1 ;;;; -*- coding: utf-8; fill-column: 78 -*-
2 changes in sbcl-1.0.25 relative to 1.0.24:
3 * incompatible change: SB-INTROSPECT:FUNCTION-ARGLIST is deprecated, to be
4 removed later. Please use SB-INTROSPECT:FUNCTION-LAMBDA-LIST instead.
5 * new feature: SB-INTROSPECT:DEFTYPE-LAMBDA-LIST allows retrieval of
6 DEFTYPE lambda lists. (thanks to Tobias Rittweiler)
7 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a
8 better name and does not return values so stale on multiprocessor systems.
9 Also, HOLDING-MUTEX-P was added for about the only sane usage of
11 * improvement: unithread builds keep track of MUTEX-VALUE.
12 * improvement: reading from a TWO-WAY-STREAM does not touch the output
13 stream anymore making it thread safe to have a concurrent reader and
14 a writer, for instance, in a pipe.
15 * improvement: GET-SETF-EXPANDER avoids adding bindings for constant
16 arguments, making compiler-macros for SETF-functions able to inspect
17 their constant arguments.
18 * improvement: COMPILE-FILE reports times with millisecond accuracy
19 (thanks to Luis Oliveira)
20 * optimization: CHAR-CODE type derivation has been improved, making
21 TYPEP elimination on subtypes of CHARACTER work better. (reported
22 by Tobias Rittweiler, patch by Paul Khuong)
23 * bug fix: setting alien structure fields of type struct by value now
24 computes the right offset for the memory copy.
25 * bug fix: compilation problem involving inlined calls to aliens with
26 result type VOID. (reported by Ken Olum)
27 * bug fix: #235a; sequential inline expasion in different policies no
28 longer reuses the functional from the previous expansion site.
29 * bug fix: DEFTYPE no longer breaks on bodies consisting of a single
30 unquoted symbol. Regression from 1.0.22.8. (reported by Ariel Badichi)
31 * improvements to the Windows port:
32 ** SB-BSD-SOCKETS now works from saved cores as well. (reported by Stephen
33 Westfold, thanks to Rudi Schlatte)
35 changes in sbcl-1.0.24 relative to 1.0.23:
36 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
37 vector of a multidimensional SIMPLE-ARRAY.
38 * new feature: the system now signals a continuable error if standard
39 readtable modification is attempted. (thanks to Tobias Rittweiler)
40 * new feature: DIRECTORY has been extended with a non-standard keyword
41 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
42 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
43 has been robustified and documented better. (thanks to Robert Goldman)
44 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
45 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
46 * tradeoff: constant FORMAT control strings are now compiled unless
47 SPACE > SPEED (previously only when SPEED > SPACE.)
48 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
50 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
52 * bug fix: lexical type declarations are now correctly reported by
53 SB-CLTL2. (reported by Larry D'Anna)
54 * bug fix: STRING-TO-OCTETS did not handle :START properly when
55 using UTF-8 as external format. (thanks to Luis Oliveira)
56 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
57 are signalled correctly. (thanks to Stas Boukarev)
58 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
59 designator. (thanks to Tobias Rittweiler)
60 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
61 designator, and returns T instead of the function. (thanks to
63 * bug fix: direct superclasses of STANDARD-CLASS and
64 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
65 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
66 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
67 when no class with the named slot yet exists no longer causes a
68 compile-time style-warning.
69 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
70 in safe code. (reported by Didier Verna)
71 * bug fix: #430; nested structure constructors can stack allocate.
72 * bug fix: on some 64-bit platforms dynamic space size was truncated
73 to #xffff0000 bytes. (reported by Benjamin Lambert)
74 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
75 REPL unusable. (reported by Daniel Herring)
76 * bug fix: return values of READ-SEQUENCE did not take :START into
77 account on file streams, regressions since 1.0.12.22. (reported by
78 Thomas Russ, patch by Paul Khuong)
79 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
80 method specializer used to confuse permuation vector optimization.
81 * bug fix: system inserted bogus implicit type declarations for local
82 special variables in DEFMETHOD bodies.
83 * bug fix: #354; duplicated frames in backtraces due to
84 non-tail-call-optimized XEPs to functions with return type NIL
86 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
87 initializes structure object slots according to DEFSTRUCT initforms,
88 and DEFSTRUCT forms :INCLUDEind structure classes defined using
89 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
90 (reported by Bruno Haible and Stephen Wilson)
91 * bug fix: #395; fill-pointer output streams used now support
92 element-type BASE-CHAR as well.
93 * bug fix: compiler error when attempting to derive return value of
94 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
96 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
98 * bug fix: Mac OS X binaries should now be portable between Leopard
101 changes in sbcl-1.0.23 relative to 1.0.22:
102 * enhancement: when disassembling method functions, disassembly
103 for the associated fast function is also produced.
104 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
106 * optimization: printing with *PRINT-PRETTY* true is now more
107 efficient as long as the object being printed doesn't require
108 special handling by the pretty printer.
109 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
110 now interact correctly with type declarations.
111 * partial bug fix: PCL detects infinite recursion during wrapper
112 validation. (thanks to Attila Lendvai)
113 * bug fix: #426; nested function calls are inlined properly.
114 Previously if FOO was an inline function, in calls of the form
115 (FOO (FOO ...)) the outer call was not inlined.
116 * bug fix: long long arguments passed in registers to alien
117 functions on PowerPC Linux and NetBSD are now handled correctly
118 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
120 changes in sbcl-1.0.22 relative to 1.0.21:
121 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
122 for the shared object in the current directory, but passes the native
123 namestring of the designated pathname to the operation system's shared
124 object loading function as-it.
125 * minor incompatible change: --disable-debugger toplevel option now takes
126 effect before processing of initialization files and --eval or --load
128 * new feature: new commandline argument: --script, which supports
129 shebang lines. See documentation for details. (based on work by
131 * new feature: SAVE-LISP-AND-DIE can save current values of
132 --dynamic-space-size and --control-stack-size in the executable core,
133 causing it to skip normal runtime option processing. See documentation
134 for details. (thanks to Zach Beane)
135 * enhancement: inoccous calls to EVAL or generic functions dispatching
136 on subclasses of eg. STREAM no longer cause compiler notes to appear.
137 * enhancement: the system no longer resignals errors from --load and
138 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
139 associated with the original error to be lost. (thanks to Ariel
141 * enhancement: :DONT-SAVE keyword argument has been added to
142 LOAD-SHARED-OBJECT for controlling interaction with
143 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
144 the effects of an earlier LOAD-SHARED-OBJECT call.
145 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
146 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
147 provided. (reported by Cedric St-Jean)
148 * bug fix: circularity handling in the reader did not treat raw
149 structure slots correctly. (reported by Cedric St-Jean)
150 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
151 file descriptors when there were none.
152 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
153 was broken. (reporteed by Willem Broekema)
154 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
155 pathnames without a directory.
156 * bug fix: COMPILE-FILE did not warn about undefined variable
157 references at toplevel, and LOAD of the resulting fasl did
159 * bug fix: functions with non-required arguments used to end up with
160 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
161 * bug fix: redefining a function with non-required arguments didn't
162 update the system's knowledge about its call signature properly.
163 * bug fix: fixed #431; incompatible alien record type redefinitions
164 are detected and handled. (reported by Neil Haven)
165 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
166 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
168 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
169 after alien stack frames.
170 * bug fix: strings passed to mk{s,d}temp lacked null termination
172 changes in sbcl-1.0.21 relative to 1.0.20:
173 * new feature: the compiler is able to track the effective type of a
174 generic function across method addition and removal.
175 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
176 appropriately typed structure slots without locking.
177 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
178 information like those gathered by TIME using a programming-friendly
180 * new feature: TIME reports time taken even if the form performs a
181 non-local transfer of control.
182 * enhancement: reduced conservativism on GENCGC platforms: on
183 average 45% less pages pinned (measured from SBCL self build).
184 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
185 mutate constant symbols or violate declaimed type of the symbol.
186 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
187 sleeping on the semaphore in heavily contested semaphores.
188 * bug fix: semaphores and condition variables were not interrupt
190 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
191 owned by other threads anymore.
192 * bug fix: FIND on lists called KEY outside the specified
193 subsequence. (reported by budden)
194 * bug fix: LOG doesn't use single-float intermediate results when
195 given mixed integer and double-float arguments, leading to better
196 precision. (reported by Bob Felts)
197 * bug fix: LOG with base zero returned values of inconsistent type.
199 changes in sbcl-1.0.20 relative to 1.0.19:
200 * minor incompatible change: OPTIMIZE qualities
201 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
202 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
203 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
204 * documentation: some slot access efficiency guidelines have been
205 added to the user manual.
206 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
207 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
208 as ASSOC and MEMEBER.
209 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
210 to more efficient EQ-comparison versions more often.
211 * optimization: enhanced derivation of DOLIST iteration variable type
213 * optimization: constant folding of simple (LIST ...) forms as DOLIST
215 * optimization: runtime lookup of function definitions can be
216 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
217 * optimization: compiler is able to derive the return type of
218 (AREF (THE STRING X) Y) as being CHARACTER.
219 * optimization: CLRHASH on empty hash-tables no longer does pointless
220 work. (thanks to Alec Berryman)
221 * optimization: REPLACE deftransforms don't punt when :START1 or
223 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
224 no longer emits a code deletion note. (reported by Andrew
226 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
227 functions with an odd number of &OPTIONAL arguments, a &REST
228 argument, and one or more &KEY arguments at the call site.
229 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
230 arguments appeared at call sites as well.
231 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
232 READ-LINE &co can no longer read from them afterwards. (reported
234 * bug fix: fixed #427: unused local aliens no longer cause compiler
235 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
237 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
238 alien-stack leakage. (reported by Andy Hefner)
239 * bug fix: PROGV signals an error when an attempt to violate declared
240 type of a variable or bind a constant is made.
241 * bug fix: SET signals an error when an attempt to violate declared
242 type of a variable is made.
243 * bug fix: restart computation during the execution of a restart
244 test function no longer causes infinite recursion. (reported by
246 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
247 signals an error. (thanks to Pierre Mai)
248 * bug fix: EXPORT left symbol unexported in conflict situations.
249 (thanks to Michael Weber)
250 * bug fix: correctly handle name conflicts involving CL:NIL.
251 (thanks to Michael Weber)
252 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
253 conflicts arising from USEing package with conflicting symbols
254 correctly. (thanks to Michael Weber)
256 changes in sbcl-1.0.19 relative to 1.0.18:
257 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
258 warnings that go otherwise unhandled will be muffled if they are
259 of the type that's the value of this variable.
260 * optimization: stack allocation is slightly more efficient on x86
262 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
263 where a raw slot always is initialized using the initform whose
264 type is not know sufficiently well a compile-time are now compiled
265 correctly. (reported by John Morrison)
266 * bug fix: compiler no longer makes erronous assumptions in the
267 presence of non-foldable SATISFIES types.
268 * bug fix: stack analysis missed cleanups of dynamic-extent
269 arguments in non-let-converted calls to local functions.
270 * improvements to the Windows port:
271 ** adjusted address spaces for building on both Win32 and
272 Win64. (thanks for John Connors)
273 * fixed some bugs revealed by Paul Dietz' test suite:
274 ** interval arithmetic during type derivation used inexact integer
275 to single-float coercions.
276 ** arithmetic operations involving large integers and single
277 floats give the same results in compiled and interpreted code.
278 ** deriving the result type of COERCE no longer signals an error
279 if the derived type of the second argument is a MEMBER type
280 containing invalid type specifiers.
281 ** ADJOIN with constant NIL as second argument works correctly.
282 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
284 changes in sbcl-1.0.18 relative to 1.0.17:
285 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
286 profiles only the current thread.
287 * minor incompatible change: changes to SYMBOL-VALUE of constants
288 defined with DEFCONSTANT now signal an error.
289 * enhancement: SB-SPROF now has support for wallclock profiling,
290 and is also able to profile specific threads. REPORT output
291 has also additional sorting options.
292 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
294 * optimization: structure allocation has been improved
295 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
296 ** out of line constructors are ~10% faster.
297 ** inline constructors are ~15% faster.
298 ** inline constructors are capable of dynamic extent allocation
299 (generally on x86 and x86-64, in some cases on other platforms
301 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
303 * optimization: file compiler is now able to coalesce non-circular
304 lists, non-base strings, and bit-vectors. Additionally, constants
305 are never referenced through SYMBOL-VALUE at runtime.
306 * optimization: code defining methods on PRINT-OBJECT (and other
307 generic functions in the COMMON-LISP package) now loads faster.
308 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
310 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
311 fasl is now deleted. (reported by Attila Lendvai)
312 * bug fix: READ-LINE always returned NIL for the last line in files.
313 (reported by Yoshinori Tahara)
314 * bug fix: more accurate disassembly annotations of foreign function
315 calls. (thanks to Andy Hefner)
316 * bug fix: trimming non-simple strings and non-string string
317 designators when the there is nothing to trim works properly.
318 (thanks to James Knight)
319 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
320 (thanks to Travis Cross)
321 * fixed some bugs revealed by Paul Dietz' test suite:
322 ** NIL is a valid function name (regression at 1.0.13.38)
323 ** FILL on lists was missing its return value (regression at 1.0.12.27)
324 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
325 fill pointers properly (regression at 1.0.12.23)
326 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
327 displacement indices properly (regression at 1.0.12.23)
329 changes in sbcl-1.0.17 relative to 1.0.16:
330 * temporary regression: user code can no longer allocate closure
331 variable storage on stack, due to bug 419 without explicitly
332 requesting it. Please consult sbcl-devel for advice if you need to
333 use this feature in the meanwhile.
334 * new feature: runtime argument --control-stack-size can be used to
335 adjust thread default control stack size.
336 * enhancement: improved TIME output
337 ** all times are reported using the measured accuracy (milliseconds
338 for real and GC times, microseconds for everything else.)
339 ** processor cycle counts on x86 and x86-64.
340 ** interpreted forms are counted for both evaluator modes.
341 ** number of lambdas converted by the compiler is reported.
342 ** CPU percentage report (computed from real and total run time.)
343 ** more comprehensive run time reporting, using a condenced format
344 ** interperted form, lambda, and page fault counts are omitted
346 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
348 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
349 * optimization: two argument forms of LAST are upto ~10% faster
350 in normal SPEED policies.
351 * optimization: NCONC no longer needs to heap cons its &REST list
352 in normal SPEED policies.
353 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
355 * bug fix: UNION and NUNION work with :TEST-NOT once more,
356 regression since 1.0.9.1. (thanks to Eric Marsden)
357 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
358 since 1.0.15.36. (thanks to Paul Khuong)
359 * bug fix: LAST when always returned the whole list when given a bignum
360 as the second argument.
361 * bug fix: dynamic extent allocation of nested lists and vectors
362 could leak to otherwise accessible parts.
363 * bug fix: invalid optimization of heap-allocated alien variable
365 * bug fix: fasl header checking is less vulnerable to different
366 platform word lengths.
367 * bug fix: more correct assembler syntax for GNU binutils
368 2.18.50.0.4 support. (thanks to Marijn Schouten)
369 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
371 * bug fix: the fopcompiler can handle LOCALLY forms (with no
372 declarations) successfully. (reported by Attila Lendvai)
374 changes in sbcl-1.0.16 relative to 1.0.15:
375 * minor incompatible change: revert the changes to sb-posix's error
376 signaling added in 1.0.14.
377 * minor incompatible change: change PROBE-FILE back to returning
378 NIL whenever we can't get a truename, as was the case before 1.0.14.
379 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
380 inherits from ERROR instead of just CONDITION.
381 * new feature: SB-INTROSPECT can provide source locations for instances
382 as well. (thanks to Tobias Rittweiler)
383 * optimization: binding special variables now generates smaller code
384 on threaded platforms.
385 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
386 and cases where no :TEST is given but the compiler can infer that
387 the element to search is of type (OR FIXNUM (NOT NUMBER)).
388 * optimization: better LOGNOT on fixnums.
389 * optimization: modular arithmetic for a particular requested width
390 is implemented using a tagged representation unless a better
391 representation is available.
392 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
393 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
394 for wellformedness and returns a specialized vector. (reported by
395 Francois-Rene Rideau)
396 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
397 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
398 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
399 unscheduling timer at the same time as another timer fires.
400 * bug fix: don't reschedule timers for dead threads.
401 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
402 * bug fix: copying output from RUN-PROGRAM to a stream signalled
403 bogus errors if select() was interrupted.
404 * enhancement: add support for fcntl's struct flock to SB-POSIX.
406 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
407 * enhancement: cleaner backtraces for interactive interrupts, as
408 well as other cases where the interesting frames used to be
409 obscured by interrupt handling frames.
410 * enhancement: untracing a whole package using (UNTRACE "FOO") is
411 now supported, and tracing a whole package using (TRACE "FOO") now
412 traces SETF-functions as well.
413 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
414 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
415 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
416 * unreadably printed representation of hash-tables now includes
418 * bug fix: partially fixed #188: type propagation from assignments
419 is now more efficient.
420 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
421 'DOUBLE-FLOAT) are not flushable.
422 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
423 full address of the object, and none of the tag bits.
424 * bug fix: readably printing hash-tables now respects other printer
425 control variables. (reported by Cedric St-Jean)
426 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
427 keyword with MAKE-HASH-TABLE.
428 * bug fix: export SB-POSIX:MKSTEMP.
429 * bug fix: SORT was not interrupt safe.
430 * bug fix: XREF accounts for the last node of each basic-block as
432 * bug fix: MAKE-INSTANCE optimizations interacted badly with
433 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
434 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
435 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
436 create obsolete instances in the case of redefinition or
437 obsoletion of a superclass. (thanks to Andy Hefner)
438 * bug fix: Support for the Alpha architecture has been revived; it had
439 suffered somewhat from lack of maintenance since sbcl-1.0.
440 * improvements to the Windows port:
441 ** The system detects the codepage to use at startup, instead of
442 using the value from the saved core. (thanks to Kei Suzuki)
444 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
445 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
446 (see documentation for details.)
