right answer.
2453 ** (SIMPLE-STRING) is a valid type specifier for sequence
2455 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2457 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2458 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2460 ** The justification version of the FORMAT ~< directive treats
2461 non-zero minpad parameter correctly.
2463 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2464 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2465 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2466 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2467 the supported interface.
2468 * new feature: Single-stepping of code compiled with DEBUG 2 or
2469 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2470 * new feature: saving cores with foreign code loaded is now
2471 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2472 Timothy Moore's work for CMUCL)
2473 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2474 parameters to * instead of NIL if no initform is supplied.
2475 (reported by Johan Bockgaard)
2476 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2477 correctly when FUNCALL appears as the car of the form. Note:
2478 despite this FUNCALL forms are not currently subject to
2479 compiler-macro expansion. (port of Raymond Toy's fix for the
2480 same from CMUCL, reported by Johan Bockgaard)
2481 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2482 (thanks for Teemu Kalvas)
2483 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2484 case instead of terminating the iteration (thanks for Julian
2485 Squires, Sean Champ and Raymond Toy)
2486 * bug fix: incorrect expansion of defgeneric that caused a style
2487 warning. (thanks for Zach Beane)
2488 * bug fix: slot accessor effective method computation works properly
2489 for classes with multiple non-standard applicable methods on
2490 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2491 * on x86 compiler supports stack allocation of results of LIST and
2492 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2493 CMUCL implementation by Gerd Moellmann)
2494 * fixed some bugs revealed by Paul Dietz' test suite:
2495 ** FORMAT strings with both the ~<~:;~> form of the justification
2496 directive and pretty-printing directives cause an error.
2498 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2499 * incompatible change: the internal functions
2500 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2501 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2502 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2503 instead of the old functions.
2504 * new feature: on platforms where "dladdr" is available foreign
2505 function names now appear in backtraces. (based on Helmut Eller's
2507 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2508 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2510 * documentation: ASDF manual is now distributed with SBCL.
2511 * bug fix: SBCL can now load its contributed modules with REQUIRE
2512 even if the system-provided entries have been removed from
2513 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2515 * bug fix: backtraces involving undefined functions or assembly
2516 routines are more informative. (thanks to Brian Downing)
2517 * bug fix: mutually referent alien structures now work correctly.
2518 (reported by Rick Taube)
2519 * bug fix: structures defined by WITH-ALIEN can be referred to
2520 within other definitions in the same WITH-ALIEN.
2521 * bug fix: division operators (MOD, TRUNCATE and the like) with
2522 constant zero divisors and integer dividends no longer generate
2524 * bug fix: provide default methods for INPUT-STREAM-P and
2525 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2526 * bug fix: improve the detection and resolution of MOP metacycles
2527 (where effective method computation depends on the generic function
2528 having its effective method computed). (reported by Bruno Haible)
2529 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2530 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2531 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2532 * bug fix: pathnames with relative directory components can be
2533 represented relative to default pathnames.
2534 * optimization: in taking the GCD of bignums, reduce the two bignums
2535 to approximately the same size (using Euclid's algorithm) before
2536 applying the more sophisticated binary GCD. (thanks to Juho
2538 * optimization: COUNT on bitvectors now operates word-at-a-time.
2539 * optimization: ASH with a positive, but not necessarily constant,
2540 (leftwards) shift, when in a modular context, is compiled to a
2542 * fixed some bugs revealed by Paul Dietz' test suite:
2543 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2544 if the corresponding argument is NIL.
2545 ** FORMAT directives accepting a minpad parameter treat negative
2547 ** literal commas as character parameters to FORMAT directives are
2549 ** literal spaces directly after ~<Newline> directives within a
2550 format-logical-block (~:< ~@:>) do not induce :FILL-style
2551 conditional newlines.
2552 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2553 as specified: it no longer includes conditional newlines).
2554 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2556 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2557 :PER-LINE-PREFIX argument does not evaluate to a string.
2559 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2560 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2561 the manual for details; for now, package locks can be disabled by
2562 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2563 if no major problems are found then it is likely that they will be
2564 compiled in unconditionally.
2565 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2566 unsupported operators on all platforms. To load a shared library
2567 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2568 object file, link it into a shared library outside of SBCL and
2569 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2571 * bug workaround: ROOM T and the GC were not working together
2572 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2573 reporting facility aren't preserved. That reporting has been
2574 disabled, so now until and unless someone figures out how to make it
2575 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2576 * minor incompatible change: as threatened around sbcl-0.8.0, the
2577 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2578 The symbols are also exported from SB-PCL for backwards
2579 compatibility, but more so than before SB-PCL should be treated as
2580 an implementation-internal package.
2581 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2583 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2584 It is known as of this release not to work on the Alpha, however.
2585 * fixed bug #167: errors signalled due to illegal syntax in method
2586 bodies are now more legible.
2587 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2588 designators and can hence be used with TYPEP.
2589 * fixed bug #333: CHECK-TYPE now ensures that the type error
2590 signalled, if any, has the right object to be accessed by
2591 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2592 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2593 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2595 * fixed bug #334: programmatic addition of slots using specialized
2596 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2597 and :ALLOCATION :CLASS slots.
2598 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2599 float exponents. (rereported by Peter Seibel)
2600 * fixed a bug: #\Space (and other whitespace characters) are no
2601 longer considered to be macro characters in standard syntax by
2602 GET-MACRO-CHARACTER.
2603 * fixed bug: initialization of condition class metaobjects no longer
2604 causes an instance of the condition to be created. (reported by
2606 * fixed bug: it is now possible to have slots such that
2607 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2608 description is neither :INSTANCE nor :CLASS.
2609 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2610 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2611 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2612 non-standard SLOT-DEFINITION-ALLOCATION values.
2613 * fixed bugs: various race conditions handling exiting threads.
2614 CL-PPCRE's thread torture test now passes.
2615 * fixed arguably-a-bug: GC time proportional to number of threads in
2616 system even when most of them are idle
2617 * optimization: improved performance of BIT and SBIT on bit-vectors.
2618 * .fasl file incompatibility: The fasl file version number has
2619 been incremented because of changes associated with package locks.
2621 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2622 * minor incompatible change: the system no longer provides
2623 optimization hints (conditions which are TYPEP
2624 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2625 mode; these hints are still emitted when higher SPEED optimization
2627 * new contrib module: a sampling profiler (profiling by statistical
2628 sampling, rather than by instrumenting functions) is available as
2629 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2630 * the behaviour of the standard function ED is now customizeable by
2631 third parties through a hook variable: see ED's documentation
2632 string for information on the protocol.
2633 * the compiler no longer emits efficiency notes for (FUNCALL X)
2634 when the type of X is uncertain under default optimization
2636 * fixed bug 276: mutating a binding of a specialized parameter to a
2637 method to something that is not TYPEP the specializer is now
2639 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2640 exhibit the required behaviour on the x86 platform. (thanks to
2641 Peter van Eynde, Eric Marsden and Bruno Haible)
2642 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2643 even for difficult arguments. (reported by Peter Graves)
2644 * fixed bug 141a: the backquote printer now descends quoted
2646 * fixed another bug in backquote printing: no more destructive
2647 modification of the form's list structure. (reported by Brian
2649 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2650 move between its address being taken and the call to
2651 interrupt_thread, fixing a crashing race condition.
2652 * the SB-POSIX contrib implementation has been adjusted so that it
2653 no longer exhibits ridiculously poor performance when constructing
2654 instances corresponding to C structs.
2656 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2657 * minor incompatible change: the sb-grovel contrib now treats C
2658 structures as alien (in the sense of SB-ALIEN) objects rather than
2659 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2660 has implications for memory management of client code
2661 (sb-grovel-returned objects must now be manually managed) and for
2662 type safety (alien objects now have full types).
2663 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2664 used to control emission of compiler diagnostics, rather than the
2665 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2666 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2667 quality should be considered deprecated.
2668 * install.sh now installs the user manual as well
2669 * (not quite a new documentable feature, but worth considering in
2670 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2671 beginnings of a semantically meaningful condition hierarchy is
2672 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2674 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2675 designator as the defaults argument.
2676 * fixed bug: Displaced arrays whose displaced-to array has become
2677 too small now cause ARRAY-DIMENSION to signal an error, providing
2678 for safer bounds-checking. (reported by Bruno Haible)
2679 * fixed bug: DEFCLASS slot definitions with identical :READER and
2680 :WRITER names now signal a reasonable error. (reported by Thomas
2682 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2684 * fixed bug: Invalid slot specification errors now print correctly.
2685 (thanks to Zach Beane)
2686 * fixed bug 320: Shared to local slot value transfers in class
2687 redefinitions now happen corrently from superclasses as
2688 well. (reported by Bruno Haible)
2689 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2691 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2692 improper lists correctly. (reported by Bruno Haible)
2693 * fixed bug 313: source-transform for <fun-name> was erroneously
2694 applied to a call of a value of a variable with name <fun-name>.
2695 (reported by Antonio Menezes Leitao)
2696 * fixed bug 307: The obsolete instance protocol ensures that
2697 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2698 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2699 remove dead unknown-values globs from the middle of the stack.
2700 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2701 easier to resume long computations after using *BREAK-ON-SIGNALS*
2702 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2703 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2704 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2705 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2706 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2707 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2709 * fixed a (fairly theoretical) bug in string printing: if
2710 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2711 does not have array-element-type equal to the most general string
2713 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2714 function name. (thanks to Nikodemus Siivola)
2715 * fixed bug: compile-time format string checker failed on
2716 non-closed ~{. (reported by Thomas F Burdick)
2717 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2718 don't warn on legal loop code involving more than one aggregate
2719 boolean. (thanks to Nikodemus Siivola)
2720 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2721 in keyword position to MAKE-PACKAGE was not being reported as
2723 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2724 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2726 * fixed bugs 280 and 312: the checking for multiple definitions in a
2727 file is less likely to become confused by uses of inline
2729 * fixed bug: the #S reader macro performs the keyword coercion
2730 specified for slot names. (reported by Kalle Niemitalo)
2731 * fixed bug: lambda lists may contain symbols whose names start with
2732 & but are not lambda-list-keywords; their occurrence triggers a
2734 * fixed bug 321: define-method-combination argument lambda lists do
2735 not cause invalid code to be generated when &optional or &aux
2736 variables have default values. (reported by Bruno Haible)
2737 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2738 have CLOS classes; this appears also to have fixed reported
2739 instability in stack exhaustion detection.
2740 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2741 MUFFLE-WARNING when no associated restart is present is now
2743 * optimization: rearranged the expansion of various defining macros
2744 so that each expands into only one top-level form in a
2745 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2747 * optimization: used a previously little-used slot in symbols to
2748 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2750 * fixed some bugs revealed by Paul Dietz' test suite:
2751 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2753 ** Fixed an optimization bug related to inheritance of initforms
2754 from local to shared slots.
2755 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2756 ** CAST optimizer forgot to flush argument derived type.
2757 ** print/read consistency on floats is now orders of magnitude
2758 more likely. (thanks also to Bruno Haible for a similar report
2760 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2761 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2762 for objects of type REAL. Make it so.
2763 ** SXHASH is specified (infelicitously) to respect similarity,
2764 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2765 it so. (thanks to Markus Ziegler)
2766 ** on the Alpha, the compiler succeeds in compiling functions
2767 returning a known number of arguments greater than 63.
2768 ** fixed handling of invalid NIL arguments in keyword position
2770 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2772 ** provided a readably-printable representation for RANDOM-STATE
2774 ** ensured that pathnames, where they have a namestring, always
2775 print using #P"..." syntax.
2777 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2778 * Support for the forthcoming 2.0 version of the NetBSD kernel
2779 running on x86 hardware has been added. (thanks to Perry
2780 E. Metzger most immediately, and others for their past work)
2781 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2782 current release version; SBCL's OpenBSD support had been broken
2783 since about the time of OpenBSD's switch to ELF binary format.)
2784 * [placeholder for DX summary]
2785 ** user code with &REST lists declared dynamic-extent, under high
2786 speed or space and low safety and debug optimization policy.
2787 * The manual has been converted to Texinfo format and the debugger
2788 chapter from the cmucl manual has been added.
2789 * A facility has been added to extract documentation strings from
2790 sbcl and store them as Texinfo-formatted snippets for inclusion in
2791 the manual (via Texinfo's @include directive)
2792 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2793 the test case to Dave Roberts)
2794 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2795 pointers. (reported by Sean Ross)
2796 * bug fix: PROFILE output is printed nicely even for large numerical
2797 values. (thanks to Zach Beane)
2798 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2799 greater than 32 handle EOF correctly.
2800 * bug fix: on X86 an immediate argument of the IMUL instruction is
2801 correctly printed in disassembly. (thanks to Lutz Euler)
2802 * bug fix: class slots in redefined classes preserve their old
2803 values. (thanks to Bruno Haible and Nikodemus Siivola)
2804 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2805 succeeds. (reported by Marco Baringer)
2806 * fixed some bugs revealed by Paul Dietz' test suite:
2807 ** READ-SEQUENCE now works on ECHO-STREAMs.
2808 ** RATIONALIZE works more according to its specification. (thanks
2811 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2812 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2813 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2814 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2815 should matter to you only if you rebind the printer control
2816 variables and then find you want different bindings in the
2817 debugger than in the ordinary execution of your program.)
