1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes in sbcl-1.0.26 relative to 1.0.25:
3 * new feature: runtime option --disable-ldb
4 * new feature: runtime option --lose-on-corruption to die at the
5 slightest hint of possibly non-recoverable errors: running out of
6 memory, stack, alien stack, binding stack, encountering a memory
7 fault, etc. In the absence of --lose-on-corruption a warning is
10 changes in sbcl-1.0.25 relative to 1.0.24:
11 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
12 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
13 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
14 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
15 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
16 better name and does not return values so stale on multiprocessor systems.
17 Also, HOLDING-MUTEX-P was added for about the only sane usage of
19 * improvement: unithread builds keep track of MUTEX-VALUE.
20 * improvement: reading from a TWO-WAY-STREAM does not touch the output
21 stream anymore making it thread safe to have a concurrent reader and
22 a writer, for instance, in a pipe.
23 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
24 arguments, making compiler-macros for SETF-functions able to inspect
25 their constant arguments.
26 * improvement: COMPILE-FILE reports times with millisecond accuracy
27 (thanks to Luis Oliveira)
28 * optimization: CHAR-CODE type derivation has been improved, making
29 TYPEP elimination on subtypes of CHARACTER work better. (reported
30 by Tobias Rittweiler, patch by Paul Khuong)
31 * bug fix: setting alien structure fields of type struct by value now
32 computes the right offset for the memory copy.
33 * bug fix: compilation problem involving inlined calls to aliens with
34 result type VOID. (reported by Ken Olum)
35 * bug fix: #235a; sequential inline expasion in different policies no
36 longer reuses the functional from the previous expansion site.
37 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
38 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
39 * bug fix: named ENUMs in multiply-referenced alien record types no longer
40 cause an implied type redefinition. Regression from 1.0.21.29.
41 * bug fix: sign-extension of small signed return values now works with C
42 code compiled with gcc 4.3 or newer. (reported by Liam Healy)
43 * improvements to the Windows port:
44 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
45 Westfold, thanks to Rudi Schlatte)
47 changes in sbcl-1.0.24 relative to 1.0.23:
48 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
49 vector of a multidimensional SIMPLE-ARRAY.
50 * new feature: the system now signals a continuable error if standard
51 readtable modification is attempted. (thanks to Tobias Rittweiler)
52 * new feature: DIRECTORY has been extended with a non-standard keyword
53 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
54 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
55 has been robustified and documented better. (thanks to Robert Goldman)
56 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
57 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
58 * tradeoff: constant FORMAT control strings are now compiled unless
59 SPACE > SPEED (previously only when SPEED > SPACE.)
60 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
62 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
64 * bug fix: lexical type declarations are now correctly reported by
65 SB-CLTL2. (reported by Larry D'Anna)
66 * bug fix: STRING-TO-OCTETS did not handle :START properly when
67 using UTF-8 as external format. (thanks to Luis Oliveira)
68 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
69 are signalled correctly. (thanks to Stas Boukarev)
70 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
71 designator. (thanks to Tobias Rittweiler)
72 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
73 designator, and returns T instead of the function. (thanks to
75 * bug fix: direct superclasses of STANDARD-CLASS and
76 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
77 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
78 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
79 when no class with the named slot yet exists no longer causes a
80 compile-time style-warning.
81 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
82 in safe code. (reported by Didier Verna)
83 * bug fix: #430; nested structure constructors can stack allocate.
84 * bug fix: on some 64-bit platforms dynamic space size was truncated
85 to #xffff0000 bytes. (reported by Benjamin Lambert)
86 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
87 REPL unusable. (reported by Daniel Herring)
88 * bug fix: return values of READ-SEQUENCE did not take :START into
89 account on file streams, regressions since 1.0.12.22. (reported by
90 Thomas Russ, patch by Paul Khuong)
91 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
92 method specializer used to confuse permuation vector optimization.
93 * bug fix: system inserted bogus implicit type declarations for local
94 special variables in DEFMETHOD bodies.
95 * bug fix: #354; duplicated frames in backtraces due to
96 non-tail-call-optimized XEPs to functions with return type NIL
98 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
99 initializes structure object slots according to DEFSTRUCT initforms,
100 and DEFSTRUCT forms :INCLUDEind structure classes defined using
101 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
102 (reported by Bruno Haible and Stephen Wilson)
103 * bug fix: #395; fill-pointer output streams used now support
104 element-type BASE-CHAR as well.
105 * bug fix: compiler error when attempting to derive return value of
106 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
108 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
110 * bug fix: Mac OS X binaries should now be portable between Leopard
113 changes in sbcl-1.0.23 relative to 1.0.22:
114 * enhancement: when disassembling method functions, disassembly
115 for the associated fast function is also produced.
116 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
118 * optimization: printing with *PRINT-PRETTY* true is now more
119 efficient as long as the object being printed doesn't require
120 special handling by the pretty printer.
121 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
122 now interact correctly with type declarations.
123 * partial bug fix: PCL detects infinite recursion during wrapper
124 validation. (thanks to Attila Lendvai)
125 * bug fix: #426; nested function calls are inlined properly.
126 Previously if FOO was an inline function, in calls of the form
127 (FOO (FOO ...)) the outer call was not inlined.
128 * bug fix: long long arguments passed in registers to alien
129 functions on PowerPC Linux and NetBSD are now handled correctly
130 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
132 changes in sbcl-1.0.22 relative to 1.0.21:
133 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
134 for the shared object in the current directory, but passes the native
135 namestring of the designated pathname to the operation system's shared
136 object loading function as-it.
137 * minor incompatible change: --disable-debugger toplevel option now takes
138 effect before processing of initialization files and --eval or --load
140 * new feature: new commandline argument: --script, which supports
141 shebang lines. See documentation for details. (based on work by
143 * new feature: SAVE-LISP-AND-DIE can save current values of
144 --dynamic-space-size and --control-stack-size in the executable core,
145 causing it to skip normal runtime option processing. See documentation
146 for details. (thanks to Zach Beane)
147 * enhancement: inoccous calls to EVAL or generic functions dispatching
148 on subclasses of eg. STREAM no longer cause compiler notes to appear.
149 * enhancement: the system no longer resignals errors from --load and
150 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
151 associated with the original error to be lost. (thanks to Ariel
153 * enhancement: :DONT-SAVE keyword argument has been added to
154 LOAD-SHARED-OBJECT for controlling interaction with
155 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
156 the effects of an earlier LOAD-SHARED-OBJECT call.
157 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
158 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
159 provided. (reported by Cedric St-Jean)
160 * bug fix: circularity handling in the reader did not treat raw
161 structure slots correctly. (reported by Cedric St-Jean)
162 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
163 file descriptors when there were none.
164 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
165 was broken. (reporteed by Willem Broekema)
166 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
167 pathnames without a directory.
168 * bug fix: COMPILE-FILE did not warn about undefined variable
169 references at toplevel, and LOAD of the resulting fasl did
171 * bug fix: functions with non-required arguments used to end up with
172 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
173 * bug fix: redefining a function with non-required arguments didn't
174 update the system's knowledge about its call signature properly.
175 * bug fix: fixed #431; incompatible alien record type redefinitions
176 are detected and handled. (reported by Neil Haven)
177 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
178 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
180 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
181 after alien stack frames.
182 * bug fix: strings passed to mk{s,d}temp lacked null termination
184 changes in sbcl-1.0.21 relative to 1.0.20:
185 * new feature: the compiler is able to track the effective type of a
186 generic function across method addition and removal.
187 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
188 appropriately typed structure slots without locking.
189 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
190 information like those gathered by TIME using a programming-friendly
192 * new feature: TIME reports time taken even if the form performs a
193 non-local transfer of control.
194 * enhancement: reduced conservativism on GENCGC platforms: on
195 average 45% less pages pinned (measured from SBCL self build).
196 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
197 mutate constant symbols or violate declaimed type of the symbol.
198 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
199 sleeping on the semaphore in heavily contested semaphores.
200 * bug fix: semaphores and condition variables were not interrupt
202 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
203 owned by other threads anymore.
204 * bug fix: FIND on lists called KEY outside the specified
205 subsequence. (reported by budden)
206 * bug fix: LOG doesn't use single-float intermediate results when
207 given mixed integer and double-float arguments, leading to better
208 precision. (reported by Bob Felts)
209 * bug fix: LOG with base zero returned values of inconsistent type.
211 changes in sbcl-1.0.20 relative to 1.0.19:
212 * minor incompatible change: OPTIMIZE qualities
213 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
214 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
215 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
216 * documentation: some slot access efficiency guidelines have been
217 added to the user manual.
218 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
219 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
220 as ASSOC and MEMEBER.
221 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
222 to more efficient EQ-comparison versions more often.
223 * optimization: enhanced derivation of DOLIST iteration variable type
225 * optimization: constant folding of simple (LIST ...) forms as DOLIST
227 * optimization: runtime lookup of function definitions can be
228 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
229 * optimization: compiler is able to derive the return type of
230 (AREF (THE STRING X) Y) as being CHARACTER.
231 * optimization: CLRHASH on empty hash-tables no longer does pointless
232 work. (thanks to Alec Berryman)
233 * optimization: REPLACE deftransforms don't punt when :START1 or
235 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
236 no longer emits a code deletion note. (reported by Andrew
238 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
239 functions with an odd number of &OPTIONAL arguments, a &REST
240 argument, and one or more &KEY arguments at the call site.
241 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
242 arguments appeared at call sites as well.
243 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
244 READ-LINE &co can no longer read from them afterwards. (reported
246 * bug fix: fixed #427: unused local aliens no longer cause compiler
247 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
249 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
250 alien-stack leakage. (reported by Andy Hefner)
251 * bug fix: PROGV signals an error when an attempt to violate declared
252 type of a variable or bind a constant is made.
253 * bug fix: SET signals an error when an attempt to violate declared
254 type of a variable is made.
255 * bug fix: restart computation during the execution of a restart
256 test function no longer causes infinite recursion. (reported by
258 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
259 signals an error. (thanks to Pierre Mai)
260 * bug fix: EXPORT left symbol unexported in conflict situations.
261 (thanks to Michael Weber)
262 * bug fix: correctly handle name conflicts involving CL:NIL.
263 (thanks to Michael Weber)
264 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
265 conflicts arising from USEing package with conflicting symbols
266 correctly. (thanks to Michael Weber)
268 changes in sbcl-1.0.19 relative to 1.0.18:
269 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
270 warnings that go otherwise unhandled will be muffled if they are
271 of the type that's the value of this variable.
272 * optimization: stack allocation is slightly more efficient on x86
274 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
275 where a raw slot always is initialized using the initform whose
276 type is not know sufficiently well a compile-time are now compiled
277 correctly. (reported by John Morrison)
278 * bug fix: compiler no longer makes erronous assumptions in the
279 presence of non-foldable SATISFIES types.
280 * bug fix: stack analysis missed cleanups of dynamic-extent
281 arguments in non-let-converted calls to local functions.
282 * improvements to the Windows port:
283 ** adjusted address spaces for building on both Win32 and
284 Win64. (thanks for John Connors)
285 * fixed some bugs revealed by Paul Dietz' test suite:
286 ** interval arithmetic during type derivation used inexact integer
287 to single-float coercions.
288 ** arithmetic operations involving large integers and single
289 floats give the same results in compiled and interpreted code.
290 ** deriving the result type of COERCE no longer signals an error
291 if the derived type of the second argument is a MEMBER type
292 containing invalid type specifiers.
293 ** ADJOIN with constant NIL as second argument works correctly.
294 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
296 changes in sbcl-1.0.18 relative to 1.0.17:
297 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
298 profiles only the current thread.
299 * minor incompatible change: changes to SYMBOL-VALUE of constants
300 defined with DEFCONSTANT now signal an error.
301 * enhancement: SB-SPROF now has support for wallclock profiling,
302 and is also able to profile specific threads. REPORT output
303 has also additional sorting options.
304 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
306 * optimization: structure allocation has been improved
307 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
308 ** out of line constructors are ~10% faster.
309 ** inline constructors are ~15% faster.
310 ** inline constructors are capable of dynamic extent allocation
311 (generally on x86 and x86-64, in some cases on other platforms
313 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
315 * optimization: file compiler is now able to coalesce non-circular
316 lists, non-base strings, and bit-vectors. Additionally, constants
317 are never referenced through SYMBOL-VALUE at runtime.
318 * optimization: code defining methods on PRINT-OBJECT (and other
319 generic functions in the COMMON-LISP package) now loads faster.
320 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
322 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
323 fasl is now deleted. (reported by Attila Lendvai)
324 * bug fix: READ-LINE always returned NIL for the last line in files.
325 (reported by Yoshinori Tahara)
326 * bug fix: more accurate disassembly annotations of foreign function
327 calls. (thanks to Andy Hefner)
328 * bug fix: trimming non-simple strings and non-string string
329 designators when the there is nothing to trim works properly.
330 (thanks to James Knight)
331 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
332 (thanks to Travis Cross)
333 * fixed some bugs revealed by Paul Dietz' test suite:
334 ** NIL is a valid function name (regression at 1.0.13.38)
335 ** FILL on lists was missing its return value (regression at 1.0.12.27)
336 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
337 fill pointers properly (regression at 1.0.12.23)
338 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
339 displacement indices properly (regression at 1.0.12.23)
341 changes in sbcl-1.0.17 relative to 1.0.16:
342 * temporary regression: user code can no longer allocate closure
343 variable storage on stack, due to bug 419 without explicitly
344 requesting it. Please consult sbcl-devel for advice if you need to
345 use this feature in the meanwhile.
346 * new feature: runtime argument --control-stack-size can be used to
347 adjust thread default control stack size.
348 * enhancement: improved TIME output
349 ** all times are reported using the measured accuracy (milliseconds
350 for real and GC times, microseconds for everything else.)
351 ** processor cycle counts on x86 and x86-64.
352 ** interpreted forms are counted for both evaluator modes.
353 ** number of lambdas converted by the compiler is reported.
354 ** CPU percentage report (computed from real and total run time.)
355 ** more comprehensive run time reporting, using a condenced format
356 ** interperted form, lambda, and page fault counts are omitted
358 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
360 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
361 * optimization: two argument forms of LAST are upto ~10% faster
362 in normal SPEED policies.
363 * optimization: NCONC no longer needs to heap cons its &REST list
364 in normal SPEED policies.
365 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
367 * bug fix: UNION and NUNION work with :TEST-NOT once more,
368 regression since 1.0.9.1. (thanks to Eric Marsden)
369 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
370 since 1.0.15.36. (thanks to Paul Khuong)
371 * bug fix: LAST when always returned the whole list when given a bignum
372 as the second argument.
373 * bug fix: dynamic extent allocation of nested lists and vectors
374 could leak to otherwise accessible parts.
375 * bug fix: invalid optimization of heap-allocated alien variable
377 * bug fix: fasl header checking is less vulnerable to different
378 platform word lengths.
379 * bug fix: more correct assembler syntax for GNU binutils
380 2.18.50.0.4 support. (thanks to Marijn Schouten)
381 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
383 * bug fix: the fopcompiler can handle LOCALLY forms (with no
384 declarations) successfully. (reported by Attila Lendvai)
386 changes in sbcl-1.0.16 relative to 1.0.15:
387 * minor incompatible change: revert the changes to sb-posix's error
388 signaling added in 1.0.14.
389 * minor incompatible change: change PROBE-FILE back to returning
390 NIL whenever we can't get a truename, as was the case before 1.0.14.
391 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
392 inherits from ERROR instead of just CONDITION.
393 * new feature: SB-INTROSPECT can provide source locations for instances
394 as well. (thanks to Tobias Rittweiler)
395 * optimization: binding special variables now generates smaller code
396 on threaded platforms.
397 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
398 and cases where no :TEST is given but the compiler can infer that
399 the element to search is of type (OR FIXNUM (NOT NUMBER)).
400 * optimization: better LOGNOT on fixnums.
401 * optimization: modular arithmetic for a particular requested width
402 is implemented using a tagged representation unless a better
403 representation is available.
404 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
405 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
406 for wellformedness and returns a specialized vector. (reported by
407 Francois-Rene Rideau)
408 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
409 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
410 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
411 unscheduling timer at the same time as another timer fires.
412 * bug fix: don't reschedule timers for dead threads.
413 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
414 * bug fix: copying output from RUN-PROGRAM to a stream signalled
415 bogus errors if select() was interrupted.
416 * enhancement: add support for fcntl's struct flock to SB-POSIX.
418 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
419 * enhancement: cleaner backtraces for interactive interrupts, as
420 well as other cases where the interesting frames used to be
421 obscured by interrupt handling frames.
422 * enhancement: untracing a whole package using (UNTRACE "FOO") is
423 now supported, and tracing a whole package using (TRACE "FOO") now
424 traces SETF-functions as well.
425 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
426 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
427 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
428 * unreadably printed representation of hash-tables now includes
430 * bug fix: partially fixed #188: type propagation from assignments
431 is now more efficient.
432 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
433 'DOUBLE-FLOAT) are not flushable.
434 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
435 full address of the object, and none of the tag bits.
436 * bug fix: readably printing hash-tables now respects other printer
437 control variables. (reported by Cedric St-Jean)
438 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
439 keyword with MAKE-HASH-TABLE.
440 * bug fix: export SB-POSIX:MKSTEMP.
441 * bug fix: SORT was not interrupt safe.
442 * bug fix: XREF accounts for the last node of each basic-block as
444 * bug fix: MAKE-INSTANCE optimizations interacted badly with
445 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
446 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
447 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
448 create obsolete instances in the case of redefinition or
449 obsoletion of a superclass. (thanks to Andy Hefner)
450 * bug fix: Support for the Alpha architecture has been revived; it had
451 suffered somewhat from lack of maintenance since sbcl-1.0.
452 * improvements to the Windows port:
453 ** The system detects the codepage to use at startup, instead of
454 using the value from the saved core. (thanks to Kei Suzuki)
456 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
457 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
458 (see documentation for details.)
459 * revived support for OpenBSD (contributed by Josh Elsasser)
460 * partially fixed bug #108: ROOM no longer suffers from occasional
461 (AVER (SAP= CURRENT END)) failures.
462 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
463 bodies no longer cause a WARNING to be signalled. (reported by
465 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
466 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
467 the bug in TRUNCATE.) (reported by Michael Weber)
468 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
469 no samples. (reported by Andy Hefner)
470 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
471 no longer appear as (NIL ...) frames in backtraces.
472 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
473 is now actually exported from SB-EXT as documented. (reported by
475 * bug fix: sb-aclrepl now correctly understands how to inspect
476 single-floats on 64-bit platforms where single-floats are not boxed.
477 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
478 yet been finalized. (reported by Levente Meszaros)
479 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
480 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
481 SBCL-specific optimize qualities.
483 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
484 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
485 an executable in the search path, and does so in the child
486 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
487 been removed; it can be found in the CVS history, for somebody who
488 needs that search behavior (see the manual).
489 * minor incompatible change: compiler policy re. weakening type
490 checks has changed: now type checks are weakened only if SAFETY < 2
492 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
493 unparsing of directory pathnames as files. Analogously,
494 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
495 filename to parse into a directory pathname.
496 * enhancement: implicit generic function creation now signals a
497 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
498 users can bind handlers for and muffle around calls to LOAD.
499 * enhancement: RUN-PROGRAM allows unicode arguments and environments
500 to be used (using the default stream external format), and allows
501 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
502 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
503 strings and vectors whose element-type or simplicity is not fully
504 known at compile-time.
505 * optimization: STRING-TRIM and related functions no longer allocate
506 a new string when no trimming needs to be performed. These functions
507 are also faster than before when the input string has been declared
509 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
510 * optimization: READ-LINE is significantly faster for files containing
512 * optimization: non-open coded uses of character comparison operators
513 (e.g. char=) no longer cons when called with more than one parameter
514 on platforms supporting dynamic-extent allocation.
515 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
516 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
517 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
519 * bug fix: some sequence functions elided bounds checking when
521 * bug fix: too liberal weakening of union-type checks when SPEED >
523 * bug fix: more bogus fixnum declarations in ROOM implementation
526 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
527 * new feature: MAKE-HASH-TABLE now experimentally accepts a
528 :SYNCHRONIZED argument, which makes the hash-table safe for
529 concurrent accesses (but not iteration.) See also:
530 SB-EXT:WITH-LOCKED-HASH-TABLE, and
531 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
532 * optimization: CONCATENATE on strings is an order of magnitude faster
533 in code compiled with (> SPEED SPACE).
534 * optimization: SUBSEQ is ~50% faster on lists.
535 * bug fix: bug 417 fixed -- source location reporting is now more robust.
536 * bug fix: SUBSEQ on a list will now correctly signal an error if if
537 END is smaller then START.
538 * bug fix: SB-PROFILE will no longer report extra consing for nested
539 calls to profiled functions.
540 * bug fix: ROOM implementation had bogus fixnum declarations which
541 could cause type-errors when calling ROOM on large images.
542 * bug fix: if file compilation is aborted, the partial fasl is now
543 deleted, and COMPILE-FILE returns NIL as the primary value.
544 * bug fix: number of thread safety issues relating to SBCL's internal
545 hash-table usage have been fixed.
546 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
547 be returned from its body when the values were being returned
548 using unknown-values return convection and the W-P-O was wrapped
549 inside an UNWIND-PROTECT.
550 * bug fix: sb-posix should now compile again under Windows, enabling
553 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
554 * incompatible change: hash-table accessor functions are no longer
555 automatically protected by locks. Concurrent accesses on the same hash-table
556 from multiple threads can give inconsistent results or even corrupt the
557 hash-table completely. Multi-threaded applications should do their own
558 locking at the correct granularity. In the current implementation it is
559 still safe to have multiple readers access the same table, but it's not
560 guaranteed that this property will be maintained in future releases.
561 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
562 and will signal an error at runtime.
563 * enhancement: SB-THREAD package now exports a semaphore interface.
564 * enhancement: CONS can now stack-allocate on x86 and
565 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
567 * enhancement: nested lists can now be stack allocated on
568 platforms providing stack allocation support.
569 * enhancement: dynamic-extent support has been extended to support
570 cases where there are multiple possible sources for the stack
572 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
573 if the mutex is uncontested on Linux.
574 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
575 as specified by the CLHS. (thanks to Tobias Rittweiler)
576 * bug fix: NaN comparison now works on x86-64.
577 * bug fix: CLOSE :ABORT T on a stream with pending output now
579 * bug fix: instances of non-standard metaclasses using standard
580 instance structure protocol sometimes missed the slot type checks
582 * bug fix: known functions can be inlined in high-debug code.
584 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
585 * minor incompatible change: the MSI installer on Windows no longer
586 associates .lisp and .fasl files with the installed SBCL.
587 * minor incompatible change: :UNIX is no longer present in *FEATURES*
588 on Windows. (thanks to Luis Oliviera)
589 * new feature: SB-CLTL2 contrib module now implements
590 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
591 * optimization: scavenging weak pointers is now more efficient,
592 requiring O(1) instead of O(N) per weak pointer to identify
593 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
594 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
595 method bodies using constant slot-names when the first argument is
596 a specializer parameter for the method.
597 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
598 names, when the class of the instance is a direct instance
599 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
600 as slow as the constant slot-name case.)
601 * optimization: member type construction is now O(N) instead
603 * optimization: UNION and NUNION are now O(N+M) for large
604 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
606 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
607 now more readable in environments like Slime which display it.
608 (thanks to Tobias C. Rittweiler)
609 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
610 was able to infer the correct argument type for the object on which
611 the CAS operation was being performed.
612 * bug fix: copy propagation interfered with parallel assignment
613 semantics in local calls. (reported by Paul Khuong)
614 * bug fix: the signed modular fixnum shift compiled to wrong code on
615 x86 and x86-64. (spotted by a slight modification to some of
618 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
619 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
620 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
621 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
623 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
624 and generic function now print the object identity for anonymous
625 instances. (thanks to Attila Lendvai)
626 * enhancement: as an extension to MOP, SBCL now supports SETF of
627 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
628 (thanks to Attila Lendvai)
629 * enhancement: sb-cover has an alternate annotation mode which puts
630 the form annotations at the car of the form, instead of annotating
631 it completely. The benefit of this mode is that it shows better
632 which forms were instrumented by the compiler.
633 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
634 by printer control variables. (reported by Dan Corkill)
635 * bug fix: system leaked memory when delayed output was performed by
636 the OS in smaller chunks then expected. (thanks to David Smith)
637 * bug fix: system leaked memory when file streams were not closed
639 * bug fix: large objects written to slow streams that were modified
640 after the write could end up with the modified state written to
641 the underlying file descriptor.