447 * revived support for OpenBSD (contributed by Josh Elsasser)
448 * partially fixed bug #108: ROOM no longer suffers from occasional
449 (AVER (SAP= CURRENT END)) failures.
450 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
451 bodies no longer cause a WARNING to be signalled. (reported by
453 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
454 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
455 the bug in TRUNCATE.) (reported by Michael Weber)
456 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
457 no samples. (reported by Andy Hefner)
458 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
459 no longer appear as (NIL ...) frames in backtraces.
460 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
461 is now actually exported from SB-EXT as documented. (reported by
463 * bug fix: sb-aclrepl now correctly understands how to inspect
464 single-floats on 64-bit platforms where single-floats are not boxed.
465 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
466 yet been finalized. (reported by Levente Meszaros)
467 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
468 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
469 SBCL-specific optimize qualities.
471 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
472 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
473 an executable in the search path, and does so in the child
474 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
475 been removed; it can be found in the CVS history, for somebody who
476 needs that search behavior (see the manual).
477 * minor incompatible change: compiler policy re. weakening type
478 checks has changed: now type checks are weakened only if SAFETY < 2
480 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
481 unparsing of directory pathnames as files. Analogously,
482 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
483 filename to parse into a directory pathname.
484 * enhancement: implicit generic function creation now signals a
485 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
486 users can bind handlers for and muffle around calls to LOAD.
487 * enhancement: RUN-PROGRAM allows unicode arguments and environments
488 to be used (using the default stream external format), and allows
489 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
490 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
491 strings and vectors whose element-type or simplicity is not fully
492 known at compile-time.
493 * optimization: STRING-TRIM and related functions no longer allocate
494 a new string when no trimming needs to be performed. These functions
495 are also faster than before when the input string has been declared
497 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
498 * optimization: READ-LINE is significantly faster for files containing
500 * optimization: non-open coded uses of character comparison operators
501 (e.g. char=) no longer cons when called with more than one parameter
502 on platforms supporting dynamic-extent allocation.
503 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
504 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
505 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
507 * bug fix: some sequence functions elided bounds checking when
509 * bug fix: too liberal weakening of union-type checks when SPEED >
511 * bug fix: more bogus fixnum declarations in ROOM implementation
514 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
515 * new feature: MAKE-HASH-TABLE now experimentally accepts a
516 :SYNCHRONIZED argument, which makes the hash-table safe for
517 concurrent accesses (but not iteration.) See also:
518 SB-EXT:WITH-LOCKED-HASH-TABLE, and
519 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
520 * optimization: CONCATENATE on strings is an order of magnitude faster
521 in code compiled with (> SPEED SPACE).
522 * optimization: SUBSEQ is ~50% faster on lists.
523 * bug fix: bug 417 fixed -- source location reporting is now more robust.
524 * bug fix: SUBSEQ on a list will now correctly signal an error if if
525 END is smaller then START.
526 * bug fix: SB-PROFILE will no longer report extra consing for nested
527 calls to profiled functions.
528 * bug fix: ROOM implementation had bogus fixnum declarations which
529 could cause type-errors when calling ROOM on large images.
530 * bug fix: if file compilation is aborted, the partial fasl is now
531 deleted, and COMPILE-FILE returns NIL as the primary value.
532 * bug fix: number of thread safety issues relating to SBCL's internal
533 hash-table usage have been fixed.
534 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
535 be returned from its body when the values were being returned
536 using unknown-values return convection and the W-P-O was wrapped
537 inside an UNWIND-PROTECT.
538 * bug fix: sb-posix should now compile again under Windows, enabling
541 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
542 * incompatible change: hash-table accessor functions are no longer
543 automatically protected by locks. Concurrent accesses on the same hash-table
544 from multiple threads can give inconsistent results or even corrupt the
545 hash-table completely. Multi-threaded applications should do their own
546 locking at the correct granularity. In the current implementation it is
547 still safe to have multiple readers access the same table, but it's not
548 guaranteed that this property will be maintained in future releases.
549 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
550 and will signal an error at runtime.
551 * enhancement: SB-THREAD package now exports a semaphore interface.
552 * enhancement: CONS can now stack-allocate on x86 and
553 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
555 * enhancement: nested lists can now be stack allocated on
556 platforms providing stack allocation support.
557 * enhancement: dynamic-extent support has been extended to support
558 cases where there are multiple possible sources for the stack
560 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
561 if the mutex is uncontested on Linux.
562 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
563 as specified by the CLHS. (thanks to Tobias Rittweiler)
564 * bug fix: NaN comparison now works on x86-64.
565 * bug fix: CLOSE :ABORT T on a stream with pending output now
567 * bug fix: instances of non-standard metaclasses using standard
568 instance structure protocol sometimes missed the slot type checks
570 * bug fix: known functions can be inlined in high-debug code.
572 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
573 * minor incompatible change: the MSI installer on Windows no longer
574 associates .lisp and .fasl files with the installed SBCL.
575 * minor incompatible change: :UNIX is no longer present in *FEATURES*
576 on Windows. (thanks to Luis Oliviera)
577 * new feature: SB-CLTL2 contrib module now implements
578 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
579 * optimization: scavenging weak pointers is now more efficient,
580 requiring O(1) instead of O(N) per weak pointer to identify
581 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
582 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
583 method bodies using constant slot-names when the first argument is
584 a specializer parameter for the method.
585 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
586 names, when the class of the instance is a direct instance
587 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
588 as slow as the constant slot-name case.)
589 * optimization: member type construction is now O(N) instead
591 * optimization: UNION and NUNION are now O(N+M) for large
592 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
594 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
595 now more readable in environments like Slime which display it.
596 (thanks to Tobias C. Rittweiler)
597 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
598 was able to infer the correct argument type for the object on which
599 the CAS operation was being performed.
600 * bug fix: copy propagation interfered with parallel assignment
601 semantics in local calls. (reported by Paul Khuong)
602 * bug fix: the signed modular fixnum shift compiled to wrong code on
603 x86 and x86-64. (spotted by a slight modification to some of
606 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
607 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
608 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
609 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
611 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
612 and generic function now print the object identity for anonymous
613 instances. (thanks to Attila Lendvai)
614 * enhancement: as an extension to MOP, SBCL now supports SETF of
615 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
616 (thanks to Attila Lendvai)
617 * enhancement: sb-cover has an alternate annotation mode which puts
618 the form annotations at the car of the form, instead of annotating
619 it completely. The benefit of this mode is that it shows better
620 which forms were instrumented by the compiler.
621 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
622 by printer control variables. (reported by Dan Corkill)
623 * bug fix: system leaked memory when delayed output was performed by
624 the OS in smaller chunks then expected. (thanks to David Smith)
625 * bug fix: system leaked memory when file streams were not closed
627 * bug fix: large objects written to slow streams that were modified
628 after the write could end up with the modified state written to
629 the underlying file descriptor.
630 * bug fix: multiple threads operating in parallel on the same stream
631 could cause buffer-overflows.
632 * bug fix: source location information is stored correctly
633 (broken since 1.0.6). This bug would generally show up as the
634 Slime debugger highlighting the wrong form when the sldb-show-source
636 * bug fix: Branch forms are again annotated as branches in the sb-cover
638 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
639 on x86-64. (reported by Gregory Vanuxem)
640 * bug fix: EQUALP could return wrong results for structures with raw
641 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
642 word). (reported by Vjacheslav Fyodorov)
643 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
646 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
647 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
648 atomic compare-and-swap operations on threaded platforms.
649 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
650 allows assigning a global minimum value to optimization qualities
651 (overriding proclamations and declarations).
652 * enhancement: closed over variables can be stack-allocated on x86
654 * performance bug fix: GETHASH and (SETF GETHASH) are once again
656 * optimization: slot definition lookup is now O(1). This speeds up
657 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
658 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
659 * optimization: ASSOC and MEMBER can now be open-coded for all
660 combinations of keyword arguments when second argument is constant
661 and SPEED >= SPACE. In other cases a specialized version is
663 * bug fix: using obsoleted structure instances with TYPEP and
664 generic functions now signals a sensible error.
665 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
666 (reported by Kristoffer Kvello)
667 * bug fix: backtrace construction is now more careful when making
668 lisp-objects from pointers on the stack, to avoid creating bogus
669 objects that can be seen by the GC.
670 * bug fix: defaulting of values in contexts expecting more than 7
671 variables now works on x86-64. (reported by Christopher Laux)
672 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
674 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
675 as the property-list of a symbol.
676 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
677 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
678 situations requested, are once again file-compileable. (reported
681 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
682 * MOP improvement: support for user-defined subclasses of
683 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
684 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
685 the expansion of DEFMETHOD, is responsible for generating a form
686 which creates a list of specializers when evaluated. Additional
687 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
688 debugging and introspective support.
689 * minor incompatible change: the (unsupported) spinlock interface
690 has changed: free spinlock now has the value NIL, and a held spinlock
691 has the owning thread as its value.
692 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
693 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
695 * enhancement: name of a socket-stream is now "a socket" instead of
697 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
698 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
699 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
700 for generic functions. (Thanks to Tobias C. Rittweiler)
701 * optimization: bignum printing speed has been improved by 20-40%
702 (depending on the bignum size.)
703 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
705 * bug fix: the cache used by the CLOS to store precomputed effective
706 methods, slot offsets, and constant return values is now thread and
708 * bug fix: generic function dispatch function updating is now thread
709 and interrupt safe (in the sense that the known issues have been
711 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
712 * bug fix: interning EQL-specializers is now thread and interrupt safe.
713 * bug fix: asdf systems with dependencies to the SB-POSIX or
714 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
715 * bug fix: interrupt safety of applicable method computation has been
718 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
719 * new contrib: sb-cover, an experimental code coverage tool, is included
721 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
722 significantly faster.
723 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
724 produce more efficient code on x86-64 if the compiler can determine
725 that (+ INDEX <constant>) does not require a bounds check and FOO
726 has an element type at least 8 bits wide.
727 * enhancement: a new, experimental synchronous timeout facility is
728 provided. Refer to SB-SYS:WITH-DEADLINE for details.
729 * enhancement: when a symbol name conflict error arises, the
730 conflicting symbols are always printed with a package prefix.
731 (thanks to Kevin Reid)
732 * enhancement: stepping is now once again supported on the SPARC and
733 MIPS platforms. (It is also now more likely to work on CheneyGC
735 * enhancement: sb-sprof can now also track and report accurate call
737 * bug fixes: the treatment of non-standard subclasses of
738 SB-MOP:SPECIALIZER is more correct.
739 * incompatible change: PURIFY no longer copies the data from the
740 dynamic space into the static and read-only spaces on platforms
741 that use the generational garbage collector
742 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
744 * bug fix: GC race condition occasionally resulting in crashes with
745 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
747 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
749 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
750 a STYLE-WARNING to be signalled (regression from 1.0.4.)
751 * bug fix: an asynchronous interrupt could previously leave the
752 system running with GC inhibited.
753 * bug fix: a DECLARE form evaluated at top-level now causes an error
754 rather than silently (or verbosely) returning NIL.
755 * bug fix: trying to dissassemble functions compiled at high DEBUG could
756 result in a error being signalled due to source form lookup errors.
757 (reported by Peter Graves)
759 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
760 * incompatible change: removed writer methods for host-ent-name,
761 host-ent-addresses -- changing the values did not update the DNS
763 * minor incompatible change: changed experimental JOIN-THREAD interface
764 * documentation: the manual now lists reader and writer methods
765 in class slot documentation sections. (thanks to Richard M Kreuter)
766 * documentation: unwinding from asyncronous events has been
767 documented as unsafe.
768 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
769 in multithreaded application code.
770 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
771 platforms. (thanks to James Anderson for the optimization hint)
772 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
774 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
775 variants no longer cons.
776 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
777 their NOT- variants no longer cons.
778 * optimization: EQUAL hash tables no longer use SXHASH for objects
779 of all data types, but instead use an EQL hash for types for which
780 EQUAL is the same as EQL.
781 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
782 are significantly faster.
783 * optimization: new STRING-OUTPUT-STREAM implementation is much
784 faster for large outputs and conses 30% less on average.
785 * enhancement: XREF information is now collected for references made
786 to global variables using SYMBOL-VALUE and a constant argument.
787 * enhancement: SIGINT now causes a specific condition
788 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
789 * bug fix: FILE-POSITION used to signal an error in some cases where
790 ANSI requires it to return NIL.
791 * bug fix: ADJUST-ARRAY is now interrupt-safe.
792 * bug fix: adding and removing fd-handlers is now interrupt-safe.
793 * bug fix: inlined calls to C now ensure 16byte stack alignment on
795 * bug fix: bad type declaration in the CLOS implementation has
796 been fixed. (reported by James Anderson)
797 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
798 (thanks to Lutz Euler)
799 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
800 required by the CLHS. (reported by Samium Gromoff)
801 * bug fix: dead unbound variable references now signal an error.
802 (reported by Marco Monteiro)
803 * bug fix: / with an unused value was being deleted in safe code.
804 (thanks to Marco Monteiro and Kevin Reid)
805 * bug fix: number of characters that can be written onto a single
806 line in a file is unlimited.
807 * bug fix: some GC deadlocks caused by asynchronous interrupts have
808 been fixed by inhibiting interrupts for when GC is disabled.
809 * bug fix: some interrupt safety issues with GETHASH, (SETF
810 GETHASH), CLRHASH and REMHASH have been fixed.
811 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
812 specifier no longer causes infinite recursion.
813 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
814 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
815 * bug fix: an &environment argument with the correct variable information
816 is passed to macros that are expanded during byte compilation.
817 (reported by Samium Gromoff)
818 * bug fix: a logic error was causing the hash function for bignums to
819 have excessive amounts of collisions. (reported by Faré Rideau)
820 * bug fix: modifying the contents of an array could change the return
821 value of SXHASH on that array, which is only allowed for strings
822 and bit vectors. (bug introduced in 0.9.16)
823 * improvement: the x86-64/darwin port now passes all tests (except
824 for the debugger tests) but should still be considered
825 experimental until this is fixed.
826 * improvement: a style-warning is signaled for CASE (etc) clauses with
827 duplicate keys. (thanks to Kevin Reid)
828 * improvement: macos/x86 and macos/x86-64 now use mach exception
829 handlers for dealing with illegal instructions (for trapping and
830 error handling) and memory protection violations (for GC).
832 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
833 * new platform: experimental support for x86-64/darwin (MacOS).
834 * incompatible change: the thread-safe (on most platforms) getaddrinfo
835 and getnameinfo sockets functions are used instead of gethostbyaddr
836 and gethostbyname, on platforms where the newer functions are available.
837 As a result, the ALIASES field of HOST-ENT will always be NIL on these
839 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
840 don't print the startup banner, but behave as if --noinform was passed
841 as a command line argument. (thanks to Kevin Reid)
842 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
843 * optimization: code using alien values with undeclared types is much faster.
844 * optimization: the compiler is now able to open code SEARCH in more cases.
845 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
846 * bug fix: using standardized COMMON-LISP special variables as loop
847 variables no longer signals bogus package lock violations. (reported
849 * bug fix: declaring local loop variables to be of a range-limited type
850 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
851 (reported by Andras Simon)
852 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
853 bugs remain on x86-64.)
854 * bug fix: the #= and ## reader macros now interact reasonably with
855 funcallable instances.
856 * bug fix: type-checks for function arguments were compiled using the
857 compiler policy settings of the wrong lexical environment (in 1.0.2
859 * bug fix: SHADOW accepts characters as string designators, as required
860 by the spec (thanks to Eric Marsden)
861 * bug fix: fixed GC safety issues when foreign functions are called with
862 non-base strings as arguments
863 * bug fix: more consistent error output for fatal-compiler-warnings, like
865 * bug fix: functions with proper names were showing up as NIL in
866 backtraces (reported by Edi Weitz, regression in 1.0.2)
868 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
869 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
870 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
871 produce more efficient code on the x86 if the compiler can determine
872 that (+ INDEX <constant>) does not require a bounds check and FOO
873 has an element type at least 8 bits wide.
874 * bug fix: references to undefined variables are handled the same way
875 in toplevel forms as in the normal compiler
876 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
878 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
879 (thanks to Jon Buller)
880 * improvement: faster compilation times for complex functions
881 * improvement: added readlink support to SB-POSIX (thanks to Richard
884 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
885 * improvement: experimental support for mach exception handling on
886 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
888 * improvement: support for GBK external format.
889 (thanks to Chun Tian (binghe))
890 * improvement: the debugger now displays variables that have been closed
891 over, in code compiled with (DEBUG 2) or higher
892 * improvement: support for executable cores on NetBSD (thanks to
894 * new feature: added a RESTART-FRAME debugger command
895 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
896 be used to provide an implementation for FILE-POSITION on Gray streams
897 (thanks to Eric Marsden)
898 * new feature: add syslog, openlog and closelog support to SB-POSIX
899 (thanks to Richard Kreuter)
900 * optimization: the function call overhead in code compiled with
901 a high DEBUG optimization setting is significantly
902 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
903 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
904 (thanks to Tony Martinez)
905 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
906 evaluated once (reported by Kevin Reid)
907 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
908 works again (thanks to Kilian Sprotte)
909 * bug fix: an error is signaled for tagbodies with duplicate tags
910 (thanks to Stephen Wilson)
911 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
912 * bug fix: Win32 port can now handle foreign code unwinding Lisp
913 stack frames from alien callbacks.
914 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
915 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
916 65k pages (thanks to David Woodhouse)
917 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
919 changes in sbcl-1.0.1 relative to sbcl-1.0:
920 * new platform: FreeBSD/x86-64, including support for threading.
921 * new feature: the compiler stores cross-referencing information
922 about function calls (who-calls), macroexpansion (who-macroexpands)
923 and special variables (who-binds, who-sets, who-references) for code
924 compiled with (< SPACE 3). This information is available through the
925 sb-introspect contrib.