2818 * The runtime build system has been tweaked to support building
2819 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2820 assembler and linker. (thanks to Nikodemus Siivola)
2821 * Unbound, undefined, undeclared variables now trigger full
2822 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2823 more often programmer error than deliberate exploitation of undefined
2825 * optimization: the hash algorithm for strings has changed to one
2826 that is less vulnerable to spurious collisions. (thanks to Juho
2828 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2829 needless bounds checking. (thanks to Juho Snellman)
2830 * optimization: implemented multiplication as a modular
2831 (UNSIGNED-BYTE 32) operation on the PPC backend.
2832 * fixed some bugs revealed by Paul Dietz' test suite:
2833 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2835 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2836 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2837 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2839 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2841 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2843 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2844 their output stream on EOF from read.
2845 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2846 have been read to end-of-file.
2847 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2849 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2850 description of determination of which consecutive characters
2852 ** Printing the "Space" character with escaping on now yields "#\\ ",
2853 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2854 ** Reading floating-point numbers with *READ-BASE* set to a number
2855 less than 10 works correctly.
2856 ** Reading floating-point numbers with *READ-BASE* set to a number
2857 more than 10 works correctly.
2858 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2859 the readtable currently in effect.
2861 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2862 * minor incompatible change: parsing of namestrings on a physical
2863 (Unix) host has changed; numbers after the final #\. in a
2864 namestring are no longer interpreted as a version field. This is
2865 intented to be largely invisible to the user, except that the
2866 meaning of the namestring "*.*.*" has changed: it now refers to a
2867 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2868 should usually be replaced by
2869 (make-pathname :name :wild :type :wild :version :wild)
2870 with the added benefit that this is more likely to be portable.
2871 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2872 option to OPEN now signals an error if the file being opened
2873 exists; this may have an impact on existing code.
2874 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2875 signals received, on the PowerPC platforms (both Linux and
2876 Darwin). (thanks to Pierre Mai for pointing out the location of
2878 * several fixes on OS X: The system now builds and runs cleanly on
2879 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2880 (thanks to Brian Mastenbrook)
2881 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2882 second-resolution: integer multiples of 1/3600 between -24 and 24.
2883 (thanks to Vincent Arkesteijn)
2884 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2885 type when called with 1 argument; PEEK-CHAR checked type of
2886 PEEK-TYPE only after having read first character from a
2887 stream. (reported by Peter Graves)
2888 * bug fix: the garbage collector now has much better locality
2889 behaviour, and in particular no longer treats all memory as being
2890 exhausted when many small objects point to each other in a deeply
2892 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2893 recognized as being TYPEP their class.
2894 * bug fix: the PUSHNEW documentation string has been corrected.
2895 (thanks to Vincent Arkesteijn)
2896 * bug fix: defaulting of the value for the last of an atypically
2897 large number of multiple values being bound was not being
2898 performed correctly on the Alpha or PPC platforms
2899 * optimization: implemented multiplication as a modular
2900 (UNSIGNED-BYTE 32) operation on the x86 backend.
2901 * optimization: SEARCH on simple-base-strings can now be open-coded.
2902 (see also contrib/compiler-extras.lisp for inspiration for
2903 teaching the compiler about the Boyer-Moore algorithm).
2904 * value, returned by MAX (and MIN) called with several EQUALP, but
2905 not EQL, arguments now does not depend on compiler settings.
2906 * fixed some bugs revealed by Paul Dietz' test suite:
2907 ** in stack analysis liveness information is propagated from
2908 non-local entry points.
2909 ** pathwise CAST removing failed when the CAST node did not start
2911 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2912 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2914 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2915 host is already defined.
2916 ** RENAME-FILE works on streams instead of signalling an internal
2918 ** PEEK-CHAR uses the current readtable when determining whether
2919 or not a character is whitespace.
2920 ** MERGE-PATHNAMES handles the case when the pathname does not
2921 specify a name while the default-pathname specifies a version.
2922 ** Pathnames now stand a better chance of respecting print/read
2924 ** Attempting to use standardized file system operators with a
2925 pathname with invalid :DIRECTORY components signals a
2927 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2929 ** DIRECTORY on logical pathnames is more correct.
2930 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2931 signal a TYPE-ERROR if their argument is not a stream.
2932 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2933 designator argument does not designate a stream.
2934 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2935 examining the synonym.
2936 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2938 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2939 element-type larger than ([UN]SIGNED-BYTE 32).
2941 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2942 * When built with the :SB-FUTEX feature, threaded builds now take
2943 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2944 for faster/more reliable mutex and condition variable support.
2945 * Incompatible change (but one you probably shouldn't have been using
2946 anyway): the interface and code for arbitrating between multiple
2947 threads in the same user session has been redesigned.
2948 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2949 error when called without an explicit environment argument.
2950 (thanks to Nikodemus Siivola)
2951 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2952 confused about their position. (thanks to Adam Warner and Gerd
2954 * bug fix: Pretty printing backquoted forms with unquotations in the
2955 argument list position of various code constructs such as LAMBDA
2956 now works correctly. (reported by Paul Dietz)
2957 * bug fix: Pretty printing unquotations no longer loses all
2958 stream position information.
2959 * optimization: performance of string output streams is now less
2960 poor for multiple small sequence writes.
2961 * optimization: performance of CSUBTYPEP in the presence of complex
2962 expressions involving CONS and NOT many times has been improved.
2963 (reported by Paul Dietz)
2964 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2966 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2967 streams for the REPL, for the debugger, and so on. (thanks to
2969 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2970 * fixed some bugs revealed by Paul Dietz' test suite:
2971 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2972 is ignored if it is too small, rather than propagating through
2973 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2974 ** extremely complex negations of CONS types were not being
2975 sufficiently canonicalized, leading to inconsistencies in
2977 ** VALUES tranformer lost derived type.
2979 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2980 * fixed a bootstrapping bug: the build process no longer assumes
2981 that the various BOOLE-related constants have the same value in
2982 host and target lisps. (noted by Paul Dietz' test suite on an
2983 SBCL binary built from CLISP)
2984 * The system can now be dynamically linked on the MIPS platform,
2985 which enables dynamic loading of foreign code from Lisp. (thanks
2986 to Ralf Baechle for discussions on the MIPS ABI)
2987 * The system now records debugging information for its own source
2988 files in a filesystem-position-independent manner, relative to
2989 the "SYS" logical host.
2990 * fixed a compiler bug: MV-LET convertion did not check references
2991 to the "max args" entry point. (reported by Brian Downing)
2992 * tweaked disassembly notes to be less confident about proclaiming
2993 some instruction as an LRA. (thanks to Brian Downing)
2994 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2995 now each have their own history, command character, and other
2996 characteristics. (thanks to David Lichteblau)
2997 * fixed some bugs revealed by Paul Dietz' test suite:
2998 ** compiler failure in compiling LOGAND expressions including a
3000 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
3001 shift greater than 32.
3002 ** FLUSH-DEST did not mark blocks for type check regeneration.
3003 ** HANDLER-CASE failed to accept declarations in handler clauses
3004 in some circumstances.
3006 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
3007 * New code in contrib/sb-introspect (still probably not entirely
3008 stable yet) provides some support for smart Lisp development
3009 environments like SLIME.
3010 * The conditions signalled for errors occurring when loading .fasl
3011 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
3012 in a way which should help ASDF recover gracefully.
3013 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
3014 stuff has been cleaned up. If you code contrib/ stuff, this might
3015 affect you, and you can look at contrib/README, contrib/STANDARDS,
3016 and/or the 0.8.4.27 diff to check.
3017 * In full calls the compiler now does not generate checks for declared
3018 argument types for all arguments.
3019 * various threading fixes
3020 ** and some experimental patches which didn't make it into
3021 the main tree for this release, but which are shipped in
3022 contrib/experimental-thread.patch as a possible fix for some
3023 failures (deadlock, spinning...) in GC-intensive multithreaded
3025 * fixed PPC build problem (source code incompatibility of different
3026 library versions): added offsetof() hackery which attempts to divine
3027 where glibc maintainers put uc_mcontext today
3028 * fixed bug 282: compiler does not trust type assertions while passing
3029 arguments to a full call.
3030 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
3031 &OPTIONAL VALUES type parameter.
3032 * fix bug 214: algorithm for noting rejected templates is now more
3033 similar to that of template seletion. (also reported by rydis on
3035 * fixed bug 141b: printing backquoted information readably and prettily
3036 inserts a space where necessary.
3037 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
3038 (CHARACTER 10) are now reported as errors, rather than propagated
3039 as unknown types. (reported by piso on #lisp)
3040 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
3041 with values NIL and :ERROR. (thanks to Milan Zamazal)
3042 * fixed bug 191c: CLOS now does proper keyword argument checking as
3043 described in CLHS 7.6.5 and 7.6.5.1.
3044 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
3045 longer raise an error; further, using a list as a for-as-arithmetic
3046 counter now raises a meaningful error.
3047 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
3048 now understood by sequence creation functions such as MAKE-SEQUENCE
3050 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
3051 a STRING-INPUT-STREAM.
3052 * compiler enhancement: SIGNUM is now better able to derive the type
3054 * type declarations inside WITH-SLOTS are checked. (reported by
3056 * fixed some bugs revealed by Paul Dietz' test suite:
3057 ** incorrect optimization of TRUNCATE for a positive first
3058 argument and negative second.
3059 ** compiler failure in let-convertion during flushing dead code.
3060 ** compiler failure while deriving type of TRUNCATE on an
3061 interval, containing 0.
3062 ** ASH of a negative bignum by a negative bignum count now returns
3064 ** intersection of CONS types now canonicalizes properly, fixing
3065 inconsistencies in SUBTYPEP.
3067 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
3068 * incompatible change: The --disable-debugger command line
3069 option now clobbers the debugger at a more fundamental
3070 level, by redefining #'INVOKE-DEBUGGER instead of by
3071 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
3072 is specified by ANSI to ignore *DEBUGGER-HOOK* and
3073 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
3074 enter the debugger REPL and then suffer recursive errors
3075 because *DEBUG-IO* is also messed up in --disable-debugger mode;
3076 while under the new system, BREAK in --disable-debugger mode
3077 terminates the system just as an unhandled error would.
3078 * fixed compiler performance when processing loops with a step >1;
3079 * bug fix: DOCUMENTATION now retrieves generic function
3080 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
3081 support has been systematized, and now supports the methods
3082 specified by ANSI, along with a default method and a method for
3083 slot documentation. (reported by Nathan Froyd)
3084 * bug fix: effective methods associated with a generic function are
3085 no longer cached over a change of that generic function's method
3086 combination. (reported by Andreas Fuchs)
3087 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
3088 names a non-existent directory. (thanks to Andreas Fuchs)
3089 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
3090 the PPC platform, lead to stack corruption; this has been fixed.
3091 (reported by Rainer Joswig)
3092 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
3093 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
3094 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
3095 arithmetic in general, is now much more reliable on the PPC
3097 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
3098 platform now returns the right answer.
3099 * optimization: restored some effective method precomputation in
3100 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
3101 precomputation is now tunable.
3102 * optimization: compiler-internal data structure use has been
3103 reviewed, and changes have been made that should improve the
3104 performance of the compiler by about 20%.
3105 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
3106 simple-base-strings and simple-bit-vectors is improved.
3107 * optimization: the optimization of 32-bit logical and arithmetic
3108 functions introduced in version 0.8.3 on the x86 has been
3109 implemented on the mips, ppc and sparc platforms; an
3110 implementation of the same facility, but for 64-bit arithmetic,
3111 has been added for the alpha.
3112 * microoptimization: the compiler is better able to make use of the
3113 x86 LEA instruction for multiplication by constants.
3114 * bug fix: in some situations compiler did not report usage of
3115 generic arithmetic in (SPEED 3) policy.
3116 * bug 145b fix: compiler used wrong type specifier while converting
3117 MEMBER-types to numeric.
3118 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
3120 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
3121 index argument now works properly on the Alpha platform.
3122 * bug fix: floating point exception treatment on the Alpha platform
3124 * bug fix: FILE-POSITION works much better on string input and
3125 output streams. (thanks to Nikodemus Siivola)
3126 * bug fix: many threading/garbage collection symptoms sorted.
3127 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
3128 might be pseudo-atomic.
3129 * internal change: Stopping for GC is now done with signals not
3130 ptrace. GC is now done in whichever thread wanted it, instead of
3132 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
3134 * bug fix: result form in DO is not contained in the implicit
3136 * incompatible change: ICR structure is changed; the value part of
3137 CONTINUATION is now called LVAR; corresponding functions are
3138 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
3139 * added type deriver for ISQRT (thanks to Robert E. Brown).
3140 * bug fix: better support for loading from the command line when an
3141 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3143 * fixed some bugs revealed by Paul Dietz' test suite:
3144 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3145 ** ROUND and FROUND now give the right answer when given very
3146 small float arguments.
3147 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3149 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3150 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3151 ** LCM should return a non-negative integer.
3152 ** PARSE-INTEGER returned the index of a terminator instead of the
3153 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3154 ** PARSE-INTEGER returned an incorrect index being applied to a
3156 ** LCM with two arguments of 0 returns 0 rather than signalling
3158 ** unsigned addition of a 32-bit constant with the high bit set no
3159 longer causes an internal compiler error on the x86.
3160 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3161 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3162 with negative last argument.
3163 ** byte specifiers with zero size and position no longer cause
3164 an error during type derivation.
3165 ** bignum multiplication on the Alpha platform now returns the
3167 * porting: The system now builds on SuSE AMD64, although it still
3168 generates a 32-bit binary.
3169 * .fasl file incompatibility: The fasl file version number has
3170 been incremented (because of the changes to internal compiler
3171 data structures referred to above).