642 * bug fix: multiple threads operating in parallel on the same stream
643 could cause buffer-overflows.
644 * bug fix: source location information is stored correctly
645 (broken since 1.0.6). This bug would generally show up as the
646 Slime debugger highlighting the wrong form when the sldb-show-source
648 * bug fix: Branch forms are again annotated as branches in the sb-cover
650 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
651 on x86-64. (reported by Gregory Vanuxem)
652 * bug fix: EQUALP could return wrong results for structures with raw
653 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
654 word). (reported by Vjacheslav Fyodorov)
655 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
658 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
659 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
660 atomic compare-and-swap operations on threaded platforms.
661 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
662 allows assigning a global minimum value to optimization qualities
663 (overriding proclamations and declarations).
664 * enhancement: closed over variables can be stack-allocated on x86
666 * performance bug fix: GETHASH and (SETF GETHASH) are once again
668 * optimization: slot definition lookup is now O(1). This speeds up
669 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
670 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
671 * optimization: ASSOC and MEMBER can now be open-coded for all
672 combinations of keyword arguments when second argument is constant
673 and SPEED >= SPACE. In other cases a specialized version is
675 * bug fix: using obsoleted structure instances with TYPEP and
676 generic functions now signals a sensible error.
677 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
678 (reported by Kristoffer Kvello)
679 * bug fix: backtrace construction is now more careful when making
680 lisp-objects from pointers on the stack, to avoid creating bogus
681 objects that can be seen by the GC.
682 * bug fix: defaulting of values in contexts expecting more than 7
683 variables now works on x86-64. (reported by Christopher Laux)
684 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
686 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
687 as the property-list of a symbol.
688 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
689 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
690 situations requested, are once again file-compileable. (reported
693 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
694 * MOP improvement: support for user-defined subclasses of
695 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
696 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
697 the expansion of DEFMETHOD, is responsible for generating a form
698 which creates a list of specializers when evaluated. Additional
699 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
700 debugging and introspective support.
701 * minor incompatible change: the (unsupported) spinlock interface
702 has changed: free spinlock now has the value NIL, and a held spinlock
703 has the owning thread as its value.
704 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
705 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
707 * enhancement: name of a socket-stream is now "a socket" instead of
709 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
710 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
711 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
712 for generic functions. (Thanks to Tobias C. Rittweiler)
713 * optimization: bignum printing speed has been improved by 20-40%
714 (depending on the bignum size.)
715 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
717 * bug fix: the cache used by the CLOS to store precomputed effective
718 methods, slot offsets, and constant return values is now thread and
720 * bug fix: generic function dispatch function updating is now thread
721 and interrupt safe (in the sense that the known issues have been
723 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
724 * bug fix: interning EQL-specializers is now thread and interrupt safe.
725 * bug fix: asdf systems with dependencies to the SB-POSIX or
726 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
727 * bug fix: interrupt safety of applicable method computation has been
730 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
731 * new contrib: sb-cover, an experimental code coverage tool, is included
733 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
734 significantly faster.
735 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
736 produce more efficient code on x86-64 if the compiler can determine
737 that (+ INDEX <constant>) does not require a bounds check and FOO
738 has an element type at least 8 bits wide.
739 * enhancement: a new, experimental synchronous timeout facility is
740 provided. Refer to SB-SYS:WITH-DEADLINE for details.
741 * enhancement: when a symbol name conflict error arises, the
742 conflicting symbols are always printed with a package prefix.
743 (thanks to Kevin Reid)
744 * enhancement: stepping is now once again supported on the SPARC and
745 MIPS platforms. (It is also now more likely to work on CheneyGC
747 * enhancement: sb-sprof can now also track and report accurate call
749 * bug fixes: the treatment of non-standard subclasses of
750 SB-MOP:SPECIALIZER is more correct.
751 * incompatible change: PURIFY no longer copies the data from the
752 dynamic space into the static and read-only spaces on platforms
753 that use the generational garbage collector
754 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
756 * bug fix: GC race condition occasionally resulting in crashes with
757 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
759 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
761 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
762 a STYLE-WARNING to be signalled (regression from 1.0.4.)
763 * bug fix: an asynchronous interrupt could previously leave the
764 system running with GC inhibited.
765 * bug fix: a DECLARE form evaluated at top-level now causes an error
766 rather than silently (or verbosely) returning NIL.
767 * bug fix: trying to dissassemble functions compiled at high DEBUG could
768 result in a error being signalled due to source form lookup errors.
769 (reported by Peter Graves)
771 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
772 * incompatible change: removed writer methods for host-ent-name,
773 host-ent-addresses -- changing the values did not update the DNS
775 * minor incompatible change: changed experimental JOIN-THREAD interface
776 * documentation: the manual now lists reader and writer methods
777 in class slot documentation sections. (thanks to Richard M Kreuter)
778 * documentation: unwinding from asyncronous events has been
779 documented as unsafe.
780 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
781 in multithreaded application code.
782 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
783 platforms. (thanks to James Anderson for the optimization hint)
784 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
786 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
787 variants no longer cons.
788 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
789 their NOT- variants no longer cons.
790 * optimization: EQUAL hash tables no longer use SXHASH for objects
791 of all data types, but instead use an EQL hash for types for which
792 EQUAL is the same as EQL.
793 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
794 are significantly faster.
795 * optimization: new STRING-OUTPUT-STREAM implementation is much
796 faster for large outputs and conses 30% less on average.
797 * enhancement: XREF information is now collected for references made
798 to global variables using SYMBOL-VALUE and a constant argument.
799 * enhancement: SIGINT now causes a specific condition
800 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
801 * bug fix: FILE-POSITION used to signal an error in some cases where
802 ANSI requires it to return NIL.
803 * bug fix: ADJUST-ARRAY is now interrupt-safe.
804 * bug fix: adding and removing fd-handlers is now interrupt-safe.
805 * bug fix: inlined calls to C now ensure 16byte stack alignment on
807 * bug fix: bad type declaration in the CLOS implementation has
808 been fixed. (reported by James Anderson)
809 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
810 (thanks to Lutz Euler)
811 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
812 required by the CLHS. (reported by Samium Gromoff)
813 * bug fix: dead unbound variable references now signal an error.
814 (reported by Marco Monteiro)
815 * bug fix: / with an unused value was being deleted in safe code.
816 (thanks to Marco Monteiro and Kevin Reid)
817 * bug fix: number of characters that can be written onto a single
818 line in a file is unlimited.
819 * bug fix: some GC deadlocks caused by asynchronous interrupts have
820 been fixed by inhibiting interrupts for when GC is disabled.
821 * bug fix: some interrupt safety issues with GETHASH, (SETF
822 GETHASH), CLRHASH and REMHASH have been fixed.
823 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
824 specifier no longer causes infinite recursion.
825 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
826 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
827 * bug fix: an &environment argument with the correct variable information
828 is passed to macros that are expanded during byte compilation.
829 (reported by Samium Gromoff)
830 * bug fix: a logic error was causing the hash function for bignums to
831 have excessive amounts of collisions. (reported by Faré Rideau)
832 * bug fix: modifying the contents of an array could change the return
833 value of SXHASH on that array, which is only allowed for strings
834 and bit vectors. (bug introduced in 0.9.16)
835 * improvement: the x86-64/darwin port now passes all tests (except
836 for the debugger tests) but should still be considered
837 experimental until this is fixed.
838 * improvement: a style-warning is signaled for CASE (etc) clauses with
839 duplicate keys. (thanks to Kevin Reid)
840 * improvement: macos/x86 and macos/x86-64 now use mach exception
841 handlers for dealing with illegal instructions (for trapping and
842 error handling) and memory protection violations (for GC).
844 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
845 * new platform: experimental support for x86-64/darwin (MacOS).
846 * incompatible change: the thread-safe (on most platforms) getaddrinfo
847 and getnameinfo sockets functions are used instead of gethostbyaddr
848 and gethostbyname, on platforms where the newer functions are available.
849 As a result, the ALIASES field of HOST-ENT will always be NIL on these
851 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
852 don't print the startup banner, but behave as if --noinform was passed
853 as a command line argument. (thanks to Kevin Reid)
854 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
855 * optimization: code using alien values with undeclared types is much faster.
856 * optimization: the compiler is now able to open code SEARCH in more cases.
857 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
858 * bug fix: using standardized COMMON-LISP special variables as loop
859 variables no longer signals bogus package lock violations. (reported
861 * bug fix: declaring local loop variables to be of a range-limited type
862 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
863 (reported by Andras Simon)
864 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
865 bugs remain on x86-64.)
866 * bug fix: the #= and ## reader macros now interact reasonably with
867 funcallable instances.
868 * bug fix: type-checks for function arguments were compiled using the
869 compiler policy settings of the wrong lexical environment (in 1.0.2
871 * bug fix: SHADOW accepts characters as string designators, as required
872 by the spec (thanks to Eric Marsden)
873 * bug fix: fixed GC safety issues when foreign functions are called with
874 non-base strings as arguments
875 * bug fix: more consistent error output for fatal-compiler-warnings, like
877 * bug fix: functions with proper names were showing up as NIL in
878 backtraces (reported by Edi Weitz, regression in 1.0.2)
880 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
881 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
882 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
883 produce more efficient code on the x86 if the compiler can determine
884 that (+ INDEX <constant>) does not require a bounds check and FOO
885 has an element type at least 8 bits wide.
886 * bug fix: references to undefined variables are handled the same way
887 in toplevel forms as in the normal compiler
888 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
890 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
891 (thanks to Jon Buller)
892 * improvement: faster compilation times for complex functions
893 * improvement: added readlink support to SB-POSIX (thanks to Richard
896 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
897 * improvement: experimental support for mach exception handling on
898 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
900 * improvement: support for GBK external format.
901 (thanks to Chun Tian (binghe))
902 * improvement: the debugger now displays variables that have been closed
903 over, in code compiled with (DEBUG 2) or higher
904 * improvement: support for executable cores on NetBSD (thanks to
906 * new feature: added a RESTART-FRAME debugger command
907 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
908 be used to provide an implementation for FILE-POSITION on Gray streams
909 (thanks to Eric Marsden)
910 * new feature: add syslog, openlog and closelog support to SB-POSIX
911 (thanks to Richard Kreuter)
912 * optimization: the function call overhead in code compiled with
913 a high DEBUG optimization setting is significantly
914 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
915 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
916 (thanks to Tony Martinez)
917 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
918 evaluated once (reported by Kevin Reid)
919 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
920 works again (thanks to Kilian Sprotte)
921 * bug fix: an error is signaled for tagbodies with duplicate tags
922 (thanks to Stephen Wilson)
923 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
924 * bug fix: Win32 port can now handle foreign code unwinding Lisp
925 stack frames from alien callbacks.
926 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
927 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
928 65k pages (thanks to David Woodhouse)
929 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
931 changes in sbcl-1.0.1 relative to sbcl-1.0:
932 * new platform: FreeBSD/x86-64, including support for threading.
933 * new feature: the compiler stores cross-referencing information
934 about function calls (who-calls), macroexpansion (who-macroexpands)
935 and special variables (who-binds, who-sets, who-references) for code
936 compiled with (< SPACE 3). This information is available through the
937 sb-introspect contrib.
938 * new feature: users may subclass SEQUENCE, and have instances of
939 these classes interoperate with standard Common Lisp functions if
940 a number of methods are defined. (This feature is experimental
941 and the interface subject to change based on feedback from SBCL
942 users and the general community)
943 * improvement: sb-sprof traces call stacks to an arbitrary depth on
944 x86/x86-64, rather than the previous fixed depth of 8
945 * improvement: another pthread back-end of mutex "pthread-futex".
946 * bug fix: non-ascii command-line arguments are processed correctly
947 (thanks to Yaroslav Kavenchuk)
948 * bug fix: non-required arguments were not passed correctly when a method
949 defined using DEFMETHOD was called from a mop-generated method using
950 CALL-NEXT-METHOD (reported by Pascal Costanza)
951 * bug fix: recursion is now permitted in accessors through
952 SLOT-UNBOUND. (reported by Pascal Costanza)
953 * bug fix: an error was signaled at startup if the HOME environment
954 variable was defined, but had an empty value (reported by Peter Van Eynde)
955 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
956 signal a PROGRAM-ERROR, not a BUG.
957 * bug fix: SB-POSIX:READDIR works when built with large file support.
958 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
960 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
961 for better automatic indentation support. (thanks to Matt Pillsbury)
962 * optimization: loading generic functions no longer takes O(n^2) time,
963 proportional to the amount of methods in the generic function
964 (reported by Todd Sabin and Jeremy Brown)
965 * optimization: the FIND and POSITION family of sequence functions
966 are significantly faster on arrays whose element types have been
968 * improvements to the Windows port:
969 ** Intermittent heap corruption problems have been fixed. (thanks
970 to Alastair Bridgewater)
971 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
973 ** Lisp is able to unwind foreign exception frames from alien
974 callbacks. (thanks to Alastair Bridgewater)
976 changes in sbcl-1.0 relative to sbcl-0.9.18:
977 * improvement: experimental support for threading on FreeBSD/x86.
978 (thanks to NIIMI Satoshi)
979 * improvement: runtime option --dynamic-space-size can be used
980 to set the size of the dynamic space reserved on startup.
981 * improvement: floating point modes in effect are now saved in
982 core, and restored on startup.
983 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
984 startup, not time since first call to GET-INTERNAL-REAL-TIME.
985 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
986 threads are not running after *SAVE-HOOKS* have run.
987 * improvement: writes to CLOS instance slots are type-checked in code
988 compiled with (SAFETY 3)
989 * improvement: floating-point exception handling on FreeBSD (thanks to
991 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
992 (thanks to Zach Beane)
993 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
995 * improvement: added support for the Shift-JIS external format.
996 (contributed by NIIMI Satoshi)
997 * improvement: callbacks are supported on Linux/PPC. (thanks to
999 * bug fix: compiler bug triggered by a (non-standard) VALUES
1000 declaration in a LET* was fixed. (reported by Kaersten Poeck)
1001 * bug fix: file compiler no longer confuses validated and already
1002 dumped structurres. (reported by Kaersten Poeck)
1003 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
1004 fill-pointer signals a type-error as required. (thanks to
1006 * bug fix: disassemly of funcallable instances works.
1007 * bug fix: single stepping on PPC.
1008 * bug fix: fix thread-safety problems in the type system (generally
1009 manifesting as nonsensical errors like "STRING is a bad type specifier
1010 for sequences" or "The value 1 is not of type FIXNUM")
1011 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1012 (reported by Josip Gracin)
1013 * bug fix: an error is signaled for attempts to displace arrays with
1014 incompatible element types (thanks to Mario Mommer)
1015 * bug fix: more correct handling of wide characters in debug info
1016 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1017 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1018 and don't cause extra consing
1019 * optimization: MAP and MAP-INTO are significantly faster on vectors
1020 whose elements types have been declared.
1021 * Improvements to SB-SPROF:
1022 ** Support for allocation profiling
1023 ** Reduced profiling overhead, especially for long profiling runs
1024 * Improvements to the Windows port:
1025 ** floating point exceptions are now reported correctly.
1026 ** stack exhaustion detection works partially.
1027 ** more accurate GET-INTERNAL-REAL-TIME.
1028 ** better RUN-PROGRAM behaviour in case of errors in the child
1030 ** PROBE-FILE now simplifies pathnames correctly.
1031 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1032 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1033 1st 1970 (although time timezone data will be unreliable).
1034 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1036 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1037 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1038 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1040 * bug fix: two potential GC deadlocks affecting threaded builds.
1041 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1042 returns T (reported by Anton Kazennikov)
1043 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1044 with non-variable places
1045 * bug fix: remove a race condition in the setting of
1046 funcallable-instance functions, this should make threaded CLOS
1047 code more stable against memory faults.
1048 * bug fix: corruption of specials when unbinding is interrupted by an
1049 asynchronous unwind (reported by Hannu Koivisto)
1050 * improvement: the debugger will now also display local variables that
1051 are only used once, for code compiled with a DEBUG optimization quality
1054 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1055 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1056 * incompatible change: External-format support for FFI calls. The
1057 SB-ALIEN:C-STRING no longer implies an ASCII
1058 external-format. Instead the string is subject to external-format
1059 conversion. Additionally return values of type SB-ALIEN:C-STRING
1060 are by default of type (SIMPLE-ARRAY CHARACTER), not
1061 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1062 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1063 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1064 * incompatible change: SB-EXT package no longer contains the
1065 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1066 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1068 * incompatible change: the single-stepper is no longer available
1069 on Alpha, Mips and Sparc platforms.
1070 * minor incompatible change: the direct superclasses of
1071 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1072 not (STANDARD-OBJECT FUNCTION). This makes the
1073 class-precedence-lists of GENERIC-FUNCTION and
1074 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1076 * minor incompatible change: the default stream external format on
1077 non-unicode SBCL is no longer detected from the locale, but is always
1079 * new feature: Add a version of evaluator that uses an interpreter instead
1080 of the compiler. EVAL still uses the compiler by default, to switch it
1081 to use the interpreter, set the value of the variable
1082 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1083 * minor incompatible change: the single-stepper REPL has been merged
1084 with the normal debugger (see the "Stepping" heading of the debugger help
1085 for more details). The debugger command STEP will no longer switch
1086 to the single-stepper REPL.
1087 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1088 for a type now works.
1089 * bug fix: loading of default sysinit file works. (thanks to Leonid
1091 * bug fix: better detection of circularities in the file-compiler.
1092 (reported by Marco Monteiro)
1093 * bug fix: the CL pathname functions now work with files that have
1094 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1095 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1096 systems with Unix98 pty semantics.
1097 * bug fix: ASDF-INSTALL will now work with bsd tar.
1098 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1100 * bug fix: timers expiring in dead threads no longer cause a
1101 type-error (reported by Paul "Nonny Mouse").
1102 * bug fix: thanks to more lightweight single-stepper instrumentation,
1103 code compiled with (DEBUG 3) will compile and execute significantly faster,
1104 and will have more accurate type-inferencing than before
1105 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1106 whose bindings are modified
1107 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1108 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1109 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1110 CL:SOFTWARE-VERSION with proper version
1112 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1113 * feature: implemented the READER-METHOD-CLASS and
1114 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1115 as specified by AMOP.
1116 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1118 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1119 (thanks to Lutz Euler)
1120 * optimization: hashing of general arrays and vectors has been
1121 improved. (reported by Any Fingerhut)
1122 * enhancement: SB-INTROSPECT is now able to find definitions of
1123 profiled functions. (thanks to Troels Henriksen)
1124 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1125 as well, allowing compiler-macros for SETF-functions to expand.
1126 * enhancement: step-instrumentation no longer wraps known
1127 single-value functions with multiple-value context, allowing
1128 better type inference.
1129 * fixed bug #337: use of MAKE-METHOD in method combination now works
1130 even in the presence of user-defined method classes. (reported by
1131 Bruno Haible and Pascal Costanza)
1132 * fixed bug #339(c): if there are applicable methods not part of any
1133 long-form method-combination group, call INVALID-METHOD-ERROR.
1134 (reported by Bruno Haible)
1135 * fixed bug #361: the :FUNCTION initarg in the protocol for
1136 initialization of methods can now be used to override
1137 internally-produced optimized functions. (reported by Bruno
1139 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1140 system-provided lambda expression no longer cause warnings about
1141 unbound #:|pv-table| symbols.
1142 * bug fix: improved the handling of type declarations and the
1143 detection of violations for keyword arguments with non-constant
1145 * bug fix: potentially erronous calls to PATHNAME and
1146 MERGE-PATHNAMES were being flushed in some cases.
1147 (reported by Richard Kreuter)
1148 * bug fix: compiled calls to TYPEP were mishandling obsolete
1149 instances. (reported by James Bielman and Attila Lendvai)
1150 * bug fix: format strings with ~> without matching ~< no longer
1151 trigger an AVER, but signal an understandable error instead.
1152 (reported by Antonio Martinez)
1153 * bug fix: specifying an output-file without a pathname-type for
1154 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1155 of input-file instead of "fasl". (reported by Robert Dodier)
1156 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1157 (reported by James Y Knight).
1158 * bug fix: compiler-macros-function did not consider the environment
1159 argument for shadowing by local functions.
1160 * bug fix: compiler-macros expansion was inhibited by local INLINE
1162 * bug fix: inline expansions of known functions were subject to
1163 step-instrumentation in high debug policies, leading to problems
1164 with type-inference.
1165 * bug fix: compiler failed to differentiate between different CONS
1166 types in some cases.
1167 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1168 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1169 * bug fix: #368: incorrect use of expressed vs. upgraded array
1171 * bug fix: #306a: more precise unions of array types.
1172 * thread-safety improvements:
1173 ** CONDITION-WAIT could return early on Linux, if the thread was
1174 interrupted and subsequently continued with SIGCONT.
1175 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1177 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1178 * added support for the ucs-2 external format. (contributed by Ivan
1180 * minor incompatible change: pretty printing of objects of type
1181 (cons symbol) is, in the default pprint-dispatch-table, now
1182 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1184 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1185 called later in a class's lifetime, possibly as late as when the
1186 first instance of the class is created. Previously,
1187 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1188 class became finalizeable.
1189 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1190 for multibyte external-format streams. (thanks to "vbzoli")
1191 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1192 with :ALLOCATION :CLASS inherited from superclasses of the
1194 * fixed bug: anonymous classes can now be created using the :NAME
1195 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1196 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1197 * fixed bug: core-files saved with :EXECUTABLE T can again be
1198 executed when SBCL_HOME isn't set. (reported by James Knight)
1199 * fixed bug: toplevel LOCALLY forms with declarations could
1200 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1201 * fixed bug: printing from several different threads using different
1202 values of *print-case* could cause invalid output, due to
1203 some internal special variables of the printer not being bound
1204 thread-locally (reported by Max Mikhanosha)
1205 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1206 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1207 (reported by Pascal Costanza)
1208 * fixed bug: FINALIZE-INHERITANCE is called from
1209 REINITIALIZE-INSTANCE on classes when the class has previously
1210 been finalized, as required by AMOP.
1211 * minor code generation optimizations:
1212 ** better register allocation in CLOS dispatching functions
1213 ** overflow detection when coercing signed bytes to fixnums on x86-64
1214 ** is now implemented with one IMUL instruction instead of three shifts
1215 ** more efficient bit-vector access on x86 and x86-64
1216 ** more efficient access to raw structure slots on x86 and x86-64
1217 * fixed some bugs revealed by Paul Dietz' test suite:
1218 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1219 return its argument.
1221 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1222 * feature: thread support on Solaris/x86, and experimental thread support
1224 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1226 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1227 default method for SLOT-UNBOUND.
1228 * minor incompatible change: prevent the user from specializing the
1229 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1230 somewhat counter to the intent of the protocol, I (CSR) think, and
1231 additionally it just doesn't work in SBCL as currently
1232 implemented, thanks to optimizations (that are always valid for
1233 the other three SLOT-VALUEish functions, but not for the setter).
1234 * minor incompatible change: the :SB-LDB feature is now enabled by
1235 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1236 the low-level debugger.
1237 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1238 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1239 down to an SB-SYS:FD-STREAM, allowing direct communication in
1240 more cases. Temporary files and pipes are still used as a fallback
1242 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1243 on the non-reentrant C functions localtime(3) and gmtime(3).
1244 * bug fix: global optimization policy was not visible in LOCALLY and
1246 * bug fix: class objects can be used as specializers in methods.
1247 (reported by Pascal Costanza)
1248 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1249 with a type error. (reported by blitz_ on #lisp)
1250 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1251 failed with a type error. (reported by Pascal Bourguignon)
1252 * bug fix: merging pathnames against defaults with :DIRECTORY
1253 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1254 (reported by James Y Knight)
1255 * bug fix: saving large (>2GB) cores on x86-64 now works
1256 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1257 constant index and a (simple-array (signed-byte 32)) array
1258 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1259 error (patch by Robert J. Macomber)
1260 * feature: TIME now displays the amount of run-time spent in GC
1261 * bug fix: The debugger now does a better job of respecting
1262 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1263 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1264 workaround for bug 403.)