926 * new feature: users may subclass SEQUENCE, and have instances of
927 these classes interoperate with standard Common Lisp functions if
928 a number of methods are defined. (This feature is experimental
929 and the interface subject to change based on feedback from SBCL
930 users and the general community)
931 * improvement: sb-sprof traces call stacks to an arbitrary depth on
932 x86/x86-64, rather than the previous fixed depth of 8
933 * improvement: another pthread back-end of mutex "pthread-futex".
934 * bug fix: non-ascii command-line arguments are processed correctly
935 (thanks to Yaroslav Kavenchuk)
936 * bug fix: non-required arguments were not passed correctly when a method
937 defined using DEFMETHOD was called from a mop-generated method using
938 CALL-NEXT-METHOD (reported by Pascal Costanza)
939 * bug fix: recursion is now permitted in accessors through
940 SLOT-UNBOUND. (reported by Pascal Costanza)
941 * bug fix: an error was signaled at startup if the HOME environment
942 variable was defined, but had an empty value (reported by Peter Van Eynde)
943 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
944 signal a PROGRAM-ERROR, not a BUG.
945 * bug fix: SB-POSIX:READDIR works when built with large file support.
946 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
948 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
949 for better automatic indentation support. (thanks to Matt Pillsbury)
950 * optimization: loading generic functions no longer takes O(n^2) time,
951 proportional to the amount of methods in the generic function
952 (reported by Todd Sabin and Jeremy Brown)
953 * optimization: the FIND and POSITION family of sequence functions
954 are significantly faster on arrays whose element types have been
956 * improvements to the Windows port:
957 ** Intermittent heap corruption problems have been fixed. (thanks
958 to Alastair Bridgewater)
959 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
961 ** Lisp is able to unwind foreign exception frames from alien
962 callbacks. (thanks to Alastair Bridgewater)
964 changes in sbcl-1.0 relative to sbcl-0.9.18:
965 * improvement: experimental support for threading on FreeBSD/x86.
966 (thanks to NIIMI Satoshi)
967 * improvement: runtime option --dynamic-space-size can be used
968 to set the size of the dynamic space reserved on startup.
969 * improvement: floating point modes in effect are now saved in
970 core, and restored on startup.
971 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
972 startup, not time since first call to GET-INTERNAL-REAL-TIME.
973 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
974 threads are not running after *SAVE-HOOKS* have run.
975 * improvement: writes to CLOS instance slots are type-checked in code
976 compiled with (SAFETY 3)
977 * improvement: floating-point exception handling on FreeBSD (thanks to
979 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
980 (thanks to Zach Beane)
981 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
983 * improvement: added support for the Shift-JIS external format.
984 (contributed by NIIMI Satoshi)
985 * improvement: callbacks are supported on Linux/PPC. (thanks to
987 * bug fix: compiler bug triggered by a (non-standard) VALUES
988 declaration in a LET* was fixed. (reported by Kaersten Poeck)
989 * bug fix: file compiler no longer confuses validated and already
990 dumped structurres. (reported by Kaersten Poeck)
991 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
992 fill-pointer signals a type-error as required. (thanks to
994 * bug fix: disassemly of funcallable instances works.
995 * bug fix: single stepping on PPC.
996 * bug fix: fix thread-safety problems in the type system (generally
997 manifesting as nonsensical errors like "STRING is a bad type specifier
998 for sequences" or "The value 1 is not of type FIXNUM")
999 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
1000 (reported by Josip Gracin)
1001 * bug fix: an error is signaled for attempts to displace arrays with
1002 incompatible element types (thanks to Mario Mommer)
1003 * bug fix: more correct handling of wide characters in debug info
1004 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1005 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1006 and don't cause extra consing
1007 * optimization: MAP and MAP-INTO are significantly faster on vectors
1008 whose elements types have been declared.
1009 * Improvements to SB-SPROF:
1010 ** Support for allocation profiling
1011 ** Reduced profiling overhead, especially for long profiling runs
1012 * Improvements to the Windows port:
1013 ** floating point exceptions are now reported correctly.
1014 ** stack exhaustion detection works partially.
1015 ** more accurate GET-INTERNAL-REAL-TIME.
1016 ** better RUN-PROGRAM behaviour in case of errors in the child
1018 ** PROBE-FILE now simplifies pathnames correctly.
1019 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1020 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1021 1st 1970 (although time timezone data will be unreliable).
1022 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1024 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1025 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1026 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1028 * bug fix: two potential GC deadlocks affecting threaded builds.
1029 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1030 returns T (reported by Anton Kazennikov)
1031 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1032 with non-variable places
1033 * bug fix: remove a race condition in the setting of
1034 funcallable-instance functions, this should make threaded CLOS
1035 code more stable against memory faults.
1036 * bug fix: corruption of specials when unbinding is interrupted by an
1037 asynchronous unwind (reported by Hannu Koivisto)
1038 * improvement: the debugger will now also display local variables that
1039 are only used once, for code compiled with a DEBUG optimization quality
1042 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1043 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1044 * incompatible change: External-format support for FFI calls. The
1045 SB-ALIEN:C-STRING no longer implies an ASCII
1046 external-format. Instead the string is subject to external-format
1047 conversion. Additionally return values of type SB-ALIEN:C-STRING
1048 are by default of type (SIMPLE-ARRAY CHARACTER), not
1049 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1050 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1051 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1052 * incompatible change: SB-EXT package no longer contains the
1053 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1054 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1056 * incompatible change: the single-stepper is no longer available
1057 on Alpha, Mips and Sparc platforms.
1058 * minor incompatible change: the direct superclasses of
1059 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1060 not (STANDARD-OBJECT FUNCTION). This makes the
1061 class-precedence-lists of GENERIC-FUNCTION and
1062 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1064 * minor incompatible change: the default stream external format on
1065 non-unicode SBCL is no longer detected from the locale, but is always
1067 * new feature: Add a version of evaluator that uses an interpreter instead
1068 of the compiler. EVAL still uses the compiler by default, to switch it
1069 to use the interpreter, set the value of the variable
1070 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1071 * minor incompatible change: the single-stepper REPL has been merged
1072 with the normal debugger (see the "Stepping" heading of the debugger help
1073 for more details). The debugger command STEP will no longer switch
1074 to the single-stepper REPL.
1075 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1076 for a type now works.
1077 * bug fix: loading of default sysinit file works. (thanks to Leonid
1079 * bug fix: better detection of circularities in the file-compiler.
1080 (reported by Marco Monteiro)
1081 * bug fix: the CL pathname functions now work with files that have
1082 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1083 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1084 systems with Unix98 pty semantics.
1085 * bug fix: ASDF-INSTALL will now work with bsd tar.
1086 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1088 * bug fix: timers expiring in dead threads no longer cause a
1089 type-error (reported by Paul "Nonny Mouse").
1090 * bug fix: thanks to more lightweight single-stepper instrumentation,
1091 code compiled with (DEBUG 3) will compile and execute significantly faster,
1092 and will have more accurate type-inferencing than before
1093 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1094 whose bindings are modified
1095 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1096 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1097 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1098 CL:SOFTWARE-VERSION with proper version
1100 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1101 * feature: implemented the READER-METHOD-CLASS and
1102 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1103 as specified by AMOP.
1104 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1106 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1107 (thanks to Lutz Euler)
1108 * optimization: hashing of general arrays and vectors has been
1109 improved. (reported by Any Fingerhut)
1110 * enhancement: SB-INTROSPECT is now able to find definitions of
1111 profiled functions. (thanks to Troels Henriksen)
1112 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1113 as well, allowing compiler-macros for SETF-functions to expand.
1114 * enhancement: step-instrumentation no longer wraps known
1115 single-value functions with multiple-value context, allowing
1116 better type inference.
1117 * fixed bug #337: use of MAKE-METHOD in method combination now works
1118 even in the presence of user-defined method classes. (reported by
1119 Bruno Haible and Pascal Costanza)
1120 * fixed bug #339(c): if there are applicable methods not part of any
1121 long-form method-combination group, call INVALID-METHOD-ERROR.
1122 (reported by Bruno Haible)
1123 * fixed bug #361: the :FUNCTION initarg in the protocol for
1124 initialization of methods can now be used to override
1125 internally-produced optimized functions. (reported by Bruno
1127 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1128 system-provided lambda expression no longer cause warnings about
1129 unbound #:|pv-table| symbols.
1130 * bug fix: improved the handling of type declarations and the
1131 detection of violations for keyword arguments with non-constant
1133 * bug fix: potentially erronous calls to PATHNAME and
1134 MERGE-PATHNAMES were being flushed in some cases.
1135 (reported by Richard Kreuter)
1136 * bug fix: compiled calls to TYPEP were mishandling obsolete
1137 instances. (reported by James Bielman and Attila Lendvai)
1138 * bug fix: format strings with ~> without matching ~< no longer
1139 trigger an AVER, but signal an understandable error instead.
1140 (reported by Antonio Martinez)
1141 * bug fix: specifying an output-file without a pathname-type for
1142 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1143 of input-file instead of "fasl". (reported by Robert Dodier)
1144 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1145 (reported by James Y Knight).
1146 * bug fix: compiler-macros-function did not consider the environment
1147 argument for shadowing by local functions.
1148 * bug fix: compiler-macros expansion was inhibited by local INLINE
1150 * bug fix: inline expansions of known functions were subject to
1151 step-instrumentation in high debug policies, leading to problems
1152 with type-inference.
1153 * bug fix: compiler failed to differentiate between different CONS
1154 types in some cases.
1155 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1156 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1157 * bug fix: #368: incorrect use of expressed vs. upgraded array
1159 * bug fix: #306a: more precise unions of array types.
1160 * thread-safety improvements:
1161 ** CONDITION-WAIT could return early on Linux, if the thread was
1162 interrupted and subsequently continued with SIGCONT.
1163 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1165 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1166 * added support for the ucs-2 external format. (contributed by Ivan
1168 * minor incompatible change: pretty printing of objects of type
1169 (cons symbol) is, in the default pprint-dispatch-table, now
1170 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1172 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1173 called later in a class's lifetime, possibly as late as when the
1174 first instance of the class is created. Previously,
1175 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1176 class became finalizeable.
1177 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1178 for multibyte external-format streams. (thanks to "vbzoli")
1179 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1180 with :ALLOCATION :CLASS inherited from superclasses of the
1182 * fixed bug: anonymous classes can now be created using the :NAME
1183 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1184 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1185 * fixed bug: core-files saved with :EXECUTABLE T can again be
1186 executed when SBCL_HOME isn't set. (reported by James Knight)
1187 * fixed bug: toplevel LOCALLY forms with declarations could
1188 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1189 * fixed bug: printing from several different threads using different
1190 values of *print-case* could cause invalid output, due to
1191 some internal special variables of the printer not being bound
1192 thread-locally (reported by Max Mikhanosha)
1193 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1194 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1195 (reported by Pascal Costanza)
1196 * fixed bug: FINALIZE-INHERITANCE is called from
1197 REINITIALIZE-INSTANCE on classes when the class has previously
1198 been finalized, as required by AMOP.
1199 * minor code generation optimizations:
1200 ** better register allocation in CLOS dispatching functions
1201 ** overflow detection when coercing signed bytes to fixnums on x86-64
1202 ** is now implemented with one IMUL instruction instead of three shifts
1203 ** more efficient bit-vector access on x86 and x86-64
1204 ** more efficient access to raw structure slots on x86 and x86-64
1205 * fixed some bugs revealed by Paul Dietz' test suite:
1206 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1207 return its argument.
1209 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1210 * feature: thread support on Solaris/x86, and experimental thread support
1212 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1214 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1215 default method for SLOT-UNBOUND.
1216 * minor incompatible change: prevent the user from specializing the
1217 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1218 somewhat counter to the intent of the protocol, I (CSR) think, and
1219 additionally it just doesn't work in SBCL as currently
1220 implemented, thanks to optimizations (that are always valid for
1221 the other three SLOT-VALUEish functions, but not for the setter).
1222 * minor incompatible change: the :SB-LDB feature is now enabled by
1223 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1224 the low-level debugger.
1225 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1226 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1227 down to an SB-SYS:FD-STREAM, allowing direct communication in
1228 more cases. Temporary files and pipes are still used as a fallback
1230 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1231 on the non-reentrant C functions localtime(3) and gmtime(3).
1232 * bug fix: global optimization policy was not visible in LOCALLY and
1234 * bug fix: class objects can be used as specializers in methods.
1235 (reported by Pascal Costanza)
1236 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1237 with a type error. (reported by blitz_ on #lisp)
1238 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1239 failed with a type error. (reported by Pascal Bourguignon)
1240 * bug fix: merging pathnames against defaults with :DIRECTORY
1241 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1242 (reported by James Y Knight)
1243 * bug fix: saving large (>2GB) cores on x86-64 now works
1244 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1245 constant index and a (simple-array (signed-byte 32)) array
1246 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1247 error (patch by Robert J. Macomber)
1248 * feature: TIME now displays the amount of run-time spent in GC
1249 * bug fix: The debugger now does a better job of respecting
1250 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1251 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1252 workaround for bug 403.)
1253 * bug fix: floating point exception handling now works on FreeBSD
1254 (thanks to NIIMI Satoshi)
1255 * fixed some bugs revealed by Paul Dietz' test suite:
1256 ** MISC.641: LET-conversion were not supposed to work in late
1258 * improvements to the Win32/x86 port:
1259 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1260 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1262 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1263 * new feature: source path information is generated for macro-expansion
1264 errors for use in IDE's like Slime (thanks to Helmut Eller)
1265 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1266 * bug fix: compiler does not loop forever on an invalid type in
1268 * improvement: compilation of most CLOS applications is significantly
1270 * optimization: added a limited bytecode compiler for simple toplevel
1271 forms, speeding up compilation and FASL loading
1272 * bug fix: the statistical profiler now properly distinguishes anonymous
1275 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1276 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1277 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1278 into subclasses of the finalized class. Now user calls to
1279 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1280 system return before any subclasses are finalized.
1281 * minor incompatible change: The reader no longer ignores errors
1282 regarding non-existent packages in #+ and #- feature tests.
1283 * new feature: command line options --no-sysinit, --no-userinit to
1284 inhibit loading the corresponding init files
1285 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1286 for sending data through UDP sockets (thanks to François-René Rideau)
1287 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1288 error is signalled instead (thanks to François-René Rideau)
1289 * minor incompatible change: Error signalling behaviour of lexical
1290 operations violating package locks has changed slightly. Refer to
1291 documentation on package locks for details.
1292 * enhancement: EVAL can process IF-expressions without resorting to the
1294 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1295 constant-folded properly due to missing out-of-line functions.
1296 (reported by Mika Pihlajamäki)
1297 * bug fix: LISTEN sometimes returned T even in cases where no data was
1298 immediately available from the stream
1299 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1300 were reversed. (reported by Levente Mészáros)
1301 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1302 when given a vector displaced to an adjustable array. (reported
1304 * bug fix: derivation of float boundaries from numbers outside the
1305 appropriate float range (reported by John Wiseman)
1306 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1308 * bug fix: precomputing cache entries for generic functions with
1309 some subclasses of specializers as yet invalid does not attempt to
1310 fill a cache line with a negative offset. (reported by Levente
1312 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1313 allow condition class objects as arguments to DOCUMENTATION and
1314 (SETF DOCUMENTATION); only find and set documentation for
1315 structure names for the STRUCTURE doc-type. (suggested by Gary
1317 * improvements to the Win32/x86 port:
1318 ** (user-homedir-pathname) and default initialization file
1319 locations now know about the user's "Documents and Settings"
1320 directory (thanks to Yaroslav Kavenchuk)
1321 ** run-program is implemented (thanks to Mike Thomas)
1322 ** sockets support (thanks to Timothy Ritchey)
1323 ** better backtrace support (thanks to Alastair Bridgewater)
1324 ** sb-grovel supported
1325 ** asdf-install and sb-posix work somewhat
1326 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1327 * improvements to the Solaris/x86 port:
1328 ** works on Solaris 11/Solaris Express
1329 ** floating-point exception handling support
1330 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1331 * fixed some bugs revealed by Paul Dietz' test suite:
1332 ** REMOVE-METHOD returns its generic function argument even when
1333 the method is not one of the generic functions' methods.
1334 ** objects declared as MEMBER types can be admissible arguments to
1335 structure accessors.
1336 ** printing characters should simply be printed by the FORMAT ~:C
1338 ** compiler failure when compiling functions with hairy constant
1339 defaults for optional parameters.
1340 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1341 function, which is already optimized.
1343 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1344 * new platform: experimental support for SBCL x86/Darwin, including
1345 MacOS X 10.4.5 on Intel.
1346 * new feature: Unicode character names are now known to the system
1347 (through CHAR-NAME and NAME-CHAR).
1348 * new feature: the filesystem location of SBCL's core file is
1349 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1350 * minor incompatible change: the contrib modules SB-POSIX and
1351 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1352 this change is to make it easier to distribute
1353 location-independent binaries.
1354 * bug fix: as implied by AMOP, standardized classes no longer have
1355 slots named by external symbols of public packages. (reported by
1357 * optimization: calling functions via constant symbols -- as in
1358 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1359 via the function object as in (FUNCALL #'FOO).
1360 * optimization: on x86 and x86-64, the calling convention for the
1361 case of dealing with an unknown number of values has been altered
1362 to be friendlier to the prediction heuristics implemented,
1363 particularly with reference to CALL and RET pairing. (thanks to
1364 Alastair Bridgewater)
1365 * optimization: on x86, the code for access to thread-local symbol
1366 values and binding/unbinding of thread-local symbols is smaller.
1367 (thanks to Alastair Bridgewater)
1368 * enhancement: CONSTANTP is now able to determine constantness of
1369 more complex forms, including calls to constant-foldable standardized
1370 functions and some special forms beyond QUOTE.
1371 * fixed bug: occasional GC crashes on Solaris/x86
1372 * optimization: x86-64 supports stack allocation of results of simple
1373 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1374 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1375 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1376 (thanks to James Knight)
1377 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1378 by the standard (thanks to Stephen Compall)
1380 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1381 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1382 be used for bundling the runtime and the core file into one
1383 executable binary. This feature is not currently supported on all SBCL
1384 platforms. (thanks to James Bielman and NIIMI Satoshi)
1385 * new feature: a generational or ephemeral garbage collector is now
1386 the default on the PowerPC platform (both Linux and Darwin). The
1387 old Cheney (stop and copy) collector is a build-time option.