3173 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3174 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3175 more accurately, on the Darwin kernel running on PowerPC hardware.
3176 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3177 * Compiler code deletion notes now signal a condition of type
3178 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3179 an associated MUFFLE-WARNING restart.
3180 * The compiler now performs limited argument count validation of
3181 constant format strings in FORMAT, and where appropriate in ERROR,
3182 CERROR and WARN. (thanks to Gerd Moellmann)
3183 * New ASDF-INSTALL contrib can be used for automatic download and
3184 installation of third-party Lisp code from CCLAN or other sites
3186 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3187 which forces another thread to execute a function supplied by the
3189 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3190 now accept and act upon their :ELEMENT-TYPE keyword argument.
3191 (reported by Martin Atzmueller, Edi Weitz)
3192 * bug fix: FILE-POSITION now accepts position designators up to
3193 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3194 is the greater. (thanks to Patrik Nordebo)
3195 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3196 Kalle Olavi Niemitalo)
3197 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3198 printable. (reported by Eric Marsden)
3199 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3200 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3201 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3202 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3203 (thanks to Stig Sandoe)
3204 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3205 UNREAD-CHAR now correctly includes the unread character in the
3206 target sequence. (thanks to Gerd Moellmann)
3207 * bug fix in threaded builds: the system can now be suspended and
3208 resumed by shell job control with minimal disruption.
3209 * bug fixes in times and timezones >2038AD
3210 * better handling of "where is GNU make?" problem in build scripts
3211 (thanks to Nikodemus Siivola)
3212 * new optimization: inside a named function any reference to a
3213 function with the same name is considered to be a self-reference;
3214 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3215 optimization quality.
3216 * new optimization on x86: logical functions and + now have
3217 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3218 used when the result is truncated to 32 bits.
3219 * VALUES declaration is partially enabled.
3220 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3221 * fixed some bugs revealed by Paul Dietz' test suite:
3222 ** The system now obeys the constraint imposed by
3223 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3224 types form a lattice under type intersection.
3225 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3226 ** ASSOC now ignores NIL elements in an alist.
3227 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3228 and (1+ MOST-POSITIVE-FIXNUM) answers.
3229 ** The addition of a method with invalid qualifiers to a generic
3230 function does not cause an error to be signalled immediately;
3231 a warning is signalled, and the error is generated only on
3232 calling the generic function.
3233 * changed .fasl file version number, in part to add type codes for
3234 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3235 obscure ANSI requirements
3237 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3238 * fixed bug 148: failure to inline-expand a local function left
3239 garbage, confusing the compiler.
3240 * fixed bugs 3cd: structure slot readers perform type check if the
3241 slot can have an invalid value (i.e. it is either not initialized
3242 or can be written with a less specific slot writer).
3243 * bug fix: the compiler now traps array references to elements off
3244 the end of an array; previously, the bounds checking in some
3245 circumstances could go off-by-one.
3246 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3248 * type declarations for array element types now obey the description
3249 on the CLHS page "Declaration TYPE", as per discussions on
3250 sbcl-help around 2003-05-08. This means that a declaration
3251 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3252 declaration, all references to BAR will be asserted or assumed
3253 (with THE, so dependent on compiler policy) to involve objects of
3254 type FOO. Note that no such declaration is implied in
3255 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3256 * declared types of functions from the "Conditions"
3257 chapter. (reported by Paul Dietz)
3258 * bug fix: CERROR accepts a function as its first argument.
3259 * bug fix: NTH an NTHCDR accept a bignum as index
3260 arguments. (reported by Adam Warner)
3261 * optimization: character compare routines now optimize comparing
3262 against a constant character. (reported by Gilbert Baumann)
3263 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3264 argument now works properly on the MIPS platform.
3265 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3266 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3267 declared to have a functional type, does not cause an error
3269 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3270 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3271 * fixed bug: if last continuation of a deleted block has a
3272 destination, this destination should be deleted too. (reported by
3274 * fixed a bug in the bootstrap process: the host compiler's values
3275 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3276 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3277 test case from Patrik Nordebo)
3278 * improved the ability of the disassembler on the PPC platform to
3279 provide helpful disassembly notes.
3280 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3281 the class in more cases than previously.
3282 * bug fix: FILE-POSITION now understands :START and :END for
3283 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3284 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3285 8) data. (thanks to David Lichteblau)
3286 * bug fix: it is possible to add a method to a generic function
3287 without lambda list.
3288 * bug fix: reader failed to signal END-OF-FILE inside an
3289 object representation. (reported by Nikodemus Siivola)
3290 * fixed some bugs revealed by Paul Dietz' test suite:
3291 ** LAST and [N]BUTLAST should accept a bignum.
3292 ** condition slot accessors are methods.
3293 ** (VECTOR NIL) is a subtype of STRING.
3295 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3296 * minor incompatible change: some nonsensical specialized lambda
3297 lists (used in DEFMETHOD) which were previously ignored now signal
3299 * minor incompatible change: the system is now aware of the types of
3300 variables in the COMMON-LISP package, and will signal errors for
3301 most violations of these type constraints (where previously they
3302 were silently accepted).
3303 * minor incompatible change: COMPILE-FILE now uses the freedom
3304 afforded (ANSI 3.2.2.3) to use derived function types for
3305 functions defined in the same file. This also permits the system
3306 to warn on static type mismatches and function
3307 redefinition. (Currently it does not work with high DEBUG level.)
3308 * minor incompatible change: VALUES declaration is disabled.
3309 * When issuing notes, the compiler now signals a condition of type
3310 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3311 restart for use in user handlers. It is expected that the
3312 COMPILER-NOTE condition will eventually become a condition
3313 supertype to a hierarchy of note types, which will then be
3314 handleable in a similar fashion. However, at the moment, no such
3315 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3316 but it's an implementation detail, not a classification for the
3318 * Changes in type checking closed the following bugs:
3319 ** type checking of unused values (192b, 194d, 203);
3320 ** template selection based on unsafe type assertions (192c, 236);
3321 ** type checking in branches (194bc).
3322 * A short form of VALUES type specifier has ANSI meaning (it has
3323 increased the number of situations when SBCL cannot perform type
3325 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3326 keywords or constants is permissible.
3327 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3328 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3329 classes. (thanks to Antonio Martinez)
3330 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3331 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3332 argument to a pretty-print stream. Instead, it leaves any such
3333 operations to DESCRIBE-OBJECT methods.
3334 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3336 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3337 lambda lists. (thanks to David Lichteblau)
3338 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3339 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3340 (thanks to Gerd Moellmann)
3341 * fixed bug 246: increased compilation speed of long
3342 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3344 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3346 * DEFCONSTANT now throws a condition of type
3347 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3348 constant to a non-EQL value; CONTINUE and ABORT restarts
3349 respectively change and preserve the value.
3350 * fixed bug 63: The code walker, part of the implementation of CLOS,
3351 is now better at handling symbol macros.
3352 * bug fix: There is no longer an internal implementation type named
3353 CL:LENGTH. (reported by Raymond Toy)
3354 * bug fix: In macro-like defining macros/special operators the
3355 implicit block does not enclose the lambda list.
3356 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3357 enclosed in parentheses) are not suitable as type specifiers, and
3358 their use properly signals an error now.
3359 * bug fix: An argument count mismatch for a type specifier in code
3360 being compiled no longer causes an unhandled error at compile
3361 time, but signals a compile-time warning.
3362 * fixed simple vector readable printing
3363 * bug fix: DESCRIBE takes more care over whether the class
3364 precedence list slot of a class is bound before accessing it.
3365 (reported by Markus Krummenacker)
3366 * bug fix: FORMATTER can successfully compile pretty-printer format
3367 strings which use variants of the ~* directive inside.
3368 * bug fix: SEARCH now applies its TEST predicate to the elements of
3369 the arguments in the correct order. (thanks to Wolfhard Buss)
3370 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3371 in type inference. (thanks to Robert E. Brown)
3372 * bug fix: Reading in symbols with an explicit package name of ""
3373 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3374 * fixed some bugs revealed by Paul Dietz' test suite:
3375 ** NIL is now allowed as a structure slot name.
3376 ** Arbitrary numbers, not just REALs, are allowed in certain
3377 circumstances in LOOP for-as-arithmetic clauses.
3378 ** Multiple class redefinitions before slot access no longer
3379 causes a type error.
3380 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3381 association between the name and a class.
3382 ** Generic functions with non-standard method-combination and over
3383 five methods all of which return constants no longer return NIL
3384 after the first few invocations. (thanks to Gerd Moellmann)
3385 ** CALL-NEXT-METHOD with no arguments now passes the original
3386 values of the arguments, even in the presence of assignment.
3387 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3389 ** DISASSEMBLE works with closures and funcallable instances.
3390 ** ADD-METHOD now returns the generic function, not the new method.
3391 ** FIND-METHOD signals an error if the lengths of the specializers
3392 is incompatible with the generic function, even if the ERRORP
3394 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3395 which its argument is a member.
3396 ** DEFCLASS only redefines the class named by its class-name
3397 argument if that name is the proper name of the class;
3398 otherwise, it creates a new class.
3399 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3400 of the UNBOUND-SLOT condition to the name of the slot.
3401 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3402 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3403 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3404 SLOT-MAKUNBOUND in the specified fashion.
3406 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3407 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3408 cross-compilation host. As a consequence, we can now bootstrap our
3409 way up to SBCL starting with a bare gcc toolchain and human-readable
3410 source code (first the source to CLISP, then the source to SBCL).
3411 * A contributed module containing a partial implementation of the
3412 simple-streams interface has been included. (thanks to Rudi
3414 * A contributed module implementing the RFC1321 Message Digest
3415 Algorithm, known as MD5, has been included.
3416 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3417 no longer has any effect, as the code controlled by this feature
3418 has been deleted. (As far as we know, no-one has ever built using
3419 this feature, and its semantics were confused in any case).
3420 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3421 work on conditions (as required by the ANSI specification),
3422 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3423 expected behaviour on conditions. Users should note, however,
3424 that such behaviour is not required by the ANSI specification,
3425 and so use of this behaviour may render their code unportable.
3426 * fixed some bugs revealed by Paul Dietz' test suite:
3427 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3429 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3430 ** SXHASH on condition objects no longer returns NIL.
3431 ** :ALLOCATION :CLASS slots are better treated; their values are
3432 updated on class redefinition, and initforms inherited from
3433 superclasses are applied.
3434 ** REMOVE-METHOD returns its generic function argument even when
3435 no method was removed.
3436 ** SHARED-INITIALIZE now initializes the values of the requested
3437 slots, including those with :ALLOCATION :CLASS.
3438 ** ALLOCATE-INSTANCE now works on structure classes defined via
3439 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3441 ** SLOT-EXISTS-P now works on conditions, as well as structures
3443 ** MAKE-LOAD-FORM now has the required methods on
3444 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3445 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3446 keyword, and now implements the SLOT-NAMES argument.
3447 ** methods with &OPTIONAL arguments no longer allow too many
3448 arguments to be passed in the call without error.
3449 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3450 option is consistent with the required arguments of the generic
3451 function lambda list.
3452 * bug fix: REQUIRE accepts a string designator. (Thanks to
3454 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3455 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3456 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3457 * bug fix: The compiler now checks for duplicated variables in macro
3459 * bug fix: SETQ on globals returns the correct value.
3460 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3461 not a COMPILER-ERROR (followed by some other strange error on
3462 choosing the CONTINUE restart).
3463 * bug fix: make.sh and friends are now more consistent in the way that
3464 they look for GNU "make".
3466 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3467 * experimental native threads support (on x86 Linux >=2.4 only).
3468 This is not compiled in by default: you need to add :SB-THREAD to
3469 the target features. See the "Beyond ANSI" chapter of the manual
3471 * fix for longstanding nonANSIism: The old distinction between
3472 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3473 The return value from CL:FIND-CLASS is now a CLOS class, and
3474 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3475 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3476 * An interface to the MetaObject Protocol, as described in Kiczales,
3477 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3478 MIT Press, 1991, is available from the SB-MOP package.
3479 * incompatible change: the SB-PCL package should now be considered
3480 a private implementation detail, and no longer a semi-private MOP
3482 * minor incompatible change: due to rearrangement for threads, the
3483 control stack and binding stack are now allocated at arbitrary
3484 addresses instead of being hardcoded per-port. Users affected by
3485 this probably have to be doing advanced things with shared
3486 libraries, and will know who they are.
3487 * minor incompatible change: Previously, all --eval forms used were
3488 processed with READ before any of them were processed with EVAL.
3489 Now each --eval form is processed with both READ and EVAL before
3490 the next --eval form is processed. (Thus package operations like
3491 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3492 work as the user might reasonably expect.)
3493 * minor incompatible change: *STANDARD-INPUT* is now only an
3494 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3496 * minor incompatible change: Y-OR-N-P is now character-oriented,
3497 not line oriented. Also, YES-OR-NO-P now works without errors.
3498 (thanks to Antonio Martinez)
3499 * sb-aclrepl module improvements: an integrated inspector, added
3500 repl features, and a bug fix to :trace command.
3501 * Known functions, which cannot be open coded by the backend, are
3502 considered to be able to check types of their arguments. (fixing
3503 a bug report by Nathan J. Froyd)
3504 * fixed a bug in computing method discriminating functions: It is
3505 now possible to define methods specialized on classes which have
3506 forward-referenced superclasses. (thanks to Gerd Moellmann)
3507 * fixed evaluation order in optional entries (reported by Gilbert
3509 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3510 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3511 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3512 required-by-AMOP NAME argument, as well as CLASS and
3513 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3514 * fixed bug 20: DEFMETHOD can define methods using names that are
3515 not the proper names of classes to designate class specializers.