1265 * bug fix: floating point exception handling now works on FreeBSD
1266 (thanks to NIIMI Satoshi)
1267 * fixed some bugs revealed by Paul Dietz' test suite:
1268 ** MISC.641: LET-conversion were not supposed to work in late
1270 * improvements to the Win32/x86 port:
1271 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1272 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1274 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1275 * new feature: source path information is generated for macro-expansion
1276 errors for use in IDE's like Slime (thanks to Helmut Eller)
1277 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1278 * bug fix: compiler does not loop forever on an invalid type in
1280 * improvement: compilation of most CLOS applications is significantly
1282 * optimization: added a limited bytecode compiler for simple toplevel
1283 forms, speeding up compilation and FASL loading
1284 * bug fix: the statistical profiler now properly distinguishes anonymous
1287 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1288 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1289 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1290 into subclasses of the finalized class. Now user calls to
1291 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1292 system return before any subclasses are finalized.
1293 * minor incompatible change: The reader no longer ignores errors
1294 regarding non-existent packages in #+ and #- feature tests.
1295 * new feature: command line options --no-sysinit, --no-userinit to
1296 inhibit loading the corresponding init files
1297 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1298 for sending data through UDP sockets (thanks to François-René Rideau)
1299 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1300 error is signalled instead (thanks to François-René Rideau)
1301 * minor incompatible change: Error signalling behaviour of lexical
1302 operations violating package locks has changed slightly. Refer to
1303 documentation on package locks for details.
1304 * enhancement: EVAL can process IF-expressions without resorting to the
1306 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1307 constant-folded properly due to missing out-of-line functions.
1308 (reported by Mika Pihlajamäki)
1309 * bug fix: LISTEN sometimes returned T even in cases where no data was
1310 immediately available from the stream
1311 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1312 were reversed. (reported by Levente Mészáros)
1313 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1314 when given a vector displaced to an adjustable array. (reported
1316 * bug fix: derivation of float boundaries from numbers outside the
1317 appropriate float range (reported by John Wiseman)
1318 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1320 * bug fix: precomputing cache entries for generic functions with
1321 some subclasses of specializers as yet invalid does not attempt to
1322 fill a cache line with a negative offset. (reported by Levente
1324 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1325 allow condition class objects as arguments to DOCUMENTATION and
1326 (SETF DOCUMENTATION); only find and set documentation for
1327 structure names for the STRUCTURE doc-type. (suggested by Gary
1329 * improvements to the Win32/x86 port:
1330 ** (user-homedir-pathname) and default initialization file
1331 locations now know about the user's "Documents and Settings"
1332 directory (thanks to Yaroslav Kavenchuk)
1333 ** run-program is implemented (thanks to Mike Thomas)
1334 ** sockets support (thanks to Timothy Ritchey)
1335 ** better backtrace support (thanks to Alastair Bridgewater)
1336 ** sb-grovel supported
1337 ** asdf-install and sb-posix work somewhat
1338 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1339 * improvements to the Solaris/x86 port:
1340 ** works on Solaris 11/Solaris Express
1341 ** floating-point exception handling support
1342 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1343 * fixed some bugs revealed by Paul Dietz' test suite:
1344 ** REMOVE-METHOD returns its generic function argument even when
1345 the method is not one of the generic functions' methods.
1346 ** objects declared as MEMBER types can be admissible arguments to
1347 structure accessors.
1348 ** printing characters should simply be printed by the FORMAT ~:C
1350 ** compiler failure when compiling functions with hairy constant
1351 defaults for optional parameters.
1352 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1353 function, which is already optimized.
1355 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1356 * new platform: experimental support for SBCL x86/Darwin, including
1357 MacOS X 10.4.5 on Intel.
1358 * new feature: Unicode character names are now known to the system
1359 (through CHAR-NAME and NAME-CHAR).
1360 * new feature: the filesystem location of SBCL's core file is
1361 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1362 * minor incompatible change: the contrib modules SB-POSIX and
1363 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1364 this change is to make it easier to distribute
1365 location-independent binaries.
1366 * bug fix: as implied by AMOP, standardized classes no longer have
1367 slots named by external symbols of public packages. (reported by
1369 * optimization: calling functions via constant symbols -- as in
1370 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1371 via the function object as in (FUNCALL #'FOO).
1372 * optimization: on x86 and x86-64, the calling convention for the
1373 case of dealing with an unknown number of values has been altered
1374 to be friendlier to the prediction heuristics implemented,
1375 particularly with reference to CALL and RET pairing. (thanks to
1376 Alastair Bridgewater)
1377 * optimization: on x86, the code for access to thread-local symbol
1378 values and binding/unbinding of thread-local symbols is smaller.
1379 (thanks to Alastair Bridgewater)
1380 * enhancement: CONSTANTP is now able to determine constantness of
1381 more complex forms, including calls to constant-foldable standardized
1382 functions and some special forms beyond QUOTE.
1383 * fixed bug: occasional GC crashes on Solaris/x86
1384 * optimization: x86-64 supports stack allocation of results of simple
1385 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1386 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1387 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1388 (thanks to James Knight)
1389 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1390 by the standard (thanks to Stephen Compall)
1392 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1393 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1394 be used for bundling the runtime and the core file into one
1395 executable binary. This feature is not currently supported on all SBCL
1396 platforms. (thanks to James Bielman and NIIMI Satoshi)
1397 * new feature: a generational or ephemeral garbage collector is now
1398 the default on the PowerPC platform (both Linux and Darwin). The
1399 old Cheney (stop and copy) collector is a build-time option.
1400 (thanks to Cyrus Harmon, after Raymond Toy)
1401 * minor incompatible change: the method by which SBCL finds its
1402 contributed modules has changed; it no longer relies on symbolic
1403 links from an $SBCL_HOME/systems directory, but searches directly
1404 in the subdirectories of $SBCL_HOME.
1405 * enhancement: the dynamic heap size on the Linux/PPC platform is
1406 markedly larger, even using the older Cheney garbage collector.
1407 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1408 certain complicated string types. (reported by Gary King)
1409 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1410 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1411 * fixed bug: DOCUMENTATION on structures no longer signals an
1412 error if no structure type of the right name exists. (reported by
1414 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1415 applicable non-standard primary or :AROUND method on
1416 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1417 no longer cause unbound variable errors. (reported by Kevin Reid)
1418 * fixed bug: metacircle resolution in cases where methods have slots
1419 added before the slots from STANDARD-METHOD. (reported by Jean
1421 * fixed bug: the Power PC assembly code for calling into lisp
1422 sometimes computed its return address wrongly (depending on the
1423 15th bit of the address, fixed at link time). This used to
1424 manifest itself in a segmentation violation while building PCL.
1425 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1426 many others over the years)
1427 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1428 the stack pointer (treated as a signed integer) is negative.
1429 (thanks to Peter van Eynde)
1431 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1432 * new platform: experimental support for the Windows operating
1433 system has been added. (thanks to Alastair Bridgewater)
1434 * fixed several bugs in and robustified the PPC FFI (including
1435 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1436 * bug fix: allow non-simple string symbol names (reported by Paul
1438 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1440 * bug fix: saving a core corrupted callbacks on x86/x86-64
1441 * bug fix: closed a loophole in metacircularity detection and
1442 grounding in the PCL implementation of CLOS.
1443 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1444 (reported by tomppa on #lisp)
1445 * bug fix: clear the direction flag on Lisp -> C transitions, as
1446 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1447 (reported by Andrej Grozin and Hendrik Maryns)
1448 * optimization: major improvements to GC efficiency on GENCGC platforms
1449 * optimization: faster implementation of EQUAL
1450 * optimization: emit more efficient opcodes for some common
1451 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1452 * optimization: several other minor code-generation improvements on x86-64
1453 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1455 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1456 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1457 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1458 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1459 by ANSI as a generic function, and for consistency (SETF
1460 GENERIC-FUNCTION-NAME) is treated likewise.
1461 * fixed bug #233.b: make constraint propagation notice when a variable
1462 value is changed after it is referenced but before it is used
1463 * fixed bug #296: no more arbitrary behaviour from filenames with
1464 odd characters as --load arguments. (reported by Adam Warner)
1465 * fixed bug #390: :CHARACTER-SET pathname components now work as
1466 expected. (reported by Tim Daly Jr)
1467 * fixed bug #391: complicated :TYPE intersections in slot
1468 definitions no longer cause an error in PCL internals.
1469 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1470 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1471 returning the number of octets which would be written to the
1472 file-stream. (thanks to Robert J. Macomber)
1473 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1474 arguments to the internal stream functions. (thanks to David
1476 * bug fix: stack exhaustion is now handled on the x86/SunOS
1477 platform. (thanks to Daisuke Homma)
1478 * bug fix: more accurate ROOM results on GENCGC platforms
1479 * optimization: improved type inference for arithmetic-for
1480 index variables in LOOP
1481 * optimization: faster floating-point SQRT on x86-64
1482 * fixed some bugs revealed by Paul Dietz' test suite:
1483 ** DOCUMENTATION returns NIL instead of "" for method combinations
1484 that don't have a docstring
1486 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1487 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1488 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1489 therefore are not customizeable by user code (as seems to be at
1490 least permitted and maybe required by AMOP). As a consolation,
1491 however, the SBCL implementation of these functions now calls
1492 REINITIALIZE-INSTANCE as specified by AMOP.
1493 * bug fix: slot-definition documentation is propagated as per ANSI
1494 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1495 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1496 argument, as do eslotd initialization methods. (from Pascal
1497 Costanza's "Closer" project)
1498 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1499 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1501 * bug fix: it is now possible to have more than one subclass of
1502 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1503 (reported by Bruno Haible, Pascal Costanza and others)
1504 * bug fix: the dependent update protocol now works for generic
1505 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1506 and Pascal Costanza)
1507 * bug fix: condition-class instances corresponding to
1508 DEFINE-CONDITION forms are now created eagerly. (reported by
1509 Kalle Olavi Niemitalo on comp.lang.lisp)
1510 * bug fix: floating point printing is more accurate in some
1511 circumstances. (thanks to Simon Alexander)
1512 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1513 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1514 * bug fix: callbacks on OS X now preserve stack-alignment.
1515 (thanks to Cyrus Harmon)
1516 * enhancement: the x86-64 disassembler is much better at
1517 disassembling SSE instructions. (thanks to Lutz Euler)
1518 * enhancement: CHECK-TYPE on a variable now allows type-inference
1519 (thanks to Vincent Arkesteijn)
1520 * optimization: improved performance of EUC-JP external format.
1521 (thanks to NIIMI Satoshi)
1522 * optimization: performance improvements to IO on file streams of
1523 :ELEMENT-TYPE CHARACTER
1524 * optimization: much faster memory allocation on x86-64
1525 * optimization: faster garbage collections (latency and throughput)
1527 * optimization: faster bignum subtraction and division on x86 and x86-64
1528 (thanks to Lutz Euler)
1529 * optimization: more accurate type-inference for loops that use a
1530 floating point index variable or a negative step.
1532 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1533 * bug fix: add a workaround to SBCL looping infinitely at startup on
1534 Linux kernels with apparently buggy implementations of personality().
1535 (thanks to Svein Ove Aas)
1536 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1537 (thanks to David Lichteblau)
1538 * bug fix: local bindings shadow global symbol macros for
1539 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1540 * optimization: non-open coded uses of numeric comparison operators
1541 (e.g. >) no longer cons when called with more than one parameter
1542 on platforms supporting dynamic-extent allocation.
1543 * enhancement: saving cores with foreign code loaded is now supported
1544 on MIPS/Linux in addition to the previously supported platforms.
1545 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1546 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1547 * bug fix: division by zero in sb-sprof when no samples were collected
1548 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1549 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1550 (thanks to Alexander Kjeldaas)
1551 * new feature: ignoring the return values of destructive functions
1552 like NREVERSE signals a compile-time style-warning.
1553 (thanks to Kevin Reid)
1554 * new feature: the alignment of alien structure fields can be
1555 explicitly specified. (thanks to Cyrus Harmon)
1556 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1557 and dump core on SIGQUIT
1559 ** incompatible change: threads do not inherit values of specials
1560 from their parents (see manual)
1561 ** bug fix: threads stacks belonging to dead threads are freed by the
1562 next exiting thread, no need to gc to collect thread stacks anymore
1563 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1564 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1565 ** bug fix: dynamic variable and thread start related gc lossage
1566 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1568 * fixed some bugs revealed by Paul Dietz' test suite:
1569 ** SUBTYPEP is slightly more accurate on heinously complicated
1570 CONS types where some of the members have uncertain (in the
1571 NIL, NIL sense) type relationships to each other.
1573 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1574 no longer purified unless :PURIFY T is explicitly specified.
1575 ** Non-purified cores are significantly smaller than before
1577 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1578 * new feature: timers based on Zach Beane's excellent timer package
1579 * added support for the following external formats: koi8-u,
1580 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1581 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1582 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1583 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1584 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1585 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1586 * incompatible change: a threaded SBCL will no longer revert to
1587 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1588 * bug fix: interrupts are disabled until startup is complete; no
1589 more sigsegvs when receiving a signal to soon
1590 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1592 * bug fix: add a workaround for the memory randomization features in
1593 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1594 workaround will only be in effect on systems with the proc filesystem
1596 * bug fix: printing objects of type HASH-TABLE signals a
1597 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1599 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1600 have been running for over 50 days. (reported by Gilbert Baumann)
1601 * bug fix: the logic for getting names of functions gets less
1602 confused when confronded with alternate-metaclass
1603 funcallable-instances. (reported by Cyrus Harmon)
1604 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1605 consistent, even on internal alternate-metaclass objects.
1606 * bug fix: flush closure information collected by physical
1607 environment analysis prepass before the main pass. (bug reported
1609 * bug fix: compiler pack phase does not modify a hashtable, which is
1610 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1611 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1612 may be expanded. (reported by Alexander <malishev> on c.l.l)
1613 * bug fix: no more style-warnings when compiling code using LAST
1614 under high-SPEED low-SPACE settings. (reported by David Wallin
1616 * fixed some incompatibilities between SBCL's MOP and the MOP
1618 ** the METAOBJECT class is now implemented;
1619 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1620 STANDARD-OBJECT, as required;
1621 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1622 now compatible (as required by VALIDATE-SUPERCLASS); there
1623 remains a consistency requirement over the presence or absence
1624 of the FUNCTION class in the superclass list of finalized
1625 classes; see the manual for more details;
1626 ** the :AROUND method for COMPUTE-SLOTS on
1627 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1628 requested slot ordering.
1630 ** bug fix: parent thread now can be gc'ed even with a live
1632 ** bug fix: binding a special with PROGV to no value is not
1633 the same as that symbol not having been bound (thanks to
1635 ** bug fix: binding specials is thread safe (thanks to
1637 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1638 and SB-SPROF do not die with 'no handler for signal XX in
1639 interrupt_handle_now(..)' anymore
1640 ** bug fix: WITH-TIMEOUT works with multiple running threads
1641 * fixed some bugs revealed by Paul Dietz' test suite:
1642 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1643 the :method-class keyword argument.
1645 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1646 * new port: the Solaris operating system on x86 processors is now
1647 mostly supported, though some rough edges in the environment
1648 remain. (thanks to Daisuke Homma)
1649 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1650 space for its heap. (thanks to Thiemo Seufer)
1651 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1652 closures. (thanks to Thiemo Seufer)
1653 * minor incompatible change: eof selects abort in the debugger.
1654 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1655 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1657 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1658 deprecated, and will go away in a future revision of SBCL.
1659 * minor incompatible change: GC-ON and GC-OFF are no longer
1660 implemented with a counter, it does not matter how many times gc
1661 is switched on or off
1662 * bug fix: discriminating functions for generic function classes
1663 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1664 make invalid assumptions about method precedence order. (reported
1666 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1667 platforms. (thanks to Thiemo Seufer)
1668 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1669 * bug fix: correct pathname printing with printer escaping is on.
1670 (thanks to Kevin Reid)
1671 * bug fix: complex VOP definitions in "user-space" no longer trigger
1672 package locks. (reported by Zach Beane)
1673 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1674 for funcallable-instance functions, and is no different from
1676 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1677 characters in the current readtable. (reported by Nicholas Neuss)
1678 * bug fix: SUBTYPEP on various CONS types returns more a more
1679 accurate acknowledgment of its certainty.
1680 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1681 lists in some cases. This partially fixes bug 384.
1682 * flush all standard streams before prompting in the REPL and the
1684 * bug fix: signal handling and triggering gc do not conflict
1685 directly anymore, in particular a high frequency sb-sprof does
1686 not prevent gc from running
1687 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1688 approximation for timezone and DST information between the
1689 universal time epoch and the smallest negative 32-bit time_t.
1690 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1691 year 1899 so long as the encoded time is non-negative.
1692 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1693 the correct answer for non-constant rotate amounts when performing
1694 an inline 32-bit rotation.
1696 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1697 there is only one thread in the session
1698 ** bug fix: memory leak for streams created in one thread and
1699 written to in another
1700 ** bug fix: lockup when compiled with gcc4
1701 ** bug fix: race that allows the gc to be triggered when gc is
1703 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1704 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1706 ** bug fix: finished threads are gc'ed properly
1707 * fixed some bugs revealed by Paul Dietz' test suite:
1708 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1709 has arguments with a different set of applicable methods from
1710 the orignal arguments.
1711 ** The type error thrown by MAP now has a correct expected-type
1713 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1714 name a compiled function.
1715 ** Three MISC tests where a large bignum was improperly coerced to
1716 a float that couldn't represent that bignum during type
1717 derivation were fixed.
1718 ** SUBTYPEP can now handle types involving the negation of a
1719 list-form FUNCTION type.
1720 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1721 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1722 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1724 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1725 * New feature: Experimental support for bivalent streams: streams
1726 opened with :element-type :default now allow character and binary
1727 (unsigned-byte 8) I/O
1728 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1729 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1731 * bug fix: no more highly sporadic "couldn't check whether ~S is
1732 readable" when reading a stream and an interrupt hits in the middle
1733 of a select system call
1734 * compiler better recognizes complex arrays (reported by Eduardo
1736 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1737 for variables on SBCL built with linkage-tables. (reported by Luis
1739 * various error reporting improvements.
1740 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1741 (Thanks to Thiemo Seufer)
1742 * bug fix: nested reader invokations work correctly
1743 * bug fix: it is possible to have simultaneous references to foreign
1744 code and foreign data with the same name.
1746 ** added x86-64 support
1747 ** incompatible change: the threading api now works with thread
1748 objects instead of thread ids
1749 ** bug fix: threads are protected from signals and interruption when
1750 starting up or going down
1751 ** bug fix: a race where an exiting thread could lose its stack to gc
1752 ** bug fix: don't halt on infinite error in threads if possible
1753 ** fixed numerous gc deadlocks introduced in the pthread merge
1754 ** bug fix: fixed thread safety issues in read and print
1755 ** bug fix: debugger doesn't hang on session lock if interrupted at
1756 an inappropriate moment
1757 ** bug fix: run-program is now thread safe(r)
1758 ** bug fix: inner with-recursive-lock no longer releases the mutex
1759 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1760 * fixed some bugs revealed by Paul Dietz' test suite:
1761 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1762 EXPECTED-TYPE slots filled.
1763 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1765 ** MAKE-CONDITION accepts classes as type-designators.
1766 ** COMPILE may never return NIL.
1767 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1768 range before calling Unix time functions
1770 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1771 * numerous signal handling fixes to increase stability
1772 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1773 * minor incompatible change: we now correctly canonize default
1774 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1775 as per the MOP, rather than the historical (INITARG INITFUNCTION
1776 INITFORM). (reported by Bruno Haible)
1777 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1778 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1779 for more information.
1780 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1781 pathname is a directory pathname.
1782 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1783 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1785 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1786 :METACLASS argument in addition to a class name. (reported by
1787 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1788 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1789 (reported by Bruno Haible; patch from Gerd Moellmann)
1790 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1792 * bug fix: more cleanups to the floating point exception handling on
1793 x86-64 (thanks to James Knight)
1794 * bug fix: the compiler does not try to constant fold calls of
1795 COERCE and BIT-* functions when they return freshly constructed
1796 objects. (reported by Nikodemus Siivola and Paul Dietz)
1797 * optimization: improved the MIPS versions of generic (in the
1798 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1799 * optimization: direct conversion of (unsigned-byte 32) to floats on
1800 the PowerPC platform.
1801 * optimization: structure instances with raw slots now use less
1802 memory, and probably show better memory locality. (thanks to
1804 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1805 are treated as requests for stack allocation on the x86-64,
1806 Alpha-32, PPC and SPARC platforms.
1807 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1808 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1809 (thanks to Tony Martinez)
1811 ** gcing a dead thread can no longer lead to lockups
1812 ** threads block signals until they are set up properly
1813 ** errno is no longer shared by threads
1814 ** interrupt-thread restores the eflags register on x86
1815 ** fixed some lockups due to gc/thread interaction
1816 * fixed some bugs revealed by Paul Dietz' test suite:
1817 ** invalid dotted lists no longer raise a read error when
1818 *READ-SUPPRESS* is T
1819 ** signal an error if a symbol that names a declaration is used
1820 as the name of a type, or vice versa
1821 ** allow using the (declare (typespec var*)) abbreviation for
1822 (declare (type typespec var*)) with all type specifiers
1823 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1824 ** always return NIL from PPRINT-POP when OBJECT is NIL
1825 ** don't signal errors when pretty-printing malformed LABELS,
1826 FLET or MACROLET forms
1827 ** declarations in a DOLIST body are also in scope for the
1829 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1831 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1834 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1835 * fixed cross-compiler leakages that prevented building a 32-bit
1836 target with a 64-bit host compiler.
1837 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1838 opened with :IF-EXISTS :OVERWRITE.
1839 * fixed bug 281: error for an invalid qualifier in a short-form method
1840 combination method is not signalled until the faulty method is called.
1841 * bug fix: iteration variable type inferrer failed to deal with open
1842 intervals. (reported by Alan Shields)
1843 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1844 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1845 or more runtime options were provided to the sbcl binary.
1846 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1847 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1848 * optimization: there's now a fast-path for fixnum arguments in the
1849 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1850 * optimization: the code generated on x86-64 is more compact thanks
1851 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1852 * bug fix: floating-point exception handling now partly works on
1853 x86-64. (Thanks to James Knight)
1854 * improvement to the MIPS backend from Thiemo Seufer: C-style
1855 64-bit long long arguments and return values to alien functions
1857 * fixed some bugs revealed by Paul Dietz' test suite:
1858 ** the type-error signalled from WARN has a filled-in DATUM slot.
1859 ** the type-error required when a stream is not associated with
1860 a file has the stream as its datum.
1861 ** type-errors on single-floats on x86-64 no longer have
1862 :INVALID-OBJECT as the datum
1863 ** the type-errors signalled for invalid function names now have
1864 a correct expected type
1865 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1866 for structures defined with a :TYPE. Documentation strings for
1867 typed structures are no longer immediately discarded
1868 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1869 broadcast streams and synonym streams. FILE-LENGTH now also works
1870 on broadcast streams.
1872 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1873 * incompatible change: the --noprogrammer option, deprecated since
1874 version 0.7.5, has been removed. Please use the equivalent
1875 --disable-debugger option instead.
1876 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1878 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1879 inoperational for a while now) has been completely removed.
1880 * null lexical environments are now printed as #<NULL-LEXENV>,
1881 significantly reducing the amount of clutter in typical backtraces.
1882 * documentation on weak pointers, finalization, and after GC hooks
1883 has been added to the manual.
1884 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1885 as other specialized array types, is much faster. SUBSEQ and
1886 COPY-SEQ on such arrays have also been sped up.
1887 * optimization: EQL is now more efficient when at least other argument
1888 is known to be of type (OR FIXNUM (NOT NUMBER)).
1889 * fixed bug: the runtime is now less vulnerable to changes in the
1890 size of the SBCL object on OS X, and virtual memory is reserved for
1891 all spaces that need to be at a fixed address.
1892 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1893 * fixed bug: finalizers and after GC hooks that cause consing are now
1895 * fixed bug: compiler error messages and summaries are now printed to
1896 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1897 * fixed inference of the upper bound of an iteration variable.
1898 (reported by Rajat Datta).
1899 * fixed bug 211e: calling local functions with duplicated constant
1900 keyword argument no longer causes a bogus style warning about an
1902 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1903 ftype declaration to be disregarded. (reported by Dave Roberts)
1904 * fixed bug 373: caused by erronous compilation of references to alien
1905 variables in the runtime on ppc/darwin.
1906 * fixed bug 376: CONJUGATE type deriver.
1907 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1908 (reported by Baughn on #lisp)
1909 * a cross-compiler bug on non-x86 platforms has been identified and
1910 fixed. (thanks to Bruno Haible)
1911 * improvements to the MIPS runtime code for increased stability.