1388 (thanks to Cyrus Harmon, after Raymond Toy)
1389 * minor incompatible change: the method by which SBCL finds its
1390 contributed modules has changed; it no longer relies on symbolic
1391 links from an $SBCL_HOME/systems directory, but searches directly
1392 in the subdirectories of $SBCL_HOME.
1393 * enhancement: the dynamic heap size on the Linux/PPC platform is
1394 markedly larger, even using the older Cheney garbage collector.
1395 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1396 certain complicated string types. (reported by Gary King)
1397 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1398 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1399 * fixed bug: DOCUMENTATION on structures no longer signals an
1400 error if no structure type of the right name exists. (reported by
1402 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1403 applicable non-standard primary or :AROUND method on
1404 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1405 no longer cause unbound variable errors. (reported by Kevin Reid)
1406 * fixed bug: metacircle resolution in cases where methods have slots
1407 added before the slots from STANDARD-METHOD. (reported by Jean
1409 * fixed bug: the Power PC assembly code for calling into lisp
1410 sometimes computed its return address wrongly (depending on the
1411 15th bit of the address, fixed at link time). This used to
1412 manifest itself in a segmentation violation while building PCL.
1413 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1414 many others over the years)
1415 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1416 the stack pointer (treated as a signed integer) is negative.
1417 (thanks to Peter van Eynde)
1419 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1420 * new platform: experimental support for the Windows operating
1421 system has been added. (thanks to Alastair Bridgewater)
1422 * fixed several bugs in and robustified the PPC FFI (including
1423 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1424 * bug fix: allow non-simple string symbol names (reported by Paul
1426 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1428 * bug fix: saving a core corrupted callbacks on x86/x86-64
1429 * bug fix: closed a loophole in metacircularity detection and
1430 grounding in the PCL implementation of CLOS.
1431 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1432 (reported by tomppa on #lisp)
1433 * bug fix: clear the direction flag on Lisp -> C transitions, as
1434 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1435 (reported by Andrej Grozin and Hendrik Maryns)
1436 * optimization: major improvements to GC efficiency on GENCGC platforms
1437 * optimization: faster implementation of EQUAL
1438 * optimization: emit more efficient opcodes for some common
1439 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1440 * optimization: several other minor code-generation improvements on x86-64
1441 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1443 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1444 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1445 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1446 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1447 by ANSI as a generic function, and for consistency (SETF
1448 GENERIC-FUNCTION-NAME) is treated likewise.
1449 * fixed bug #233.b: make constraint propagation notice when a variable
1450 value is changed after it is referenced but before it is used
1451 * fixed bug #296: no more arbitrary behaviour from filenames with
1452 odd characters as --load arguments. (reported by Adam Warner)
1453 * fixed bug #390: :CHARACTER-SET pathname components now work as
1454 expected. (reported by Tim Daly Jr)
1455 * fixed bug #391: complicated :TYPE intersections in slot
1456 definitions no longer cause an error in PCL internals.
1457 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1458 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1459 returning the number of octets which would be written to the
1460 file-stream. (thanks to Robert J. Macomber)
1461 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1462 arguments to the internal stream functions. (thanks to David
1464 * bug fix: stack exhaustion is now handled on the x86/SunOS
1465 platform. (thanks to Daisuke Homma)
1466 * bug fix: more accurate ROOM results on GENCGC platforms
1467 * optimization: improved type inference for arithmetic-for
1468 index variables in LOOP
1469 * optimization: faster floating-point SQRT on x86-64
1470 * fixed some bugs revealed by Paul Dietz' test suite:
1471 ** DOCUMENTATION returns NIL instead of "" for method combinations
1472 that don't have a docstring
1474 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1475 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1476 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1477 therefore are not customizeable by user code (as seems to be at
1478 least permitted and maybe required by AMOP). As a consolation,
1479 however, the SBCL implementation of these functions now calls
1480 REINITIALIZE-INSTANCE as specified by AMOP.
1481 * bug fix: slot-definition documentation is propagated as per ANSI
1482 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1483 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1484 argument, as do eslotd initialization methods. (from Pascal
1485 Costanza's "Closer" project)
1486 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1487 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1489 * bug fix: it is now possible to have more than one subclass of
1490 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1491 (reported by Bruno Haible, Pascal Costanza and others)
1492 * bug fix: the dependent update protocol now works for generic
1493 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1494 and Pascal Costanza)
1495 * bug fix: condition-class instances corresponding to
1496 DEFINE-CONDITION forms are now created eagerly. (reported by
1497 Kalle Olavi Niemitalo on comp.lang.lisp)
1498 * bug fix: floating point printing is more accurate in some
1499 circumstances. (thanks to Simon Alexander)
1500 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1501 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1502 * bug fix: callbacks on OS X now preserve stack-alignment.
1503 (thanks to Cyrus Harmon)
1504 * enhancement: the x86-64 disassembler is much better at
1505 disassembling SSE instructions. (thanks to Lutz Euler)
1506 * enhancement: CHECK-TYPE on a variable now allows type-inference
1507 (thanks to Vincent Arkesteijn)
1508 * optimization: improved performance of EUC-JP external format.
1509 (thanks to NIIMI Satoshi)
1510 * optimization: performance improvements to IO on file streams of
1511 :ELEMENT-TYPE CHARACTER
1512 * optimization: much faster memory allocation on x86-64
1513 * optimization: faster garbage collections (latency and throughput)
1515 * optimization: faster bignum subtraction and division on x86 and x86-64
1516 (thanks to Lutz Euler)
1517 * optimization: more accurate type-inference for loops that use a
1518 floating point index variable or a negative step.
1520 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1521 * bug fix: add a workaround to SBCL looping infinitely at startup on
1522 Linux kernels with apparently buggy implementations of personality().
1523 (thanks to Svein Ove Aas)
1524 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1525 (thanks to David Lichteblau)
1526 * bug fix: local bindings shadow global symbol macros for
1527 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1528 * optimization: non-open coded uses of numeric comparison operators
1529 (e.g. >) no longer cons when called with more than one parameter
1530 on platforms supporting dynamic-extent allocation.
1531 * enhancement: saving cores with foreign code loaded is now supported
1532 on MIPS/Linux in addition to the previously supported platforms.
1533 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1534 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1535 * bug fix: division by zero in sb-sprof when no samples were collected
1536 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1537 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1538 (thanks to Alexander Kjeldaas)
1539 * new feature: ignoring the return values of destructive functions
1540 like NREVERSE signals a compile-time style-warning.
1541 (thanks to Kevin Reid)
1542 * new feature: the alignment of alien structure fields can be
1543 explicitly specified. (thanks to Cyrus Harmon)
1544 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1545 and dump core on SIGQUIT
1547 ** incompatible change: threads do not inherit values of specials
1548 from their parents (see manual)
1549 ** bug fix: threads stacks belonging to dead threads are freed by the
1550 next exiting thread, no need to gc to collect thread stacks anymore
1551 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1552 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1553 ** bug fix: dynamic variable and thread start related gc lossage
1554 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1556 * fixed some bugs revealed by Paul Dietz' test suite:
1557 ** SUBTYPEP is slightly more accurate on heinously complicated
1558 CONS types where some of the members have uncertain (in the
1559 NIL, NIL sense) type relationships to each other.
1561 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1562 no longer purified unless :PURIFY T is explicitly specified.
1563 ** Non-purified cores are significantly smaller than before
1565 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1566 * new feature: timers based on Zach Beane's excellent timer package
1567 * added support for the following external formats: koi8-u,
1568 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1569 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1570 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1571 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1572 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1573 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1574 * incompatible change: a threaded SBCL will no longer revert to
1575 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1576 * bug fix: interrupts are disabled until startup is complete; no
1577 more sigsegvs when receiving a signal to soon
1578 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1580 * bug fix: add a workaround for the memory randomization features in
1581 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1582 workaround will only be in effect on systems with the proc filesystem
1584 * bug fix: printing objects of type HASH-TABLE signals a
1585 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1587 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1588 have been running for over 50 days. (reported by Gilbert Baumann)
1589 * bug fix: the logic for getting names of functions gets less
1590 confused when confronded with alternate-metaclass
1591 funcallable-instances. (reported by Cyrus Harmon)
1592 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1593 consistent, even on internal alternate-metaclass objects.
1594 * bug fix: flush closure information collected by physical
1595 environment analysis prepass before the main pass. (bug reported
1597 * bug fix: compiler pack phase does not modify a hashtable, which is
1598 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1599 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1600 may be expanded. (reported by Alexander <malishev> on c.l.l)
1601 * bug fix: no more style-warnings when compiling code using LAST
1602 under high-SPEED low-SPACE settings. (reported by David Wallin
1604 * fixed some incompatibilities between SBCL's MOP and the MOP
1606 ** the METAOBJECT class is now implemented;
1607 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1608 STANDARD-OBJECT, as required;
1609 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1610 now compatible (as required by VALIDATE-SUPERCLASS); there
1611 remains a consistency requirement over the presence or absence
1612 of the FUNCTION class in the superclass list of finalized
1613 classes; see the manual for more details;
1614 ** the :AROUND method for COMPUTE-SLOTS on
1615 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1616 requested slot ordering.
1618 ** bug fix: parent thread now can be gc'ed even with a live
1620 ** bug fix: binding a special with PROGV to no value is not
1621 the same as that symbol not having been bound (thanks to
1623 ** bug fix: binding specials is thread safe (thanks to
1625 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1626 and SB-SPROF do not die with 'no handler for signal XX in
1627 interrupt_handle_now(..)' anymore
1628 ** bug fix: WITH-TIMEOUT works with multiple running threads
1629 * fixed some bugs revealed by Paul Dietz' test suite:
1630 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1631 the :method-class keyword argument.
1633 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1634 * new port: the Solaris operating system on x86 processors is now
1635 mostly supported, though some rough edges in the environment
1636 remain. (thanks to Daisuke Homma)
1637 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1638 space for its heap. (thanks to Thiemo Seufer)
1639 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1640 closures. (thanks to Thiemo Seufer)
1641 * minor incompatible change: eof selects abort in the debugger.
1642 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1643 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1645 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1646 deprecated, and will go away in a future revision of SBCL.
1647 * minor incompatible change: GC-ON and GC-OFF are no longer
1648 implemented with a counter, it does not matter how many times gc
1649 is switched on or off
1650 * bug fix: discriminating functions for generic function classes
1651 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1652 make invalid assumptions about method precedence order. (reported
1654 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1655 platforms. (thanks to Thiemo Seufer)
1656 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1657 * bug fix: correct pathname printing with printer escaping is on.
1658 (thanks to Kevin Reid)
1659 * bug fix: complex VOP definitions in "user-space" no longer trigger
1660 package locks. (reported by Zach Beane)
1661 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1662 for funcallable-instance functions, and is no different from
1664 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1665 characters in the current readtable. (reported by Nicholas Neuss)
1666 * bug fix: SUBTYPEP on various CONS types returns more a more
1667 accurate acknowledgment of its certainty.
1668 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1669 lists in some cases. This partially fixes bug 384.
1670 * flush all standard streams before prompting in the REPL and the
1672 * bug fix: signal handling and triggering gc do not conflict
1673 directly anymore, in particular a high frequency sb-sprof does
1674 not prevent gc from running
1675 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1676 approximation for timezone and DST information between the
1677 universal time epoch and the smallest negative 32-bit time_t.
1678 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1679 year 1899 so long as the encoded time is non-negative.
1680 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1681 the correct answer for non-constant rotate amounts when performing
1682 an inline 32-bit rotation.
1684 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1685 there is only one thread in the session
1686 ** bug fix: memory leak for streams created in one thread and
1687 written to in another
1688 ** bug fix: lockup when compiled with gcc4
1689 ** bug fix: race that allows the gc to be triggered when gc is
1691 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1692 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1694 ** bug fix: finished threads are gc'ed properly
1695 * fixed some bugs revealed by Paul Dietz' test suite:
1696 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1697 has arguments with a different set of applicable methods from
1698 the orignal arguments.
1699 ** The type error thrown by MAP now has a correct expected-type
1701 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1702 name a compiled function.
1703 ** Three MISC tests where a large bignum was improperly coerced to
1704 a float that couldn't represent that bignum during type
1705 derivation were fixed.
1706 ** SUBTYPEP can now handle types involving the negation of a
1707 list-form FUNCTION type.
1708 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1709 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1710 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1712 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1713 * New feature: Experimental support for bivalent streams: streams
1714 opened with :element-type :default now allow character and binary
1715 (unsigned-byte 8) I/O
1716 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1717 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1719 * bug fix: no more highly sporadic "couldn't check whether ~S is
1720 readable" when reading a stream and an interrupt hits in the middle
1721 of a select system call
1722 * compiler better recognizes complex arrays (reported by Eduardo
1724 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1725 for variables on SBCL built with linkage-tables. (reported by Luis
1727 * various error reporting improvements.
1728 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1729 (Thanks to Thiemo Seufer)
1730 * bug fix: nested reader invokations work correctly
1731 * bug fix: it is possible to have simultaneous references to foreign
1732 code and foreign data with the same name.
1734 ** added x86-64 support
1735 ** incompatible change: the threading api now works with thread
1736 objects instead of thread ids
1737 ** bug fix: threads are protected from signals and interruption when
1738 starting up or going down
1739 ** bug fix: a race where an exiting thread could lose its stack to gc
1740 ** bug fix: don't halt on infinite error in threads if possible
1741 ** fixed numerous gc deadlocks introduced in the pthread merge
1742 ** bug fix: fixed thread safety issues in read and print
1743 ** bug fix: debugger doesn't hang on session lock if interrupted at
1744 an inappropriate moment
1745 ** bug fix: run-program is now thread safe(r)
1746 ** bug fix: inner with-recursive-lock no longer releases the mutex
1747 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1748 * fixed some bugs revealed by Paul Dietz' test suite:
1749 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1750 EXPECTED-TYPE slots filled.
1751 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1753 ** MAKE-CONDITION accepts classes as type-designators.
1754 ** COMPILE may never return NIL.
1755 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1756 range before calling Unix time functions
1758 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1759 * numerous signal handling fixes to increase stability
1760 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1761 * minor incompatible change: we now correctly canonize default
1762 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1763 as per the MOP, rather than the historical (INITARG INITFUNCTION
1764 INITFORM). (reported by Bruno Haible)
1765 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1766 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1767 for more information.
1768 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1769 pathname is a directory pathname.
1770 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1771 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1773 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1774 :METACLASS argument in addition to a class name. (reported by
1775 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1776 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1777 (reported by Bruno Haible; patch from Gerd Moellmann)
1778 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1780 * bug fix: more cleanups to the floating point exception handling on
1781 x86-64 (thanks to James Knight)
1782 * bug fix: the compiler does not try to constant fold calls of
1783 COERCE and BIT-* functions when they return freshly constructed
1784 objects. (reported by Nikodemus Siivola and Paul Dietz)
1785 * optimization: improved the MIPS versions of generic (in the
1786 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1787 * optimization: direct conversion of (unsigned-byte 32) to floats on
1788 the PowerPC platform.
1789 * optimization: structure instances with raw slots now use less
1790 memory, and probably show better memory locality. (thanks to
1792 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1793 are treated as requests for stack allocation on the x86-64,
1794 Alpha-32, PPC and SPARC platforms.
1795 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1796 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1797 (thanks to Tony Martinez)
1799 ** gcing a dead thread can no longer lead to lockups
1800 ** threads block signals until they are set up properly
1801 ** errno is no longer shared by threads
1802 ** interrupt-thread restores the eflags register on x86
1803 ** fixed some lockups due to gc/thread interaction
1804 * fixed some bugs revealed by Paul Dietz' test suite:
1805 ** invalid dotted lists no longer raise a read error when
1806 *READ-SUPPRESS* is T
1807 ** signal an error if a symbol that names a declaration is used
1808 as the name of a type, or vice versa
1809 ** allow using the (declare (typespec var*)) abbreviation for
1810 (declare (type typespec var*)) with all type specifiers
1811 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1812 ** always return NIL from PPRINT-POP when OBJECT is NIL
1813 ** don't signal errors when pretty-printing malformed LABELS,
1814 FLET or MACROLET forms
1815 ** declarations in a DOLIST body are also in scope for the
1817 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1819 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1822 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1823 * fixed cross-compiler leakages that prevented building a 32-bit
1824 target with a 64-bit host compiler.
1825 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1826 opened with :IF-EXISTS :OVERWRITE.
1827 * fixed bug 281: error for an invalid qualifier in a short-form method
1828 combination method is not signalled until the faulty method is called.
1829 * bug fix: iteration variable type inferrer failed to deal with open
1830 intervals. (reported by Alan Shields)
1831 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1832 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1833 or more runtime options were provided to the sbcl binary.
1834 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1835 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1836 * optimization: there's now a fast-path for fixnum arguments in the
1837 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1838 * optimization: the code generated on x86-64 is more compact thanks
1839 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1840 * bug fix: floating-point exception handling now partly works on
1841 x86-64. (Thanks to James Knight)
1842 * improvement to the MIPS backend from Thiemo Seufer: C-style
1843 64-bit long long arguments and return values to alien functions
1845 * fixed some bugs revealed by Paul Dietz' test suite:
1846 ** the type-error signalled from WARN has a filled-in DATUM slot.
1847 ** the type-error required when a stream is not associated with
1848 a file has the stream as its datum.
1849 ** type-errors on single-floats on x86-64 no longer have
1850 :INVALID-OBJECT as the datum
1851 ** the type-errors signalled for invalid function names now have
1852 a correct expected type
1853 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1854 for structures defined with a :TYPE. Documentation strings for
1855 typed structures are no longer immediately discarded
1856 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1857 broadcast streams and synonym streams. FILE-LENGTH now also works
1858 on broadcast streams.
1860 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1861 * incompatible change: the --noprogrammer option, deprecated since
1862 version 0.7.5, has been removed. Please use the equivalent
1863 --disable-debugger option instead.