3516 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3517 Unix file descriptors, instead of blowing up. (thanks to Antonio
3519 * Garbage collection refactoring: user-visible change is that a
3520 call to the GC function during WITHOUT-GCING will not do garbage
3521 collection until the end of the WITHOUT-GCING. If you were doing
3522 this you were probably losing anyway.
3523 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3524 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3525 TYPEP the latter but not the former.
3526 * The compiler issues a full WARNING for calls to undefined functions
3527 with names from the CL package.
3528 * MAP-INTO for a vector destination is open coded. (reported by
3529 Brian Downing on c.l.l)
3530 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3531 documentation string.
3532 * fixed some bugs revealed by Paul Dietz' test suite:
3533 ** COPY-ALIST now signals an error if its argument is a dotted
3535 ** Condition slots are now accessed more correctly in the presence
3536 of multiple initargs for a given slot.
3537 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3538 exclude restarts of the same name associated with a different
3540 ** DEFCLASS of forward-referenced classes with another
3541 forward-referenced class in the superclasses list no longer
3543 ** Condition slots are now initialized once each, not multiple
3544 times. (thanks to Gerd Moellmann)
3545 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3547 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3548 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3549 arguments contain duplicated elements.
3550 ** RESTART-CASE understands local macros.
3551 ** RESTART-CASE associates exactly its own restarts with a condition.
3552 ** ENDP in safe mode checks its argument to be of type LIST.
3553 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3554 in question is unbound.
3555 ** Optimized MAKE-INSTANCE functions no longer cause internal
3556 assertion failures in the presence of duplicate initargs.
3557 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3558 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3559 propagated even in the ctor optimized implementation of
3561 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3563 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3564 * a better implementation of SXHASH on (simple) bit vectors,
3565 measured both in execution speed and in distribution of results
3566 over the positive fixnums, has been installed. Likewise, a better
3567 implementation of EQUAL for simple bit vectors is now available.
3568 * fixed CEILING optimization for a divisor of form 2^k.
3569 * fixed bug 240 (emitting extra style warnings "using the lexical
3570 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3571 by Antonio Martinez)
3572 * fixed SXHASH, giving different results for NIL depending on type
3573 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3574 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3575 variable. (found by Rolf Wester)
3576 * SBCL does not ignore type declarations for special
3577 variables. (reported by rif on c.l.l 2003-03-05)
3578 * some bug fixes in contrib/sb-aclrepl/
3579 * fixed some bugs revealed by Paul Dietz' test suite:
3580 ** a bug in the CONS type specifier, whereby the CAR and CDR
3581 types got intertwined, has been fixed;
3582 ** the type system is now able to reason about the interaction
3583 between INTEGER and RATIO types more completely;
3584 ** APPEND, [N]REVERSE and NRECONC check that those their
3585 arguments, which must be proper lists, are really so;
3586 ** An array specialized to be unable to hold elements has been
3587 implemented, as required -- yes, really -- by ANSI;
3588 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3589 on malformed property lists;
3591 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3592 * incompatible packaging change: in line with Unix convention,
3593 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3594 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3595 * REQUIRE and PROVIDE are now optionally capable of doing something
3596 useful. See the documentation string for REQUIRE.
3597 * infrastructure for a managed SBCL contrib system: contributed
3598 modules in this release include:
3599 ** the ASDF system definition facility;
3600 ** an interface to the BSD Sockets API;
3601 ** an ACL-like convenience interface to the repl;
3602 (thanks to Kevin Rosenberg)
3603 ** an implementation of ROTATE-BYTE, with efficient implementation
3605 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3606 gives a closer approximation to the right answer than previously.
3607 (thanks to Raymond Toy)
3608 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3609 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3610 optional environment arguments, as required by ANSI.
3611 * fixed bugs in other functions taking environment objects, allowing
3612 calls with an explicit NIL environment argument to be compiled
3614 * fixed bug 228: primary return values from
3615 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3616 COMPILE or FUNCTION.
3617 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3618 :TYPE will no longer signal errors on innocuous objects.
3619 * fixed bug 231b: SETQ is better at respecting type declarations in
3620 the lexical environment.
3621 * fixed a bug in DEFCLASS: classes named by symbols with no or
3622 unprintable packages can now be defined.
3623 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3624 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3625 * fixed compiler failure related to checking types of functions.
3626 (reported by Robert E. Brown)
3627 * the compiler is now much more consistent in its error-checking
3628 treatment of bounding index arguments to sequence functions: in
3629 (SAFETY 3) code, errors will be signalled in almost all cases if
3630 invalid sequence bounding indices are passed to functions defined
3631 by ANSI to operate on sequences.
3632 * fixed a bug in the build procedure: documentation of SBCL-specific
3633 packages is now preserved and available in the final Lisp image.
3634 * lifted FDEFINITION lookup out of loops in the implementation of
3635 many list operations. (thanks to Robert E. Brown)
3636 * fixed a bug in the reader: the #n# reader macro now works for
3637 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3638 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3639 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3640 E. Brown and Tony Martinez respectively)
3641 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3642 SLOT-BOUNDP: in particular, it is now able to optimize them much
3643 better, and is now not vulnerable to having packages renamed.
3644 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3645 always called when a slot is not present in an instance. (thanks
3647 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3648 longer signal an error on unreading a character following EOF on
3649 the previous constituent stream. (thanks to Tony Martinez)
3650 * fixed some bugs revealed by Paul Dietz' test suite:
3651 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3652 not just nonnegative fixnums;
3653 ** the logical bit-array operators such as BIT-AND now accept an
3654 explicit NIL for their "opt-arg" argument (to indicate a
3655 freshly-consed result bit-array);
3656 ** ELT now signals an error on an invalid sequence index in safe
3658 ** the type system is now cleverer about negations of numeric
3659 types, and consequently understands the BIGNUM and RATIO types
3661 ** the type system is now cleverer about the interaction between
3662 INTEGER and RATIO types: while bugs still remain, many more
3663 cases are accurately computed;
3664 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3665 if it is in the last clause;
3666 ** CONSTANTLY now correctly returns a side-effect-free function in
3668 ** DECLARE is no longer treated as a special-operator; in
3669 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3670 * incremented fasl file version number due to the change in the
3671 DEFSTRUCT-SLOT-DESCRIPTION structure.
3673 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3674 * minor incompatible change: code processed by the "interpreter" or
3675 EVAL now has a compilation optimization policy of (DEBUG 2)
3676 (changed from (DEBUG 1)) to improve debuggability of interactive
3677 development, and to allow the use of the debug RETURN command in
3679 * an experimental implementation of the RETURN command for the
3680 debugger has been included. (thanks to Frederik Kuivinen)
3681 * fixed bug 62: constraints were not propagated into a loop.
3682 * fixed bug in embedded calls of SORT (reported and investigated by
3684 * fixed some bugs revealed by Paul Dietz' test suite:
3685 ** printing and reading of arrays with some dimensions having
3686 length 0 (thanks to Gerd Moellmann);
3687 ** BOA constructor with &AUX argument without a default value does
3688 not cause a type error;
3689 ** CONSTANTP now returns true for all self-evaluating objects.
3691 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3692 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3693 accessors that are related by inheritance, as specified in the
3694 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3695 Valtteri Vuorikoski)
3696 * The compiler is now able to inline functions that were defined in
3697 a complex lexical environment (e.g. inside a MACROLET).
3698 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3700 * Support for the upcoming FreeBSD-5.0 release has been included.
3701 (thanks to Dag-Erling Smorgrav)
3702 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3703 effect when it is not in a toplevel context.
3704 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3705 stand a better chance of being correct. (thanks to Gerd
3707 * fixed bug in COERCE, which now signals an error on coercing a
3708 rational to a bounded real type which excludes the expected
3710 * The compiler is now able to derive types more accurately from the
3711 COERCE and COMPILE functions.
3712 * fixed bug 223: functional binding is considered to be constant
3713 only for symbols in the CL package.
3714 * fixed bug 231: SETQ did not check the type of a variable being set
3715 (reported by Robert E. Brown)
3716 * A new optimization for MAKE-INSTANCE has been included, fixing
3717 various bugs (including relating to :ALLOCATION :CLASS slots and
3718 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3720 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3721 ** As required by ANSI, LOOP now disallows anonymous collection
3722 clauses such as COLLECT I in conjunction with aggregate boolean
3723 clauses such as THEREIS (= I 1);
3724 ** LOOP now signals an error when any variable is reused in the
3725 same loop (including the potentially useful construct analogous
3726 to WITH A = 1 WITH A = (1+ A);
3727 ** IT is only a special loop symbol within the first clause of a
3728 conditional loop clause;
3729 ** LOOP with a typed iteration variable over a hashtable now
3730 signals a type error iff it should.
3731 * fixed some other bugs revealed by Paul Dietz' test suite:
3732 ** FILE-STREAM now names the class previously known as FD-STREAM;
3733 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3734 argument) no longer signals an error;
3735 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3736 of the slot symbol, rather than using the current package
3737 ((:CONC-NAME "") continues to intern the slot's name in the
3739 * incremented fasl file version number, because of the incompatible
3740 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3741 of the new implementation of DEFINE-COMPILER-MACRO.
3743 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3744 * Support for building SBCL for MIPS platforms running in
3745 little-endian mode has now been checked in, and basic
3746 functionality on said platforms verified.
3747 * minor incompatible change: PCL now records the pathname of a file
3748 in which methods and the like are defined, rather than its
3750 * minor incompatible change: TRUENAME now considers the truename of
3751 a file naming a directory to be the pathname with :DIRECTORY
3752 component indicating that directory.
3753 * minor incompatible change: a NAMED clause in the extended form of
3754 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3755 reason for the previous behaviour is unclear.
3756 * more systematization and improvement of CLOS and MOP conformance
3757 in PCL (thanks to Gerd Moellman and Pierre Mai):
3758 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3760 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3761 primary methods with no specializers;
3762 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3764 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3765 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3766 FORWARD-REFERENCED-CLASSes; error reporting on
3767 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3769 ** SXHASH on CLOS instances now uses a slot internal to the
3770 instance to return different numbers on distinct instances,
3771 while preserving the same return value through invocations of
3773 ** DEFMETHOD signals errors when methods with longer incongruent
3774 lambda lists are added to generic functions;
3775 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3776 CLASS, as specified in AMOP;
3777 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3778 on the order returned by the primary method for classes of
3779 class STANDARD-CLASS;
3780 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3781 * fixed some bugs shown by Paul Dietz' test suite:
3782 ** DOLIST puts its body in TAGBODY;
3783 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3785 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3786 value producing form;
3787 ** if more variables are given to PROGV than values, extra
3788 variables are bound and made to have no value;
3789 ** NSUBSTITUTE on list arguments gets the right answer with
3791 ** ELT signals an error of type TYPE-ERROR when the index argument
3792 is not a valid sequence index;
3793 ** LOOP signals (at macroexpansion time) an error of type
3794 PROGRAM-ERROR when duplicate variable names are found;
3795 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3796 ** FUNCALL of special-operators now cause an error of type
3798 ** PSETQ now works as required in the presence of side-effecting
3799 symbol-macro places;
3800 ** NCONC accepts any object as its last argument;
3801 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3803 ** loop-for-as-package does not require a package to be explicitely
3805 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3807 * fixed bug 166: compiler preserves "there is a way to go"
3808 invariant when deleting code.
3809 * fixed bug 172: macro lambda lists with required arguments after
3810 &REST arguments now cause an error to be signalled. (thanks to
3812 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3813 bug. (thanks to Matthew Danish)
3814 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3816 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3817 arguments are lexically rebound. (thanks to Gerd Moellmann and
3819 * fixed bug 194: error messages are now more informative when there
3820 is no primary method applicable in a call to a generic
3821 function. (thanks to Gerd Moellmann)
3822 * fixed bug in command line argument checking (thanks to Julian
3824 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3825 (thanks to Matthew Danish)
3826 * incremented fasl file version number, because of the
3827 SXHASH-related changes in the layout of CLOS data structures
3829 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3830 * minor incompatible change: The runtime (the Unix executable named
3831 "sbcl") is now much pickier about the .core files it will load.
3832 Essentially it now requires .core files to descend from the same
3833 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3834 as the runtime does. (The intent is to prevent the crashes which
3835 can occur, and which can even be reported as mysterious failures,
3836 when people patch the sources or change the build parameters
3837 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3838 sbcl and .core files.)
3839 * fixed bug: VALUES-LIST is no longer optimized away.
3840 * fixed bug 142: The FFI conversion of C string values to Lisp
3841 string values no longer conses excessively. (thanks to Nathan
3842 Froyd porting Raymond Toy's fix to CMU CL)
3843 * began to systematize and improve MOP conformance in PCL (thanks to
3844 Nathan Froyd, Gerd Moellman and Pierre Mai):
3845 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3847 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3848 ** FINALIZE-INHERITANCE is now called on class finalization;
3849 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3850 argument precedence order.
3851 * fixed bug 202: The compiler no longer fails on functions whose
3852 derived types contradict their declared type.
3853 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3854 so it can be non-toplevel.
3855 * The fasl file version number has changed (because of the new
3856 implementation of DEFMACRO).
3857 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3858 safe code, that any length requirement by their type-specifier
3859 argument is valid. The exceptions to this are described in bug
3861 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3862 functions now check the types of their inputs as required by ANSI.