1912 (thanks to Thiemo Seufer)
1913 * increased the maximimum compact environment size to allow
1914 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1915 * partial workaround for bug 135: don't name the function we're
1916 calling for hairy cases of EVAL, so as not to accumulate one environment
1917 entry per call to EVAL. (reported by Kevin Reid)
1918 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1919 * optimization: type testing for non-vector arrays should be faster.
1920 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1922 * bug fix: setting 31st element of a bit vector to zero did not work
1924 * bug fix: redefining a class definition which failed due to a
1925 previous accessor / function clash now works (but see BUGS entry
1926 #380 for more problems in this area). (thanks to Zach Beane)
1927 * the long form of DEFINE-METHOD-COMBINATION disables method group
1928 checking when given a single method group with pattern *. (thanks
1930 * on x86 compiler supports stack allocation of results of simple
1931 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1932 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1933 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1934 * fixed some bugs related to Unicode integration:
1935 ** the restarts for recovering from input and output encoding
1936 errors only appear when there is in fact such an error to
1938 * fixed some bugs revealed by Paul Dietz' test suite:
1939 ** MISC.549 and similar: late transformation of unsafe type
1940 assertions into derived types caused unexpected code
1942 ** SCALE-FLOAT type deriver is less wrong.
1943 ** type derivers for EXP, LOG and similar functions compute result
1944 types for complex arguments better.
1945 ** (MISC.563) CONJUGATE type deriver works for very restricted
1947 ** out-of-line type testers for character strings are available.
1948 ** EQUAL compiler transform understands specialness of objects
1950 ** accessing double-floats stored on the stack now works on x86-64.
1951 ** debugger internals could sometimes create invalid lispobjs,
1952 resulting in GC crashes.
1953 ** MISC.548: type check weakening can convert required type into
1955 ** initialization forms for bindings are not in scope of free special
1958 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1959 * incompatible change: thread support for non-NPTL systems has
1960 been removed - locking is buggy and unreliable. A threaded
1961 SBCL build will now warn at startup and refuse to create
1962 new threads, unless futex support is detected in the kernel
1963 * incompatible change: the top level REPL now has only an ABORT
1964 restart associated with it, not TOPLEVEL and ABORT as it used to.
1965 TOP and TOPLEVEL are now available as debugger commands for
1966 returning to the top level.
1967 * incompatible change: forms evaluated in the REPL now use the
1968 global optimization policy.
1969 * incompatible change: user- and system-initialization files are
1970 no longer processed with LOAD, but by READ and EVAL; hence the
1971 global optimization policy, startup package, readtable, etc,
1973 * The .fasl file format number has been incremented because of
1974 various incompatible changes.
1975 * internal entry point details and argument counts no longer appear
1976 in backtraces unless explicitly requested by setting
1977 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1978 * built-in and standard functions no longer have names like "top
1979 level local call to FOO".
1980 * fixed bug 32: functions defined in non-null lexical environments
1981 now have more legible printed representation
1982 * fixed bug 33: functions defined in non-null lexical environemnts
1983 are now more amenable to inspection by INSPECT.
1984 * workaround for bug 354: XEPs no longer appear in backtraces unless
1985 explicitly requested.
1986 * fixed bug: receiving the signal which results from attempting to
1987 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1988 SIGBUS on other *BSDs) on architectures where the C stack is also
1989 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1990 to debugger confusion or wild instability. (reported by Cheuksan
1992 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1993 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1994 (reported by Lutz Euler)
1995 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1996 compiled code. (reported by Paul Dietz)
1997 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1998 specialized parameter to a method to something that is not TYPEP
1999 the specializer is now possible.
2000 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
2001 face of package deletion.
2002 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
2003 pathname. (reported by David Tolpin on c.l.l)
2004 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
2005 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
2006 STRING-STREAM and FILE-STREAM, respectively.
2007 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
2008 than previously; in addition, its test suite is now run on build.
2009 (thanks to Luke Gorrie)
2010 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
2011 * optimization: added a immediate representation for single-floats
2013 * fixed some bugs revealed by Paul Dietz' test suite:
2014 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2015 ** attempting to create a package with a colliding nickname causes
2016 correctable errors to be signalled.
2017 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2018 ** several x86-64 backend bugs related to sign-extension of immediate
2021 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2022 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2023 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2025 * fixed loading of multiply forward-referenced layouts.
2026 (thanks to Cheuksan Wang)
2027 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2028 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2029 (see COMPILE-FILE documentation for details of :PRINT option)
2030 * fixed bugs 19 and 317: fixed-format floating point printing is
2031 more accurate. This also fixes a bug reported by Adam Warner
2032 related to the ~@F format directive.
2033 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2035 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2036 dispatch table if the from-char is a dispatch macro character.
2037 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2038 by the wordsize no longer ignore the last word. (reported by Lutz
2040 * fixed bug in type checking of dynamic-extent variables. (reported
2042 * optimization: sequence traversal functions use their freedom to
2043 coerce function designators to functions.
2044 * optimization: code with many calls to CLOS methods specialized on
2045 CLOS classes has had redundant type checks removed.
2046 * optimization: added declarations to speed up operations that access
2047 the internal character database (for example STRING-UPCASE)
2048 * optimization: comparison operations between floats and sufficiently small
2049 fixnums no longer create extra rationals
2050 * fixed some bugs related to Unicode integration:
2051 ** portions of multibyte characters at the end of buffers for
2052 character-based file input are correctly transferred to the
2053 start of the buffer at the next read.
2054 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2055 passing it through to OPEN.
2056 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2057 argument given to internal calls to OPEN.
2058 ** the built-in comment readers (introduced by character sequences
2059 ";" and "#|") are more forgiving to encoding errors; they will
2060 STYLE-WARN and then attempt to resync the stream at a character
2061 boundary. (thanks to Teemu Kalvas)
2062 * fixed some bugs in the x86-64 port:
2063 ** Negative short int return values from c-calls are sign-extended
2065 ** The stack is aligned to 16-bytes for c-calls, as required by
2066 the ABI. (thanks to Cheuksan Wang)
2067 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2068 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2069 for enums. (thanks to Vincent Arkesteijn)
2070 ** Multiple small optimizations and bugfixes for floating point
2072 * fixed some bugs revealed by Paul Dietz' test suite:
2073 ** Space, Tab, Linefeed, Return and Page have the invalid
2074 secondary constituent character trait.
2075 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2077 ** WITH-INPUT-FROM-STRING should only update the index place on
2079 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2080 works more reliably.
2081 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2082 with one-dimensional arrays with fill pointers.
2083 ** TYPE-OF failed on a complex with an integer realpart and a
2085 ** compiler failure during type inference for the code of form
2086 (IF (EQL X (THE ...)) ...) (MISC.535).
2088 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2089 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2090 platform. The port supports 61-bit fixnums, large memory spaces
2091 and reloading shared object files.
2092 * enhancement: saving cores with foreign code loaded is now
2093 supported on ppc/Darwin in addition to the previously supported
2095 * enhancement: the statistical profiler now walks deeper into the
2096 call stack for more meaningful call-graphs and accrued time
2097 reports (x86/x86-64 only). It also now reports time spent in
2099 * enhancement: it is now possible to trace most individual methods
2100 of a generic function in addition to tracing the generic function
2102 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2104 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2105 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2106 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2107 produces an error. (thanks to Vincent Arkesteijn)
2108 * bug fix: NAMESTRING on pathnames with :WILD components in their
2109 directories works correctly. (thanks to Artem V. Andreev)
2110 * fixed bug 125: compiler preserves identity of closures. (reported
2112 * bug fixed: functions with &REST arguments sometimes failed with
2113 "Undefined function" when compiled with (DEBUG 3). (reported by
2115 * bug fix: overflow during compiling of setting element of a bit
2116 vector with constant index and value. (reported by Timmy Douglas)
2117 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2118 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2119 * on x86 compiler partially supports stack allocation of dynamic-extent
2121 * GO and RETURN-FROM do not check the extent of their exit points
2122 when compiled with SAFETY 0.
2123 * fixed some bugs related to Unicode integration:
2124 ** encoding and decoding errors are now much more robustly
2125 handled; it should now be possible to recover even from invalid
2126 input or output to the terminal. (thanks to Teemu Kalvas)
2127 ** provided a first cut at implementing STRING-TO-OCTETS and
2128 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2129 ** altered the SB-MD5 contributed module slightly, changing the
2130 interface just enough for it to be supportable for builds where
2131 lisp characters are not eight bits.
2132 * fixed some bugs revealed by Paul Dietz' test suite:
2133 ** the FORMATTER-generated functions for ~V[ conditionals require
2134 the correct number of arguments.
2135 ** READ-FROM-STRING returns the mandated second value when applied
2136 to displaced strings.
2137 ** the #\Rubout and #\Backspace characters are treated as invalid
2138 constituent characters by the tokenizer.
2140 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2141 * new feature: reloading changed shared object files with
2142 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2143 * new feature: references to foreign variables and functions
2144 can now be compiled and loaded before the corresponding shared
2145 object file is loaded, as long as the foreign definitions are
2146 available at runtime.
2147 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2148 supported. (thanks to Dan Debertin)
2149 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2150 just keywords. (thanks to Vincent Arkesteijn)
2151 * fixed bug #331: structure-class instances corresponding to
2152 DEFSTRUCT forms are now created eagerly.
2153 * fixed bug #345: backtraces from calls to undefined functions work
2154 on x86 as well. Related bug #61 is now also partially fixed on x86
2155 (backtraces from throws to unknown catch tags.)
2156 * bug fix: lambda-list parsing is now stricter vrt. order and number
2157 of lambda-list keywords.
2158 * bug fix: as specified by AMOP, an error is signalled if a
2159 class-option appears multiple times in a DEFCLASS form. (reported
2161 * bug fix: RANDOM can be compiled when the compiler derives the type
2162 of its numeric argument as a disjoint set of small integers.
2163 (reported by Paul Dietz)
2164 * bug fix: starting a core saved with shared objects loaded when
2165 those objects are not available no longer causes threaded SBCL to
2166 hang. (reported by Sean Ross)
2167 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2169 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2170 argument, which must always be NIL. (reported by Kalle Niemitalo)
2171 * bug fix: printing 1.0d+23 no longer results in an error.
2172 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2173 * bug fix: structure slot setters preserve evaluation order. (thanks
2175 * fixed some bugs related to Unicode integration:
2176 ** RUN-PROGRAM can allow its child to take input from a Lisp
2177 stream. (reported by Stefan Scholl)
2178 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2179 locales. (reported by Ken Causey)
2180 * fixed some bugs revealed by Paul Dietz' test suite:
2181 ** INCF, DECF and REMF evaluate their place form as specified in
2183 ** FORMATTER expands ~{ ~} iteration directives with V or #
2184 parameters correctly.
2185 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2186 consequent uses no arguments correctly.
2187 ** the system has a partial understanding of the (COMPLEX RATIO)
2189 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2192 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2193 * new feature: a build-time option (controlled by the :SB-UNICODE
2194 keyword feature, enabled by default) for building the system with
2195 support for the entire 21-bit character space defined by the
2197 * new feature: the system now has rudimentary external-format
2198 support; the primary user-visible change at this time is that
2199 characters with the high bit set (such as lower-case-e-acute) will
2200 print correctly to a terminal in a UTF-8 environment.
2201 * minor incompatible change: BASE-CHAR no longer names a class;
2202 however, CHARACTER continues to do so, as required by ANSI.
2203 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2204 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2205 has been moved to the SB-EXT package (temporarily re-exported from
2207 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2208 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2210 * fixed bug #351: better error-handlind and reporting for malformed
2212 * fixed bug #350: bignum-printing is now more memory-efficient,
2213 allowing printing of very large bignums, eg. (expt 2 10000000).
2214 (reported by Bruno Haible)
2215 * fixed bug #302: better primitive-type selection for intersection
2217 * fixed bug #308: non-graphic characters now all have names, as
2218 required. (reported by Bruno Haible)
2219 * bug fix: redefining a class with different superclasses now correctly
2220 removes it from the direct-subclasses of its previous superclasses.
2221 (reported by David Morse)
2222 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2223 new value now works. (reported by Bruno Haible)
2224 * bug fix: correct canonicalization of multiple non-standard slot
2225 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2226 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2227 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2228 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2229 now exists, an signals an error.
2230 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2231 during macroexpansion for non-iteration variables as well. (reported
2232 by Bruno Haible for CMUCL)
2233 * bug fix: Cyclic structures and unprintable objects in compiler
2234 messages no longer cause errors. (reported by Bruno Haible)
2235 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2236 and READ-FROM-STRING all now return a primary value of NIL if
2237 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2238 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2239 (reported by Bruno Haible for CMUCL)
2240 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2241 do not designate a condition. (reported by Bruno Haible for
2243 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2244 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2245 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2246 permitting resolution in favour of any of the conflicting symbols.
2247 (reported by Bruno Haible for CMUCL)
2248 * bug fix: EQUAL compiler optimizations is less aggressive on
2249 strings which can potentially compare true despite having distinct
2250 specialized array element types.
2251 * bug fix: unit enumerations can be defined without dividing by
2252 zero. (thanks to Vincent Arkesteijn)
2253 * FORMAT compile-time argument count checking has been enhanced.
2254 (report from Bruno Haible for CMUCL)
2255 * a partial workaround for the bug 262: the compiler does not try to
2256 inline-expand a local function doing RETURN-FROM from a deleted
2257 BLOCK. (thanks to Peter Denno for the bug report and to David
2258 Wragg for the simple test case)
2259 * fixed some bugs revealed by Paul Dietz' test suite:
2260 ** RENAME-PACKAGE allows all package designators as new package
2262 ** constraint propagation and lambda variable substitution are
2263 more cautious in dealing with partially deleted code.
2264 ** compiler sometimes lost reoptimization passes.
2265 ** CERROR, given a condition as condition designator, passes its
2266 remaining arguments to the continue format control without
2268 ** Case-altering FORMAT directives work correctly on non-ASCII
2270 ** The REAL type specifier handles bounds outside the floating
2271 point ranges without signalling FLOATING-POINT-OVERFLOW.
2272 ** Functions with IR1-transformations can create intercomponent
2273 references to global functions.
2274 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2276 ** FORMAT ~R treats a nil value for its first parameter correctly.
2278 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2279 * enhancement: saving cores with foreign code loaded is now
2280 supported on x86/NetBSD and sparc/Linux in addition to the previously
2281 supported platforms.
2282 * bug fix: on some platforms repeated installations caused multiple
2283 copies of HTML documentation to be installed -- should not happen
2284 any more. (reported by Stefan Scholl)
2285 * bug fix: parsing self-recursive alien record types multiple times
2286 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2287 original patch by Helmut Eller for CMUCL)
2288 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2289 (thanks to Richard Kreuter)
2290 * bug fix: defining classes whose accessors are methods on existing
2291 generic functions in other (locked) packages no longer signals
2292 bogus package lock violations. (reported by François-René Rideau)
2293 * bug fix: special variables as DEFMETHOD parameters no longer have
2294 associated bogus type declarations. (reported by David Wragg and
2296 * bug fix: read-write consistency on streams of element-type
2297 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2298 * bug fix: redefiniton of the only method of a generic function with
2299 no DEFGENERIC no longer emits a full WARNING. In addition,
2300 redefinition of generic functions with no DEFGENERIC to an
2301 incompatible lambda list now signals an error. (thanks to Zach
2303 * bug fix: DEFGENERIC now works even when there's a function of the
2304 same name in an enclosing lexical environment. (thanks to Zach
2306 * fixed compiler failure, caused by instrumenting code during
2307 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2308 * optimization: added loop analysis and improved register allocation
2309 * fixed some bugs revealed by Paul Dietz' test suite:
2310 ** POSITION on displaced vectors with non-zero displacement
2311 returns the right answer.
2312 ** (SIMPLE-STRING) is a valid type specifier for sequence
2314 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2316 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2317 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2319 ** The justification version of the FORMAT ~< directive treats
2320 non-zero minpad parameter correctly.
2322 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2323 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2324 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2325 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2326 the supported interface.
2327 * new feature: Single-stepping of code compiled with DEBUG 2 or
2328 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2329 * new feature: saving cores with foreign code loaded is now
2330 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2331 Timothy Moore's work for CMUCL)
2332 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2333 parameters to * instead of NIL if no initform is supplied.
2334 (reported by Johan Bockgaard)
2335 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2336 correctly when FUNCALL appears as the car of the form. Note:
2337 despite this FUNCALL forms are not currently subject to
2338 compiler-macro expansion. (port of Raymond Toy's fix for the
2339 same from CMUCL, reported by Johan Bockgaard)
2340 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2341 (thanks for Teemu Kalvas)
2342 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2343 case instead of terminating the iteration (thanks for Julian
2344 Squires, Sean Champ and Raymond Toy)
2345 * bug fix: incorrect expansion of defgeneric that caused a style
2346 warning. (thanks for Zach Beane)
2347 * bug fix: slot accessor effective method computation works properly
2348 for classes with multiple non-standard applicable methods on
2349 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2350 * on x86 compiler supports stack allocation of results of LIST and
2351 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2352 CMUCL implementation by Gerd Moellmann)
2353 * fixed some bugs revealed by Paul Dietz' test suite:
2354 ** FORMAT strings with both the ~<~:;~> form of the justification
2355 directive and pretty-printing directives cause an error.
2357 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2358 * incompatible change: the internal functions
2359 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2360 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2361 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2362 instead of the old functions.
2363 * new feature: on platforms where "dladdr" is available foreign
2364 function names now appear in backtraces. (based on Helmut Eller's
2366 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2367 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2369 * documentation: ASDF manual is now distributed with SBCL.
2370 * bug fix: SBCL can now load its contributed modules with REQUIRE
2371 even if the system-provided entries have been removed from
2372 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2374 * bug fix: backtraces involving undefined functions or assembly
2375 routines are more informative. (thanks to Brian Downing)
2376 * bug fix: mutually referent alien structures now work correctly.
2377 (reported by Rick Taube)
2378 * bug fix: structures defined by WITH-ALIEN can be referred to
2379 within other definitions in the same WITH-ALIEN.
2380 * bug fix: division operators (MOD, TRUNCATE and the like) with
2381 constant zero divisors and integer dividends no longer generate
2383 * bug fix: provide default methods for INPUT-STREAM-P and
2384 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2385 * bug fix: improve the detection and resolution of MOP metacycles
2386 (where effective method computation depends on the generic function
2387 having its effective method computed). (reported by Bruno Haible)
2388 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2389 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2390 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2391 * bug fix: pathnames with relative directory components can be
2392 represented relative to default pathnames.
2393 * optimization: in taking the GCD of bignums, reduce the two bignums
2394 to approximately the same size (using Euclid's algorithm) before
2395 applying the more sophisticated binary GCD. (thanks to Juho
2397 * optimization: COUNT on bitvectors now operates word-at-a-time.
2398 * optimization: ASH with a positive, but not necessarily constant,
2399 (leftwards) shift, when in a modular context, is compiled to a
2401 * fixed some bugs revealed by Paul Dietz' test suite:
2402 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2403 if the corresponding argument is NIL.
2404 ** FORMAT directives accepting a minpad parameter treat negative
2406 ** literal commas as character parameters to FORMAT directives are
2408 ** literal spaces directly after ~<Newline> directives within a
2409 format-logical-block (~:< ~@:>) do not induce :FILL-style
2410 conditional newlines.
2411 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2412 as specified: it no longer includes conditional newlines).
2413 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2415 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2416 :PER-LINE-PREFIX argument does not evaluate to a string.
2418 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2419 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2420 the manual for details; for now, package locks can be disabled by
2421 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2422 if no major problems are found then it is likely that they will be
2423 compiled in unconditionally.
2424 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2425 unsupported operators on all platforms. To load a shared library
2426 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2427 object file, link it into a shared library outside of SBCL and
2428 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2430 * bug workaround: ROOM T and the GC were not working together
2431 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2432 reporting facility aren't preserved. That reporting has been
2433 disabled, so now until and unless someone figures out how to make it
2434 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2435 * minor incompatible change: as threatened around sbcl-0.8.0, the
2436 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2437 The symbols are also exported from SB-PCL for backwards
2438 compatibility, but more so than before SB-PCL should be treated as
2439 an implementation-internal package.
2440 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2442 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2443 It is known as of this release not to work on the Alpha, however.
2444 * fixed bug #167: errors signalled due to illegal syntax in method
2445 bodies are now more legible.
2446 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2447 designators and can hence be used with TYPEP.
2448 * fixed bug #333: CHECK-TYPE now ensures that the type error
2449 signalled, if any, has the right object to be accessed by
2450 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2451 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2452 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2454 * fixed bug #334: programmatic addition of slots using specialized
2455 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2456 and :ALLOCATION :CLASS slots.
2457 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2458 float exponents. (rereported by Peter Seibel)
2459 * fixed a bug: #\Space (and other whitespace characters) are no
2460 longer considered to be macro characters in standard syntax by
2461 GET-MACRO-CHARACTER.
2462 * fixed bug: initialization of condition class metaobjects no longer
2463 causes an instance of the condition to be created. (reported by
2465 * fixed bug: it is now possible to have slots such that
2466 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2467 description is neither :INSTANCE nor :CLASS.
2468 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2469 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2470 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2471 non-standard SLOT-DEFINITION-ALLOCATION values.
2472 * fixed bugs: various race conditions handling exiting threads.
2473 CL-PPCRE's thread torture test now passes.
2474 * fixed arguably-a-bug: GC time proportional to number of threads in
2475 system even when most of them are idle
2476 * optimization: improved performance of BIT and SBIT on bit-vectors.
2477 * .fasl file incompatibility: The fasl file version number has
2478 been incremented because of changes associated with package locks.
2480 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2481 * minor incompatible change: the system no longer provides
2482 optimization hints (conditions which are TYPEP
2483 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2484 mode; these hints are still emitted when higher SPEED optimization
2486 * new contrib module: a sampling profiler (profiling by statistical
2487 sampling, rather than by instrumenting functions) is available as
2488 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2489 * the behaviour of the standard function ED is now customizeable by
2490 third parties through a hook variable: see ED's documentation
2491 string for information on the protocol.
2492 * the compiler no longer emits efficiency notes for (FUNCALL X)
2493 when the type of X is uncertain under default optimization
2495 * fixed bug 276: mutating a binding of a specialized parameter to a
2496 method to something that is not TYPEP the specializer is now
2498 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2499 exhibit the required behaviour on the x86 platform. (thanks to
2500 Peter van Eynde, Eric Marsden and Bruno Haible)
2501 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2502 even for difficult arguments. (reported by Peter Graves)
2503 * fixed bug 141a: the backquote printer now descends quoted
2505 * fixed another bug in backquote printing: no more destructive
2506 modification of the form's list structure. (reported by Brian
2508 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2509 move between its address being taken and the call to
2510 interrupt_thread, fixing a crashing race condition.
2511 * the SB-POSIX contrib implementation has been adjusted so that it
2512 no longer exhibits ridiculously poor performance when constructing
2513 instances corresponding to C structs.
2515 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2516 * minor incompatible change: the sb-grovel contrib now treats C
2517 structures as alien (in the sense of SB-ALIEN) objects rather than
2518 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2519 has implications for memory management of client code
2520 (sb-grovel-returned objects must now be manually managed) and for
2521 type safety (alien objects now have full types).
2522 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2523 used to control emission of compiler diagnostics, rather than the
2524 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2525 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2526 quality should be considered deprecated.
2527 * install.sh now installs the user manual as well
2528 * (not quite a new documentable feature, but worth considering in
2529 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2530 beginnings of a semantically meaningful condition hierarchy is
2531 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2533 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2534 designator as the defaults argument.
2535 * fixed bug: Displaced arrays whose displaced-to array has become
2536 too small now cause ARRAY-DIMENSION to signal an error, providing
2537 for safer bounds-checking. (reported by Bruno Haible)
2538 * fixed bug: DEFCLASS slot definitions with identical :READER and
2539 :WRITER names now signal a reasonable error. (reported by Thomas
2541 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2543 * fixed bug: Invalid slot specification errors now print correctly.
2544 (thanks to Zach Beane)
2545 * fixed bug 320: Shared to local slot value transfers in class
2546 redefinitions now happen corrently from superclasses as
2547 well. (reported by Bruno Haible)
2548 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2550 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2551 improper lists correctly. (reported by Bruno Haible)
2552 * fixed bug 313: source-transform for <fun-name> was erroneously
2553 applied to a call of a value of a variable with name <fun-name>.