1864 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1866 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1867 inoperational for a while now) has been completely removed.
1868 * null lexical environments are now printed as #<NULL-LEXENV>,
1869 significantly reducing the amount of clutter in typical backtraces.
1870 * documentation on weak pointers, finalization, and after GC hooks
1871 has been added to the manual.
1872 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1873 as other specialized array types, is much faster. SUBSEQ and
1874 COPY-SEQ on such arrays have also been sped up.
1875 * optimization: EQL is now more efficient when at least other argument
1876 is known to be of type (OR FIXNUM (NOT NUMBER)).
1877 * fixed bug: the runtime is now less vulnerable to changes in the
1878 size of the SBCL object on OS X, and virtual memory is reserved for
1879 all spaces that need to be at a fixed address.
1880 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1881 * fixed bug: finalizers and after GC hooks that cause consing are now
1883 * fixed bug: compiler error messages and summaries are now printed to
1884 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1885 * fixed inference of the upper bound of an iteration variable.
1886 (reported by Rajat Datta).
1887 * fixed bug 211e: calling local functions with duplicated constant
1888 keyword argument no longer causes a bogus style warning about an
1890 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1891 ftype declaration to be disregarded. (reported by Dave Roberts)
1892 * fixed bug 373: caused by erronous compilation of references to alien
1893 variables in the runtime on ppc/darwin.
1894 * fixed bug 376: CONJUGATE type deriver.
1895 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1896 (reported by Baughn on #lisp)
1897 * a cross-compiler bug on non-x86 platforms has been identified and
1898 fixed. (thanks to Bruno Haible)
1899 * improvements to the MIPS runtime code for increased stability.
1900 (thanks to Thiemo Seufer)
1901 * increased the maximimum compact environment size to allow
1902 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1903 * partial workaround for bug 135: don't name the function we're
1904 calling for hairy cases of EVAL, so as not to accumulate one environment
1905 entry per call to EVAL. (reported by Kevin Reid)
1906 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1907 * optimization: type testing for non-vector arrays should be faster.
1908 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1910 * bug fix: setting 31st element of a bit vector to zero did not work
1912 * bug fix: redefining a class definition which failed due to a
1913 previous accessor / function clash now works (but see BUGS entry
1914 #380 for more problems in this area). (thanks to Zach Beane)
1915 * the long form of DEFINE-METHOD-COMBINATION disables method group
1916 checking when given a single method group with pattern *. (thanks
1918 * on x86 compiler supports stack allocation of results of simple
1919 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1920 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1921 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1922 * fixed some bugs related to Unicode integration:
1923 ** the restarts for recovering from input and output encoding
1924 errors only appear when there is in fact such an error to
1926 * fixed some bugs revealed by Paul Dietz' test suite:
1927 ** MISC.549 and similar: late transformation of unsafe type
1928 assertions into derived types caused unexpected code
1930 ** SCALE-FLOAT type deriver is less wrong.
1931 ** type derivers for EXP, LOG and similar functions compute result
1932 types for complex arguments better.
1933 ** (MISC.563) CONJUGATE type deriver works for very restricted
1935 ** out-of-line type testers for character strings are available.
1936 ** EQUAL compiler transform understands specialness of objects
1938 ** accessing double-floats stored on the stack now works on x86-64.
1939 ** debugger internals could sometimes create invalid lispobjs,
1940 resulting in GC crashes.
1941 ** MISC.548: type check weakening can convert required type into
1943 ** initialization forms for bindings are not in scope of free special
1946 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1947 * incompatible change: thread support for non-NPTL systems has
1948 been removed - locking is buggy and unreliable. A threaded
1949 SBCL build will now warn at startup and refuse to create
1950 new threads, unless futex support is detected in the kernel
1951 * incompatible change: the top level REPL now has only an ABORT
1952 restart associated with it, not TOPLEVEL and ABORT as it used to.
1953 TOP and TOPLEVEL are now available as debugger commands for
1954 returning to the top level.
1955 * incompatible change: forms evaluated in the REPL now use the
1956 global optimization policy.
1957 * incompatible change: user- and system-initialization files are
1958 no longer processed with LOAD, but by READ and EVAL; hence the
1959 global optimization policy, startup package, readtable, etc,
1961 * The .fasl file format number has been incremented because of
1962 various incompatible changes.
1963 * internal entry point details and argument counts no longer appear
1964 in backtraces unless explicitly requested by setting
1965 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1966 * built-in and standard functions no longer have names like "top
1967 level local call to FOO".
1968 * fixed bug 32: functions defined in non-null lexical environments
1969 now have more legible printed representation
1970 * fixed bug 33: functions defined in non-null lexical environemnts
1971 are now more amenable to inspection by INSPECT.
1972 * workaround for bug 354: XEPs no longer appear in backtraces unless
1973 explicitly requested.
1974 * fixed bug: receiving the signal which results from attempting to
1975 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1976 SIGBUS on other *BSDs) on architectures where the C stack is also
1977 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1978 to debugger confusion or wild instability. (reported by Cheuksan
1980 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1981 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1982 (reported by Lutz Euler)
1983 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1984 compiled code. (reported by Paul Dietz)
1985 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1986 specialized parameter to a method to something that is not TYPEP
1987 the specializer is now possible.
1988 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1989 face of package deletion.
1990 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1991 pathname. (reported by David Tolpin on c.l.l)
1992 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1993 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1994 STRING-STREAM and FILE-STREAM, respectively.
1995 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1996 than previously; in addition, its test suite is now run on build.
1997 (thanks to Luke Gorrie)
1998 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1999 * optimization: added a immediate representation for single-floats
2001 * fixed some bugs revealed by Paul Dietz' test suite:
2002 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2003 ** attempting to create a package with a colliding nickname causes
2004 correctable errors to be signalled.
2005 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2006 ** several x86-64 backend bugs related to sign-extension of immediate
2009 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2010 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2011 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2013 * fixed loading of multiply forward-referenced layouts.
2014 (thanks to Cheuksan Wang)
2015 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2016 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2017 (see COMPILE-FILE documentation for details of :PRINT option)
2018 * fixed bugs 19 and 317: fixed-format floating point printing is
2019 more accurate. This also fixes a bug reported by Adam Warner
2020 related to the ~@F format directive.
2021 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2023 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2024 dispatch table if the from-char is a dispatch macro character.
2025 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2026 by the wordsize no longer ignore the last word. (reported by Lutz
2028 * fixed bug in type checking of dynamic-extent variables. (reported
2030 * optimization: sequence traversal functions use their freedom to
2031 coerce function designators to functions.
2032 * optimization: code with many calls to CLOS methods specialized on
2033 CLOS classes has had redundant type checks removed.
2034 * optimization: added declarations to speed up operations that access
2035 the internal character database (for example STRING-UPCASE)
2036 * optimization: comparison operations between floats and sufficiently small
2037 fixnums no longer create extra rationals
2038 * fixed some bugs related to Unicode integration:
2039 ** portions of multibyte characters at the end of buffers for
2040 character-based file input are correctly transferred to the
2041 start of the buffer at the next read.
2042 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2043 passing it through to OPEN.
2044 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2045 argument given to internal calls to OPEN.
2046 ** the built-in comment readers (introduced by character sequences
2047 ";" and "#|") are more forgiving to encoding errors; they will
2048 STYLE-WARN and then attempt to resync the stream at a character
2049 boundary. (thanks to Teemu Kalvas)
2050 * fixed some bugs in the x86-64 port:
2051 ** Negative short int return values from c-calls are sign-extended
2053 ** The stack is aligned to 16-bytes for c-calls, as required by
2054 the ABI. (thanks to Cheuksan Wang)
2055 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2056 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2057 for enums. (thanks to Vincent Arkesteijn)
2058 ** Multiple small optimizations and bugfixes for floating point
2060 * fixed some bugs revealed by Paul Dietz' test suite:
2061 ** Space, Tab, Linefeed, Return and Page have the invalid
2062 secondary constituent character trait.
2063 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2065 ** WITH-INPUT-FROM-STRING should only update the index place on
2067 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2068 works more reliably.
2069 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2070 with one-dimensional arrays with fill pointers.
2071 ** TYPE-OF failed on a complex with an integer realpart and a
2073 ** compiler failure during type inference for the code of form
2074 (IF (EQL X (THE ...)) ...) (MISC.535).
2076 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2077 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2078 platform. The port supports 61-bit fixnums, large memory spaces
2079 and reloading shared object files.
2080 * enhancement: saving cores with foreign code loaded is now
2081 supported on ppc/Darwin in addition to the previously supported
2083 * enhancement: the statistical profiler now walks deeper into the
2084 call stack for more meaningful call-graphs and accrued time
2085 reports (x86/x86-64 only). It also now reports time spent in
2087 * enhancement: it is now possible to trace most individual methods
2088 of a generic function in addition to tracing the generic function
2090 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2092 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2093 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2094 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2095 produces an error. (thanks to Vincent Arkesteijn)
2096 * bug fix: NAMESTRING on pathnames with :WILD components in their
2097 directories works correctly. (thanks to Artem V. Andreev)
2098 * fixed bug 125: compiler preserves identity of closures. (reported
2100 * bug fixed: functions with &REST arguments sometimes failed with
2101 "Undefined function" when compiled with (DEBUG 3). (reported by
2103 * bug fix: overflow during compiling of setting element of a bit
2104 vector with constant index and value. (reported by Timmy Douglas)
2105 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2106 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2107 * on x86 compiler partially supports stack allocation of dynamic-extent
2109 * GO and RETURN-FROM do not check the extent of their exit points
2110 when compiled with SAFETY 0.
2111 * fixed some bugs related to Unicode integration:
2112 ** encoding and decoding errors are now much more robustly
2113 handled; it should now be possible to recover even from invalid
2114 input or output to the terminal. (thanks to Teemu Kalvas)
2115 ** provided a first cut at implementing STRING-TO-OCTETS and
2116 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2117 ** altered the SB-MD5 contributed module slightly, changing the
2118 interface just enough for it to be supportable for builds where
2119 lisp characters are not eight bits.
2120 * fixed some bugs revealed by Paul Dietz' test suite:
2121 ** the FORMATTER-generated functions for ~V[ conditionals require
2122 the correct number of arguments.
2123 ** READ-FROM-STRING returns the mandated second value when applied
2124 to displaced strings.
2125 ** the #\Rubout and #\Backspace characters are treated as invalid
2126 constituent characters by the tokenizer.
2128 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2129 * new feature: reloading changed shared object files with
2130 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2131 * new feature: references to foreign variables and functions
2132 can now be compiled and loaded before the corresponding shared
2133 object file is loaded, as long as the foreign definitions are
2134 available at runtime.
2135 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2136 supported. (thanks to Dan Debertin)
2137 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2138 just keywords. (thanks to Vincent Arkesteijn)
2139 * fixed bug #331: structure-class instances corresponding to
2140 DEFSTRUCT forms are now created eagerly.
2141 * fixed bug #345: backtraces from calls to undefined functions work
2142 on x86 as well. Related bug #61 is now also partially fixed on x86
2143 (backtraces from throws to unknown catch tags.)
2144 * bug fix: lambda-list parsing is now stricter vrt. order and number
2145 of lambda-list keywords.
2146 * bug fix: as specified by AMOP, an error is signalled if a
2147 class-option appears multiple times in a DEFCLASS form. (reported
2149 * bug fix: RANDOM can be compiled when the compiler derives the type
2150 of its numeric argument as a disjoint set of small integers.
2151 (reported by Paul Dietz)
2152 * bug fix: starting a core saved with shared objects loaded when
2153 those objects are not available no longer causes threaded SBCL to
2154 hang. (reported by Sean Ross)
2155 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2157 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2158 argument, which must always be NIL. (reported by Kalle Niemitalo)
2159 * bug fix: printing 1.0d+23 no longer results in an error.
2160 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2161 * bug fix: structure slot setters preserve evaluation order. (thanks
2163 * fixed some bugs related to Unicode integration:
2164 ** RUN-PROGRAM can allow its child to take input from a Lisp
2165 stream. (reported by Stefan Scholl)
2166 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2167 locales. (reported by Ken Causey)
2168 * fixed some bugs revealed by Paul Dietz' test suite:
2169 ** INCF, DECF and REMF evaluate their place form as specified in
2171 ** FORMATTER expands ~{ ~} iteration directives with V or #
2172 parameters correctly.
2173 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2174 consequent uses no arguments correctly.
2175 ** the system has a partial understanding of the (COMPLEX RATIO)
2177 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2180 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2181 * new feature: a build-time option (controlled by the :SB-UNICODE
2182 keyword feature, enabled by default) for building the system with
2183 support for the entire 21-bit character space defined by the
2185 * new feature: the system now has rudimentary external-format
2186 support; the primary user-visible change at this time is that
2187 characters with the high bit set (such as lower-case-e-acute) will
2188 print correctly to a terminal in a UTF-8 environment.
2189 * minor incompatible change: BASE-CHAR no longer names a class;
2190 however, CHARACTER continues to do so, as required by ANSI.
2191 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2192 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2193 has been moved to the SB-EXT package (temporarily re-exported from
2195 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2196 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2198 * fixed bug #351: better error-handlind and reporting for malformed
2200 * fixed bug #350: bignum-printing is now more memory-efficient,
2201 allowing printing of very large bignums, eg. (expt 2 10000000).
2202 (reported by Bruno Haible)
2203 * fixed bug #302: better primitive-type selection for intersection
2205 * fixed bug #308: non-graphic characters now all have names, as
2206 required. (reported by Bruno Haible)
2207 * bug fix: redefining a class with different superclasses now correctly
2208 removes it from the direct-subclasses of its previous superclasses.
2209 (reported by David Morse)
2210 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2211 new value now works. (reported by Bruno Haible)
2212 * bug fix: correct canonicalization of multiple non-standard slot
2213 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2214 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2215 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2216 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2217 now exists, an signals an error.
2218 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2219 during macroexpansion for non-iteration variables as well. (reported
2220 by Bruno Haible for CMUCL)
2221 * bug fix: Cyclic structures and unprintable objects in compiler
2222 messages no longer cause errors. (reported by Bruno Haible)
2223 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2224 and READ-FROM-STRING all now return a primary value of NIL if
2225 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2226 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2227 (reported by Bruno Haible for CMUCL)
2228 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2229 do not designate a condition. (reported by Bruno Haible for
2231 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2232 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2233 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2234 permitting resolution in favour of any of the conflicting symbols.
2235 (reported by Bruno Haible for CMUCL)
2236 * bug fix: EQUAL compiler optimizations is less aggressive on
2237 strings which can potentially compare true despite having distinct
2238 specialized array element types.
2239 * bug fix: unit enumerations can be defined without dividing by
2240 zero. (thanks to Vincent Arkesteijn)
2241 * FORMAT compile-time argument count checking has been enhanced.
2242 (report from Bruno Haible for CMUCL)
2243 * a partial workaround for the bug 262: the compiler does not try to
2244 inline-expand a local function doing RETURN-FROM from a deleted
2245 BLOCK. (thanks to Peter Denno for the bug report and to David
2246 Wragg for the simple test case)
2247 * fixed some bugs revealed by Paul Dietz' test suite:
2248 ** RENAME-PACKAGE allows all package designators as new package
2250 ** constraint propagation and lambda variable substitution are
2251 more cautious in dealing with partially deleted code.
2252 ** compiler sometimes lost reoptimization passes.
2253 ** CERROR, given a condition as condition designator, passes its
2254 remaining arguments to the continue format control without
2256 ** Case-altering FORMAT directives work correctly on non-ASCII
2258 ** The REAL type specifier handles bounds outside the floating
2259 point ranges without signalling FLOATING-POINT-OVERFLOW.
2260 ** Functions with IR1-transformations can create intercomponent
2261 references to global functions.
2262 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2264 ** FORMAT ~R treats a nil value for its first parameter correctly.
2266 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2267 * enhancement: saving cores with foreign code loaded is now
2268 supported on x86/NetBSD and sparc/Linux in addition to the previously
2269 supported platforms.
2270 * bug fix: on some platforms repeated installations caused multiple
2271 copies of HTML documentation to be installed -- should not happen
2272 any more. (reported by Stefan Scholl)
2273 * bug fix: parsing self-recursive alien record types multiple times
2274 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2275 original patch by Helmut Eller for CMUCL)
2276 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2277 (thanks to Richard Kreuter)
2278 * bug fix: defining classes whose accessors are methods on existing
2279 generic functions in other (locked) packages no longer signals
2280 bogus package lock violations. (reported by François-René Rideau)
2281 * bug fix: special variables as DEFMETHOD parameters no longer have
2282 associated bogus type declarations. (reported by David Wragg and
2284 * bug fix: read-write consistency on streams of element-type
2285 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2286 * bug fix: redefiniton of the only method of a generic function with
2287 no DEFGENERIC no longer emits a full WARNING. In addition,
2288 redefinition of generic functions with no DEFGENERIC to an
2289 incompatible lambda list now signals an error. (thanks to Zach
2291 * bug fix: DEFGENERIC now works even when there's a function of the
2292 same name in an enclosing lexical environment. (thanks to Zach
2294 * fixed compiler failure, caused by instrumenting code during
2295 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2296 * optimization: added loop analysis and improved register allocation
2297 * fixed some bugs revealed by Paul Dietz' test suite:
2298 ** POSITION on displaced vectors with non-zero displacement
2299 returns the right answer.
2300 ** (SIMPLE-STRING) is a valid type specifier for sequence
2302 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2304 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2305 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2307 ** The justification version of the FORMAT ~< directive treats
2308 non-zero minpad parameter correctly.
2310 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2311 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2312 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2313 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2314 the supported interface.
2315 * new feature: Single-stepping of code compiled with DEBUG 2 or
2316 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2317 * new feature: saving cores with foreign code loaded is now
2318 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2319 Timothy Moore's work for CMUCL)
2320 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2321 parameters to * instead of NIL if no initform is supplied.