3863 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3864 introduced symbol is DECLAREd to be SPECIAL.
3865 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3866 * fixed bug: :COUNT argument to sequence functions may be negative
3867 * fixed bug: body of DO-SYMBOLS may contain declarations
3868 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3869 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3870 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3871 symbol macro only once
3872 * fixed printing of call frame when argument list is unavailable
3873 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3874 * compiler no longer signals WARNING on unknown keyword
3877 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3878 * A beta-quality port to the mips architecture running Linux,
3879 based on the old CMUCL backend, has been made. It has been tested
3880 on a big-endian kernel, and works sufficiently well to be able to
3881 rebuild itself; it has not been tested in little-endian mode.
3882 * fixed an inconsistency between gencgc.c and purify.c which made
3883 dumping/loading .core files unreliable
3884 * fixed bug 120a: The compiler now deals correctly with IFs where
3885 the consequent is the same as the alternative, instead of
3886 misderiving the return type. (thanks to Alexey Dejneka)
3887 * fixed bug 113: Logical pathnames are now dumpable (the logical
3888 host is resolved at load-time, throwing an error if it is not
3890 * fixed bug 174: FORMAT's error message is slightly clearer when a
3891 non-printing character is used in a format directive.
3892 * fixed several bugs in compiler checking of type declarations, i.e.
3893 violations of the Python "declarations are assertions" principle
3894 (thanks to Alexey Dejneka)
3895 * fixed several bugs in PCL's error checking (thanks to Gerd
3897 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3899 * fixed bug in compilation of functions as first class values
3900 (thanks to Antonio Martinez-Shotton)
3901 * The compiler's handling TYPE-ERRORs which it can prove will
3902 inevitably happen at runtime has been cleaned up and corrected
3903 in several ways. (thanks to Alexey Dejneka)
3904 * improved argument type checking for various basic arithmetic
3905 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3906 lavished on them in the past that they can be compiled in many
3907 ways in different special cases
3908 * fixed bug 181: compiler checks validity of user supplied type
3910 * cleaned up code flushing in optimization: Function calls which
3911 should signal errors for safety purposes (e.g. which ANSI says
3912 should signal errors when their arguments are of incorrect type)
3913 are no longer optimized away.
3914 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3915 * incremented fasl file version number, because changes in the
3916 implementation of sequence functions like COERCE caused
3917 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3918 in old inline expansions) to become undefined. (Actually these
3919 changes were later undone, so we might very well be binary
3920 compatible with 0.7.7 after all, but leaving the version number
3921 incremented seemed like the simplest and most conservative
3924 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3925 * An alpha-quality port to the parisc architecture running Linux,
3926 based on the old CMUCL backend, has been made. This, even more so
3927 than the other backends, should be considered still a work in
3928 progress; known problems include that the Linux kernel in 64-bit
3929 mode does not propagate the correct sigcontext structure to
3930 userspace, and consequently SBCL on a parisc64 kernel will not
3932 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3933 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3934 suppressed.) Also now that the compiler is looking at declarations
3935 in the environment, it checks optimization declarations as well,
3936 and suppresses inlining when (> DEBUG SPEED).
3937 * More fixes have been made to treatment of floating point exception
3938 treatment and other Unix signals. In particular, floating point
3939 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3940 * The detection and handling of control stack exhaustion (infinite
3941 or very deeply nested recursion) has changed. Stack exhaustion
3942 detection is now done by write-protecting pages at the OS level
3943 and applies at all optimization settings; when found, a
3944 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3945 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3946 be caught using IGNORE-ERRORS.
3947 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3948 correctly when transforming recursive LABELS functions to LETs.
3949 (thanks to Alexey Dejneka)
3950 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3951 that are names of constants or global variables.
3952 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3953 alien routines with docstrings.
3954 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3955 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3957 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3958 (thanks to Eric Marsden)
3959 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3960 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3961 * Bug fix: The SPARC backend can now compile functions involving
3962 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3963 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3964 to a functions expecting a PATHNAME-DESIGNATOR.
3965 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3966 lambda lists. (thanks to Alexey Dejneka)
3967 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3968 functions so that EXPORTing the name of the class doesn't cause
3969 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3970 to Antonio Martinez for reporting this)
3971 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3972 constant-folded without failing an assertion. (thanks to Einar
3973 Floystad Dorum for reporting this)
3974 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3975 types are now dealt with more correctly by the compiler.
3976 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3977 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3978 behaviour analogous to RENAME-FILE. This puts its behaviour more
3979 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3981 * The fasl file version number has changed again. (because of the
3982 bug fix involving the names of PCL MAKE-INSTANCE functions)
3984 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3985 * bug fix: Floating point exceptions are treated much more
3986 consistently on the x86/Linux and PPC/Linux platforms.
3987 * Array initialization with :INITIAL-ELEMENT is now much faster for
3988 cases when the compiler cannot open code the array creation, but
3989 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3990 array accesses have also seen a speed increase.
3991 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3992 specified. (This was at the root of some bad interactions between
3993 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3995 * bug fix: Internal error arguments for undefined functions are now
3996 computed correctly on the PPC/Linux platform.
3997 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3998 Raymond Toy's patch for CMU CL)
3999 * Support for the Solaris 9 operating environment has been included
4000 (thanks to Daniel Merritt)
4001 * A very ugly but hopefully complete draft of the missing FFI chapter
4002 of the manual has been created by reformatting the corresponding
4003 CMU CL manual chapter into (currently very ugly and incoherent)
4004 DocBook and bringing it up to date for SBCL behavior. Thus, the
4005 manual is now essentially complete, at least by my extreme
4006 once-and-only-once standards, whereby it's acceptable to refer to
4007 the doc strings of SB-EXT functions as the primary documentation.
4008 * The fasl file version number has changed again, due to cleanup of
4009 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
4010 is no longer a static symbol.)
4012 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
4013 * SBCL now builds with OpenMCL (version 0.12) as the
4014 cross-compilation host; also, more progress has been made toward
4015 bootstrapping under CLISP.
4016 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
4018 * bug 158 fixed: The compiler can now deal with integer loop
4019 increments different from 1; fixing this turned out also to fix
4021 * bug 169 fixed: no more bogus warnings about using lexical bindings
4022 despite the presence of perfectly good SPECIAL declarations (thanks
4023 to David Lichteblau)
4024 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
4025 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
4026 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
4028 * bug 180 fixed: Method combination specifications no longer ignore
4029 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
4030 * bug fix: Structure type predicate functions now check their argument
4031 count as they should.
4032 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
4033 correctly. (thanks to Pierre Mai)
4034 * minor incompatible change: The --noprogrammer option is deprecated
4035 in favor of the new --disable-debugger option, which is very similar.
4036 (The major difference is that it takes effect at a slightly different
4037 time at startup, causing handling of errors in --sysinit and
4038 --userinit files will be affected differently.) The
4039 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
4040 been added to allow this functionality to be controlled from ordinary
4041 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
4042 maintainers, who might want to run non-interactive scripts to
4043 build SBCL cores which will later be used interactively.)
4044 * minor incompatible change: The LOAD function no longer, when given
4045 a wild pathname to load, loads all files matching that pathname.
4046 Instead, an error of type FILE-ERROR is signalled.
4048 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
4049 * bug 147 fixed: The compiler preserves its block link/count
4050 invariants more correctly now so that it doesn't crash. (thanks
4052 * Dynamic loading of object files in OpenBSD is now supported. (thanks
4054 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
4055 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
4056 no-value-for-this-character as NIL (as specified by ANSI).
4057 * HOST-NAMESTRING on physical pathnames now returns a string that is
4058 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
4059 (thanks to Christophe Rhodes)
4060 * The Alpha port handles icache flushing more correctly. (thanks to
4062 * More progress has been made toward bootstrapping under CLISP. (thanks
4063 to Christophe Rhodes)
4064 * The fasl file format has changed again, because dynamic loading
4065 on OpenBSD (which has non-ELF object files) motivated some cleanups
4066 in the way that foreign symbols are transformed and passed around.
4067 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
4068 is no longer treated as whitespace by the reader, but instead as
4069 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
4070 |A
\7fB|, instead of A as it used to.
4072 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
4073 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
4074 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
4075 * SBCL now runs on the PPC archtiecture under Linux. It actually did
4076 this as of 0.7.1.45, but was left out of the previous news section
4077 (thanks to Dan Barlow)
4078 * SBCL now runs on the Solaris operating system on SPARC architectures
4079 (thanks to Christophe Rhodes's port of the CMUCL runtime)
4080 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
4081 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
4082 * SPARC backend cleanups, allowing builds of cores optimized for V8
4083 and V9 SPARCS, and also emission of code targeted to a particular
4084 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
4086 * SBCL is closer to bootstrapping under CLISP, thanks to various
4087 fixes by Christophe Rhodes.
4088 * The fasl file format has changed again, to allow the compiler's
4089 INFO database to support symbol macros.
4090 * The user manual (in doc/) is formatted into HTML more nicely.
4091 (thanks to coreythomas)
4092 * The system is smarter about SUBTYPEP relationships, especially
4093 those involving NOT types (including types such as ATOM which are
4094 represented internally using NOT types). Thus SUBTYPEP is less
4095 likely to return (VALUES NIL NIL) in general, and in particular
4096 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
4098 * The fasl file format has changed again, because the internal
4099 representation of types now includes a new slot to support the new
4100 SUBTYPEP-of-NOT-types logic.
4101 * (not a change in the main branch of SBCL, but a related prototype
4102 which can hopefully be merged into the main branch of SBCL in the
4103 future:) Brian Spilsbury has produced a Unicode-enabled variant of
4104 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
4105 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
4106 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
4107 dispatch macro character combinations. (thanks to Alexey Dejneka)
4108 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
4109 that superficially look like logical namestrings correctly.
4110 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
4112 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
4113 default supplied-p) for &optional and &key arguments. (thanks to
4116 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
4117 * incompatible change: The compiler is now less aggressive about
4118 tail call optimization, doing it only when (> SPACE DEBUG) or
4119 (> SPEED DEBUG). (This is an incompatible change because there are
4120 programs which relied on the old CMU-CL-style behavior to optimize
4121 away their unbounded recursion which will now die of stack overflow.)
4122 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
4123 for non-GENCGC systems has been increased to 20M (since that
4124 seems much closer to the likely performance optimum for modern
4125 systems than the old 4M value was)
4126 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
4127 and *DEBUG-PRINT-LEVEL*
4128 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
4129 of CMU CL's support for SPARC, and various endianness and other
4130 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
4131 * new syntactic sugar for the Unix command line: --load foo.bar is now
4132 an alternate notation for --eval '(load "foo.bar")'.
4134 ** The system now detects stack overflow and handles it gracefully,
4135 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
4136 optimization settings. (This is a good thing in general, and
4137 its introduction in this version should be particularly timely
4138 for anyone whose code fails because of suppression of tail
4140 ** The system now hunts for the C variable "environ" in a more
4141 devious way, to avoid segfaults when the C library version differs
4142 between compile time and run time. (thanks to Christophe Rhodes)
4143 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4144 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4145 ** The compiler no longer issues bogus style warnings for undefined
4146 classes in the same source file as the DEFCLASSes which defined
4147 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4149 ** fixes in CONDITION class precedence list for undefined function
4150 errors (thanks to Alexei Dejneka)
4151 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4152 correctly. (thanks to Dan Barlow)
4153 ** portability fixes aiming at bootstrapping under CLISP (thanks
4154 to Dave McDonald and Christophe Rhodes)
4155 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4156 ** fixes in type translation and and type inference (thanks to
4158 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4159 ** various fixes in the new ports (thanks to Dan Barlow)
4160 * several changes related to debugging:
4161 ** suppression of tail recursion, as noted above
4162 ** stack overflow detection, as noted above
4163 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4164 is now the default. (For some time encapsulation has been more
4165 reliable than the breakpoint-based :ENCAPSULATE NIL
4166 implementation, at least on X86 systems; and I just noticed that
4167 encapsulation also seems closer to the spirit of the ANSI
4170 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4172 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4173 up properly again. (There was a packaging bug in 0.7.0 which
4174 left their definitions in SB-SYS::LOAD-FOREIGN and
4175 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4176 vital for most things which interface to C-level interfaces,
4177 like extensions working with sockets or databases or
4178 Perl-compatible regexes or whatever, and the need to fix
4179 this bug is the main reason that 0.7.1 was released so
4181 ** DEFGENERIC is now choosier about the methods it redefines, so that
4182 reLOADing a previously-LOADed file containing DEFGENERICs does
4183 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4184 cycle now works as it should. (thanks to Alexey Dejneka)
4185 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4186 fixed by Christophe Rhodes. (He actually submitted this patch
4187 months ago, and I delayed until after 0.7.0.)
4188 ** Bug 111 (internal compiler confusion about runtime checks on
4189 FUNCTION types) was fixed by Alexey Dejneka.
4190 * Some internal cleanups (getting rid of variables which aren't
4191 needed now that the byte interpreter is gone) caused the fasl
4192 file format number to change again.
4194 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4195 * major incompatible change: The default fasl file extension, i.e. the
4196 default extension for files produced by COMPILE-FILE, has changed
4197 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4199 ** There are many changes in the implementation of the compiler.
4200 SBCL is now essentially a compiler-only implementation of ANSI
4201 Common Lisp. EVAL still "interprets" a few special cases, but
4202 almost all the interesting cases are handled by creating
4203 a LAMBDA expression, calling COMPILE on it, then calling
4204 FUNCALL on the result.
4205 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4206 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4207 Since the code is newer, there might still be some new bugs
4208 (though not as many as before Martin Atzmueller's fixes:-). But
4209 the new code is substantially simpler and clearer, and hopefully
4210 any remaining bugs will be simpler, less fundamental, and more
4211 fixable then the bugs in the old code.