2554 (reported by Antonio Menezes Leitao)
2555 * fixed bug 307: The obsolete instance protocol ensures that
2556 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2557 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2558 remove dead unknown-values globs from the middle of the stack.
2559 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2560 easier to resume long computations after using *BREAK-ON-SIGNALS*
2561 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2562 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2563 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2564 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2565 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2566 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2568 * fixed a (fairly theoretical) bug in string printing: if
2569 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2570 does not have array-element-type equal to the most general string
2572 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2573 function name. (thanks to Nikodemus Siivola)
2574 * fixed bug: compile-time format string checker failed on
2575 non-closed ~{. (reported by Thomas F Burdick)
2576 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2577 don't warn on legal loop code involving more than one aggregate
2578 boolean. (thanks to Nikodemus Siivola)
2579 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2580 in keyword position to MAKE-PACKAGE was not being reported as
2582 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2583 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2585 * fixed bugs 280 and 312: the checking for multiple definitions in a
2586 file is less likely to become confused by uses of inline
2588 * fixed bug: the #S reader macro performs the keyword coercion
2589 specified for slot names. (reported by Kalle Niemitalo)
2590 * fixed bug: lambda lists may contain symbols whose names start with
2591 & but are not lambda-list-keywords; their occurrence triggers a
2593 * fixed bug 321: define-method-combination argument lambda lists do
2594 not cause invalid code to be generated when &optional or &aux
2595 variables have default values. (reported by Bruno Haible)
2596 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2597 have CLOS classes; this appears also to have fixed reported
2598 instability in stack exhaustion detection.
2599 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2600 MUFFLE-WARNING when no associated restart is present is now
2602 * optimization: rearranged the expansion of various defining macros
2603 so that each expands into only one top-level form in a
2604 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2606 * optimization: used a previously little-used slot in symbols to
2607 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2609 * fixed some bugs revealed by Paul Dietz' test suite:
2610 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2612 ** Fixed an optimization bug related to inheritance of initforms
2613 from local to shared slots.
2614 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2615 ** CAST optimizer forgot to flush argument derived type.
2616 ** print/read consistency on floats is now orders of magnitude
2617 more likely. (thanks also to Bruno Haible for a similar report
2619 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2620 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2621 for objects of type REAL. Make it so.
2622 ** SXHASH is specified (infelicitously) to respect similarity,
2623 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2624 it so. (thanks to Markus Ziegler)
2625 ** on the Alpha, the compiler succeeds in compiling functions
2626 returning a known number of arguments greater than 63.
2627 ** fixed handling of invalid NIL arguments in keyword position
2629 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2631 ** provided a readably-printable representation for RANDOM-STATE
2633 ** ensured that pathnames, where they have a namestring, always
2634 print using #P"..." syntax.
2636 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2637 * Support for the forthcoming 2.0 version of the NetBSD kernel
2638 running on x86 hardware has been added. (thanks to Perry
2639 E. Metzger most immediately, and others for their past work)
2640 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2641 current release version; SBCL's OpenBSD support had been broken
2642 since about the time of OpenBSD's switch to ELF binary format.)
2643 * [placeholder for DX summary]
2644 ** user code with &REST lists declared dynamic-extent, under high
2645 speed or space and low safety and debug optimization policy.
2646 * The manual has been converted to Texinfo format and the debugger
2647 chapter from the cmucl manual has been added.
2648 * A facility has been added to extract documentation strings from
2649 sbcl and store them as Texinfo-formatted snippets for inclusion in
2650 the manual (via Texinfo's @include directive)
2651 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2652 the test case to Dave Roberts)
2653 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2654 pointers. (reported by Sean Ross)
2655 * bug fix: PROFILE output is printed nicely even for large numerical
2656 values. (thanks to Zach Beane)
2657 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2658 greater than 32 handle EOF correctly.
2659 * bug fix: on X86 an immediate argument of the IMUL instruction is
2660 correctly printed in disassembly. (thanks to Lutz Euler)
2661 * bug fix: class slots in redefined classes preserve their old
2662 values. (thanks to Bruno Haible and Nikodemus Siivola)
2663 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2664 succeeds. (reported by Marco Baringer)
2665 * fixed some bugs revealed by Paul Dietz' test suite:
2666 ** READ-SEQUENCE now works on ECHO-STREAMs.
2667 ** RATIONALIZE works more according to its specification. (thanks
2670 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2671 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2672 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2673 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2674 should matter to you only if you rebind the printer control
2675 variables and then find you want different bindings in the
2676 debugger than in the ordinary execution of your program.)
2677 * The runtime build system has been tweaked to support building
2678 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2679 assembler and linker. (thanks to Nikodemus Siivola)
2680 * Unbound, undefined, undeclared variables now trigger full
2681 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2682 more often programmer error than deliberate exploitation of undefined
2684 * optimization: the hash algorithm for strings has changed to one
2685 that is less vulnerable to spurious collisions. (thanks to Juho
2687 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2688 needless bounds checking. (thanks to Juho Snellman)
2689 * optimization: implemented multiplication as a modular
2690 (UNSIGNED-BYTE 32) operation on the PPC backend.
2691 * fixed some bugs revealed by Paul Dietz' test suite:
2692 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2694 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2695 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2696 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2698 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2700 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2702 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2703 their output stream on EOF from read.
2704 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2705 have been read to end-of-file.
2706 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2708 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2709 description of determination of which consecutive characters
2711 ** Printing the "Space" character with escaping on now yields "#\\ ",
2712 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2713 ** Reading floating-point numbers with *READ-BASE* set to a number
2714 less than 10 works correctly.
2715 ** Reading floating-point numbers with *READ-BASE* set to a number
2716 more than 10 works correctly.
2717 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2718 the readtable currently in effect.
2720 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2721 * minor incompatible change: parsing of namestrings on a physical
2722 (Unix) host has changed; numbers after the final #\. in a
2723 namestring are no longer interpreted as a version field. This is
2724 intented to be largely invisible to the user, except that the
2725 meaning of the namestring "*.*.*" has changed: it now refers to a
2726 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2727 should usually be replaced by
2728 (make-pathname :name :wild :type :wild :version :wild)
2729 with the added benefit that this is more likely to be portable.
2730 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2731 option to OPEN now signals an error if the file being opened
2732 exists; this may have an impact on existing code.
2733 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2734 signals received, on the PowerPC platforms (both Linux and
2735 Darwin). (thanks to Pierre Mai for pointing out the location of
2737 * several fixes on OS X: The system now builds and runs cleanly on
2738 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2739 (thanks to Brian Mastenbrook)
2740 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2741 second-resolution: integer multiples of 1/3600 between -24 and 24.
2742 (thanks to Vincent Arkesteijn)
2743 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2744 type when called with 1 argument; PEEK-CHAR checked type of
2745 PEEK-TYPE only after having read first character from a
2746 stream. (reported by Peter Graves)
2747 * bug fix: the garbage collector now has much better locality
2748 behaviour, and in particular no longer treats all memory as being
2749 exhausted when many small objects point to each other in a deeply
2751 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2752 recognized as being TYPEP their class.
2753 * bug fix: the PUSHNEW documentation string has been corrected.
2754 (thanks to Vincent Arkesteijn)
2755 * bug fix: defaulting of the value for the last of an atypically
2756 large number of multiple values being bound was not being
2757 performed correctly on the Alpha or PPC platforms
2758 * optimization: implemented multiplication as a modular
2759 (UNSIGNED-BYTE 32) operation on the x86 backend.
2760 * optimization: SEARCH on simple-base-strings can now be open-coded.
2761 (see also contrib/compiler-extras.lisp for inspiration for
2762 teaching the compiler about the Boyer-Moore algorithm).
2763 * value, returned by MAX (and MIN) called with several EQUALP, but
2764 not EQL, arguments now does not depend on compiler settings.
2765 * fixed some bugs revealed by Paul Dietz' test suite:
2766 ** in stack analysis liveness information is propagated from
2767 non-local entry points.
2768 ** pathwise CAST removing failed when the CAST node did not start
2770 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2771 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2773 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2774 host is already defined.
2775 ** RENAME-FILE works on streams instead of signalling an internal
2777 ** PEEK-CHAR uses the current readtable when determining whether
2778 or not a character is whitespace.
2779 ** MERGE-PATHNAMES handles the case when the pathname does not
2780 specify a name while the default-pathname specifies a version.
2781 ** Pathnames now stand a better chance of respecting print/read
2783 ** Attempting to use standardized file system operators with a
2784 pathname with invalid :DIRECTORY components signals a
2786 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2788 ** DIRECTORY on logical pathnames is more correct.
2789 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2790 signal a TYPE-ERROR if their argument is not a stream.
2791 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2792 designator argument does not designate a stream.
2793 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2794 examining the synonym.
2795 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2797 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2798 element-type larger than ([UN]SIGNED-BYTE 32).
2800 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2801 * When built with the :SB-FUTEX feature, threaded builds now take
2802 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2803 for faster/more reliable mutex and condition variable support.
2804 * Incompatible change (but one you probably shouldn't have been using
2805 anyway): the interface and code for arbitrating between multiple
2806 threads in the same user session has been redesigned.
2807 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2808 error when called without an explicit environment argument.
2809 (thanks to Nikodemus Siivola)
2810 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2811 confused about their position. (thanks to Adam Warner and Gerd
2813 * bug fix: Pretty printing backquoted forms with unquotations in the
2814 argument list position of various code constructs such as LAMBDA
2815 now works correctly. (reported by Paul Dietz)
2816 * bug fix: Pretty printing unquotations no longer loses all
2817 stream position information.
2818 * optimization: performance of string output streams is now less
2819 poor for multiple small sequence writes.
2820 * optimization: performance of CSUBTYPEP in the presence of complex
2821 expressions involving CONS and NOT many times has been improved.
2822 (reported by Paul Dietz)
2823 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2825 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2826 streams for the REPL, for the debugger, and so on. (thanks to
2828 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2829 * fixed some bugs revealed by Paul Dietz' test suite:
2830 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2831 is ignored if it is too small, rather than propagating through
2832 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2833 ** extremely complex negations of CONS types were not being
2834 sufficiently canonicalized, leading to inconsistencies in
2836 ** VALUES tranformer lost derived type.
2838 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2839 * fixed a bootstrapping bug: the build process no longer assumes
2840 that the various BOOLE-related constants have the same value in
2841 host and target lisps. (noted by Paul Dietz' test suite on an
2842 SBCL binary built from CLISP)
2843 * The system can now be dynamically linked on the MIPS platform,
2844 which enables dynamic loading of foreign code from Lisp. (thanks
2845 to Ralf Baechle for discussions on the MIPS ABI)
2846 * The system now records debugging information for its own source
2847 files in a filesystem-position-independent manner, relative to
2848 the "SYS" logical host.
2849 * fixed a compiler bug: MV-LET convertion did not check references
2850 to the "max args" entry point. (reported by Brian Downing)
2851 * tweaked disassembly notes to be less confident about proclaiming
2852 some instruction as an LRA. (thanks to Brian Downing)
2853 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2854 now each have their own history, command character, and other
2855 characteristics. (thanks to David Lichteblau)
2856 * fixed some bugs revealed by Paul Dietz' test suite:
2857 ** compiler failure in compiling LOGAND expressions including a
2859 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2860 shift greater than 32.
2861 ** FLUSH-DEST did not mark blocks for type check regeneration.
2862 ** HANDLER-CASE failed to accept declarations in handler clauses
2863 in some circumstances.
2865 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2866 * New code in contrib/sb-introspect (still probably not entirely
2867 stable yet) provides some support for smart Lisp development
2868 environments like SLIME.
2869 * The conditions signalled for errors occurring when loading .fasl
2870 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2871 in a way which should help ASDF recover gracefully.
2872 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2873 stuff has been cleaned up. If you code contrib/ stuff, this might
2874 affect you, and you can look at contrib/README, contrib/STANDARDS,
2875 and/or the 0.8.4.27 diff to check.
2876 * In full calls the compiler now does not generate checks for declared
2877 argument types for all arguments.
2878 * various threading fixes
2879 ** and some experimental patches which didn't make it into
2880 the main tree for this release, but which are shipped in
2881 contrib/experimental-thread.patch as a possible fix for some
2882 failures (deadlock, spinning...) in GC-intensive multithreaded
2884 * fixed PPC build problem (source code incompatibility of different
2885 library versions): added offsetof() hackery which attempts to divine
2886 where glibc maintainers put uc_mcontext today
2887 * fixed bug 282: compiler does not trust type assertions while passing
2888 arguments to a full call.
2889 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2890 &OPTIONAL VALUES type parameter.
2891 * fix bug 214: algorithm for noting rejected templates is now more
2892 similar to that of template seletion. (also reported by rydis on
2894 * fixed bug 141b: printing backquoted information readably and prettily
2895 inserts a space where necessary.
2896 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2897 (CHARACTER 10) are now reported as errors, rather than propagated
2898 as unknown types. (reported by piso on #lisp)
2899 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2900 with values NIL and :ERROR. (thanks to Milan Zamazal)
2901 * fixed bug 191c: CLOS now does proper keyword argument checking as
2902 described in CLHS 7.6.5 and 7.6.5.1.
2903 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2904 longer raise an error; further, using a list as a for-as-arithmetic
2905 counter now raises a meaningful error.
2906 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2907 now understood by sequence creation functions such as MAKE-SEQUENCE
2909 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2910 a STRING-INPUT-STREAM.
2911 * compiler enhancement: SIGNUM is now better able to derive the type
2913 * type declarations inside WITH-SLOTS are checked. (reported by
2915 * fixed some bugs revealed by Paul Dietz' test suite:
2916 ** incorrect optimization of TRUNCATE for a positive first
2917 argument and negative second.
2918 ** compiler failure in let-convertion during flushing dead code.
2919 ** compiler failure while deriving type of TRUNCATE on an
2920 interval, containing 0.
2921 ** ASH of a negative bignum by a negative bignum count now returns
2923 ** intersection of CONS types now canonicalizes properly, fixing
2924 inconsistencies in SUBTYPEP.
2926 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2927 * incompatible change: The --disable-debugger command line
2928 option now clobbers the debugger at a more fundamental
2929 level, by redefining #'INVOKE-DEBUGGER instead of by
2930 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2931 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2932 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2933 enter the debugger REPL and then suffer recursive errors
2934 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2935 while under the new system, BREAK in --disable-debugger mode
2936 terminates the system just as an unhandled error would.
2937 * fixed compiler performance when processing loops with a step >1;
2938 * bug fix: DOCUMENTATION now retrieves generic function
2939 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2940 support has been systematized, and now supports the methods
2941 specified by ANSI, along with a default method and a method for
2942 slot documentation. (reported by Nathan Froyd)
2943 * bug fix: effective methods associated with a generic function are
2944 no longer cached over a change of that generic function's method
2945 combination. (reported by Andreas Fuchs)
2946 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2947 names a non-existent directory. (thanks to Andreas Fuchs)
2948 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2949 the PPC platform, lead to stack corruption; this has been fixed.
2950 (reported by Rainer Joswig)
2951 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2952 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2953 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2954 arithmetic in general, is now much more reliable on the PPC
2956 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2957 platform now returns the right answer.
2958 * optimization: restored some effective method precomputation in
2959 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2960 precomputation is now tunable.
2961 * optimization: compiler-internal data structure use has been
2962 reviewed, and changes have been made that should improve the
2963 performance of the compiler by about 20%.
2964 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2965 simple-base-strings and simple-bit-vectors is improved.
2966 * optimization: the optimization of 32-bit logical and arithmetic
2967 functions introduced in version 0.8.3 on the x86 has been
2968 implemented on the mips, ppc and sparc platforms; an
2969 implementation of the same facility, but for 64-bit arithmetic,
2970 has been added for the alpha.
2971 * microoptimization: the compiler is better able to make use of the
2972 x86 LEA instruction for multiplication by constants.
2973 * bug fix: in some situations compiler did not report usage of
2974 generic arithmetic in (SPEED 3) policy.
2975 * bug 145b fix: compiler used wrong type specifier while converting
2976 MEMBER-types to numeric.
2977 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2979 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2980 index argument now works properly on the Alpha platform.
2981 * bug fix: floating point exception treatment on the Alpha platform
2983 * bug fix: FILE-POSITION works much better on string input and
2984 output streams. (thanks to Nikodemus Siivola)
2985 * bug fix: many threading/garbage collection symptoms sorted.
2986 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2987 might be pseudo-atomic.
2988 * internal change: Stopping for GC is now done with signals not
2989 ptrace. GC is now done in whichever thread wanted it, instead of
2991 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2993 * bug fix: result form in DO is not contained in the implicit
2995 * incompatible change: ICR structure is changed; the value part of
2996 CONTINUATION is now called LVAR; corresponding functions are
2997 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2998 * added type deriver for ISQRT (thanks to Robert E. Brown).
2999 * bug fix: better support for loading from the command line when an
3000 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
3002 * fixed some bugs revealed by Paul Dietz' test suite:
3003 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
3004 ** ROUND and FROUND now give the right answer when given very
3005 small float arguments.
3006 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
3008 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
3009 ** (GCD 0 <negative-integer>) returned <negative-integer>.
3010 ** LCM should return a non-negative integer.
3011 ** PARSE-INTEGER returned the index of a terminator instead of the
3012 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3013 ** PARSE-INTEGER returned an incorrect index being applied to a
3015 ** LCM with two arguments of 0 returns 0 rather than signalling
3017 ** unsigned addition of a 32-bit constant with the high bit set no
3018 longer causes an internal compiler error on the x86.
3019 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3020 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3021 with negative last argument.
3022 ** byte specifiers with zero size and position no longer cause
3023 an error during type derivation.
3024 ** bignum multiplication on the Alpha platform now returns the
3026 * porting: The system now builds on SuSE AMD64, although it still
3027 generates a 32-bit binary.
3028 * .fasl file incompatibility: The fasl file version number has
3029 been incremented (because of the changes to internal compiler
3030 data structures referred to above).
3032 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3033 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3034 more accurately, on the Darwin kernel running on PowerPC hardware.
3035 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3036 * Compiler code deletion notes now signal a condition of type
3037 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3038 an associated MUFFLE-WARNING restart.
3039 * The compiler now performs limited argument count validation of
3040 constant format strings in FORMAT, and where appropriate in ERROR,
3041 CERROR and WARN. (thanks to Gerd Moellmann)
3042 * New ASDF-INSTALL contrib can be used for automatic download and
3043 installation of third-party Lisp code from CCLAN or other sites
3045 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3046 which forces another thread to execute a function supplied by the
3048 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3049 now accept and act upon their :ELEMENT-TYPE keyword argument.
3050 (reported by Martin Atzmueller, Edi Weitz)
3051 * bug fix: FILE-POSITION now accepts position designators up to
3052 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3053 is the greater. (thanks to Patrik Nordebo)
3054 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3055 Kalle Olavi Niemitalo)
3056 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3057 printable. (reported by Eric Marsden)
3058 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3059 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3060 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3061 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3062 (thanks to Stig Sandoe)
3063 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3064 UNREAD-CHAR now correctly includes the unread character in the
3065 target sequence. (thanks to Gerd Moellmann)
3066 * bug fix in threaded builds: the system can now be suspended and
3067 resumed by shell job control with minimal disruption.
3068 * bug fixes in times and timezones >2038AD
3069 * better handling of "where is GNU make?" problem in build scripts
3070 (thanks to Nikodemus Siivola)
3071 * new optimization: inside a named function any reference to a
3072 function with the same name is considered to be a self-reference;
3073 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3074 optimization quality.
3075 * new optimization on x86: logical functions and + now have
3076 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3077 used when the result is truncated to 32 bits.
3078 * VALUES declaration is partially enabled.
3079 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3080 * fixed some bugs revealed by Paul Dietz' test suite:
3081 ** The system now obeys the constraint imposed by
3082 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3083 types form a lattice under type intersection.
3084 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3085 ** ASSOC now ignores NIL elements in an alist.
3086 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3087 and (1+ MOST-POSITIVE-FIXNUM) answers.
3088 ** The addition of a method with invalid qualifiers to a generic
3089 function does not cause an error to be signalled immediately;
3090 a warning is signalled, and the error is generated only on
3091 calling the generic function.
3092 * changed .fasl file version number, in part to add type codes for
3093 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3094 obscure ANSI requirements
3096 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3097 * fixed bug 148: failure to inline-expand a local function left
3098 garbage, confusing the compiler.
3099 * fixed bugs 3cd: structure slot readers perform type check if the
3100 slot can have an invalid value (i.e. it is either not initialized
3101 or can be written with a less specific slot writer).
3102 * bug fix: the compiler now traps array references to elements off
3103 the end of an array; previously, the bounds checking in some
3104 circumstances could go off-by-one.
3105 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3107 * type declarations for array element types now obey the description
3108 on the CLHS page "Declaration TYPE", as per discussions on
3109 sbcl-help around 2003-05-08. This means that a declaration
3110 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3111 declaration, all references to BAR will be asserted or assumed
3112 (with THE, so dependent on compiler policy) to involve objects of
3113 type FOO. Note that no such declaration is implied in
3114 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3115 * declared types of functions from the "Conditions"
3116 chapter. (reported by Paul Dietz)
3117 * bug fix: CERROR accepts a function as its first argument.
3118 * bug fix: NTH an NTHCDR accept a bignum as index
3119 arguments. (reported by Adam Warner)
3120 * optimization: character compare routines now optimize comparing
3121 against a constant character. (reported by Gilbert Baumann)
3122 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3123 argument now works properly on the MIPS platform.
3124 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3125 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3126 declared to have a functional type, does not cause an error
3128 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3129 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3130 * fixed bug: if last continuation of a deleted block has a
3131 destination, this destination should be deleted too. (reported by
3133 * fixed a bug in the bootstrap process: the host compiler's values
3134 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3135 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3136 test case from Patrik Nordebo)
3137 * improved the ability of the disassembler on the PPC platform to
3138 provide helpful disassembly notes.
3139 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3140 the class in more cases than previously.
3141 * bug fix: FILE-POSITION now understands :START and :END for
3142 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3143 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3144 8) data. (thanks to David Lichteblau)
3145 * bug fix: it is possible to add a method to a generic function
3146 without lambda list.
3147 * bug fix: reader failed to signal END-OF-FILE inside an
3148 object representation. (reported by Nikodemus Siivola)
3149 * fixed some bugs revealed by Paul Dietz' test suite:
3150 ** LAST and [N]BUTLAST should accept a bignum.
3151 ** condition slot accessors are methods.
3152 ** (VECTOR NIL) is a subtype of STRING.
3154 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3155 * minor incompatible change: some nonsensical specialized lambda
3156 lists (used in DEFMETHOD) which were previously ignored now signal
3158 * minor incompatible change: the system is now aware of the types of
3159 variables in the COMMON-LISP package, and will signal errors for
3160 most violations of these type constraints (where previously they
3161 were silently accepted).
3162 * minor incompatible change: COMPILE-FILE now uses the freedom
3163 afforded (ANSI 3.2.2.3) to use derived function types for
3164 functions defined in the same file. This also permits the system
3165 to warn on static type mismatches and function
3166 redefinition. (Currently it does not work with high DEBUG level.)
3167 * minor incompatible change: VALUES declaration is disabled.
3168 * When issuing notes, the compiler now signals a condition of type
3169 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3170 restart for use in user handlers. It is expected that the
3171 COMPILER-NOTE condition will eventually become a condition
3172 supertype to a hierarchy of note types, which will then be
3173 handleable in a similar fashion. However, at the moment, no such
3174 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3175 but it's an implementation detail, not a classification for the
3177 * Changes in type checking closed the following bugs:
3178 ** type checking of unused values (192b, 194d, 203);
3179 ** template selection based on unsafe type assertions (192c, 236);
3180 ** type checking in branches (194bc).
3181 * A short form of VALUES type specifier has ANSI meaning (it has
3182 increased the number of situations when SBCL cannot perform type
3184 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3185 keywords or constants is permissible.
3186 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3187 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3188 classes. (thanks to Antonio Martinez)
3189 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3190 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3191 argument to a pretty-print stream. Instead, it leaves any such
3192 operations to DESCRIBE-OBJECT methods.
3193 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3195 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3196 lambda lists. (thanks to David Lichteblau)
3197 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3198 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3199 (thanks to Gerd Moellmann)
3200 * fixed bug 246: increased compilation speed of long
3201 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3203 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3205 * DEFCONSTANT now throws a condition of type
3206 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3207 constant to a non-EQL value; CONTINUE and ABORT restarts
3208 respectively change and preserve the value.