2322 (reported by Johan Bockgaard)
2323 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2324 correctly when FUNCALL appears as the car of the form. Note:
2325 despite this FUNCALL forms are not currently subject to
2326 compiler-macro expansion. (port of Raymond Toy's fix for the
2327 same from CMUCL, reported by Johan Bockgaard)
2328 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2329 (thanks for Teemu Kalvas)
2330 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2331 case instead of terminating the iteration (thanks for Julian
2332 Squires, Sean Champ and Raymond Toy)
2333 * bug fix: incorrect expansion of defgeneric that caused a style
2334 warning. (thanks for Zach Beane)
2335 * bug fix: slot accessor effective method computation works properly
2336 for classes with multiple non-standard applicable methods on
2337 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2338 * on x86 compiler supports stack allocation of results of LIST and
2339 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2340 CMUCL implementation by Gerd Moellmann)
2341 * fixed some bugs revealed by Paul Dietz' test suite:
2342 ** FORMAT strings with both the ~<~:;~> form of the justification
2343 directive and pretty-printing directives cause an error.
2345 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2346 * incompatible change: the internal functions
2347 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2348 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2349 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2350 instead of the old functions.
2351 * new feature: on platforms where "dladdr" is available foreign
2352 function names now appear in backtraces. (based on Helmut Eller's
2354 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2355 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2357 * documentation: ASDF manual is now distributed with SBCL.
2358 * bug fix: SBCL can now load its contributed modules with REQUIRE
2359 even if the system-provided entries have been removed from
2360 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2362 * bug fix: backtraces involving undefined functions or assembly
2363 routines are more informative. (thanks to Brian Downing)
2364 * bug fix: mutually referent alien structures now work correctly.
2365 (reported by Rick Taube)
2366 * bug fix: structures defined by WITH-ALIEN can be referred to
2367 within other definitions in the same WITH-ALIEN.
2368 * bug fix: division operators (MOD, TRUNCATE and the like) with
2369 constant zero divisors and integer dividends no longer generate
2371 * bug fix: provide default methods for INPUT-STREAM-P and
2372 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2373 * bug fix: improve the detection and resolution of MOP metacycles
2374 (where effective method computation depends on the generic function
2375 having its effective method computed). (reported by Bruno Haible)
2376 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2377 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2378 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2379 * bug fix: pathnames with relative directory components can be
2380 represented relative to default pathnames.
2381 * optimization: in taking the GCD of bignums, reduce the two bignums
2382 to approximately the same size (using Euclid's algorithm) before
2383 applying the more sophisticated binary GCD. (thanks to Juho
2385 * optimization: COUNT on bitvectors now operates word-at-a-time.
2386 * optimization: ASH with a positive, but not necessarily constant,
2387 (leftwards) shift, when in a modular context, is compiled to a
2389 * fixed some bugs revealed by Paul Dietz' test suite:
2390 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2391 if the corresponding argument is NIL.
2392 ** FORMAT directives accepting a minpad parameter treat negative
2394 ** literal commas as character parameters to FORMAT directives are
2396 ** literal spaces directly after ~<Newline> directives within a
2397 format-logical-block (~:< ~@:>) do not induce :FILL-style
2398 conditional newlines.
2399 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2400 as specified: it no longer includes conditional newlines).
2401 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2403 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2404 :PER-LINE-PREFIX argument does not evaluate to a string.
2406 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2407 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2408 the manual for details; for now, package locks can be disabled by
2409 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2410 if no major problems are found then it is likely that they will be
2411 compiled in unconditionally.
2412 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2413 unsupported operators on all platforms. To load a shared library
2414 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2415 object file, link it into a shared library outside of SBCL and
2416 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2418 * bug workaround: ROOM T and the GC were not working together
2419 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2420 reporting facility aren't preserved. That reporting has been
2421 disabled, so now until and unless someone figures out how to make it
2422 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2423 * minor incompatible change: as threatened around sbcl-0.8.0, the
2424 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2425 The symbols are also exported from SB-PCL for backwards
2426 compatibility, but more so than before SB-PCL should be treated as
2427 an implementation-internal package.
2428 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2430 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2431 It is known as of this release not to work on the Alpha, however.
2432 * fixed bug #167: errors signalled due to illegal syntax in method
2433 bodies are now more legible.
2434 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2435 designators and can hence be used with TYPEP.
2436 * fixed bug #333: CHECK-TYPE now ensures that the type error
2437 signalled, if any, has the right object to be accessed by
2438 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2439 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2440 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2442 * fixed bug #334: programmatic addition of slots using specialized
2443 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2444 and :ALLOCATION :CLASS slots.
2445 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2446 float exponents. (rereported by Peter Seibel)
2447 * fixed a bug: #\Space (and other whitespace characters) are no
2448 longer considered to be macro characters in standard syntax by
2449 GET-MACRO-CHARACTER.
2450 * fixed bug: initialization of condition class metaobjects no longer
2451 causes an instance of the condition to be created. (reported by
2453 * fixed bug: it is now possible to have slots such that
2454 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2455 description is neither :INSTANCE nor :CLASS.
2456 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2457 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2458 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2459 non-standard SLOT-DEFINITION-ALLOCATION values.
2460 * fixed bugs: various race conditions handling exiting threads.
2461 CL-PPCRE's thread torture test now passes.
2462 * fixed arguably-a-bug: GC time proportional to number of threads in
2463 system even when most of them are idle
2464 * optimization: improved performance of BIT and SBIT on bit-vectors.
2465 * .fasl file incompatibility: The fasl file version number has
2466 been incremented because of changes associated with package locks.
2468 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2469 * minor incompatible change: the system no longer provides
2470 optimization hints (conditions which are TYPEP
2471 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2472 mode; these hints are still emitted when higher SPEED optimization
2474 * new contrib module: a sampling profiler (profiling by statistical
2475 sampling, rather than by instrumenting functions) is available as
2476 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2477 * the behaviour of the standard function ED is now customizeable by
2478 third parties through a hook variable: see ED's documentation
2479 string for information on the protocol.
2480 * the compiler no longer emits efficiency notes for (FUNCALL X)
2481 when the type of X is uncertain under default optimization
2483 * fixed bug 276: mutating a binding of a specialized parameter to a
2484 method to something that is not TYPEP the specializer is now
2486 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2487 exhibit the required behaviour on the x86 platform. (thanks to
2488 Peter van Eynde, Eric Marsden and Bruno Haible)
2489 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2490 even for difficult arguments. (reported by Peter Graves)
2491 * fixed bug 141a: the backquote printer now descends quoted
2493 * fixed another bug in backquote printing: no more destructive
2494 modification of the form's list structure. (reported by Brian
2496 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2497 move between its address being taken and the call to
2498 interrupt_thread, fixing a crashing race condition.
2499 * the SB-POSIX contrib implementation has been adjusted so that it
2500 no longer exhibits ridiculously poor performance when constructing
2501 instances corresponding to C structs.
2503 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2504 * minor incompatible change: the sb-grovel contrib now treats C
2505 structures as alien (in the sense of SB-ALIEN) objects rather than
2506 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2507 has implications for memory management of client code
2508 (sb-grovel-returned objects must now be manually managed) and for
2509 type safety (alien objects now have full types).
2510 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2511 used to control emission of compiler diagnostics, rather than the
2512 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2513 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2514 quality should be considered deprecated.
2515 * install.sh now installs the user manual as well
2516 * (not quite a new documentable feature, but worth considering in
2517 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2518 beginnings of a semantically meaningful condition hierarchy is
2519 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2521 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2522 designator as the defaults argument.
2523 * fixed bug: Displaced arrays whose displaced-to array has become
2524 too small now cause ARRAY-DIMENSION to signal an error, providing
2525 for safer bounds-checking. (reported by Bruno Haible)
2526 * fixed bug: DEFCLASS slot definitions with identical :READER and
2527 :WRITER names now signal a reasonable error. (reported by Thomas
2529 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2531 * fixed bug: Invalid slot specification errors now print correctly.
2532 (thanks to Zach Beane)
2533 * fixed bug 320: Shared to local slot value transfers in class
2534 redefinitions now happen corrently from superclasses as
2535 well. (reported by Bruno Haible)
2536 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2538 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2539 improper lists correctly. (reported by Bruno Haible)
2540 * fixed bug 313: source-transform for <fun-name> was erroneously
2541 applied to a call of a value of a variable with name <fun-name>.
2542 (reported by Antonio Menezes Leitao)
2543 * fixed bug 307: The obsolete instance protocol ensures that
2544 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2545 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2546 remove dead unknown-values globs from the middle of the stack.
2547 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2548 easier to resume long computations after using *BREAK-ON-SIGNALS*
2549 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2550 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2551 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2552 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2553 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2554 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2556 * fixed a (fairly theoretical) bug in string printing: if
2557 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2558 does not have array-element-type equal to the most general string
2560 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2561 function name. (thanks to Nikodemus Siivola)
2562 * fixed bug: compile-time format string checker failed on
2563 non-closed ~{. (reported by Thomas F Burdick)
2564 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2565 don't warn on legal loop code involving more than one aggregate
2566 boolean. (thanks to Nikodemus Siivola)
2567 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2568 in keyword position to MAKE-PACKAGE was not being reported as
2570 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2571 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2573 * fixed bugs 280 and 312: the checking for multiple definitions in a
2574 file is less likely to become confused by uses of inline
2576 * fixed bug: the #S reader macro performs the keyword coercion
2577 specified for slot names. (reported by Kalle Niemitalo)
2578 * fixed bug: lambda lists may contain symbols whose names start with
2579 & but are not lambda-list-keywords; their occurrence triggers a
2581 * fixed bug 321: define-method-combination argument lambda lists do
2582 not cause invalid code to be generated when &optional or &aux
2583 variables have default values. (reported by Bruno Haible)
2584 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2585 have CLOS classes; this appears also to have fixed reported
2586 instability in stack exhaustion detection.
2587 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2588 MUFFLE-WARNING when no associated restart is present is now
2590 * optimization: rearranged the expansion of various defining macros
2591 so that each expands into only one top-level form in a
2592 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2594 * optimization: used a previously little-used slot in symbols to
2595 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2597 * fixed some bugs revealed by Paul Dietz' test suite:
2598 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2600 ** Fixed an optimization bug related to inheritance of initforms
2601 from local to shared slots.
2602 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2603 ** CAST optimizer forgot to flush argument derived type.
2604 ** print/read consistency on floats is now orders of magnitude
2605 more likely. (thanks also to Bruno Haible for a similar report
2607 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2608 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2609 for objects of type REAL. Make it so.
2610 ** SXHASH is specified (infelicitously) to respect similarity,
2611 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2612 it so. (thanks to Markus Ziegler)
2613 ** on the Alpha, the compiler succeeds in compiling functions
2614 returning a known number of arguments greater than 63.
2615 ** fixed handling of invalid NIL arguments in keyword position
2617 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2619 ** provided a readably-printable representation for RANDOM-STATE
2621 ** ensured that pathnames, where they have a namestring, always
2622 print using #P"..." syntax.
2624 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2625 * Support for the forthcoming 2.0 version of the NetBSD kernel
2626 running on x86 hardware has been added. (thanks to Perry
2627 E. Metzger most immediately, and others for their past work)
2628 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2629 current release version; SBCL's OpenBSD support had been broken
2630 since about the time of OpenBSD's switch to ELF binary format.)
2631 * [placeholder for DX summary]
2632 ** user code with &REST lists declared dynamic-extent, under high
2633 speed or space and low safety and debug optimization policy.
2634 * The manual has been converted to Texinfo format and the debugger
2635 chapter from the cmucl manual has been added.
2636 * A facility has been added to extract documentation strings from
2637 sbcl and store them as Texinfo-formatted snippets for inclusion in
2638 the manual (via Texinfo's @include directive)
2639 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2640 the test case to Dave Roberts)
2641 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2642 pointers. (reported by Sean Ross)
2643 * bug fix: PROFILE output is printed nicely even for large numerical
2644 values. (thanks to Zach Beane)
2645 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2646 greater than 32 handle EOF correctly.
2647 * bug fix: on X86 an immediate argument of the IMUL instruction is
2648 correctly printed in disassembly. (thanks to Lutz Euler)
2649 * bug fix: class slots in redefined classes preserve their old
2650 values. (thanks to Bruno Haible and Nikodemus Siivola)
2651 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2652 succeeds. (reported by Marco Baringer)
2653 * fixed some bugs revealed by Paul Dietz' test suite:
2654 ** READ-SEQUENCE now works on ECHO-STREAMs.
2655 ** RATIONALIZE works more according to its specification. (thanks
2658 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2659 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2660 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2661 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2662 should matter to you only if you rebind the printer control
2663 variables and then find you want different bindings in the
2664 debugger than in the ordinary execution of your program.)
2665 * The runtime build system has been tweaked to support building
2666 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2667 assembler and linker. (thanks to Nikodemus Siivola)
2668 * Unbound, undefined, undeclared variables now trigger full
2669 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2670 more often programmer error than deliberate exploitation of undefined
2672 * optimization: the hash algorithm for strings has changed to one
2673 that is less vulnerable to spurious collisions. (thanks to Juho
2675 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2676 needless bounds checking. (thanks to Juho Snellman)
2677 * optimization: implemented multiplication as a modular
2678 (UNSIGNED-BYTE 32) operation on the PPC backend.
2679 * fixed some bugs revealed by Paul Dietz' test suite:
2680 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2682 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2683 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2684 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2686 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2688 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2690 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2691 their output stream on EOF from read.
2692 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2693 have been read to end-of-file.
2694 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2696 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2697 description of determination of which consecutive characters
2699 ** Printing the "Space" character with escaping on now yields "#\\ ",
2700 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2701 ** Reading floating-point numbers with *READ-BASE* set to a number
2702 less than 10 works correctly.
2703 ** Reading floating-point numbers with *READ-BASE* set to a number
2704 more than 10 works correctly.
2705 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2706 the readtable currently in effect.
2708 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2709 * minor incompatible change: parsing of namestrings on a physical
2710 (Unix) host has changed; numbers after the final #\. in a
2711 namestring are no longer interpreted as a version field. This is
2712 intented to be largely invisible to the user, except that the
2713 meaning of the namestring "*.*.*" has changed: it now refers to a
2714 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2715 should usually be replaced by
2716 (make-pathname :name :wild :type :wild :version :wild)
2717 with the added benefit that this is more likely to be portable.
2718 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2719 option to OPEN now signals an error if the file being opened
2720 exists; this may have an impact on existing code.
2721 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2722 signals received, on the PowerPC platforms (both Linux and
2723 Darwin). (thanks to Pierre Mai for pointing out the location of
2725 * several fixes on OS X: The system now builds and runs cleanly on
2726 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2727 (thanks to Brian Mastenbrook)
2728 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2729 second-resolution: integer multiples of 1/3600 between -24 and 24.
2730 (thanks to Vincent Arkesteijn)
2731 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2732 type when called with 1 argument; PEEK-CHAR checked type of
2733 PEEK-TYPE only after having read first character from a
2734 stream. (reported by Peter Graves)
2735 * bug fix: the garbage collector now has much better locality
2736 behaviour, and in particular no longer treats all memory as being
2737 exhausted when many small objects point to each other in a deeply
2739 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2740 recognized as being TYPEP their class.
2741 * bug fix: the PUSHNEW documentation string has been corrected.
2742 (thanks to Vincent Arkesteijn)
2743 * bug fix: defaulting of the value for the last of an atypically
2744 large number of multiple values being bound was not being
2745 performed correctly on the Alpha or PPC platforms
2746 * optimization: implemented multiplication as a modular
2747 (UNSIGNED-BYTE 32) operation on the x86 backend.
2748 * optimization: SEARCH on simple-base-strings can now be open-coded.
2749 (see also contrib/compiler-extras.lisp for inspiration for
2750 teaching the compiler about the Boyer-Moore algorithm).
2751 * value, returned by MAX (and MIN) called with several EQUALP, but
2752 not EQL, arguments now does not depend on compiler settings.
2753 * fixed some bugs revealed by Paul Dietz' test suite:
2754 ** in stack analysis liveness information is propagated from
2755 non-local entry points.
2756 ** pathwise CAST removing failed when the CAST node did not start
2758 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2759 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2761 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2762 host is already defined.
2763 ** RENAME-FILE works on streams instead of signalling an internal
2765 ** PEEK-CHAR uses the current readtable when determining whether
2766 or not a character is whitespace.
2767 ** MERGE-PATHNAMES handles the case when the pathname does not
2768 specify a name while the default-pathname specifies a version.
2769 ** Pathnames now stand a better chance of respecting print/read
2771 ** Attempting to use standardized file system operators with a
2772 pathname with invalid :DIRECTORY components signals a
2774 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2776 ** DIRECTORY on logical pathnames is more correct.
2777 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2778 signal a TYPE-ERROR if their argument is not a stream.
2779 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2780 designator argument does not designate a stream.
2781 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2782 examining the synonym.
2783 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2785 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2786 element-type larger than ([UN]SIGNED-BYTE 32).
2788 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2789 * When built with the :SB-FUTEX feature, threaded builds now take
2790 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2791 for faster/more reliable mutex and condition variable support.
2792 * Incompatible change (but one you probably shouldn't have been using
2793 anyway): the interface and code for arbitrating between multiple
2794 threads in the same user session has been redesigned.
2795 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2796 error when called without an explicit environment argument.
2797 (thanks to Nikodemus Siivola)
2798 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2799 confused about their position. (thanks to Adam Warner and Gerd
2801 * bug fix: Pretty printing backquoted forms with unquotations in the
2802 argument list position of various code constructs such as LAMBDA
2803 now works correctly. (reported by Paul Dietz)
2804 * bug fix: Pretty printing unquotations no longer loses all
2805 stream position information.
2806 * optimization: performance of string output streams is now less
2807 poor for multiple small sequence writes.
2808 * optimization: performance of CSUBTYPEP in the presence of complex
2809 expressions involving CONS and NOT many times has been improved.