4212 ** The revised compiler is still a little unsteady on its feet.
4214 *** The debugging information it produces (particularly the names
4215 of FUNCTION objects) is sometimes much less useful than what
4216 the old compiler produced.
4217 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4218 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4219 within a MACROLET) has been temporarily weakened.
4220 ** There are new compiler optimizations for various functions:
4221 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4222 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4223 *** the math functions TRUNCATE, FLOOR, and CEILING
4224 *** the function-of-all-trades COERCE
4225 Mostly these should be transparent, but there's one
4226 potentially-annoying problem (bug 117): when the compiler
4227 inline-expands a function and does type analysis on the result,
4228 it can create control paths which have type mismatches, and
4229 when it can't prove that those control paths aren't taken,
4230 it will issue WARNINGs about the type mismatches. This is
4231 a particular problem in practice for the new sequence functions.
4232 It's not clear how this should be fixed, and for now, a
4233 workaround is given in the entry for 117 in the BUGS file.
4234 ** (Because of the interaction between the two previous items --
4235 occasional inlining problems and new inline expansions -- some
4236 of the new sequence function optimizations won't really kick in
4237 completely until debugging information, and then inlining, are
4238 straightened out in some future version.)
4239 * minor incompatible changes:
4240 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4241 DIRECTORY no longer implicitly promotes NIL slots of its
4242 pathname argument to :WILD. In particular, when you ask for the
4243 contents of a directory (which you used to be able to do without
4244 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4245 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4246 ** changes in behavior that ANSI explicitly defines to be
4247 implementation dependent:
4248 *** The new compiler-only implementation still conforms with ANSI,
4249 but acts a little different than before. Besides the obvious
4250 changes in performance tradeoffs (that the cost per form passed
4251 to EVAL has gone up, and the cost per form executed by EVAL
4252 has gone down), the behavior of the system changes a little
4253 because there are no longer any interpreted function objects.
4254 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4255 e.g. doing COMPILE on the output of interactive DEFUN is
4257 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4259 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4260 has changed from (:CL) to NIL.
4261 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4262 e.g. #\Newline and #\Tab, don't have names specified in the
4263 ANSI Common Lisp standard, is now based on their ASCII symbolic
4264 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4265 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4266 are no longer used for output.
4267 ** changes in internal implementation constants:
4268 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4269 4 million. (If your application spends a lot of time GCing and
4270 you have a lot of RAM, you might want to experiment with
4271 increasing it even more.)
4272 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4273 However, almost all old code should still continue to work without
4274 immediate update, as SB-C-CALL is now a (deprecated) nickname
4276 ** Old operator names in the style DEF-FOO are now deprecated in
4277 favor of new corresponding names DEFINE-FOO, for consistency with
4278 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4279 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4280 internal symbols, but a few supported extensions like
4281 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4282 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4283 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4284 etc. as you get deeper into recursive calls to the debugger
4285 command loop, instead of the old "5]", "5]]", "5]]]"
4286 sequence. (I was motivated to do this when squabbles between
4287 ILISP and SBCL left me very deeply nested in the debugger. In the
4288 short term, this change will probably provoke more ILISP/SBCL
4289 squabbles, but hopefully it will be an improvement in the long run.)
4290 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4291 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4292 function, since declaiming return types involving aliens is
4293 (1) annoyingly messy to do by hand and (2) vital to efficient
4294 compilation of code which calls such functions.
4295 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4296 longer reexported by the SB-EXT package. They're solely useful
4297 for alien code, so it seems more logical that you should get
4298 them from the SB-ALIEN package, not in SB-EXT.
4299 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4300 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4301 features. Instead, the code that they used to control is always
4302 built into the system.
4303 * many other bug fixes
4304 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4305 advantage of the new EVAL-WHEN stuff and to clean them up in
4306 general, and they are now more ANSI-compliant in a number of
4307 ways. Martin Atzmueller is responsible for a lot of this.
4308 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4310 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4311 *** correct ERROR type for various file operations
4312 *** some fixes for Lisp streams
4313 *** DEFMETHOD syntax checking
4314 *** changing old weird representation of debug information as
4315 strings (which, among their other deficiencies, don't transform
4316 correctly when you rename packages, and don't change their
4317 print representation when you change things like *PACKAGE*
4318 and *PRINT-LENGTH*) to symbols and lists of symbols
4319 He also made several improvements and fixed several bugs in DESCRIBE.
4320 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4322 *** misbehavior of WRITE-STRING/WRITE-LINE
4323 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4324 and several other LOOP problems as well
4325 *** DIRECTORY when similar filenames are present
4326 *** DEFGENERIC with :METHOD options
4327 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4328 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4329 *** argument ordering in FIND with :TEST option
4330 *** mishandled package designator argument in APROPOS-LIST
4331 *** various problems in the backquote readmacro
4332 *** a bug in APROPOS
4333 *** probably some others that I'm not describing very well here,
4334 since the CVS log documents them by reference to sbcl-devel
4335 messages, and the SourceForge archives aren't working well.:-(
4336 ** Dan Barlow improved the Alpha port (and is making progress on the
4337 PPC port, for those of you who think different).
4338 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4339 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4340 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4341 continues to work on the SPARC port (for those of you in a position
4342 to look down upon our little PC-compatible boxes from a great height).
4343 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4344 and uses that copy, rather than the current dynamic value, when
4345 it's trying to decide whether to truncate output. Thus e.g.
4346 (let ((*print-lines* 50))
4347 (pprint-logical-block (stream nil)
4349 (let ((*print-lines* 8))
4350 (print (aref possiblybigthings i) stream)))))
4351 should now truncate the logical block only at 50 lines, instead of
4352 often truncating it at 8 lines, as it did before.
4353 * The doc/cmucl/ directory, containing old CMU CL documentation
4354 from the time of the fork, is no longer part of the base system.
4355 SourceForge has shut down its anonymous FTP service, and with it
4356 my original plan for distributing the old CMU CL documentation
4357 there. For now, if you need these files you can download an old
4358 SBCL source release and extract them from it.
4359 * The fasl file version number changed again, for dozens of reasons,
4360 some of which are apparent above.
4362 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4363 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4364 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4365 patch, so that DISASSEMBLE output is much nicer.
4366 * The code in the SB-PROFILE package now seems reasonably stable.
4367 I still haven't decided what the final interface should look like
4368 (I'd like PROFILE to interact cleanly with TRACE, since both
4369 facilities use function encapsulation) but if you have a need
4370 for profiling now, you can probably use it successfully with
4371 the current CMU-CL-style interface.
4372 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4373 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4374 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4375 link now returns the dangling link itself, and for similar
4376 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4377 link itself. (In these cases the old code signalled an error and
4378 looped endlessly, respectively.) Thus, DIRECTORY now works even
4379 in the presence of dangling and cyclic symbolic links.
4380 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4381 is now a supported extension again, since the consensus on
4382 sbcl-devel was that it can be useful for ordinary development
4383 work, not just for debugging SBCL itself.
4384 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4385 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4386 that currently-defined functions might be redefined later with
4387 different return types.
4388 * Hash tables can be printed readably, as inspired by CMU CL code
4389 of Eric Marsden and SBCL code of Martin Atzmueller.
4390 * better error handling in CLOS method combination, thanks to
4391 Martin Atzmueller porting Pierre Mai's CMU CL patches
4392 * more overflow fixes for >16Mbyte I/O buffers
4393 * A bug in READ has been fixed, so that now a single Ctrl-D
4394 character suffices to cause end-of-file on character streams.
4395 In particular, now you only need one Ctrl-D at the command
4396 line (not two) to exit SBCL.
4397 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4399 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4400 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4401 does the right thing.
4402 * The implementation of some type tests, especially for CONDITION
4403 types, is now tidier and maybe faster, due to CMU CL code
4404 originally by Douglas Crosher, ported by Martin Atzmueller.
4405 * Some math functions have been fixed, and there are new
4406 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4407 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4408 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4409 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4410 likely to be built into sbcl-0.7.0.)
4411 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4412 CL, ported to SBCL by Martin Atzmueller)
4413 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4414 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4415 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4416 * There's a new slam.sh hack to shorten the edit/compile/debug
4417 cycle for low-level changes to SBCL itself, and a new
4418 :SB-AFTER-XC-CORE target feature to control the generation of
4419 the after-xc.core file needed by slam.sh.
4420 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4421 COMPILE-FILE is no longer supported, so that now every function
4422 gets an entry point, so that block compilation looks a little
4423 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4424 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4425 returns the number of bytes consed since the system started,
4426 rather than the number consed since the first time the function
4427 was called. (The new definition parallels ANSI functions like
4428 CL:GET-INTERNAL-RUN-TIME.)
4429 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4430 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4431 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4432 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4433 * Fasl file version numbers are now independent of the target CPU,
4434 since historically most system changes which required version
4435 number changes have affected all CPUs equally. Similarly,
4436 the byte fasl file version is now equal to the ordinary
4439 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4440 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4441 is no longer recognized. I apologize for this, because it was
4442 listed in SB-EXT as a supported extension, but I found that
4443 its existing behavior was poorly specified, as well as incorrectly
4444 specified, and it looked like too much of a mess to straighten it
4445 out. I have enough on my hands trying to get ANSI stuff to work..
4446 * many patches ported from CMU CL by Martin Atzmueller, with
4447 half a dozen bug fixes in pretty-printing and the debugger, and
4448 half a dozen others elsewhere
4449 * fixed bug 13: Floating point infinities are now supported again.
4450 They might still be a little bit flaky, but thanks to bug reports
4451 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4452 as flaky as they were.
4453 * The --noprogrammer command line option is now supported. (Its
4454 behavior is slightly different in detail from what the old man
4455 page claimed it would do, but it's still appropriate under the
4456 same circumstances that the man page talks about.)
4457 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4458 are now supported, and enabled by default. Thus, the compiler can
4459 handle many floating point and complex operations much less
4460 inefficiently. (Thus e.g. you can implement a complex FFT
4462 * The compiler now detects type mismatches between DECLAIM FTYPE
4463 and DEFUN better, and implements CHECK-TYPE more correctly, and
4464 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4465 thanks to patches from Martin Atzmueller.
4466 * various fixes to make the cross-compiler more portable to
4467 ANSI-conforming-but-different cross-compilation hosts (notably
4468 Lispworks for Windows, following bug reports from Arthur Lemmens)
4469 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4470 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4471 fixed thanks to Pierre Mai's CMU CL patches.
4472 * fixes to keep the system from overflowing internal counters when
4473 it tries to use i/o buffers larger than 16M bytes
4474 * fixed bug 45a: Various internal functions required to support
4475 complex special functions have been merged from CMU CL sources.
4476 (When I was first setting up SBCL, I misunderstood a compile-time
4477 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4478 * improved support for type intersection and union, fixing bug 12
4479 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4480 more obscure bugs as well
4481 * some steps toward byte-compiling non-performance-critical
4482 parts of the system, courtesy of patches from Martin Atzmueller
4483 * Christophe Rhodes has made some debian packages of sbcl at
4484 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4485 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4486 stable, but are nonetheless usable. When he's ready, I'd be happy
4487 to add them to the SourceForge "File Releases" section. (And if
4488 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4489 * new fasl file format version number (because of changes in
4490 internal representation of (OR ..) types to accommodate the new
4491 support for (AND ..) types, among other things)
4493 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4494 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4496 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4498 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4499 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4500 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4501 * DESCRIBE now works on structure objects again.
4502 * Most function call argument type mismatches are now handled as
4503 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4504 know whether the function will be redefined before the call is
4505 executed. (The compiler could flag local calls with full WARNINGs,
4506 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4507 it doesn't keep track of enough information to know whether calls
4508 are local in this sense.)
4509 * Compiler output is now more verbose, with messages truncated
4510 later than before. (There should be some supported way for users
4511 to override the default verbosity, but I haven't decided how to
4512 provide it yet, so this behavior is still controlled by the internal
4513 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4514 src/compiler/ir1util.lisp.)
4515 * Fasl file format version numbers have increased again, because
4516 support for the Gray streams extension changes the layout of the
4517 system's STREAM objects.
4518 * The Gray subclassable streams extension now works, thanks to a
4519 patch from Martin Atzmueller.
4520 * The full LOAD-FOREIGN extension (not just the primitive
4521 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4522 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4523 but like most other programs, it defaults to copying the Unix
4524 environment from the original process instead of starting the
4525 new process in an empty environment.
4526 * Extensions which manipulate the Unix environment now support
4527 an :ENVIRONMENT keyword option which doesn't smash case or
4528 do other bad things. The CMU-CL-style :ENV option is retained
4529 for porting convenience.
4530 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4531 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4533 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4535 * A patch from Martin Atzmueller seems to have solved the SIGINT
4536 problem, and as far as we know, signal-handling now works cleanly.
4537 (If you find any new bugs, please report them!)
4538 * The system no longer defaults Lisp source file names to types
4539 ".l", ".cl", or ".lsp", but only to ".lisp".
4540 * The compiler no longer uses special default file extensions for
4541 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4542 seems to expect a single default extension for all compiled code,
4543 and there's no compelling reason to try to stretch the standard
4544 to allow two different extensions.) Instead, byte-compiled files
4545 default to the same extension as native-compiled files.