3209 * fixed bug 63: The code walker, part of the implementation of CLOS,
3210 is now better at handling symbol macros.
3211 * bug fix: There is no longer an internal implementation type named
3212 CL:LENGTH. (reported by Raymond Toy)
3213 * bug fix: In macro-like defining macros/special operators the
3214 implicit block does not enclose the lambda list.
3215 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3216 enclosed in parentheses) are not suitable as type specifiers, and
3217 their use properly signals an error now.
3218 * bug fix: An argument count mismatch for a type specifier in code
3219 being compiled no longer causes an unhandled error at compile
3220 time, but signals a compile-time warning.
3221 * fixed simple vector readable printing
3222 * bug fix: DESCRIBE takes more care over whether the class
3223 precedence list slot of a class is bound before accessing it.
3224 (reported by Markus Krummenacker)
3225 * bug fix: FORMATTER can successfully compile pretty-printer format
3226 strings which use variants of the ~* directive inside.
3227 * bug fix: SEARCH now applies its TEST predicate to the elements of
3228 the arguments in the correct order. (thanks to Wolfhard Buss)
3229 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3230 in type inference. (thanks to Robert E. Brown)
3231 * bug fix: Reading in symbols with an explicit package name of ""
3232 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3233 * fixed some bugs revealed by Paul Dietz' test suite:
3234 ** NIL is now allowed as a structure slot name.
3235 ** Arbitrary numbers, not just REALs, are allowed in certain
3236 circumstances in LOOP for-as-arithmetic clauses.
3237 ** Multiple class redefinitions before slot access no longer
3238 causes a type error.
3239 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3240 association between the name and a class.
3241 ** Generic functions with non-standard method-combination and over
3242 five methods all of which return constants no longer return NIL
3243 after the first few invocations. (thanks to Gerd Moellmann)
3244 ** CALL-NEXT-METHOD with no arguments now passes the original
3245 values of the arguments, even in the presence of assignment.
3246 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3248 ** DISASSEMBLE works with closures and funcallable instances.
3249 ** ADD-METHOD now returns the generic function, not the new method.
3250 ** FIND-METHOD signals an error if the lengths of the specializers
3251 is incompatible with the generic function, even if the ERRORP
3253 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3254 which its argument is a member.
3255 ** DEFCLASS only redefines the class named by its class-name
3256 argument if that name is the proper name of the class;
3257 otherwise, it creates a new class.
3258 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3259 of the UNBOUND-SLOT condition to the name of the slot.
3260 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3261 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3262 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3263 SLOT-MAKUNBOUND in the specified fashion.
3265 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3266 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3267 cross-compilation host. As a consequence, we can now bootstrap our
3268 way up to SBCL starting with a bare gcc toolchain and human-readable
3269 source code (first the source to CLISP, then the source to SBCL).
3270 * A contributed module containing a partial implementation of the
3271 simple-streams interface has been included. (thanks to Rudi
3273 * A contributed module implementing the RFC1321 Message Digest
3274 Algorithm, known as MD5, has been included.
3275 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3276 no longer has any effect, as the code controlled by this feature
3277 has been deleted. (As far as we know, no-one has ever built using
3278 this feature, and its semantics were confused in any case).
3279 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3280 work on conditions (as required by the ANSI specification),
3281 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3282 expected behaviour on conditions. Users should note, however,
3283 that such behaviour is not required by the ANSI specification,
3284 and so use of this behaviour may render their code unportable.
3285 * fixed some bugs revealed by Paul Dietz' test suite:
3286 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3288 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3289 ** SXHASH on condition objects no longer returns NIL.
3290 ** :ALLOCATION :CLASS slots are better treated; their values are
3291 updated on class redefinition, and initforms inherited from
3292 superclasses are applied.
3293 ** REMOVE-METHOD returns its generic function argument even when
3294 no method was removed.
3295 ** SHARED-INITIALIZE now initializes the values of the requested
3296 slots, including those with :ALLOCATION :CLASS.
3297 ** ALLOCATE-INSTANCE now works on structure classes defined via
3298 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3300 ** SLOT-EXISTS-P now works on conditions, as well as structures
3302 ** MAKE-LOAD-FORM now has the required methods on
3303 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3304 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3305 keyword, and now implements the SLOT-NAMES argument.
3306 ** methods with &OPTIONAL arguments no longer allow too many
3307 arguments to be passed in the call without error.
3308 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3309 option is consistent with the required arguments of the generic
3310 function lambda list.
3311 * bug fix: REQUIRE accepts a string designator. (Thanks to
3313 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3314 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3315 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3316 * bug fix: The compiler now checks for duplicated variables in macro
3318 * bug fix: SETQ on globals returns the correct value.
3319 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3320 not a COMPILER-ERROR (followed by some other strange error on
3321 choosing the CONTINUE restart).
3322 * bug fix: make.sh and friends are now more consistent in the way that
3323 they look for GNU "make".
3325 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3326 * experimental native threads support (on x86 Linux >=2.4 only).
3327 This is not compiled in by default: you need to add :SB-THREAD to
3328 the target features. See the "Beyond ANSI" chapter of the manual
3330 * fix for longstanding nonANSIism: The old distinction between
3331 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3332 The return value from CL:FIND-CLASS is now a CLOS class, and
3333 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3334 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3335 * An interface to the MetaObject Protocol, as described in Kiczales,
3336 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3337 MIT Press, 1991, is available from the SB-MOP package.
3338 * incompatible change: the SB-PCL package should now be considered
3339 a private implementation detail, and no longer a semi-private MOP
3341 * minor incompatible change: due to rearrangement for threads, the
3342 control stack and binding stack are now allocated at arbitrary
3343 addresses instead of being hardcoded per-port. Users affected by
3344 this probably have to be doing advanced things with shared
3345 libraries, and will know who they are.
3346 * minor incompatible change: Previously, all --eval forms used were
3347 processed with READ before any of them were processed with EVAL.
3348 Now each --eval form is processed with both READ and EVAL before
3349 the next --eval form is processed. (Thus package operations like
3350 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3351 work as the user might reasonably expect.)
3352 * minor incompatible change: *STANDARD-INPUT* is now only an
3353 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3355 * minor incompatible change: Y-OR-N-P is now character-oriented,
3356 not line oriented. Also, YES-OR-NO-P now works without errors.
3357 (thanks to Antonio Martinez)
3358 * sb-aclrepl module improvements: an integrated inspector, added
3359 repl features, and a bug fix to :trace command.
3360 * Known functions, which cannot be open coded by the backend, are
3361 considered to be able to check types of their arguments. (fixing
3362 a bug report by Nathan J. Froyd)
3363 * fixed a bug in computing method discriminating functions: It is
3364 now possible to define methods specialized on classes which have
3365 forward-referenced superclasses. (thanks to Gerd Moellmann)
3366 * fixed evaluation order in optional entries (reported by Gilbert
3368 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3369 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3370 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3371 required-by-AMOP NAME argument, as well as CLASS and
3372 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3373 * fixed bug 20: DEFMETHOD can define methods using names that are
3374 not the proper names of classes to designate class specializers.
3375 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3376 Unix file descriptors, instead of blowing up. (thanks to Antonio
3378 * Garbage collection refactoring: user-visible change is that a
3379 call to the GC function during WITHOUT-GCING will not do garbage
3380 collection until the end of the WITHOUT-GCING. If you were doing
3381 this you were probably losing anyway.
3382 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3383 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3384 TYPEP the latter but not the former.
3385 * The compiler issues a full WARNING for calls to undefined functions
3386 with names from the CL package.
3387 * MAP-INTO for a vector destination is open coded. (reported by
3388 Brian Downing on c.l.l)
3389 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3390 documentation string.
3391 * fixed some bugs revealed by Paul Dietz' test suite:
3392 ** COPY-ALIST now signals an error if its argument is a dotted
3394 ** Condition slots are now accessed more correctly in the presence
3395 of multiple initargs for a given slot.
3396 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3397 exclude restarts of the same name associated with a different
3399 ** DEFCLASS of forward-referenced classes with another
3400 forward-referenced class in the superclasses list no longer
3402 ** Condition slots are now initialized once each, not multiple
3403 times. (thanks to Gerd Moellmann)
3404 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3406 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3407 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3408 arguments contain duplicated elements.
3409 ** RESTART-CASE understands local macros.
3410 ** RESTART-CASE associates exactly its own restarts with a condition.
3411 ** ENDP in safe mode checks its argument to be of type LIST.
3412 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3413 in question is unbound.
3414 ** Optimized MAKE-INSTANCE functions no longer cause internal
3415 assertion failures in the presence of duplicate initargs.
3416 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3417 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3418 propagated even in the ctor optimized implementation of
3420 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3422 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3423 * a better implementation of SXHASH on (simple) bit vectors,
3424 measured both in execution speed and in distribution of results
3425 over the positive fixnums, has been installed. Likewise, a better
3426 implementation of EQUAL for simple bit vectors is now available.
3427 * fixed CEILING optimization for a divisor of form 2^k.
3428 * fixed bug 240 (emitting extra style warnings "using the lexical
3429 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3430 by Antonio Martinez)
3431 * fixed SXHASH, giving different results for NIL depending on type
3432 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3433 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3434 variable. (found by Rolf Wester)
3435 * SBCL does not ignore type declarations for special
3436 variables. (reported by rif on c.l.l 2003-03-05)
3437 * some bug fixes in contrib/sb-aclrepl/
3438 * fixed some bugs revealed by Paul Dietz' test suite:
3439 ** a bug in the CONS type specifier, whereby the CAR and CDR
3440 types got intertwined, has been fixed;
3441 ** the type system is now able to reason about the interaction
3442 between INTEGER and RATIO types more completely;
3443 ** APPEND, [N]REVERSE and NRECONC check that those their
3444 arguments, which must be proper lists, are really so;
3445 ** An array specialized to be unable to hold elements has been
3446 implemented, as required -- yes, really -- by ANSI;
3447 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3448 on malformed property lists;
3450 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3451 * incompatible packaging change: in line with Unix convention,
3452 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3453 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3454 * REQUIRE and PROVIDE are now optionally capable of doing something
3455 useful. See the documentation string for REQUIRE.
3456 * infrastructure for a managed SBCL contrib system: contributed
3457 modules in this release include:
3458 ** the ASDF system definition facility;
3459 ** an interface to the BSD Sockets API;
3460 ** an ACL-like convenience interface to the repl;
3461 (thanks to Kevin Rosenberg)
3462 ** an implementation of ROTATE-BYTE, with efficient implementation
3464 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3465 gives a closer approximation to the right answer than previously.
3466 (thanks to Raymond Toy)
3467 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3468 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3469 optional environment arguments, as required by ANSI.
3470 * fixed bugs in other functions taking environment objects, allowing
3471 calls with an explicit NIL environment argument to be compiled
3473 * fixed bug 228: primary return values from
3474 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3475 COMPILE or FUNCTION.
3476 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3477 :TYPE will no longer signal errors on innocuous objects.
3478 * fixed bug 231b: SETQ is better at respecting type declarations in
3479 the lexical environment.
3480 * fixed a bug in DEFCLASS: classes named by symbols with no or
3481 unprintable packages can now be defined.
3482 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3483 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3484 * fixed compiler failure related to checking types of functions.
3485 (reported by Robert E. Brown)
3486 * the compiler is now much more consistent in its error-checking
3487 treatment of bounding index arguments to sequence functions: in
3488 (SAFETY 3) code, errors will be signalled in almost all cases if
3489 invalid sequence bounding indices are passed to functions defined
3490 by ANSI to operate on sequences.
3491 * fixed a bug in the build procedure: documentation of SBCL-specific
3492 packages is now preserved and available in the final Lisp image.
3493 * lifted FDEFINITION lookup out of loops in the implementation of
3494 many list operations. (thanks to Robert E. Brown)
3495 * fixed a bug in the reader: the #n# reader macro now works for
3496 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3497 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3498 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3499 E. Brown and Tony Martinez respectively)
3500 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3501 SLOT-BOUNDP: in particular, it is now able to optimize them much
3502 better, and is now not vulnerable to having packages renamed.
3503 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3504 always called when a slot is not present in an instance. (thanks
3506 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3507 longer signal an error on unreading a character following EOF on
3508 the previous constituent stream. (thanks to Tony Martinez)
3509 * fixed some bugs revealed by Paul Dietz' test suite:
3510 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3511 not just nonnegative fixnums;
3512 ** the logical bit-array operators such as BIT-AND now accept an
3513 explicit NIL for their "opt-arg" argument (to indicate a
3514 freshly-consed result bit-array);
3515 ** ELT now signals an error on an invalid sequence index in safe
3517 ** the type system is now cleverer about negations of numeric
3518 types, and consequently understands the BIGNUM and RATIO types
3520 ** the type system is now cleverer about the interaction between
3521 INTEGER and RATIO types: while bugs still remain, many more
3522 cases are accurately computed;
3523 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3524 if it is in the last clause;
3525 ** CONSTANTLY now correctly returns a side-effect-free function in
3527 ** DECLARE is no longer treated as a special-operator; in
3528 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3529 * incremented fasl file version number due to the change in the
3530 DEFSTRUCT-SLOT-DESCRIPTION structure.
3532 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3533 * minor incompatible change: code processed by the "interpreter" or
3534 EVAL now has a compilation optimization policy of (DEBUG 2)
3535 (changed from (DEBUG 1)) to improve debuggability of interactive
3536 development, and to allow the use of the debug RETURN command in
3538 * an experimental implementation of the RETURN command for the
3539 debugger has been included. (thanks to Frederik Kuivinen)
3540 * fixed bug 62: constraints were not propagated into a loop.
3541 * fixed bug in embedded calls of SORT (reported and investigated by
3543 * fixed some bugs revealed by Paul Dietz' test suite:
3544 ** printing and reading of arrays with some dimensions having
3545 length 0 (thanks to Gerd Moellmann);
3546 ** BOA constructor with &AUX argument without a default value does
3547 not cause a type error;
3548 ** CONSTANTP now returns true for all self-evaluating objects.
3550 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3551 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3552 accessors that are related by inheritance, as specified in the
3553 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3554 Valtteri Vuorikoski)
3555 * The compiler is now able to inline functions that were defined in
3556 a complex lexical environment (e.g. inside a MACROLET).
3557 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3559 * Support for the upcoming FreeBSD-5.0 release has been included.
3560 (thanks to Dag-Erling Smorgrav)
3561 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3562 effect when it is not in a toplevel context.
3563 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3564 stand a better chance of being correct. (thanks to Gerd
3566 * fixed bug in COERCE, which now signals an error on coercing a
3567 rational to a bounded real type which excludes the expected
3569 * The compiler is now able to derive types more accurately from the
3570 COERCE and COMPILE functions.
3571 * fixed bug 223: functional binding is considered to be constant
3572 only for symbols in the CL package.
3573 * fixed bug 231: SETQ did not check the type of a variable being set
3574 (reported by Robert E. Brown)
3575 * A new optimization for MAKE-INSTANCE has been included, fixing
3576 various bugs (including relating to :ALLOCATION :CLASS slots and
3577 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3579 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3580 ** As required by ANSI, LOOP now disallows anonymous collection
3581 clauses such as COLLECT I in conjunction with aggregate boolean
3582 clauses such as THEREIS (= I 1);
3583 ** LOOP now signals an error when any variable is reused in the
3584 same loop (including the potentially useful construct analogous
3585 to WITH A = 1 WITH A = (1+ A);
3586 ** IT is only a special loop symbol within the first clause of a
3587 conditional loop clause;
3588 ** LOOP with a typed iteration variable over a hashtable now
3589 signals a type error iff it should.
3590 * fixed some other bugs revealed by Paul Dietz' test suite:
3591 ** FILE-STREAM now names the class previously known as FD-STREAM;
3592 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3593 argument) no longer signals an error;
3594 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3595 of the slot symbol, rather than using the current package
3596 ((:CONC-NAME "") continues to intern the slot's name in the
3598 * incremented fasl file version number, because of the incompatible
3599 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3600 of the new implementation of DEFINE-COMPILER-MACRO.
3602 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3603 * Support for building SBCL for MIPS platforms running in
3604 little-endian mode has now been checked in, and basic
3605 functionality on said platforms verified.
3606 * minor incompatible change: PCL now records the pathname of a file
3607 in which methods and the like are defined, rather than its
3609 * minor incompatible change: TRUENAME now considers the truename of
3610 a file naming a directory to be the pathname with :DIRECTORY
3611 component indicating that directory.
3612 * minor incompatible change: a NAMED clause in the extended form of
3613 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3614 reason for the previous behaviour is unclear.
3615 * more systematization and improvement of CLOS and MOP conformance
3616 in PCL (thanks to Gerd Moellman and Pierre Mai):
3617 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3619 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3620 primary methods with no specializers;
3621 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3623 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3624 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3625 FORWARD-REFERENCED-CLASSes; error reporting on
3626 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3628 ** SXHASH on CLOS instances now uses a slot internal to the
3629 instance to return different numbers on distinct instances,
3630 while preserving the same return value through invocations of
3632 ** DEFMETHOD signals errors when methods with longer incongruent
3633 lambda lists are added to generic functions;
3634 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3635 CLASS, as specified in AMOP;
3636 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3637 on the order returned by the primary method for classes of
3638 class STANDARD-CLASS;
3639 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3640 * fixed some bugs shown by Paul Dietz' test suite:
3641 ** DOLIST puts its body in TAGBODY;
3642 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3644 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3645 value producing form;
3646 ** if more variables are given to PROGV than values, extra
3647 variables are bound and made to have no value;
3648 ** NSUBSTITUTE on list arguments gets the right answer with
3650 ** ELT signals an error of type TYPE-ERROR when the index argument
3651 is not a valid sequence index;
3652 ** LOOP signals (at macroexpansion time) an error of type
3653 PROGRAM-ERROR when duplicate variable names are found;
3654 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3655 ** FUNCALL of special-operators now cause an error of type
3657 ** PSETQ now works as required in the presence of side-effecting
3658 symbol-macro places;
3659 ** NCONC accepts any object as its last argument;
3660 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3662 ** loop-for-as-package does not require a package to be explicitely
3664 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3666 * fixed bug 166: compiler preserves "there is a way to go"
3667 invariant when deleting code.
3668 * fixed bug 172: macro lambda lists with required arguments after
3669 &REST arguments now cause an error to be signalled. (thanks to
3671 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3672 bug. (thanks to Matthew Danish)
3673 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3675 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3676 arguments are lexically rebound. (thanks to Gerd Moellmann and
3678 * fixed bug 194: error messages are now more informative when there
3679 is no primary method applicable in a call to a generic
3680 function. (thanks to Gerd Moellmann)
3681 * fixed bug in command line argument checking (thanks to Julian
3683 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3684 (thanks to Matthew Danish)
3685 * incremented fasl file version number, because of the
3686 SXHASH-related changes in the layout of CLOS data structures
3688 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3689 * minor incompatible change: The runtime (the Unix executable named
3690 "sbcl") is now much pickier about the .core files it will load.
3691 Essentially it now requires .core files to descend from the same
3692 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3693 as the runtime does. (The intent is to prevent the crashes which
3694 can occur, and which can even be reported as mysterious failures,
3695 when people patch the sources or change the build parameters
3696 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3697 sbcl and .core files.)
3698 * fixed bug: VALUES-LIST is no longer optimized away.
3699 * fixed bug 142: The FFI conversion of C string values to Lisp
3700 string values no longer conses excessively. (thanks to Nathan
3701 Froyd porting Raymond Toy's fix to CMU CL)
3702 * began to systematize and improve MOP conformance in PCL (thanks to
3703 Nathan Froyd, Gerd Moellman and Pierre Mai):
3704 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3706 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3707 ** FINALIZE-INHERITANCE is now called on class finalization;
3708 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3709 argument precedence order.
3710 * fixed bug 202: The compiler no longer fails on functions whose
3711 derived types contradict their declared type.
3712 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3713 so it can be non-toplevel.
3714 * The fasl file version number has changed (because of the new
3715 implementation of DEFMACRO).
3716 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3717 safe code, that any length requirement by their type-specifier
3718 argument is valid. The exceptions to this are described in bug
3720 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3721 functions now check the types of their inputs as required by ANSI.
3722 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3723 introduced symbol is DECLAREd to be SPECIAL.
3724 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3725 * fixed bug: :COUNT argument to sequence functions may be negative
3726 * fixed bug: body of DO-SYMBOLS may contain declarations
3727 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3728 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3729 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3730 symbol macro only once
3731 * fixed printing of call frame when argument list is unavailable
3732 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3733 * compiler no longer signals WARNING on unknown keyword
3736 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3737 * A beta-quality port to the mips architecture running Linux,
3738 based on the old CMUCL backend, has been made. It has been tested
3739 on a big-endian kernel, and works sufficiently well to be able to
3740 rebuild itself; it has not been tested in little-endian mode.
3741 * fixed an inconsistency between gencgc.c and purify.c which made
3742 dumping/loading .core files unreliable
3743 * fixed bug 120a: The compiler now deals correctly with IFs where
3744 the consequent is the same as the alternative, instead of
3745 misderiving the return type. (thanks to Alexey Dejneka)
3746 * fixed bug 113: Logical pathnames are now dumpable (the logical
3747 host is resolved at load-time, throwing an error if it is not
3749 * fixed bug 174: FORMAT's error message is slightly clearer when a
3750 non-printing character is used in a format directive.
3751 * fixed several bugs in compiler checking of type declarations, i.e.
3752 violations of the Python "declarations are assertions" principle
3753 (thanks to Alexey Dejneka)
3754 * fixed several bugs in PCL's error checking (thanks to Gerd
3756 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3758 * fixed bug in compilation of functions as first class values
3759 (thanks to Antonio Martinez-Shotton)
3760 * The compiler's handling TYPE-ERRORs which it can prove will
3761 inevitably happen at runtime has been cleaned up and corrected
3762 in several ways. (thanks to Alexey Dejneka)
3763 * improved argument type checking for various basic arithmetic
3764 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3765 lavished on them in the past that they can be compiled in many
3766 ways in different special cases
3767 * fixed bug 181: compiler checks validity of user supplied type
3769 * cleaned up code flushing in optimization: Function calls which
3770 should signal errors for safety purposes (e.g. which ANSI says
3771 should signal errors when their arguments are of incorrect type)
3772 are no longer optimized away.
3773 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3774 * incremented fasl file version number, because changes in the
3775 implementation of sequence functions like COERCE caused
3776 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3777 in old inline expansions) to become undefined. (Actually these
3778 changes were later undone, so we might very well be binary
3779 compatible with 0.7.7 after all, but leaving the version number
3780 incremented seemed like the simplest and most conservative
3783 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3784 * An alpha-quality port to the parisc architecture running Linux,
3785 based on the old CMUCL backend, has been made. This, even more so
3786 than the other backends, should be considered still a work in
3787 progress; known problems include that the Linux kernel in 64-bit
3788 mode does not propagate the correct sigcontext structure to
3789 userspace, and consequently SBCL on a parisc64 kernel will not
3791 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3792 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3793 suppressed.) Also now that the compiler is looking at declarations
3794 in the environment, it checks optimization declarations as well,
3795 and suppresses inlining when (> DEBUG SPEED).
3796 * More fixes have been made to treatment of floating point exception
3797 treatment and other Unix signals. In particular, floating point
3798 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3799 * The detection and handling of control stack exhaustion (infinite
3800 or very deeply nested recursion) has changed. Stack exhaustion
3801 detection is now done by write-protecting pages at the OS level
3802 and applies at all optimization settings; when found, a
3803 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3804 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3805 be caught using IGNORE-ERRORS.
3806 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3807 correctly when transforming recursive LABELS functions to LETs.
3808 (thanks to Alexey Dejneka)
3809 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3810 that are names of constants or global variables.
3811 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3812 alien routines with docstrings.
3813 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3814 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3816 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3817 (thanks to Eric Marsden)
3818 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3819 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3820 * Bug fix: The SPARC backend can now compile functions involving
3821 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3822 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3823 to a functions expecting a PATHNAME-DESIGNATOR.
3824 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3825 lambda lists. (thanks to Alexey Dejneka)
3826 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3827 functions so that EXPORTing the name of the class doesn't cause
3828 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3829 to Antonio Martinez for reporting this)
3830 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3831 constant-folded without failing an assertion. (thanks to Einar
3832 Floystad Dorum for reporting this)
3833 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3834 types are now dealt with more correctly by the compiler.