2810 (reported by Paul Dietz)
2811 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2813 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2814 streams for the REPL, for the debugger, and so on. (thanks to
2816 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2817 * fixed some bugs revealed by Paul Dietz' test suite:
2818 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2819 is ignored if it is too small, rather than propagating through
2820 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2821 ** extremely complex negations of CONS types were not being
2822 sufficiently canonicalized, leading to inconsistencies in
2824 ** VALUES tranformer lost derived type.
2826 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2827 * fixed a bootstrapping bug: the build process no longer assumes
2828 that the various BOOLE-related constants have the same value in
2829 host and target lisps. (noted by Paul Dietz' test suite on an
2830 SBCL binary built from CLISP)
2831 * The system can now be dynamically linked on the MIPS platform,
2832 which enables dynamic loading of foreign code from Lisp. (thanks
2833 to Ralf Baechle for discussions on the MIPS ABI)
2834 * The system now records debugging information for its own source
2835 files in a filesystem-position-independent manner, relative to
2836 the "SYS" logical host.
2837 * fixed a compiler bug: MV-LET convertion did not check references
2838 to the "max args" entry point. (reported by Brian Downing)
2839 * tweaked disassembly notes to be less confident about proclaiming
2840 some instruction as an LRA. (thanks to Brian Downing)
2841 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2842 now each have their own history, command character, and other
2843 characteristics. (thanks to David Lichteblau)
2844 * fixed some bugs revealed by Paul Dietz' test suite:
2845 ** compiler failure in compiling LOGAND expressions including a
2847 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2848 shift greater than 32.
2849 ** FLUSH-DEST did not mark blocks for type check regeneration.
2850 ** HANDLER-CASE failed to accept declarations in handler clauses
2851 in some circumstances.
2853 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2854 * New code in contrib/sb-introspect (still probably not entirely
2855 stable yet) provides some support for smart Lisp development
2856 environments like SLIME.
2857 * The conditions signalled for errors occurring when loading .fasl
2858 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2859 in a way which should help ASDF recover gracefully.
2860 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2861 stuff has been cleaned up. If you code contrib/ stuff, this might
2862 affect you, and you can look at contrib/README, contrib/STANDARDS,
2863 and/or the 0.8.4.27 diff to check.
2864 * In full calls the compiler now does not generate checks for declared
2865 argument types for all arguments.
2866 * various threading fixes
2867 ** and some experimental patches which didn't make it into
2868 the main tree for this release, but which are shipped in
2869 contrib/experimental-thread.patch as a possible fix for some
2870 failures (deadlock, spinning...) in GC-intensive multithreaded
2872 * fixed PPC build problem (source code incompatibility of different
2873 library versions): added offsetof() hackery which attempts to divine
2874 where glibc maintainers put uc_mcontext today
2875 * fixed bug 282: compiler does not trust type assertions while passing
2876 arguments to a full call.
2877 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2878 &OPTIONAL VALUES type parameter.
2879 * fix bug 214: algorithm for noting rejected templates is now more
2880 similar to that of template seletion. (also reported by rydis on
2882 * fixed bug 141b: printing backquoted information readably and prettily
2883 inserts a space where necessary.
2884 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2885 (CHARACTER 10) are now reported as errors, rather than propagated
2886 as unknown types. (reported by piso on #lisp)
2887 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2888 with values NIL and :ERROR. (thanks to Milan Zamazal)
2889 * fixed bug 191c: CLOS now does proper keyword argument checking as
2890 described in CLHS 7.6.5 and 7.6.5.1.
2891 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2892 longer raise an error; further, using a list as a for-as-arithmetic
2893 counter now raises a meaningful error.
2894 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2895 now understood by sequence creation functions such as MAKE-SEQUENCE
2897 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2898 a STRING-INPUT-STREAM.
2899 * compiler enhancement: SIGNUM is now better able to derive the type
2901 * type declarations inside WITH-SLOTS are checked. (reported by
2903 * fixed some bugs revealed by Paul Dietz' test suite:
2904 ** incorrect optimization of TRUNCATE for a positive first
2905 argument and negative second.
2906 ** compiler failure in let-convertion during flushing dead code.
2907 ** compiler failure while deriving type of TRUNCATE on an
2908 interval, containing 0.
2909 ** ASH of a negative bignum by a negative bignum count now returns
2911 ** intersection of CONS types now canonicalizes properly, fixing
2912 inconsistencies in SUBTYPEP.
2914 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2915 * incompatible change: The --disable-debugger command line
2916 option now clobbers the debugger at a more fundamental
2917 level, by redefining #'INVOKE-DEBUGGER instead of by
2918 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2919 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2920 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2921 enter the debugger REPL and then suffer recursive errors
2922 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2923 while under the new system, BREAK in --disable-debugger mode
2924 terminates the system just as an unhandled error would.
2925 * fixed compiler performance when processing loops with a step >1;
2926 * bug fix: DOCUMENTATION now retrieves generic function
2927 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2928 support has been systematized, and now supports the methods
2929 specified by ANSI, along with a default method and a method for
2930 slot documentation. (reported by Nathan Froyd)
2931 * bug fix: effective methods associated with a generic function are
2932 no longer cached over a change of that generic function's method
2933 combination. (reported by Andreas Fuchs)
2934 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2935 names a non-existent directory. (thanks to Andreas Fuchs)
2936 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2937 the PPC platform, lead to stack corruption; this has been fixed.
2938 (reported by Rainer Joswig)
2939 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2940 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2941 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2942 arithmetic in general, is now much more reliable on the PPC
2944 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2945 platform now returns the right answer.
2946 * optimization: restored some effective method precomputation in
2947 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2948 precomputation is now tunable.
2949 * optimization: compiler-internal data structure use has been
2950 reviewed, and changes have been made that should improve the
2951 performance of the compiler by about 20%.
2952 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2953 simple-base-strings and simple-bit-vectors is improved.
2954 * optimization: the optimization of 32-bit logical and arithmetic
2955 functions introduced in version 0.8.3 on the x86 has been
2956 implemented on the mips, ppc and sparc platforms; an
2957 implementation of the same facility, but for 64-bit arithmetic,
2958 has been added for the alpha.
2959 * microoptimization: the compiler is better able to make use of the
2960 x86 LEA instruction for multiplication by constants.
2961 * bug fix: in some situations compiler did not report usage of
2962 generic arithmetic in (SPEED 3) policy.
2963 * bug 145b fix: compiler used wrong type specifier while converting
2964 MEMBER-types to numeric.
2965 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2967 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2968 index argument now works properly on the Alpha platform.
2969 * bug fix: floating point exception treatment on the Alpha platform
2971 * bug fix: FILE-POSITION works much better on string input and
2972 output streams. (thanks to Nikodemus Siivola)
2973 * bug fix: many threading/garbage collection symptoms sorted.
2974 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2975 might be pseudo-atomic.
2976 * internal change: Stopping for GC is now done with signals not
2977 ptrace. GC is now done in whichever thread wanted it, instead of
2979 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2981 * bug fix: result form in DO is not contained in the implicit
2983 * incompatible change: ICR structure is changed; the value part of
2984 CONTINUATION is now called LVAR; corresponding functions are
2985 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2986 * added type deriver for ISQRT (thanks to Robert E. Brown).
2987 * bug fix: better support for loading from the command line when an
2988 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2990 * fixed some bugs revealed by Paul Dietz' test suite:
2991 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2992 ** ROUND and FROUND now give the right answer when given very
2993 small float arguments.
2994 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2996 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2997 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2998 ** LCM should return a non-negative integer.
2999 ** PARSE-INTEGER returned the index of a terminator instead of the
3000 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
3001 ** PARSE-INTEGER returned an incorrect index being applied to a
3003 ** LCM with two arguments of 0 returns 0 rather than signalling
3005 ** unsigned addition of a 32-bit constant with the high bit set no
3006 longer causes an internal compiler error on the x86.
3007 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3008 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3009 with negative last argument.
3010 ** byte specifiers with zero size and position no longer cause
3011 an error during type derivation.
3012 ** bignum multiplication on the Alpha platform now returns the
3014 * porting: The system now builds on SuSE AMD64, although it still
3015 generates a 32-bit binary.
3016 * .fasl file incompatibility: The fasl file version number has
3017 been incremented (because of the changes to internal compiler
3018 data structures referred to above).
3020 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3021 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3022 more accurately, on the Darwin kernel running on PowerPC hardware.
3023 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3024 * Compiler code deletion notes now signal a condition of type
3025 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3026 an associated MUFFLE-WARNING restart.
3027 * The compiler now performs limited argument count validation of
3028 constant format strings in FORMAT, and where appropriate in ERROR,
3029 CERROR and WARN. (thanks to Gerd Moellmann)
3030 * New ASDF-INSTALL contrib can be used for automatic download and
3031 installation of third-party Lisp code from CCLAN or other sites
3033 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3034 which forces another thread to execute a function supplied by the
3036 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3037 now accept and act upon their :ELEMENT-TYPE keyword argument.
3038 (reported by Martin Atzmueller, Edi Weitz)
3039 * bug fix: FILE-POSITION now accepts position designators up to
3040 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3041 is the greater. (thanks to Patrik Nordebo)
3042 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3043 Kalle Olavi Niemitalo)
3044 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3045 printable. (reported by Eric Marsden)
3046 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3047 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3048 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3049 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3050 (thanks to Stig Sandoe)
3051 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3052 UNREAD-CHAR now correctly includes the unread character in the
3053 target sequence. (thanks to Gerd Moellmann)
3054 * bug fix in threaded builds: the system can now be suspended and
3055 resumed by shell job control with minimal disruption.
3056 * bug fixes in times and timezones >2038AD
3057 * better handling of "where is GNU make?" problem in build scripts
3058 (thanks to Nikodemus Siivola)
3059 * new optimization: inside a named function any reference to a
3060 function with the same name is considered to be a self-reference;
3061 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3062 optimization quality.
3063 * new optimization on x86: logical functions and + now have
3064 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3065 used when the result is truncated to 32 bits.
3066 * VALUES declaration is partially enabled.
3067 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3068 * fixed some bugs revealed by Paul Dietz' test suite:
3069 ** The system now obeys the constraint imposed by
3070 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3071 types form a lattice under type intersection.
3072 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3073 ** ASSOC now ignores NIL elements in an alist.
3074 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3075 and (1+ MOST-POSITIVE-FIXNUM) answers.
3076 ** The addition of a method with invalid qualifiers to a generic
3077 function does not cause an error to be signalled immediately;
3078 a warning is signalled, and the error is generated only on
3079 calling the generic function.
3080 * changed .fasl file version number, in part to add type codes for
3081 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3082 obscure ANSI requirements
3084 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3085 * fixed bug 148: failure to inline-expand a local function left
3086 garbage, confusing the compiler.
3087 * fixed bugs 3cd: structure slot readers perform type check if the
3088 slot can have an invalid value (i.e. it is either not initialized
3089 or can be written with a less specific slot writer).
3090 * bug fix: the compiler now traps array references to elements off
3091 the end of an array; previously, the bounds checking in some
3092 circumstances could go off-by-one.
3093 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3095 * type declarations for array element types now obey the description
3096 on the CLHS page "Declaration TYPE", as per discussions on
3097 sbcl-help around 2003-05-08. This means that a declaration
3098 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3099 declaration, all references to BAR will be asserted or assumed
3100 (with THE, so dependent on compiler policy) to involve objects of
3101 type FOO. Note that no such declaration is implied in
3102 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3103 * declared types of functions from the "Conditions"
3104 chapter. (reported by Paul Dietz)
3105 * bug fix: CERROR accepts a function as its first argument.
3106 * bug fix: NTH an NTHCDR accept a bignum as index
3107 arguments. (reported by Adam Warner)
3108 * optimization: character compare routines now optimize comparing
3109 against a constant character. (reported by Gilbert Baumann)
3110 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3111 argument now works properly on the MIPS platform.
3112 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3113 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3114 declared to have a functional type, does not cause an error
3116 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3117 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3118 * fixed bug: if last continuation of a deleted block has a
3119 destination, this destination should be deleted too. (reported by
3121 * fixed a bug in the bootstrap process: the host compiler's values
3122 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3123 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3124 test case from Patrik Nordebo)
3125 * improved the ability of the disassembler on the PPC platform to
3126 provide helpful disassembly notes.
3127 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3128 the class in more cases than previously.
3129 * bug fix: FILE-POSITION now understands :START and :END for
3130 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3131 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3132 8) data. (thanks to David Lichteblau)
3133 * bug fix: it is possible to add a method to a generic function
3134 without lambda list.
3135 * bug fix: reader failed to signal END-OF-FILE inside an
3136 object representation. (reported by Nikodemus Siivola)
3137 * fixed some bugs revealed by Paul Dietz' test suite:
3138 ** LAST and [N]BUTLAST should accept a bignum.
3139 ** condition slot accessors are methods.
3140 ** (VECTOR NIL) is a subtype of STRING.
3142 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3143 * minor incompatible change: some nonsensical specialized lambda
3144 lists (used in DEFMETHOD) which were previously ignored now signal
3146 * minor incompatible change: the system is now aware of the types of
3147 variables in the COMMON-LISP package, and will signal errors for
3148 most violations of these type constraints (where previously they
3149 were silently accepted).
3150 * minor incompatible change: COMPILE-FILE now uses the freedom
3151 afforded (ANSI 3.2.2.3) to use derived function types for
3152 functions defined in the same file. This also permits the system
3153 to warn on static type mismatches and function
3154 redefinition. (Currently it does not work with high DEBUG level.)
3155 * minor incompatible change: VALUES declaration is disabled.
3156 * When issuing notes, the compiler now signals a condition of type
3157 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3158 restart for use in user handlers. It is expected that the
3159 COMPILER-NOTE condition will eventually become a condition
3160 supertype to a hierarchy of note types, which will then be
3161 handleable in a similar fashion. However, at the moment, no such
3162 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3163 but it's an implementation detail, not a classification for the
3165 * Changes in type checking closed the following bugs:
3166 ** type checking of unused values (192b, 194d, 203);
3167 ** template selection based on unsafe type assertions (192c, 236);
3168 ** type checking in branches (194bc).
3169 * A short form of VALUES type specifier has ANSI meaning (it has
3170 increased the number of situations when SBCL cannot perform type
3172 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3173 keywords or constants is permissible.
3174 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3175 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3176 classes. (thanks to Antonio Martinez)
3177 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3178 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3179 argument to a pretty-print stream. Instead, it leaves any such
3180 operations to DESCRIBE-OBJECT methods.
3181 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3183 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3184 lambda lists. (thanks to David Lichteblau)
3185 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3186 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3187 (thanks to Gerd Moellmann)
3188 * fixed bug 246: increased compilation speed of long
3189 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3191 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3193 * DEFCONSTANT now throws a condition of type
3194 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3195 constant to a non-EQL value; CONTINUE and ABORT restarts
3196 respectively change and preserve the value.
3197 * fixed bug 63: The code walker, part of the implementation of CLOS,
3198 is now better at handling symbol macros.
3199 * bug fix: There is no longer an internal implementation type named
3200 CL:LENGTH. (reported by Raymond Toy)
3201 * bug fix: In macro-like defining macros/special operators the
3202 implicit block does not enclose the lambda list.
3203 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3204 enclosed in parentheses) are not suitable as type specifiers, and
3205 their use properly signals an error now.
3206 * bug fix: An argument count mismatch for a type specifier in code
3207 being compiled no longer causes an unhandled error at compile
3208 time, but signals a compile-time warning.
3209 * fixed simple vector readable printing
3210 * bug fix: DESCRIBE takes more care over whether the class
3211 precedence list slot of a class is bound before accessing it.
3212 (reported by Markus Krummenacker)
3213 * bug fix: FORMATTER can successfully compile pretty-printer format
3214 strings which use variants of the ~* directive inside.
3215 * bug fix: SEARCH now applies its TEST predicate to the elements of
3216 the arguments in the correct order. (thanks to Wolfhard Buss)
3217 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3218 in type inference. (thanks to Robert E. Brown)
3219 * bug fix: Reading in symbols with an explicit package name of ""
3220 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3221 * fixed some bugs revealed by Paul Dietz' test suite:
3222 ** NIL is now allowed as a structure slot name.
3223 ** Arbitrary numbers, not just REALs, are allowed in certain
3224 circumstances in LOOP for-as-arithmetic clauses.
3225 ** Multiple class redefinitions before slot access no longer
3226 causes a type error.
3227 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3228 association between the name and a class.
3229 ** Generic functions with non-standard method-combination and over
3230 five methods all of which return constants no longer return NIL
3231 after the first few invocations. (thanks to Gerd Moellmann)
3232 ** CALL-NEXT-METHOD with no arguments now passes the original
3233 values of the arguments, even in the presence of assignment.
3234 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3236 ** DISASSEMBLE works with closures and funcallable instances.
3237 ** ADD-METHOD now returns the generic function, not the new method.
3238 ** FIND-METHOD signals an error if the lengths of the specializers
3239 is incompatible with the generic function, even if the ERRORP
3241 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3242 which its argument is a member.
3243 ** DEFCLASS only redefines the class named by its class-name
3244 argument if that name is the proper name of the class;
3245 otherwise, it creates a new class.
3246 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3247 of the UNBOUND-SLOT condition to the name of the slot.
3248 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3249 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3250 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3251 SLOT-MAKUNBOUND in the specified fashion.
3253 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3254 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3255 cross-compilation host. As a consequence, we can now bootstrap our
3256 way up to SBCL starting with a bare gcc toolchain and human-readable
3257 source code (first the source to CLISP, then the source to SBCL).
3258 * A contributed module containing a partial implementation of the
3259 simple-streams interface has been included. (thanks to Rudi
3261 * A contributed module implementing the RFC1321 Message Digest
3262 Algorithm, known as MD5, has been included.
3263 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3264 no longer has any effect, as the code controlled by this feature
3265 has been deleted. (As far as we know, no-one has ever built using
3266 this feature, and its semantics were confused in any case).
3267 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3268 work on conditions (as required by the ANSI specification),
3269 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3270 expected behaviour on conditions. Users should note, however,
3271 that such behaviour is not required by the ANSI specification,
3272 and so use of this behaviour may render their code unportable.