4546 * Fasl file format version numbers have increased again, because
4547 a rearrangement of internal implementation packages made some
4548 dumped symbols in old fasl files unreadable in new cores.
4549 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4550 many fewer weird special cases.
4551 * Bug #17 (differing COMPILE-FILE behavior between logical and
4552 physical pathnames) has been fixed, and some related misbehavior too,
4553 thanks to a patch from Martin Atzmueller.
4554 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4555 by Tim Moore, ported to SBCL by Martin Atzmueller.
4556 * Martin Atzmueller fixed several filesystem-related problems,
4557 including bug #36, in part by porting CMU CL patches, which were
4558 written in part by Paul Werkowski.
4559 * More compiler warnings in src/runtime/ are gone, thanks to
4560 more patches from Martin Atzmueller.
4561 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4564 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4566 * DESCRIBE now works on CONDITION objects.
4567 * The debugger now handles errors which arise when trying to print
4568 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4570 * The build system now uses an additional file, customize-target-features.lisp,
4571 to allow local modifications to the target *FEATURES* list. (The point of
4572 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4573 debugging features enabled, without having to worry about propagating it
4574 into everyone's system when I do a "cvs update".) When no
4575 customize-target-features.lisp file exists, the target *FEATURES* list
4576 should be constructed the same way as before.
4577 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4578 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4579 warn about behavior which is undefined under the ANSI spec. Note: This
4580 is specified by ANSI, but it's not very popular with programmers.
4581 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4582 macro in the SBCL sources for an example of a workaround which you
4583 might use to make portable ANSI-standard code which does what you want.
4584 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4585 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4586 when it's used as a non-toplevel form. (This is required in order
4587 to implement the DEFCONSTANT-EQX macro.)
4588 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4589 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4590 funny things, due to bugs in the implementation of EVAL-WHEN
4591 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4592 non-toplevel DEFCONSTANTs are uncommon.)
4593 * The core file version number and fasl file version number have been
4594 incremented, because the old noncompliant DEFCONSTANT behavior involved
4595 calling functions which no longer exist, and because I also took the
4596 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4597 * fixed bug 1 (error handling before read-eval-print loop starts), and
4598 redid debugger restarts and related debugger commands somewhat while
4600 ** The QUIT debugger command is gone, since it did something
4601 rather different than the SB-EXT:QUIT command, and since it never
4602 worked properly outside the main toplevel read/eval/print loop.
4603 Invoking the new TOPLEVEL restart provides the same functionality.
4604 ** The GO debugger command is also gone, since you can just invoke
4605 the CONTINUE restart directly instead.
4606 ** The TOP debugger command is also gone, since it's redundant with the
4607 FRAME 0 command, and since it interfered with abbreviations for the
4609 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4611 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4612 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4613 some porting work by Martin Atzmueller)
4614 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4615 with a boatload of patches which he ported to SBCL. Now that those
4617 ** The system tries to make sure that its low-priority messages
4618 are prefixed by semicolons, to help people who like to use
4619 syntax highlighting in their ILISP buffer. (This patch
4620 was originally due to Raymond Toy.)
4621 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4622 patches originally written by Raymond Toy.
4623 ** The compiler understands coercion between single-value and
4624 multiple-VALUES type expressions better, getting rid of some very
4625 weird behavior, thanks to patches originally by Robert MacLachlan
4626 and Douglas Crosher.
4627 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4628 lambda lists, thanks to a patch originally by Pierre Mai.
4629 ** The system no longer bogusly warns about "abbreviated type
4631 ** The compiler gets less confused by inlining and RETURN-FROM,
4632 thanks to some patches originally by Tim Moore.
4633 ** The system no longer hangs when dumping circular lists to fasl
4634 files, thanks to a patch originally from Douglas Crosher.
4635 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4636 undefined function error.
4637 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4638 in structure slot initforms) for now, documented workaround instead:-|
4639 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4640 * fixed bug 5: added stubs for various Gray stream functions called
4641 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4642 installed, at least appropriate type errors are generated
4643 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4644 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4645 consistently in DEFMETHOD forms.
4646 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4647 it doesn't seem to affect SBCL after all
4648 * The C runtime system now builds with better optimization and many
4649 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4651 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4653 * The system is now under CVS at SourceForge (instead of the
4654 CVS repository on my home machine).
4655 * The new signal handling code has been tweaked to treat register
4656 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4657 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4658 downstream, e.g. in debug-int.lisp, has implicit dependencies
4659 on the unsignedness of integer representation of machine words,
4660 and that caused the system to bomb out with infinite regress
4661 when trying to recover from type errors involving signed values,
4662 e.g. (BUTLAST '(1 2 3) -1).)
4663 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4664 (This was one of the bugs Peter Van Eynde reported back in July.)
4665 * The system now uses code inspired by Colin Walters' O(N)
4666 implementation of MAP (from the cmucl-imp@cons.org mailing
4667 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4668 inline the MAP operation, and there is more than one
4669 sequence argument to the MAP call (so that it can't just
4670 do ETYPECASE once and for all based on the type of the
4671 single sequence argument). (The old non-inline implementation
4672 of the general M-argument sequence-of-length-N case required
4673 O(M*N*N) time when any of the sequence arguments were LISTs.)
4674 * The QUIT :UNIX-CODE keyword argument has been renamed to
4675 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4676 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4678 * Raymond Wiker's patches to port dynamic loading from Linux to
4679 FreeBSD have been added.
4680 * The BUGS file is now more nearly up to date, thanks in large part
4681 to Martin Atzmueller's review of it.
4682 * The debugger now flushes standard output streams before it begins
4683 its output ("debugger invoked" and so forth).
4684 * The core version number and fasl file version number have both
4685 been incremented, because of incompatible changes in the layout
4687 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4688 used to not be called for a saved Lisp image.)
4689 * Martin Atzmueller's version of a patch to fix a compiler crash,
4690 as posted on sbcl-devel 13 September 2000, has been installed.
4691 * Instead of installing Martin Atzmueller's patch for the
4692 compiler transform for SUBSEQ, I deleted the compiler transform,
4693 and transforms for some similar consing operations.
4694 * A bug in signal handling which kept TRACE from working on OpenBSD
4696 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4697 away by constant folding
4698 * The system now defines its address space constants in one place
4699 (in the Lisp sources), and propagates them automatically elsewhere
4700 (through GENESIS and the sbcl.h file). Therefore, patching the
4701 address map is less unnecessarily tedious and error-prone. The
4702 Lisp names of address space constants have also been systematized.
4703 * CVS tags like dollar-Header-dollar have been removed from
4704 the sources, because they have never saved me trouble and
4705 they've been source of trouble working with patches and other
4706 diff-related operations.
4707 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4708 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4710 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4712 * The system has been ported to OpenBSD.
4713 * The system now compiles with a simple "sh make.sh" on the systems
4714 that it's supported on. I.e., now you no longer need to tweak
4715 text in the target-features.lisp-expr and symlinks in src/runtime/
4716 by hand, the make.sh takes care of it for you.
4717 * The system is no longer so grossly inefficient when compiling code
4718 involving vectors implemented as general (not simple) vectors (VECTOR T),
4719 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4720 which dares to use the various sequence functions on non-simple
4721 vectors, takes less of a performance hit.
4722 * There is now a primitive type predicate VECTOR-T-P
4723 to test for the (VECTOR T) type, so that e.g.
4724 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4725 can now be compiled with some semblance of efficiency. (The old code
4726 turned the type declaration into a full call to %TYPEP at runtime!)
4727 * AREF on (VECTOR T) is still not fast, since it's still compiled
4728 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4729 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4730 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4731 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4732 without making full calls to SUBTYPEP at runtime.
4733 (Some analogous efficiency issues for non-simple vectors specialized to
4734 element types other than T, or for non-simple multidimensional arrays,
4735 have not been addressed. They could almost certainly be handled the
4736 same way if anyone is motivated to do so.)
4737 * The changes in array handling break binary compatibility, so
4738 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4739 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4740 instead of (VALUES T T).
4741 * By following the instructions that Dan Barlow posted to sbcl-devel
4742 on 2 July 2000, I was able to enable primitive dynamic object
4743 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4744 functionality is not implemented (since it calls ld to resolve
4745 library references automatically, requiring RUN-PROGRAM for its
4746 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4747 try to resolve library references) is now supported.
4748 * The system now flushes the standard output streams when it terminates,
4749 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4750 them at several other probably-convenient times, e.g. in each pass of
4751 the toplevel read-eval-print loop, and after evaluating a form given
4752 as an "--eval" command-line option. (These changes were motivated by a
4753 discussion of stream flushing issues on cmucl-imp in August 2000.)
4754 * The source transform for TYPEP of array types no longer assumes
4755 that an array whose element type is a not-yet-defined type
4756 is implemented as an array of T, but instead punts, so that the
4757 type will be interpreted at runtime.
4758 * There is now some support for cross-compiling in make.sh: each of
4759 the phases of make.sh has its own script. (This should be transparent
4760 to people doing ordinary, non-cross-compile builds.)
4761 * Since my laptop doesn't have hundreds of megabytes of memory like
4762 my desktop machine, I became more motivated to do some items on
4763 my to-do list in order to reduce the size of the system a little:
4764 ** Arrange for various needed-only-at-cold-init things to be
4765 uninterned after cold init. To support this, those things have
4766 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4767 symbols with such names are now uninterned after cold init).
4768 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4769 fasl files for cold load.
4770 ** Remove the old compiler structure pooling code (which used to
4771 be conditional on the target feature :SB-ALLOC) completely.
4772 ** Redo the representation of some data in cold init to be more compact.
4773 (I also looked into supporting byte compiled code at bootstrap time,
4774 which would probably reduce the size of the system a lot, but that
4775 looked too complicated, so I punted for now.)
4776 * The maximum signal nesting depth in the src/runtime/ support code has
4777 been reduced from 4096 to 256. (I don't know any reason for the very
4778 large old value. If the new smaller value turns out to break something,
4779 I'll probably just bump it back up.)
4780 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4782 * Many, many bugs reported by Peter Van Eynde have been added to
4783 the BUGS list; some have even been fixed.
4784 * While enabling dynamic object file loading, I tried to make the
4785 code easier to understand, renaming various functions and variables
4786 with less ambiguous names, and changing some function calling
4787 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4788 * While trying to figure out how to do the OpenBSD port, I tried to
4789 clean up some of the code in src/runtime/. In particular, I dropped
4790 support for non-POSIX signal handling, added various comments,
4791 tweaked the code to reduce the number of compilation warnings, and
4792 renamed some files to increase consistency.
4793 * To support the new automatic configuration functionality in make.sh,
4794 the source file target-features.lisp-expr has been replaced with the
4795 source file base-target-features.lisp-expr and the machine-generated
4796 file local-target-features.lisp-expr.
4797 * fixed a stupid quoting error in make.sh so that using CMU CL
4798 "lisp -batch" as cross-compilation host works again
4800 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4802 * DESCRIBE no longer tries to call itself recursively to describe
4803 bound/fbound values, so that it no longer fails on symbols which are
4804 bound to themselves (like keywords, T, and NIL).
4805 * DESCRIBE now works on generic functions.
4806 * The printer now prints less-screwed-up representations of closures
4807 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4808 * A private symbol is used instead of the :EMPTY keyword previously
4809 used to mark empty slots in hash tables. Thus
4810 (DEFVAR *HT* (MAKE-HASH-TABLE))
4811 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4812 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4813 now does what ANSI says that it should. (You can still get
4814 similar noncompliant behavior if bang on the hash table
4815 implementation with all the symbols you get back from
4816 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4817 This breaks binary compatibility, since tests for equality to
4818 :EMPTY are wired into things like the macroexpansion of
4819 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4821 * There's now a minimal placeholder implementation for CL:STEP,
4822 as required by ANSI.
4823 * An obscure bug in the interaction of the normal compiler, the byte
4824 compiler, inlining, and structure predicates has been patched
4825 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4826 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4828 * Missing ordinary arguments in a macro call are now detected even
4829 when the macro lambda list contains &KEY or &REST.
4830 * The debugger no longer complains about encountering the top of the
4831 stack when you type "FRAME 0" to explicitly instruct it to go to
4832 the top of the stack. And it now prints the frame you request even
4833 if it's the current frame (instead of saying "You are here.").
4834 * As specified by ANSI, the system now always prints keywords
4835 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4837 * The default initial SIZE of HASH-TABLEs is now smaller.
4838 * Type information from CLOS class dispatch is now propagated
4839 into DEFMETHOD bodies, so that e.g.
4840 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4842 is now basically equivalent to
4843 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4844 (DECLARE (TYPE SINGLE-FLOAT X))
4846 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4847 operation, without having to do run-time type dispatch.
4848 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4849 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4850 * Since I don't seem to be making big file reorganizations very often
4851 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4852 is overflowing my ability to conveniently back them up), I've finally
4853 checked the system into CVS. (The CVS repository is on my home system,
4854 not at SourceForge -- putting it on SourceForge might come later.)
4855 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4856 high-level GC-NOTIFY-FOO functions send their output. (There's
4857 still very little control of where low-level verbose GC functions
4858 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4859 less than it used to -- the GC-NOTIFY-FOO functions are now under
4860 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4861 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4862 in only one place in the source code, and propagates it automatically
4863 everywhere that it's needed. Thus e.g. when I bump the version from
4864 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4865 * The C source files now include boilerplate legalese and documentation
4866 at the head of each file (just as the Lisp source files already did).
4867 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4868 to his page will be replaced with a link to his new CLiki service.