3835 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3836 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3837 behaviour analogous to RENAME-FILE. This puts its behaviour more
3838 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3840 * The fasl file version number has changed again. (because of the
3841 bug fix involving the names of PCL MAKE-INSTANCE functions)
3843 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3844 * bug fix: Floating point exceptions are treated much more
3845 consistently on the x86/Linux and PPC/Linux platforms.
3846 * Array initialization with :INITIAL-ELEMENT is now much faster for
3847 cases when the compiler cannot open code the array creation, but
3848 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3849 array accesses have also seen a speed increase.
3850 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3851 specified. (This was at the root of some bad interactions between
3852 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3854 * bug fix: Internal error arguments for undefined functions are now
3855 computed correctly on the PPC/Linux platform.
3856 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3857 Raymond Toy's patch for CMU CL)
3858 * Support for the Solaris 9 operating environment has been included
3859 (thanks to Daniel Merritt)
3860 * A very ugly but hopefully complete draft of the missing FFI chapter
3861 of the manual has been created by reformatting the corresponding
3862 CMU CL manual chapter into (currently very ugly and incoherent)
3863 DocBook and bringing it up to date for SBCL behavior. Thus, the
3864 manual is now essentially complete, at least by my extreme
3865 once-and-only-once standards, whereby it's acceptable to refer to
3866 the doc strings of SB-EXT functions as the primary documentation.
3867 * The fasl file version number has changed again, due to cleanup of
3868 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3869 is no longer a static symbol.)
3871 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3872 * SBCL now builds with OpenMCL (version 0.12) as the
3873 cross-compilation host; also, more progress has been made toward
3874 bootstrapping under CLISP.
3875 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3877 * bug 158 fixed: The compiler can now deal with integer loop
3878 increments different from 1; fixing this turned out also to fix
3880 * bug 169 fixed: no more bogus warnings about using lexical bindings
3881 despite the presence of perfectly good SPECIAL declarations (thanks
3882 to David Lichteblau)
3883 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3884 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3885 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3887 * bug 180 fixed: Method combination specifications no longer ignore
3888 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3889 * bug fix: Structure type predicate functions now check their argument
3890 count as they should.
3891 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3892 correctly. (thanks to Pierre Mai)
3893 * minor incompatible change: The --noprogrammer option is deprecated
3894 in favor of the new --disable-debugger option, which is very similar.
3895 (The major difference is that it takes effect at a slightly different
3896 time at startup, causing handling of errors in --sysinit and
3897 --userinit files will be affected differently.) The
3898 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3899 been added to allow this functionality to be controlled from ordinary
3900 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3901 maintainers, who might want to run non-interactive scripts to
3902 build SBCL cores which will later be used interactively.)
3903 * minor incompatible change: The LOAD function no longer, when given
3904 a wild pathname to load, loads all files matching that pathname.
3905 Instead, an error of type FILE-ERROR is signalled.
3907 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3908 * bug 147 fixed: The compiler preserves its block link/count
3909 invariants more correctly now so that it doesn't crash. (thanks
3911 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3913 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3914 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3915 no-value-for-this-character as NIL (as specified by ANSI).
3916 * HOST-NAMESTRING on physical pathnames now returns a string that is
3917 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3918 (thanks to Christophe Rhodes)
3919 * The Alpha port handles icache flushing more correctly. (thanks to
3921 * More progress has been made toward bootstrapping under CLISP. (thanks
3922 to Christophe Rhodes)
3923 * The fasl file format has changed again, because dynamic loading
3924 on OpenBSD (which has non-ELF object files) motivated some cleanups
3925 in the way that foreign symbols are transformed and passed around.
3926 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3927 is no longer treated as whitespace by the reader, but instead as
3928 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3929 |A
\7fB|, instead of A as it used to.
3931 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3932 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3933 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3934 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3935 this as of 0.7.1.45, but was left out of the previous news section
3936 (thanks to Dan Barlow)
3937 * SBCL now runs on the Solaris operating system on SPARC architectures
3938 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3939 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3940 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3941 * SPARC backend cleanups, allowing builds of cores optimized for V8
3942 and V9 SPARCS, and also emission of code targeted to a particular
3943 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3945 * SBCL is closer to bootstrapping under CLISP, thanks to various
3946 fixes by Christophe Rhodes.
3947 * The fasl file format has changed again, to allow the compiler's
3948 INFO database to support symbol macros.
3949 * The user manual (in doc/) is formatted into HTML more nicely.
3950 (thanks to coreythomas)
3951 * The system is smarter about SUBTYPEP relationships, especially
3952 those involving NOT types (including types such as ATOM which are
3953 represented internally using NOT types). Thus SUBTYPEP is less
3954 likely to return (VALUES NIL NIL) in general, and in particular
3955 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3957 * The fasl file format has changed again, because the internal
3958 representation of types now includes a new slot to support the new
3959 SUBTYPEP-of-NOT-types logic.
3960 * (not a change in the main branch of SBCL, but a related prototype
3961 which can hopefully be merged into the main branch of SBCL in the
3962 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3963 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3964 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3965 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3966 dispatch macro character combinations. (thanks to Alexey Dejneka)
3967 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3968 that superficially look like logical namestrings correctly.
3969 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3971 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3972 default supplied-p) for &optional and &key arguments. (thanks to
3975 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3976 * incompatible change: The compiler is now less aggressive about
3977 tail call optimization, doing it only when (> SPACE DEBUG) or
3978 (> SPEED DEBUG). (This is an incompatible change because there are
3979 programs which relied on the old CMU-CL-style behavior to optimize
3980 away their unbounded recursion which will now die of stack overflow.)
3981 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3982 for non-GENCGC systems has been increased to 20M (since that
3983 seems much closer to the likely performance optimum for modern
3984 systems than the old 4M value was)
3985 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3986 and *DEBUG-PRINT-LEVEL*
3987 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3988 of CMU CL's support for SPARC, and various endianness and other
3989 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3990 * new syntactic sugar for the Unix command line: --load foo.bar is now
3991 an alternate notation for --eval '(load "foo.bar")'.
3993 ** The system now detects stack overflow and handles it gracefully,
3994 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3995 optimization settings. (This is a good thing in general, and
3996 its introduction in this version should be particularly timely
3997 for anyone whose code fails because of suppression of tail
3999 ** The system now hunts for the C variable "environ" in a more
4000 devious way, to avoid segfaults when the C library version differs
4001 between compile time and run time. (thanks to Christophe Rhodes)
4002 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
4003 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
4004 ** The compiler no longer issues bogus style warnings for undefined
4005 classes in the same source file as the DEFCLASSes which defined
4006 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
4008 ** fixes in CONDITION class precedence list for undefined function
4009 errors (thanks to Alexei Dejneka)
4010 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
4011 correctly. (thanks to Dan Barlow)
4012 ** portability fixes aiming at bootstrapping under CLISP (thanks
4013 to Dave McDonald and Christophe Rhodes)
4014 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4015 ** fixes in type translation and and type inference (thanks to
4017 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4018 ** various fixes in the new ports (thanks to Dan Barlow)
4019 * several changes related to debugging:
4020 ** suppression of tail recursion, as noted above
4021 ** stack overflow detection, as noted above
4022 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4023 is now the default. (For some time encapsulation has been more
4024 reliable than the breakpoint-based :ENCAPSULATE NIL
4025 implementation, at least on X86 systems; and I just noticed that
4026 encapsulation also seems closer to the spirit of the ANSI
4029 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4031 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4032 up properly again. (There was a packaging bug in 0.7.0 which
4033 left their definitions in SB-SYS::LOAD-FOREIGN and
4034 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4035 vital for most things which interface to C-level interfaces,
4036 like extensions working with sockets or databases or
4037 Perl-compatible regexes or whatever, and the need to fix
4038 this bug is the main reason that 0.7.1 was released so
4040 ** DEFGENERIC is now choosier about the methods it redefines, so that
4041 reLOADing a previously-LOADed file containing DEFGENERICs does
4042 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4043 cycle now works as it should. (thanks to Alexey Dejneka)
4044 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4045 fixed by Christophe Rhodes. (He actually submitted this patch
4046 months ago, and I delayed until after 0.7.0.)
4047 ** Bug 111 (internal compiler confusion about runtime checks on
4048 FUNCTION types) was fixed by Alexey Dejneka.
4049 * Some internal cleanups (getting rid of variables which aren't
4050 needed now that the byte interpreter is gone) caused the fasl
4051 file format number to change again.
4053 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4054 * major incompatible change: The default fasl file extension, i.e. the
4055 default extension for files produced by COMPILE-FILE, has changed
4056 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4058 ** There are many changes in the implementation of the compiler.
4059 SBCL is now essentially a compiler-only implementation of ANSI
4060 Common Lisp. EVAL still "interprets" a few special cases, but
4061 almost all the interesting cases are handled by creating
4062 a LAMBDA expression, calling COMPILE on it, then calling
4063 FUNCALL on the result.
4064 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4065 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4066 Since the code is newer, there might still be some new bugs
4067 (though not as many as before Martin Atzmueller's fixes:-). But
4068 the new code is substantially simpler and clearer, and hopefully
4069 any remaining bugs will be simpler, less fundamental, and more
4070 fixable then the bugs in the old code.
4071 ** The revised compiler is still a little unsteady on its feet.
4073 *** The debugging information it produces (particularly the names
4074 of FUNCTION objects) is sometimes much less useful than what
4075 the old compiler produced.
4076 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4077 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4078 within a MACROLET) has been temporarily weakened.
4079 ** There are new compiler optimizations for various functions:
4080 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4081 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4082 *** the math functions TRUNCATE, FLOOR, and CEILING
4083 *** the function-of-all-trades COERCE
4084 Mostly these should be transparent, but there's one
4085 potentially-annoying problem (bug 117): when the compiler
4086 inline-expands a function and does type analysis on the result,
4087 it can create control paths which have type mismatches, and
4088 when it can't prove that those control paths aren't taken,
4089 it will issue WARNINGs about the type mismatches. This is
4090 a particular problem in practice for the new sequence functions.
4091 It's not clear how this should be fixed, and for now, a
4092 workaround is given in the entry for 117 in the BUGS file.
4093 ** (Because of the interaction between the two previous items --
4094 occasional inlining problems and new inline expansions -- some
4095 of the new sequence function optimizations won't really kick in
4096 completely until debugging information, and then inlining, are
4097 straightened out in some future version.)
4098 * minor incompatible changes:
4099 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4100 DIRECTORY no longer implicitly promotes NIL slots of its
4101 pathname argument to :WILD. In particular, when you ask for the
4102 contents of a directory (which you used to be able to do without
4103 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4104 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4105 ** changes in behavior that ANSI explicitly defines to be
4106 implementation dependent:
4107 *** The new compiler-only implementation still conforms with ANSI,
4108 but acts a little different than before. Besides the obvious
4109 changes in performance tradeoffs (that the cost per form passed
4110 to EVAL has gone up, and the cost per form executed by EVAL
4111 has gone down), the behavior of the system changes a little
4112 because there are no longer any interpreted function objects.
4113 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4114 e.g. doing COMPILE on the output of interactive DEFUN is
4116 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4118 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4119 has changed from (:CL) to NIL.
4120 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4121 e.g. #\Newline and #\Tab, don't have names specified in the
4122 ANSI Common Lisp standard, is now based on their ASCII symbolic
4123 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4124 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4125 are no longer used for output.
4126 ** changes in internal implementation constants:
4127 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4128 4 million. (If your application spends a lot of time GCing and
4129 you have a lot of RAM, you might want to experiment with
4130 increasing it even more.)
4131 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4132 However, almost all old code should still continue to work without
4133 immediate update, as SB-C-CALL is now a (deprecated) nickname
4135 ** Old operator names in the style DEF-FOO are now deprecated in
4136 favor of new corresponding names DEFINE-FOO, for consistency with
4137 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4138 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4139 internal symbols, but a few supported extensions like
4140 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4141 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4142 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4143 etc. as you get deeper into recursive calls to the debugger
4144 command loop, instead of the old "5]", "5]]", "5]]]"
4145 sequence. (I was motivated to do this when squabbles between
4146 ILISP and SBCL left me very deeply nested in the debugger. In the
4147 short term, this change will probably provoke more ILISP/SBCL
4148 squabbles, but hopefully it will be an improvement in the long run.)
4149 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4150 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4151 function, since declaiming return types involving aliens is
4152 (1) annoyingly messy to do by hand and (2) vital to efficient
4153 compilation of code which calls such functions.
4154 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4155 longer reexported by the SB-EXT package. They're solely useful
4156 for alien code, so it seems more logical that you should get
4157 them from the SB-ALIEN package, not in SB-EXT.
4158 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4159 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4160 features. Instead, the code that they used to control is always
4161 built into the system.
4162 * many other bug fixes
4163 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4164 advantage of the new EVAL-WHEN stuff and to clean them up in
4165 general, and they are now more ANSI-compliant in a number of
4166 ways. Martin Atzmueller is responsible for a lot of this.
4167 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4169 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4170 *** correct ERROR type for various file operations
4171 *** some fixes for Lisp streams
4172 *** DEFMETHOD syntax checking
4173 *** changing old weird representation of debug information as
4174 strings (which, among their other deficiencies, don't transform
4175 correctly when you rename packages, and don't change their
4176 print representation when you change things like *PACKAGE*
4177 and *PRINT-LENGTH*) to symbols and lists of symbols
4178 He also made several improvements and fixed several bugs in DESCRIBE.
4179 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4181 *** misbehavior of WRITE-STRING/WRITE-LINE
4182 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4183 and several other LOOP problems as well
4184 *** DIRECTORY when similar filenames are present
4185 *** DEFGENERIC with :METHOD options
4186 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4187 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4188 *** argument ordering in FIND with :TEST option
4189 *** mishandled package designator argument in APROPOS-LIST
4190 *** various problems in the backquote readmacro
4191 *** a bug in APROPOS
4192 *** probably some others that I'm not describing very well here,
4193 since the CVS log documents them by reference to sbcl-devel
4194 messages, and the SourceForge archives aren't working well.:-(
4195 ** Dan Barlow improved the Alpha port (and is making progress on the
4196 PPC port, for those of you who think different).
4197 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4198 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4199 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4200 continues to work on the SPARC port (for those of you in a position
4201 to look down upon our little PC-compatible boxes from a great height).
4202 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4203 and uses that copy, rather than the current dynamic value, when
4204 it's trying to decide whether to truncate output. Thus e.g.
4205 (let ((*print-lines* 50))
4206 (pprint-logical-block (stream nil)
4208 (let ((*print-lines* 8))
4209 (print (aref possiblybigthings i) stream)))))
4210 should now truncate the logical block only at 50 lines, instead of
4211 often truncating it at 8 lines, as it did before.
4212 * The doc/cmucl/ directory, containing old CMU CL documentation
4213 from the time of the fork, is no longer part of the base system.
4214 SourceForge has shut down its anonymous FTP service, and with it
4215 my original plan for distributing the old CMU CL documentation
4216 there. For now, if you need these files you can download an old
4217 SBCL source release and extract them from it.
4218 * The fasl file version number changed again, for dozens of reasons,
4219 some of which are apparent above.
4221 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4222 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4223 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4224 patch, so that DISASSEMBLE output is much nicer.
4225 * The code in the SB-PROFILE package now seems reasonably stable.
4226 I still haven't decided what the final interface should look like
4227 (I'd like PROFILE to interact cleanly with TRACE, since both
4228 facilities use function encapsulation) but if you have a need
4229 for profiling now, you can probably use it successfully with
4230 the current CMU-CL-style interface.
4231 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4232 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4233 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4234 link now returns the dangling link itself, and for similar
4235 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4236 link itself. (In these cases the old code signalled an error and
4237 looped endlessly, respectively.) Thus, DIRECTORY now works even
4238 in the presence of dangling and cyclic symbolic links.
4239 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4240 is now a supported extension again, since the consensus on
4241 sbcl-devel was that it can be useful for ordinary development
4242 work, not just for debugging SBCL itself.
4243 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4244 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4245 that currently-defined functions might be redefined later with
4246 different return types.
4247 * Hash tables can be printed readably, as inspired by CMU CL code
4248 of Eric Marsden and SBCL code of Martin Atzmueller.
4249 * better error handling in CLOS method combination, thanks to
4250 Martin Atzmueller porting Pierre Mai's CMU CL patches
4251 * more overflow fixes for >16Mbyte I/O buffers
4252 * A bug in READ has been fixed, so that now a single Ctrl-D
4253 character suffices to cause end-of-file on character streams.
4254 In particular, now you only need one Ctrl-D at the command
4255 line (not two) to exit SBCL.
4256 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4258 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4259 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4260 does the right thing.
4261 * The implementation of some type tests, especially for CONDITION
4262 types, is now tidier and maybe faster, due to CMU CL code
4263 originally by Douglas Crosher, ported by Martin Atzmueller.
4264 * Some math functions have been fixed, and there are new
4265 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4266 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4267 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4268 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4269 likely to be built into sbcl-0.7.0.)
4270 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4271 CL, ported to SBCL by Martin Atzmueller)
4272 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4273 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4274 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4275 * There's a new slam.sh hack to shorten the edit/compile/debug
4276 cycle for low-level changes to SBCL itself, and a new
4277 :SB-AFTER-XC-CORE target feature to control the generation of
4278 the after-xc.core file needed by slam.sh.
4279 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4280 COMPILE-FILE is no longer supported, so that now every function
4281 gets an entry point, so that block compilation looks a little
4282 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4283 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4284 returns the number of bytes consed since the system started,
4285 rather than the number consed since the first time the function
4286 was called. (The new definition parallels ANSI functions like
4287 CL:GET-INTERNAL-RUN-TIME.)
4288 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4289 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4290 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4291 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4292 * Fasl file version numbers are now independent of the target CPU,
4293 since historically most system changes which required version
4294 number changes have affected all CPUs equally. Similarly,
4295 the byte fasl file version is now equal to the ordinary
4298 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4299 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4300 is no longer recognized. I apologize for this, because it was
4301 listed in SB-EXT as a supported extension, but I found that
4302 its existing behavior was poorly specified, as well as incorrectly
4303 specified, and it looked like too much of a mess to straighten it
4304 out. I have enough on my hands trying to get ANSI stuff to work..
4305 * many patches ported from CMU CL by Martin Atzmueller, with
4306 half a dozen bug fixes in pretty-printing and the debugger, and
4307 half a dozen others elsewhere
4308 * fixed bug 13: Floating point infinities are now supported again.
4309 They might still be a little bit flaky, but thanks to bug reports
4310 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4311 as flaky as they were.
4312 * The --noprogrammer command line option is now supported. (Its
4313 behavior is slightly different in detail from what the old man
4314 page claimed it would do, but it's still appropriate under the
4315 same circumstances that the man page talks about.)
4316 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4317 are now supported, and enabled by default. Thus, the compiler can
4318 handle many floating point and complex operations much less
4319 inefficiently. (Thus e.g. you can implement a complex FFT
4321 * The compiler now detects type mismatches between DECLAIM FTYPE
4322 and DEFUN better, and implements CHECK-TYPE more correctly, and
4323 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4324 thanks to patches from Martin Atzmueller.
4325 * various fixes to make the cross-compiler more portable to
4326 ANSI-conforming-but-different cross-compilation hosts (notably
4327 Lispworks for Windows, following bug reports from Arthur Lemmens)
4328 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4329 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4330 fixed thanks to Pierre Mai's CMU CL patches.
4331 * fixes to keep the system from overflowing internal counters when
4332 it tries to use i/o buffers larger than 16M bytes
4333 * fixed bug 45a: Various internal functions required to support
4334 complex special functions have been merged from CMU CL sources.
4335 (When I was first setting up SBCL, I misunderstood a compile-time
4336 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4337 * improved support for type intersection and union, fixing bug 12
4338 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4339 more obscure bugs as well
4340 * some steps toward byte-compiling non-performance-critical
4341 parts of the system, courtesy of patches from Martin Atzmueller
4342 * Christophe Rhodes has made some debian packages of sbcl at
4343 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4344 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4345 stable, but are nonetheless usable. When he's ready, I'd be happy
4346 to add them to the SourceForge "File Releases" section. (And if
4347 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4348 * new fasl file format version number (because of changes in
4349 internal representation of (OR ..) types to accommodate the new
4350 support for (AND ..) types, among other things)
4352 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4353 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4355 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4357 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4358 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4359 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4360 * DESCRIBE now works on structure objects again.
4361 * Most function call argument type mismatches are now handled as
4362 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4363 know whether the function will be redefined before the call is
4364 executed. (The compiler could flag local calls with full WARNINGs,
4365 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4366 it doesn't keep track of enough information to know whether calls
4367 are local in this sense.)
4368 * Compiler output is now more verbose, with messages truncated
4369 later than before. (There should be some supported way for users
4370 to override the default verbosity, but I haven't decided how to
4371 provide it yet, so this behavior is still controlled by the internal
4372 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4373 src/compiler/ir1util.lisp.)
4374 * Fasl file format version numbers have increased again, because
4375 support for the Gray streams extension changes the layout of the
4376 system's STREAM objects.
4377 * The Gray subclassable streams extension now works, thanks to a
4378 patch from Martin Atzmueller.
4379 * The full LOAD-FOREIGN extension (not just the primitive
4380 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4381 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4382 but like most other programs, it defaults to copying the Unix
4383 environment from the original process instead of starting the
4384 new process in an empty environment.
4385 * Extensions which manipulate the Unix environment now support
4386 an :ENVIRONMENT keyword option which doesn't smash case or
4387 do other bad things. The CMU-CL-style :ENV option is retained
4388 for porting convenience.
4389 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4390 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4392 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4394 * A patch from Martin Atzmueller seems to have solved the SIGINT
4395 problem, and as far as we know, signal-handling now works cleanly.
4396 (If you find any new bugs, please report them!)
4397 * The system no longer defaults Lisp source file names to types
4398 ".l", ".cl", or ".lsp", but only to ".lisp".
4399 * The compiler no longer uses special default file extensions for
4400 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4401 seems to expect a single default extension for all compiled code,
4402 and there's no compelling reason to try to stretch the standard
4403 to allow two different extensions.) Instead, byte-compiled files
4404 default to the same extension as native-compiled files.
4405 * Fasl file format version numbers have increased again, because
4406 a rearrangement of internal implementation packages made some
4407 dumped symbols in old fasl files unreadable in new cores.
4408 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4409 many fewer weird special cases.
4410 * Bug #17 (differing COMPILE-FILE behavior between logical and
4411 physical pathnames) has been fixed, and some related misbehavior too,
4412 thanks to a patch from Martin Atzmueller.
4413 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4414 by Tim Moore, ported to SBCL by Martin Atzmueller.
4415 * Martin Atzmueller fixed several filesystem-related problems,
4416 including bug #36, in part by porting CMU CL patches, which were
4417 written in part by Paul Werkowski.
4418 * More compiler warnings in src/runtime/ are gone, thanks to
4419 more patches from Martin Atzmueller.
4420 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4423 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4425 * DESCRIBE now works on CONDITION objects.
4426 * The debugger now handles errors which arise when trying to print
4427 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4429 * The build system now uses an additional file, customize-target-features.lisp,
4430 to allow local modifications to the target *FEATURES* list. (The point of
4431 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4432 debugging features enabled, without having to worry about propagating it
4433 into everyone's system when I do a "cvs update".) When no
4434 customize-target-features.lisp file exists, the target *FEATURES* list
4435 should be constructed the same way as before.
4436 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4437 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4438 warn about behavior which is undefined under the ANSI spec. Note: This
4439 is specified by ANSI, but it's not very popular with programmers.
4440 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4441 macro in the SBCL sources for an example of a workaround which you
4442 might use to make portable ANSI-standard code which does what you want.
4443 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4444 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4445 when it's used as a non-toplevel form. (This is required in order
4446 to implement the DEFCONSTANT-EQX macro.)
4447 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4448 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4449 funny things, due to bugs in the implementation of EVAL-WHEN
4450 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4451 non-toplevel DEFCONSTANTs are uncommon.)
4452 * The core file version number and fasl file version number have been
4453 incremented, because the old noncompliant DEFCONSTANT behavior involved
4454 calling functions which no longer exist, and because I also took the
4455 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4456 * fixed bug 1 (error handling before read-eval-print loop starts), and
4457 redid debugger restarts and related debugger commands somewhat while
4459 ** The QUIT debugger command is gone, since it did something
4460 rather different than the SB-EXT:QUIT command, and since it never
4461 worked properly outside the main toplevel read/eval/print loop.
4462 Invoking the new TOPLEVEL restart provides the same functionality.
4463 ** The GO debugger command is also gone, since you can just invoke
4464 the CONTINUE restart directly instead.
4465 ** The TOP debugger command is also gone, since it's redundant with the
4466 FRAME 0 command, and since it interfered with abbreviations for the
4468 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4470 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4471 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4472 some porting work by Martin Atzmueller)
4473 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4474 with a boatload of patches which he ported to SBCL. Now that those
4476 ** The system tries to make sure that its low-priority messages
4477 are prefixed by semicolons, to help people who like to use
4478 syntax highlighting in their ILISP buffer. (This patch
4479 was originally due to Raymond Toy.)
4480 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4481 patches originally written by Raymond Toy.
4482 ** The compiler understands coercion between single-value and
4483 multiple-VALUES type expressions better, getting rid of some very
4484 weird behavior, thanks to patches originally by Robert MacLachlan
4485 and Douglas Crosher.
4486 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4487 lambda lists, thanks to a patch originally by Pierre Mai.
4488 ** The system no longer bogusly warns about "abbreviated type
4490 ** The compiler gets less confused by inlining and RETURN-FROM,
4491 thanks to some patches originally by Tim Moore.
4492 ** The system no longer hangs when dumping circular lists to fasl
4493 files, thanks to a patch originally from Douglas Crosher.
4494 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4495 undefined function error.
4496 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4497 in structure slot initforms) for now, documented workaround instead:-|
4498 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4499 * fixed bug 5: added stubs for various Gray stream functions called
4500 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4501 installed, at least appropriate type errors are generated
4502 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4503 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4504 consistently in DEFMETHOD forms.
4505 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4506 it doesn't seem to affect SBCL after all
4507 * The C runtime system now builds with better optimization and many
4508 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4510 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4512 * The system is now under CVS at SourceForge (instead of the
4513 CVS repository on my home machine).
4514 * The new signal handling code has been tweaked to treat register
4515 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4516 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4517 downstream, e.g. in debug-int.lisp, has implicit dependencies
4518 on the unsignedness of integer representation of machine words,
4519 and that caused the system to bomb out with infinite regress
4520 when trying to recover from type errors involving signed values,
4521 e.g. (BUTLAST '(1 2 3) -1).)
4522 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4523 (This was one of the bugs Peter Van Eynde reported back in July.)
4524 * The system now uses code inspired by Colin Walters' O(N)
4525 implementation of MAP (from the cmucl-imp@cons.org mailing
4526 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4527 inline the MAP operation, and there is more than one
4528 sequence argument to the MAP call (so that it can't just
4529 do ETYPECASE once and for all based on the type of the
4530 single sequence argument). (The old non-inline implementation
4531 of the general M-argument sequence-of-length-N case required
4532 O(M*N*N) time when any of the sequence arguments were LISTs.)
4533 * The QUIT :UNIX-CODE keyword argument has been renamed to
4534 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4535 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4537 * Raymond Wiker's patches to port dynamic loading from Linux to
4538 FreeBSD have been added.
4539 * The BUGS file is now more nearly up to date, thanks in large part
4540 to Martin Atzmueller's review of it.
4541 * The debugger now flushes standard output streams before it begins
4542 its output ("debugger invoked" and so forth).
4543 * The core version number and fasl file version number have both
4544 been incremented, because of incompatible changes in the layout
4546 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4547 used to not be called for a saved Lisp image.)
4548 * Martin Atzmueller's version of a patch to fix a compiler crash,
4549 as posted on sbcl-devel 13 September 2000, has been installed.
4550 * Instead of installing Martin Atzmueller's patch for the
4551 compiler transform for SUBSEQ, I deleted the compiler transform,
4552 and transforms for some similar consing operations.
4553 * A bug in signal handling which kept TRACE from working on OpenBSD
4555 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4556 away by constant folding
4557 * The system now defines its address space constants in one place
4558 (in the Lisp sources), and propagates them automatically elsewhere
4559 (through GENESIS and the sbcl.h file). Therefore, patching the
4560 address map is less unnecessarily tedious and error-prone. The
4561 Lisp names of address space constants have also been systematized.
4562 * CVS tags like dollar-Header-dollar have been removed from
4563 the sources, because they have never saved me trouble and
4564 they've been source of trouble working with patches and other
4565 diff-related operations.
4566 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4567 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4569 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4571 * The system has been ported to OpenBSD.
4572 * The system now compiles with a simple "sh make.sh" on the systems
4573 that it's supported on. I.e., now you no longer need to tweak
4574 text in the target-features.lisp-expr and symlinks in src/runtime/
4575 by hand, the make.sh takes care of it for you.
4576 * The system is no longer so grossly inefficient when compiling code
4577 involving vectors implemented as general (not simple) vectors (VECTOR T),
4578 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4579 which dares to use the various sequence functions on non-simple
4580 vectors, takes less of a performance hit.
4581 * There is now a primitive type predicate VECTOR-T-P
4582 to test for the (VECTOR T) type, so that e.g.
4583 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4584 can now be compiled with some semblance of efficiency. (The old code
4585 turned the type declaration into a full call to %TYPEP at runtime!)
4586 * AREF on (VECTOR T) is still not fast, since it's still compiled
4587 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4588 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4589 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4590 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4591 without making full calls to SUBTYPEP at runtime.
4592 (Some analogous efficiency issues for non-simple vectors specialized to
4593 element types other than T, or for non-simple multidimensional arrays,
4594 have not been addressed. They could almost certainly be handled the
4595 same way if anyone is motivated to do so.)
4596 * The changes in array handling break binary compatibility, so
4597 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4598 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4599 instead of (VALUES T T).
4600 * By following the instructions that Dan Barlow posted to sbcl-devel
4601 on 2 July 2000, I was able to enable primitive dynamic object
4602 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4603 functionality is not implemented (since it calls ld to resolve
4604 library references automatically, requiring RUN-PROGRAM for its
4605 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4606 try to resolve library references) is now supported.
4607 * The system now flushes the standard output streams when it terminates,
4608 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4609 them at several other probably-convenient times, e.g. in each pass of
4610 the toplevel read-eval-print loop, and after evaluating a form given
4611 as an "--eval" command-line option. (These changes were motivated by a
4612 discussion of stream flushing issues on cmucl-imp in August 2000.)
4613 * The source transform for TYPEP of array types no longer assumes
4614 that an array whose element type is a not-yet-defined type
4615 is implemented as an array of T, but instead punts, so that the
4616 type will be interpreted at runtime.
4617 * There is now some support for cross-compiling in make.sh: each of
4618 the phases of make.sh has its own script. (This should be transparent
4619 to people doing ordinary, non-cross-compile builds.)
4620 * Since my laptop doesn't have hundreds of megabytes of memory like
4621 my desktop machine, I became more motivated to do some items on
4622 my to-do list in order to reduce the size of the system a little:
4623 ** Arrange for various needed-only-at-cold-init things to be
4624 uninterned after cold init. To support this, those things have
4625 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4626 symbols with such names are now uninterned after cold init).
4627 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4628 fasl files for cold load.
4629 ** Remove the old compiler structure pooling code (which used to
4630 be conditional on the target feature :SB-ALLOC) completely.
4631 ** Redo the representation of some data in cold init to be more compact.
4632 (I also looked into supporting byte compiled code at bootstrap time,
4633 which would probably reduce the size of the system a lot, but that
4634 looked too complicated, so I punted for now.)
4635 * The maximum signal nesting depth in the src/runtime/ support code has
4636 been reduced from 4096 to 256. (I don't know any reason for the very
4637 large old value. If the new smaller value turns out to break something,
4638 I'll probably just bump it back up.)
4639 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4641 * Many, many bugs reported by Peter Van Eynde have been added to
4642 the BUGS list; some have even been fixed.
4643 * While enabling dynamic object file loading, I tried to make the
4644 code easier to understand, renaming various functions and variables
4645 with less ambiguous names, and changing some function calling
4646 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4647 * While trying to figure out how to do the OpenBSD port, I tried to
4648 clean up some of the code in src/runtime/. In particular, I dropped
4649 support for non-POSIX signal handling, added various comments,
4650 tweaked the code to reduce the number of compilation warnings, and
4651 renamed some files to increase consistency.
4652 * To support the new automatic configuration functionality in make.sh,
4653 the source file target-features.lisp-expr has been replaced with the
4654 source file base-target-features.lisp-expr and the machine-generated
4655 file local-target-features.lisp-expr.
4656 * fixed a stupid quoting error in make.sh so that using CMU CL
4657 "lisp -batch" as cross-compilation host works again
4659 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4661 * DESCRIBE no longer tries to call itself recursively to describe
4662 bound/fbound values, so that it no longer fails on symbols which are
4663 bound to themselves (like keywords, T, and NIL).
4664 * DESCRIBE now works on generic functions.
4665 * The printer now prints less-screwed-up representations of closures
4666 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4667 * A private symbol is used instead of the :EMPTY keyword previously
4668 used to mark empty slots in hash tables. Thus
4669 (DEFVAR *HT* (MAKE-HASH-TABLE))
4670 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4671 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4672 now does what ANSI says that it should. (You can still get
4673 similar noncompliant behavior if bang on the hash table
4674 implementation with all the symbols you get back from
4675 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4676 This breaks binary compatibility, since tests for equality to
4677 :EMPTY are wired into things like the macroexpansion of
4678 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4680 * There's now a minimal placeholder implementation for CL:STEP,
4681 as required by ANSI.
4682 * An obscure bug in the interaction of the normal compiler, the byte
4683 compiler, inlining, and structure predicates has been patched
4684 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4685 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4687 * Missing ordinary arguments in a macro call are now detected even
4688 when the macro lambda list contains &KEY or &REST.
4689 * The debugger no longer complains about encountering the top of the
4690 stack when you type "FRAME 0" to explicitly instruct it to go to
4691 the top of the stack. And it now prints the frame you request even
4692 if it's the current frame (instead of saying "You are here.").
4693 * As specified by ANSI, the system now always prints keywords
4694 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4696 * The default initial SIZE of HASH-TABLEs is now smaller.
4697 * Type information from CLOS class dispatch is now propagated
4698 into DEFMETHOD bodies, so that e.g.
4699 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4701 is now basically equivalent to
4702 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4703 (DECLARE (TYPE SINGLE-FLOAT X))
4705 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4706 operation, without having to do run-time type dispatch.
4707 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4708 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4709 * Since I don't seem to be making big file reorganizations very often
4710 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4711 is overflowing my ability to conveniently back them up), I've finally
4712 checked the system into CVS. (The CVS repository is on my home system,
4713 not at SourceForge -- putting it on SourceForge might come later.)
4714 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4715 high-level GC-NOTIFY-FOO functions send their output. (There's
4716 still very little control of where low-level verbose GC functions
4717 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4718 less than it used to -- the GC-NOTIFY-FOO functions are now under
4719 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4720 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4721 in only one place in the source code, and propagates it automatically
4722 everywhere that it's needed. Thus e.g. when I bump the version from
4723 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4724 * The C source files now include boilerplate legalese and documentation
4725 at the head of each file (just as the Lisp source files already did).
4726 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4727 to his page will be replaced with a link to his new CLiki service.
4729 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4731 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4732 * The build process now looks for GNU make under the default name "gmake",
4733 instead of "make" as it used to. If GNU make is not available as "gmake"
4734 on your system, you can change this default behavior by setting the
4735 GNUMAKE environment variable.
4736 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4737 can build without error under CMU CL.
4739 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4741 * There is now a partial SBCL user manual (with some new text and some
4742 text cribbed from the CMU CL manual).
4743 * The beginnings of a profiler have been added (starting with the
4744 CMU CL profiler and simplifying and cleaning up). Eventually the
4745 main interface should be through the TRACE macro, but for now,
4746 it's still accessed through vaguely CMU-CL-style functions and macros
4747 exported from the package SB-PROFILE.
4748 * Some problems left over from porting CMU CL to the new
4749 cross-compilation bootstrap process have been cleaned up:
4750 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4751 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4752 part when getting rid of the compiler *BACKEND* stuff.)
4753 ** The value of *NULL-TYPE* was screwed up, because it was
4754 being initialized before the type system knew the final
4755 definition of the 'NULL type. This screwed up several key
4756 optimizations in the compiler, causing inefficiency in all sorts
4757 of places. (I found it because I wanted to understand why
4758 GET-INTERNAL-RUN-TIME was consing.)
4759 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4760 PROCLAIM FTYPE information. Unfortunately this broke binary
4761 compatibility again, since now the forms output by DEFGENERIC
4762 to refer to functions which didn't exist in 0.6.3.
4763 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4764 can use the new (as of 0.6.3) transform for SOME into MAP into
4766 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4767 values to (INTEGER 0 1000000), so that the time code will no longer
4768 occasionally get blown up by Linux returning 1000000 microseconds
4769 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4770 its output conform to the ANSI spec. (Alas, this makes its output
4771 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4772 * A full call to MAP NIL with a single sequence argument no longer conses.
4773 * fixes to problems pointed out by Martin Atzmueller:
4774 * The manual page no longer talks about multiprocessing as though
4775 it were currently supported.
4776 * The ILISP support patches have been removed from the distribution,
4777 because as of version 5.10.1, ILISP now supports SBCL without us
4778 having to maintain patches.
4779 * added a modified version of Raymond Toy's recent CMU CL patch for
4780 EQUALP comparison of HASH-TABLE
4782 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4784 * The system still can't cross-compile itself with
4785 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4786 entails), but at least it can compile more of itself that way
4787 than it used to be able to, and various buglets which were uncovered
4788 by trying to cross-compile itself that way have now been fixed.
4789 * This release breaks binary compatibility again. This time
4790 at least I've incremented the FASL file format version to 2, so that the
4791 problem can be detected reliably instead of just causing weird errors.
4792 * various new style warnings:
4793 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4794 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4795 ** using the lexical binding of a variable named in the *FOO* style
4796 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4797 as specified by ANSI.
4798 * *RANDOM-STATE* is no longer automatically initialized from
4799 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4800 default behavior of the system is to repeat its behavior every time
4801 it's run. If you'd like to change this behavior, you can always
4802 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4803 old convention there was no comparably easy way to get the system to
4804 repeat its behavior every time it was run.
4805 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4806 FTYPE declarations has been removed, in favor of their ANSI
4807 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4808 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4809 the types of their sequence arguments can be determined at compile time.
4810 This is done through a new open code expansion for MAP which eliminates
4811 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4812 argument types can be determined at compile time.
4813 * The optimizer now transforms COERCE into an identity operation when it
4814 can prove that the coerced object is already of the correct type. (This
4815 can be a win for machine generated code, including the output of other
4816 optimization transforms, such as the MAP transform above.)
4817 * Credit information has been moved from source file headers into CREDITS.
4818 * Source file headers have been made more standard.
4819 * The CASE macro now compiles without complaining even when it has
4822 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4824 * (Note that the way that the PCL macroexpansions were rewritten
4825 to accommodate the change in DEFGENERIC below breaks binary
4826 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4827 not run under sbcl-0.6.2. Once we get out of alpha releases,
4828 i.e. hit release 1.0.0, we'll probably try to maintain binary
4829 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4830 and sbcl-1.4.4. Until then, however, it might be fairly common
4831 for maintenance releases to break binary compatibility.)
4832 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4834 * The reporting of unhandled signals has been changed to print some
4835 explanatory text as well as the report form. (Previously only
4836 the report form was printed.)
4837 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4838 it defines, so that the compiler no longer issues undefined function
4839 warnings for compiled-but-not-yet-loaded generic functions.
4840 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4841 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4842 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4843 section of the ANSI spec are supported.
4844 * The "" nickname for the "KEYWORD" package has been removed.
4845 The reader still handles symbol tokens which begin with a package marker
4846 as keywords, but it doesn't expose its mechanism for doing so in the
4847 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4848 * The system now issues STYLE-WARNINGs for contradictory TYPE
4849 proclamations. (Warnings for contradictory FTYPE proclamations would
4850 be nice too, but those can't be done usefully unless the type system
4851 is made smarter about FUNCTION types.)
4852 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4853 systematized, so that "*target-*.lisp is supposed to exist only on the
4854 target and imply that there's a related file which exists on the
4855 host, and *host-*.lisp is supposed to exist only on the host and imply
4856 that there's a related file which exists on the target. This involves a
4857 lot of renaming. Hopefully the acute confusion caused by the renaming
4858 will be justified by the reduction in chronic confusion..
4859 ** runtime-type.lisp -> early-target-type.lisp
4860 ** target-type.lisp -> late-target-type.lisp
4861 ** early-host-format.lisp -> early-format.lisp
4862 ** late-host-format.lisp -> late-format.lisp
4863 ** host-error.lisp -> misc-error.lisp
4864 ** early-error.lisp -> early-target-error.lisp
4865 ** late-error.lisp -> late-target-error.lisp
4866 ** host-defboot.lisp -> early-defboot.lisp
4867 ** code/misc.lisp -> code/target-misc.lisp
4868 ** code/host-misc.lisp -> code/misc.lisp
4869 ** code/numbers.lisp -> code/target-numbers.lisp
4870 ** code/early-numbers.lisp -> numbers.lisp
4871 ** early-host-type.lisp -> early-type.lisp
4872 ** late-host-type.lisp -> late-type.lisp
4873 ** host-typep.lisp -> typep.lisp
4874 ** load.lisp -> target-load.lisp
4875 ** host-load.lisp -> load.lisp
4876 ** host-disassem.lisp -> disassem.lisp
4877 ** host-insts.lisp -> insts.lisp
4878 ** byte-comp.lisp -> target-byte-comp.lisp
4879 ** host-byte-comp.lisp -> byte-comp.lisp
4880 ** host-signal.lisp -> signal.lisp
4881 ** host-defstruct.lisp -> defstruct.lisp
4882 ** late-target-type.lisp -> deftypes-for-target.lisp
4883 Furthermore, several other previously target-only files foo.lisp (e.g.
4884 hash-table.lisp and random.lisp) have been split into a target-and-host
4885 foo.lisp file and a target-only target-foo.lisp file, with their key type
4886 definitions in the target-and-host part, so that the cross-compiler will
4887 know more about target types.
4888 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4889 gone away. In their place are various *BACKEND-FOO* variables
4890 corresponding to the slots of the old structure.
4891 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4892 into the target SBCL has been fixed.
4893 * The chill.lisp system for loading cold code into a running SBCL
4895 * Support for the CMU CL "scavenger hook" extension has been removed.
4896 (It was undocumented and unused in the CMU CL sources that SBCL was
4897 derived from, and stale in sbcl-0.6.1.)
4898 * Various errors in the cross-compiler type system were detected
4899 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4900 (enabling various consistency checks). Many of them were fixed,
4901 but some hard problems remain, so the compiler is back to
4902 running without *TYPE-SYSTEM-INITIALIZED* for now.
4903 * As part of the cross-compiler type system cleanup, I implemented
4904 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4905 * I have started adding UNCROSS calls throughout the type system
4906 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4907 on cross-compiler input files will be able to go away, and various
4909 * CONSTANTP now returns true for quoted forms (as explicitly required
4912 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4914 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4915 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4916 (MAKE-LIST -1). In the longer term, it ought to become true
4917 that declarations are assertions even at SAFETY 1. For now, it's not
4918 quite true even at SAFETY 3, but it's at least more nearly true..
4919 (Note that this change seems to increases the size of the system by
4920 O(5%) and to decrease the speed of the compiler by 20% or more.)
4921 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4922 for the problem of printing dozens of lines of distracting information
4923 about low-level system machinery as part of the top stack frame
4924 on entry to the debugger when an undefined function was called.
4925 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4926 is not reset to COMMON-LISP-USER.
4927 * Compilation of stuff related to dyncount.lisp has been made conditional
4928 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4929 smaller. The various dyncount-related symbols have been moved into
4930 a new "SB-DYNCOUNT" package.
4931 * tty-inspect.lisp has been renamed to inspect.lisp.
4932 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4933 feature has gone away. (When we eventually port to other flavors of
4934 libc and/or Unix, we'll try to make the differences between flavors
4935 invisible at the user level.)
4936 * Various other *FEATURES* tags, and/or their associated conditionals,
4937 have been removed if obsolescent, or given better documentation, or
4938 sometimes given more-mnemonic names.
4940 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4942 * tidied up "make.sh" script
4943 * tidied up system directory structure
4944 * better "clean.sh" behavior
4945 * added doc/FOR-CMUCL-DEVELOPERS
4946 * many many small tweaks to output format, e.g. removing possibly-confusing
4947 trailing #\. character in DESCRIBE-INSTANCE
4948 * (EQUALP #\A 'A) no longer signals an error.
4949 * new hashing code, including EQUALP hashing
4950 * tidied up Lisp initialization and toplevel
4951 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4952 * command line argument processing
4953 * added POSIX-GETENV function to deal with Unix-ish environment variables
4954 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4955 terminating SBCL on EOF
4956 * non-verbose GC by default
4957 * There is no more "sbcl" shell script; the sbcl file is now the C
4958 runtime executable (just like CMU CL).
4959 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4961 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4962 symbol and package information as Lisp native symbol and package objects
4963 instead of strings naming symbols and strings naming packages. This way,
4964 whenever packages are renamed (as in warm init), debug information is
4965 transformed along with everything else.
4966 * tweaked the optimization policy declarations which control the building
4967 of SBCL itself. Now, among other things, the system no longer saves
4968 source location debugging information. (This helps two problems at once
4969 by reducing SBCL size and by keeping SBCL from trying to look for its
4970 sources -- which may not exist -- when reporting errors.)
4971 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4972 debugging and testing purposes
4973 * cleaned up printing, making the printer call PRINT-OBJECT for
4974 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4975 methods, giving nearly-ANSI behavior
4976 * converted almost all special variables to use *FOO* naming convention
4977 * deleted PARSE-TIME functionality, since it can be done portably
4978 * moved some files out of cold init into warm init
4979 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4981 * regularized formatting of source files
4982 * added an install.sh script
4983 * fixed ridiculous memory usage of cross-compiler by making
4984 compiler/alloc.lisp not try to do pooling unless it can hook
4985 itself into the GC of the cross-compilation host. Now the system
4986 builds nicely on my old laptop.
4987 * added :SB-ALLOC in target-features.lisp-expr
4988 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4989 was not implemented)
4990 * re-did condition handling and note reporting in the compiler. Notes
4991 are no longer handled by signalling conditions. Style warnings
4992 and warnings are handled more correctly and reported in such a way
4993 that it's easy to find one or the other in your output (so that you
4994 can e.g. figure out which of many problems caused COMPILE-FILE to
4996 * changed the severity of several compiler warnings from full WARNING
4997 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4998 compiler note reporting so that it doesn't use the condition system
4999 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
5000 COMPILE-FILE command)
5001 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
5002 function. As a consequence, START-BLOCK and END-BLOCK declarations are
5003 no longer supported, since their implementation was deeply intertwingled
5004 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
5005 * removed bogus "support" for compiler macros named (SETF FOO), and
5006 removed the compiler macro for SETF INFO (but only after making a fool
5007 of myself on the cmucl-imp mailing list by posting a bogus patch for
5008 DEFINE-COMPILER-MACRO..)
5009 * Compiled files containing forms which have side effects on the Lisp
5010 reader (such as DEFPACKAGE forms) are now handled more correctly.
5011 (Compiler queuing of top level lambdas has been suppressed by setting
5012 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5013 * deleted various currently-unused source files, e.g. gengc.lisp. They
5014 may be added back at some point e.g. when porting to other architectures,
5015 but until they are it's distracting to distribute them and to try to
5017 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5018 there were so many of them they're just distractions, and UNCROSS is
5019 known to be able to handle the current sources
5020 * moved PROFILE functionality into TRACE, so that it will be clear
5021 how the wrapping and unwrapping of functions when you profile them
5022 interacts with the wrapping and unwrapping of functions when you
5023 trace them. (Actually, the functionality isn't there yet, but at least
5024 the interface specification is there. Hopefully, the functionality will
5025 arrive with some future maintenance release.)
5026 * removed host-oops.lisp
5027 * changed signature of QUIT function to allow UNIX-CODE argument
5028 * fixed READ-SEQUENCE bug
5029 * tweaked verbose GC output so that it looks more like the progress
5030 output that ANSI specifies for functions like LOAD
5031 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5032 * added <http://sbcl.sourceforge.com> to the banner information printed by