4870 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4872 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4873 * The build process now looks for GNU make under the default name "gmake",
4874 instead of "make" as it used to. If GNU make is not available as "gmake"
4875 on your system, you can change this default behavior by setting the
4876 GNUMAKE environment variable.
4877 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4878 can build without error under CMU CL.
4880 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4882 * There is now a partial SBCL user manual (with some new text and some
4883 text cribbed from the CMU CL manual).
4884 * The beginnings of a profiler have been added (starting with the
4885 CMU CL profiler and simplifying and cleaning up). Eventually the
4886 main interface should be through the TRACE macro, but for now,
4887 it's still accessed through vaguely CMU-CL-style functions and macros
4888 exported from the package SB-PROFILE.
4889 * Some problems left over from porting CMU CL to the new
4890 cross-compilation bootstrap process have been cleaned up:
4891 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4892 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4893 part when getting rid of the compiler *BACKEND* stuff.)
4894 ** The value of *NULL-TYPE* was screwed up, because it was
4895 being initialized before the type system knew the final
4896 definition of the 'NULL type. This screwed up several key
4897 optimizations in the compiler, causing inefficiency in all sorts
4898 of places. (I found it because I wanted to understand why
4899 GET-INTERNAL-RUN-TIME was consing.)
4900 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4901 PROCLAIM FTYPE information. Unfortunately this broke binary
4902 compatibility again, since now the forms output by DEFGENERIC
4903 to refer to functions which didn't exist in 0.6.3.
4904 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4905 can use the new (as of 0.6.3) transform for SOME into MAP into
4907 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4908 values to (INTEGER 0 1000000), so that the time code will no longer
4909 occasionally get blown up by Linux returning 1000000 microseconds
4910 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4911 its output conform to the ANSI spec. (Alas, this makes its output
4912 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4913 * A full call to MAP NIL with a single sequence argument no longer conses.
4914 * fixes to problems pointed out by Martin Atzmueller:
4915 * The manual page no longer talks about multiprocessing as though
4916 it were currently supported.
4917 * The ILISP support patches have been removed from the distribution,
4918 because as of version 5.10.1, ILISP now supports SBCL without us
4919 having to maintain patches.
4920 * added a modified version of Raymond Toy's recent CMU CL patch for
4921 EQUALP comparison of HASH-TABLE
4923 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4925 * The system still can't cross-compile itself with
4926 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4927 entails), but at least it can compile more of itself that way
4928 than it used to be able to, and various buglets which were uncovered
4929 by trying to cross-compile itself that way have now been fixed.
4930 * This release breaks binary compatibility again. This time
4931 at least I've incremented the FASL file format version to 2, so that the
4932 problem can be detected reliably instead of just causing weird errors.
4933 * various new style warnings:
4934 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4935 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4936 ** using the lexical binding of a variable named in the *FOO* style
4937 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4938 as specified by ANSI.
4939 * *RANDOM-STATE* is no longer automatically initialized from
4940 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4941 default behavior of the system is to repeat its behavior every time
4942 it's run. If you'd like to change this behavior, you can always
4943 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4944 old convention there was no comparably easy way to get the system to
4945 repeat its behavior every time it was run.
4946 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4947 FTYPE declarations has been removed, in favor of their ANSI
4948 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4949 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4950 the types of their sequence arguments can be determined at compile time.
4951 This is done through a new open code expansion for MAP which eliminates
4952 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4953 argument types can be determined at compile time.
4954 * The optimizer now transforms COERCE into an identity operation when it
4955 can prove that the coerced object is already of the correct type. (This
4956 can be a win for machine generated code, including the output of other
4957 optimization transforms, such as the MAP transform above.)
4958 * Credit information has been moved from source file headers into CREDITS.
4959 * Source file headers have been made more standard.
4960 * The CASE macro now compiles without complaining even when it has
4963 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4965 * (Note that the way that the PCL macroexpansions were rewritten
4966 to accommodate the change in DEFGENERIC below breaks binary
4967 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4968 not run under sbcl-0.6.2. Once we get out of alpha releases,
4969 i.e. hit release 1.0.0, we'll probably try to maintain binary
4970 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4971 and sbcl-1.4.4. Until then, however, it might be fairly common
4972 for maintenance releases to break binary compatibility.)
4973 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4975 * The reporting of unhandled signals has been changed to print some
4976 explanatory text as well as the report form. (Previously only
4977 the report form was printed.)
4978 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4979 it defines, so that the compiler no longer issues undefined function
4980 warnings for compiled-but-not-yet-loaded generic functions.
4981 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4982 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4983 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4984 section of the ANSI spec are supported.
4985 * The "" nickname for the "KEYWORD" package has been removed.
4986 The reader still handles symbol tokens which begin with a package marker
4987 as keywords, but it doesn't expose its mechanism for doing so in the
4988 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4989 * The system now issues STYLE-WARNINGs for contradictory TYPE
4990 proclamations. (Warnings for contradictory FTYPE proclamations would
4991 be nice too, but those can't be done usefully unless the type system
4992 is made smarter about FUNCTION types.)
4993 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4994 systematized, so that "*target-*.lisp is supposed to exist only on the
4995 target and imply that there's a related file which exists on the
4996 host, and *host-*.lisp is supposed to exist only on the host and imply
4997 that there's a related file which exists on the target. This involves a
4998 lot of renaming. Hopefully the acute confusion caused by the renaming
4999 will be justified by the reduction in chronic confusion..
5000 ** runtime-type.lisp -> early-target-type.lisp
5001 ** target-type.lisp -> late-target-type.lisp
5002 ** early-host-format.lisp -> early-format.lisp
5003 ** late-host-format.lisp -> late-format.lisp
5004 ** host-error.lisp -> misc-error.lisp
5005 ** early-error.lisp -> early-target-error.lisp
5006 ** late-error.lisp -> late-target-error.lisp
5007 ** host-defboot.lisp -> early-defboot.lisp
5008 ** code/misc.lisp -> code/target-misc.lisp
5009 ** code/host-misc.lisp -> code/misc.lisp
5010 ** code/numbers.lisp -> code/target-numbers.lisp
5011 ** code/early-numbers.lisp -> numbers.lisp
5012 ** early-host-type.lisp -> early-type.lisp
5013 ** late-host-type.lisp -> late-type.lisp
5014 ** host-typep.lisp -> typep.lisp
5015 ** load.lisp -> target-load.lisp
5016 ** host-load.lisp -> load.lisp
5017 ** host-disassem.lisp -> disassem.lisp
5018 ** host-insts.lisp -> insts.lisp
5019 ** byte-comp.lisp -> target-byte-comp.lisp
5020 ** host-byte-comp.lisp -> byte-comp.lisp
5021 ** host-signal.lisp -> signal.lisp
5022 ** host-defstruct.lisp -> defstruct.lisp
5023 ** late-target-type.lisp -> deftypes-for-target.lisp
5024 Furthermore, several other previously target-only files foo.lisp (e.g.
5025 hash-table.lisp and random.lisp) have been split into a target-and-host
5026 foo.lisp file and a target-only target-foo.lisp file, with their key type
5027 definitions in the target-and-host part, so that the cross-compiler will
5028 know more about target types.
5029 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
5030 gone away. In their place are various *BACKEND-FOO* variables
5031 corresponding to the slots of the old structure.
5032 * A bug which caused the SB-COLD bootstrap-time package to be propagated
5033 into the target SBCL has been fixed.
5034 * The chill.lisp system for loading cold code into a running SBCL
5036 * Support for the CMU CL "scavenger hook" extension has been removed.
5037 (It was undocumented and unused in the CMU CL sources that SBCL was
5038 derived from, and stale in sbcl-0.6.1.)
5039 * Various errors in the cross-compiler type system were detected
5040 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
5041 (enabling various consistency checks). Many of them were fixed,
5042 but some hard problems remain, so the compiler is back to
5043 running without *TYPE-SYSTEM-INITIALIZED* for now.
5044 * As part of the cross-compiler type system cleanup, I implemented
5045 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
5046 * I have started adding UNCROSS calls throughout the type system
5047 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
5048 on cross-compiler input files will be able to go away, and various
5050 * CONSTANTP now returns true for quoted forms (as explicitly required
5053 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
5055 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
5056 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
5057 (MAKE-LIST -1). In the longer term, it ought to become true
5058 that declarations are assertions even at SAFETY 1. For now, it's not
5059 quite true even at SAFETY 3, but it's at least more nearly true..
5060 (Note that this change seems to increases the size of the system by
5061 O(5%) and to decrease the speed of the compiler by 20% or more.)
5062 * changed ALIEN printing to be much more abbreviated, as a short-term fix
5063 for the problem of printing dozens of lines of distracting information
5064 about low-level system machinery as part of the top stack frame
5065 on entry to the debugger when an undefined function was called.
5066 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
5067 is not reset to COMMON-LISP-USER.
5068 * Compilation of stuff related to dyncount.lisp has been made conditional
5069 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
5070 smaller. The various dyncount-related symbols have been moved into
5071 a new "SB-DYNCOUNT" package.
5072 * tty-inspect.lisp has been renamed to inspect.lisp.
5073 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
5074 feature has gone away. (When we eventually port to other flavors of
5075 libc and/or Unix, we'll try to make the differences between flavors
5076 invisible at the user level.)
5077 * Various other *FEATURES* tags, and/or their associated conditionals,
5078 have been removed if obsolescent, or given better documentation, or
5079 sometimes given more-mnemonic names.
5081 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
5083 * tidied up "make.sh" script
5084 * tidied up system directory structure
5085 * better "clean.sh" behavior
5086 * added doc/FOR-CMUCL-DEVELOPERS
5087 * many many small tweaks to output format, e.g. removing possibly-confusing
5088 trailing #\. character in DESCRIBE-INSTANCE
5089 * (EQUALP #\A 'A) no longer signals an error.
5090 * new hashing code, including EQUALP hashing
5091 * tidied up Lisp initialization and toplevel
5092 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
5093 * command line argument processing
5094 * added POSIX-GETENV function to deal with Unix-ish environment variables
5095 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
5096 terminating SBCL on EOF
5097 * non-verbose GC by default
5098 * There is no more "sbcl" shell script; the sbcl file is now the C
5099 runtime executable (just like CMU CL).
5100 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
5102 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
5103 symbol and package information as Lisp native symbol and package objects
5104 instead of strings naming symbols and strings naming packages. This way,
5105 whenever packages are renamed (as in warm init), debug information is
5106 transformed along with everything else.
5107 * tweaked the optimization policy declarations which control the building
5108 of SBCL itself. Now, among other things, the system no longer saves
5109 source location debugging information. (This helps two problems at once
5110 by reducing SBCL size and by keeping SBCL from trying to look for its
5111 sources -- which may not exist -- when reporting errors.)
5112 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
5113 debugging and testing purposes
5114 * cleaned up printing, making the printer call PRINT-OBJECT for
5115 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
5116 methods, giving nearly-ANSI behavior
5117 * converted almost all special variables to use *FOO* naming convention
5118 * deleted PARSE-TIME functionality, since it can be done portably
5119 * moved some files out of cold init into warm init
5120 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
5122 * regularized formatting of source files
5123 * added an install.sh script
5124 * fixed ridiculous memory usage of cross-compiler by making
5125 compiler/alloc.lisp not try to do pooling unless it can hook
5126 itself into the GC of the cross-compilation host. Now the system
5127 builds nicely on my old laptop.
5128 * added :SB-ALLOC in target-features.lisp-expr
5129 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
5130 was not implemented)
5131 * re-did condition handling and note reporting in the compiler. Notes
5132 are no longer handled by signalling conditions. Style warnings
5133 and warnings are handled more correctly and reported in such a way
5134 that it's easy to find one or the other in your output (so that you
5135 can e.g. figure out which of many problems caused COMPILE-FILE to
5137 * changed the severity of several compiler warnings from full WARNING
5138 to STYLE-WARNING in order to conform with the ANSI spec; also changed
5139 compiler note reporting so that it doesn't use the condition system
5140 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5141 COMPILE-FILE command)
5142 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5143 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5144 no longer supported, since their implementation was deeply intertwingled
5145 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5146 * removed bogus "support" for compiler macros named (SETF FOO), and
5147 removed the compiler macro for SETF INFO (but only after making a fool
5148 of myself on the cmucl-imp mailing list by posting a bogus patch for
5149 DEFINE-COMPILER-MACRO..)
5150 * Compiled files containing forms which have side effects on the Lisp
5151 reader (such as DEFPACKAGE forms) are now handled more correctly.
5152 (Compiler queuing of top level lambdas has been suppressed by setting
5153 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5154 * deleted various currently-unused source files, e.g. gengc.lisp. They
5155 may be added back at some point e.g. when porting to other architectures,
5156 but until they are it's distracting to distribute them and to try to
5158 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5159 there were so many of them they're just distractions, and UNCROSS is
5160 known to be able to handle the current sources
5161 * moved PROFILE functionality into TRACE, so that it will be clear
5162 how the wrapping and unwrapping of functions when you profile them
5163 interacts with the wrapping and unwrapping of functions when you
5164 trace them. (Actually, the functionality isn't there yet, but at least
5165 the interface specification is there. Hopefully, the functionality will
5166 arrive with some future maintenance release.)
5167 * removed host-oops.lisp
5168 * changed signature of QUIT function to allow UNIX-CODE argument
5169 * fixed READ-SEQUENCE bug
5170 * tweaked verbose GC output so that it looks more like the progress
5171 output that ANSI specifies for functions like LOAD
5172 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5173 * added <http://sbcl.sourceforge.com> to the banner information printed by