3273 * fixed some bugs revealed by Paul Dietz' test suite:
3274 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3276 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3277 ** SXHASH on condition objects no longer returns NIL.
3278 ** :ALLOCATION :CLASS slots are better treated; their values are
3279 updated on class redefinition, and initforms inherited from
3280 superclasses are applied.
3281 ** REMOVE-METHOD returns its generic function argument even when
3282 no method was removed.
3283 ** SHARED-INITIALIZE now initializes the values of the requested
3284 slots, including those with :ALLOCATION :CLASS.
3285 ** ALLOCATE-INSTANCE now works on structure classes defined via
3286 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3288 ** SLOT-EXISTS-P now works on conditions, as well as structures
3290 ** MAKE-LOAD-FORM now has the required methods on
3291 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3292 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3293 keyword, and now implements the SLOT-NAMES argument.
3294 ** methods with &OPTIONAL arguments no longer allow too many
3295 arguments to be passed in the call without error.
3296 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3297 option is consistent with the required arguments of the generic
3298 function lambda list.
3299 * bug fix: REQUIRE accepts a string designator. (Thanks to
3301 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3302 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3303 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3304 * bug fix: The compiler now checks for duplicated variables in macro
3306 * bug fix: SETQ on globals returns the correct value.
3307 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3308 not a COMPILER-ERROR (followed by some other strange error on
3309 choosing the CONTINUE restart).
3310 * bug fix: make.sh and friends are now more consistent in the way that
3311 they look for GNU "make".
3313 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3314 * experimental native threads support (on x86 Linux >=2.4 only).
3315 This is not compiled in by default: you need to add :SB-THREAD to
3316 the target features. See the "Beyond ANSI" chapter of the manual
3318 * fix for longstanding nonANSIism: The old distinction between
3319 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3320 The return value from CL:FIND-CLASS is now a CLOS class, and
3321 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3322 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3323 * An interface to the MetaObject Protocol, as described in Kiczales,
3324 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3325 MIT Press, 1991, is available from the SB-MOP package.
3326 * incompatible change: the SB-PCL package should now be considered
3327 a private implementation detail, and no longer a semi-private MOP
3329 * minor incompatible change: due to rearrangement for threads, the
3330 control stack and binding stack are now allocated at arbitrary
3331 addresses instead of being hardcoded per-port. Users affected by
3332 this probably have to be doing advanced things with shared
3333 libraries, and will know who they are.
3334 * minor incompatible change: Previously, all --eval forms used were
3335 processed with READ before any of them were processed with EVAL.
3336 Now each --eval form is processed with both READ and EVAL before
3337 the next --eval form is processed. (Thus package operations like
3338 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3339 work as the user might reasonably expect.)
3340 * minor incompatible change: *STANDARD-INPUT* is now only an
3341 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3343 * minor incompatible change: Y-OR-N-P is now character-oriented,
3344 not line oriented. Also, YES-OR-NO-P now works without errors.
3345 (thanks to Antonio Martinez)
3346 * sb-aclrepl module improvements: an integrated inspector, added
3347 repl features, and a bug fix to :trace command.
3348 * Known functions, which cannot be open coded by the backend, are
3349 considered to be able to check types of their arguments. (fixing
3350 a bug report by Nathan J. Froyd)
3351 * fixed a bug in computing method discriminating functions: It is
3352 now possible to define methods specialized on classes which have
3353 forward-referenced superclasses. (thanks to Gerd Moellmann)
3354 * fixed evaluation order in optional entries (reported by Gilbert
3356 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3357 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3358 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3359 required-by-AMOP NAME argument, as well as CLASS and
3360 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3361 * fixed bug 20: DEFMETHOD can define methods using names that are
3362 not the proper names of classes to designate class specializers.
3363 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3364 Unix file descriptors, instead of blowing up. (thanks to Antonio
3366 * Garbage collection refactoring: user-visible change is that a
3367 call to the GC function during WITHOUT-GCING will not do garbage
3368 collection until the end of the WITHOUT-GCING. If you were doing
3369 this you were probably losing anyway.
3370 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3371 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3372 TYPEP the latter but not the former.
3373 * The compiler issues a full WARNING for calls to undefined functions
3374 with names from the CL package.
3375 * MAP-INTO for a vector destination is open coded. (reported by
3376 Brian Downing on c.l.l)
3377 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3378 documentation string.
3379 * fixed some bugs revealed by Paul Dietz' test suite:
3380 ** COPY-ALIST now signals an error if its argument is a dotted
3382 ** Condition slots are now accessed more correctly in the presence
3383 of multiple initargs for a given slot.
3384 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3385 exclude restarts of the same name associated with a different
3387 ** DEFCLASS of forward-referenced classes with another
3388 forward-referenced class in the superclasses list no longer
3390 ** Condition slots are now initialized once each, not multiple
3391 times. (thanks to Gerd Moellmann)
3392 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3394 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3395 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3396 arguments contain duplicated elements.
3397 ** RESTART-CASE understands local macros.
3398 ** RESTART-CASE associates exactly its own restarts with a condition.
3399 ** ENDP in safe mode checks its argument to be of type LIST.
3400 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3401 in question is unbound.
3402 ** Optimized MAKE-INSTANCE functions no longer cause internal
3403 assertion failures in the presence of duplicate initargs.
3404 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3405 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3406 propagated even in the ctor optimized implementation of
3408 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3410 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3411 * a better implementation of SXHASH on (simple) bit vectors,
3412 measured both in execution speed and in distribution of results
3413 over the positive fixnums, has been installed. Likewise, a better
3414 implementation of EQUAL for simple bit vectors is now available.
3415 * fixed CEILING optimization for a divisor of form 2^k.
3416 * fixed bug 240 (emitting extra style warnings "using the lexical
3417 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3418 by Antonio Martinez)
3419 * fixed SXHASH, giving different results for NIL depending on type
3420 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3421 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3422 variable. (found by Rolf Wester)
3423 * SBCL does not ignore type declarations for special
3424 variables. (reported by rif on c.l.l 2003-03-05)
3425 * some bug fixes in contrib/sb-aclrepl/
3426 * fixed some bugs revealed by Paul Dietz' test suite:
3427 ** a bug in the CONS type specifier, whereby the CAR and CDR
3428 types got intertwined, has been fixed;
3429 ** the type system is now able to reason about the interaction
3430 between INTEGER and RATIO types more completely;
3431 ** APPEND, [N]REVERSE and NRECONC check that those their
3432 arguments, which must be proper lists, are really so;
3433 ** An array specialized to be unable to hold elements has been
3434 implemented, as required -- yes, really -- by ANSI;
3435 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3436 on malformed property lists;
3438 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3439 * incompatible packaging change: in line with Unix convention,
3440 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3441 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3442 * REQUIRE and PROVIDE are now optionally capable of doing something
3443 useful. See the documentation string for REQUIRE.
3444 * infrastructure for a managed SBCL contrib system: contributed
3445 modules in this release include:
3446 ** the ASDF system definition facility;
3447 ** an interface to the BSD Sockets API;
3448 ** an ACL-like convenience interface to the repl;
3449 (thanks to Kevin Rosenberg)
3450 ** an implementation of ROTATE-BYTE, with efficient implementation
3452 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3453 gives a closer approximation to the right answer than previously.
3454 (thanks to Raymond Toy)
3455 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3456 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3457 optional environment arguments, as required by ANSI.
3458 * fixed bugs in other functions taking environment objects, allowing
3459 calls with an explicit NIL environment argument to be compiled
3461 * fixed bug 228: primary return values from
3462 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3463 COMPILE or FUNCTION.
3464 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3465 :TYPE will no longer signal errors on innocuous objects.
3466 * fixed bug 231b: SETQ is better at respecting type declarations in
3467 the lexical environment.
3468 * fixed a bug in DEFCLASS: classes named by symbols with no or
3469 unprintable packages can now be defined.
3470 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3471 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3472 * fixed compiler failure related to checking types of functions.
3473 (reported by Robert E. Brown)
3474 * the compiler is now much more consistent in its error-checking
3475 treatment of bounding index arguments to sequence functions: in
3476 (SAFETY 3) code, errors will be signalled in almost all cases if
3477 invalid sequence bounding indices are passed to functions defined
3478 by ANSI to operate on sequences.
3479 * fixed a bug in the build procedure: documentation of SBCL-specific
3480 packages is now preserved and available in the final Lisp image.
3481 * lifted FDEFINITION lookup out of loops in the implementation of
3482 many list operations. (thanks to Robert E. Brown)
3483 * fixed a bug in the reader: the #n# reader macro now works for
3484 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3485 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3486 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3487 E. Brown and Tony Martinez respectively)
3488 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3489 SLOT-BOUNDP: in particular, it is now able to optimize them much
3490 better, and is now not vulnerable to having packages renamed.
3491 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3492 always called when a slot is not present in an instance. (thanks
3494 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3495 longer signal an error on unreading a character following EOF on
3496 the previous constituent stream. (thanks to Tony Martinez)
3497 * fixed some bugs revealed by Paul Dietz' test suite:
3498 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3499 not just nonnegative fixnums;
3500 ** the logical bit-array operators such as BIT-AND now accept an
3501 explicit NIL for their "opt-arg" argument (to indicate a
3502 freshly-consed result bit-array);
3503 ** ELT now signals an error on an invalid sequence index in safe
3505 ** the type system is now cleverer about negations of numeric
3506 types, and consequently understands the BIGNUM and RATIO types
3508 ** the type system is now cleverer about the interaction between
3509 INTEGER and RATIO types: while bugs still remain, many more
3510 cases are accurately computed;
3511 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3512 if it is in the last clause;
3513 ** CONSTANTLY now correctly returns a side-effect-free function in
3515 ** DECLARE is no longer treated as a special-operator; in
3516 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3517 * incremented fasl file version number due to the change in the
3518 DEFSTRUCT-SLOT-DESCRIPTION structure.
3520 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3521 * minor incompatible change: code processed by the "interpreter" or
3522 EVAL now has a compilation optimization policy of (DEBUG 2)
3523 (changed from (DEBUG 1)) to improve debuggability of interactive
3524 development, and to allow the use of the debug RETURN command in
3526 * an experimental implementation of the RETURN command for the
3527 debugger has been included. (thanks to Frederik Kuivinen)
3528 * fixed bug 62: constraints were not propagated into a loop.
3529 * fixed bug in embedded calls of SORT (reported and investigated by
3531 * fixed some bugs revealed by Paul Dietz' test suite:
3532 ** printing and reading of arrays with some dimensions having
3533 length 0 (thanks to Gerd Moellmann);
3534 ** BOA constructor with &AUX argument without a default value does
3535 not cause a type error;
3536 ** CONSTANTP now returns true for all self-evaluating objects.
3538 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3539 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3540 accessors that are related by inheritance, as specified in the
3541 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3542 Valtteri Vuorikoski)
3543 * The compiler is now able to inline functions that were defined in
3544 a complex lexical environment (e.g. inside a MACROLET).
3545 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3547 * Support for the upcoming FreeBSD-5.0 release has been included.
3548 (thanks to Dag-Erling Smorgrav)
3549 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3550 effect when it is not in a toplevel context.
3551 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3552 stand a better chance of being correct. (thanks to Gerd
3554 * fixed bug in COERCE, which now signals an error on coercing a
3555 rational to a bounded real type which excludes the expected
3557 * The compiler is now able to derive types more accurately from the
3558 COERCE and COMPILE functions.
3559 * fixed bug 223: functional binding is considered to be constant
3560 only for symbols in the CL package.
3561 * fixed bug 231: SETQ did not check the type of a variable being set
3562 (reported by Robert E. Brown)
3563 * A new optimization for MAKE-INSTANCE has been included, fixing
3564 various bugs (including relating to :ALLOCATION :CLASS slots and
3565 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3567 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3568 ** As required by ANSI, LOOP now disallows anonymous collection
3569 clauses such as COLLECT I in conjunction with aggregate boolean
3570 clauses such as THEREIS (= I 1);
3571 ** LOOP now signals an error when any variable is reused in the
3572 same loop (including the potentially useful construct analogous
3573 to WITH A = 1 WITH A = (1+ A);
3574 ** IT is only a special loop symbol within the first clause of a
3575 conditional loop clause;
3576 ** LOOP with a typed iteration variable over a hashtable now
3577 signals a type error iff it should.
3578 * fixed some other bugs revealed by Paul Dietz' test suite:
3579 ** FILE-STREAM now names the class previously known as FD-STREAM;
3580 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3581 argument) no longer signals an error;
3582 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3583 of the slot symbol, rather than using the current package
3584 ((:CONC-NAME "") continues to intern the slot's name in the
3586 * incremented fasl file version number, because of the incompatible
3587 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3588 of the new implementation of DEFINE-COMPILER-MACRO.
3590 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3591 * Support for building SBCL for MIPS platforms running in
3592 little-endian mode has now been checked in, and basic
3593 functionality on said platforms verified.
3594 * minor incompatible change: PCL now records the pathname of a file
3595 in which methods and the like are defined, rather than its
3597 * minor incompatible change: TRUENAME now considers the truename of
3598 a file naming a directory to be the pathname with :DIRECTORY
3599 component indicating that directory.
3600 * minor incompatible change: a NAMED clause in the extended form of
3601 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3602 reason for the previous behaviour is unclear.
3603 * more systematization and improvement of CLOS and MOP conformance
3604 in PCL (thanks to Gerd Moellman and Pierre Mai):
3605 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3607 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3608 primary methods with no specializers;
3609 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3611 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3612 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3613 FORWARD-REFERENCED-CLASSes; error reporting on
3614 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3616 ** SXHASH on CLOS instances now uses a slot internal to the
3617 instance to return different numbers on distinct instances,
3618 while preserving the same return value through invocations of
3620 ** DEFMETHOD signals errors when methods with longer incongruent
3621 lambda lists are added to generic functions;
3622 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3623 CLASS, as specified in AMOP;
3624 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3625 on the order returned by the primary method for classes of
3626 class STANDARD-CLASS;
3627 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3628 * fixed some bugs shown by Paul Dietz' test suite:
3629 ** DOLIST puts its body in TAGBODY;
3630 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3632 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3633 value producing form;
3634 ** if more variables are given to PROGV than values, extra
3635 variables are bound and made to have no value;
3636 ** NSUBSTITUTE on list arguments gets the right answer with
3638 ** ELT signals an error of type TYPE-ERROR when the index argument
3639 is not a valid sequence index;
3640 ** LOOP signals (at macroexpansion time) an error of type
3641 PROGRAM-ERROR when duplicate variable names are found;
3642 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3643 ** FUNCALL of special-operators now cause an error of type
3645 ** PSETQ now works as required in the presence of side-effecting
3646 symbol-macro places;
3647 ** NCONC accepts any object as its last argument;
3648 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3650 ** loop-for-as-package does not require a package to be explicitely
3652 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3654 * fixed bug 166: compiler preserves "there is a way to go"
3655 invariant when deleting code.
3656 * fixed bug 172: macro lambda lists with required arguments after
3657 &REST arguments now cause an error to be signalled. (thanks to
3659 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3660 bug. (thanks to Matthew Danish)
3661 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3663 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3664 arguments are lexically rebound. (thanks to Gerd Moellmann and
3666 * fixed bug 194: error messages are now more informative when there
3667 is no primary method applicable in a call to a generic
3668 function. (thanks to Gerd Moellmann)
3669 * fixed bug in command line argument checking (thanks to Julian
3671 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3672 (thanks to Matthew Danish)
3673 * incremented fasl file version number, because of the
3674 SXHASH-related changes in the layout of CLOS data structures
3676 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3677 * minor incompatible change: The runtime (the Unix executable named
3678 "sbcl") is now much pickier about the .core files it will load.
3679 Essentially it now requires .core files to descend from the same
3680 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3681 as the runtime does. (The intent is to prevent the crashes which
3682 can occur, and which can even be reported as mysterious failures,
3683 when people patch the sources or change the build parameters
3684 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3685 sbcl and .core files.)
3686 * fixed bug: VALUES-LIST is no longer optimized away.
3687 * fixed bug 142: The FFI conversion of C string values to Lisp
3688 string values no longer conses excessively. (thanks to Nathan
3689 Froyd porting Raymond Toy's fix to CMU CL)
3690 * began to systematize and improve MOP conformance in PCL (thanks to
3691 Nathan Froyd, Gerd Moellman and Pierre Mai):
3692 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3694 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3695 ** FINALIZE-INHERITANCE is now called on class finalization;
3696 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3697 argument precedence order.
3698 * fixed bug 202: The compiler no longer fails on functions whose
3699 derived types contradict their declared type.
3700 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3701 so it can be non-toplevel.
3702 * The fasl file version number has changed (because of the new
3703 implementation of DEFMACRO).
3704 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3705 safe code, that any length requirement by their type-specifier
3706 argument is valid. The exceptions to this are described in bug
3708 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3709 functions now check the types of their inputs as required by ANSI.
3710 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3711 introduced symbol is DECLAREd to be SPECIAL.
3712 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3713 * fixed bug: :COUNT argument to sequence functions may be negative
3714 * fixed bug: body of DO-SYMBOLS may contain declarations
3715 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3716 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3717 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3718 symbol macro only once
3719 * fixed printing of call frame when argument list is unavailable
3720 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3721 * compiler no longer signals WARNING on unknown keyword
3724 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3725 * A beta-quality port to the mips architecture running Linux,
3726 based on the old CMUCL backend, has been made. It has been tested
3727 on a big-endian kernel, and works sufficiently well to be able to
3728 rebuild itself; it has not been tested in little-endian mode.
3729 * fixed an inconsistency between gencgc.c and purify.c which made
3730 dumping/loading .core files unreliable
3731 * fixed bug 120a: The compiler now deals correctly with IFs where
3732 the consequent is the same as the alternative, instead of
3733 misderiving the return type. (thanks to Alexey Dejneka)
3734 * fixed bug 113: Logical pathnames are now dumpable (the logical
3735 host is resolved at load-time, throwing an error if it is not
3737 * fixed bug 174: FORMAT's error message is slightly clearer when a
3738 non-printing character is used in a format directive.
3739 * fixed several bugs in compiler checking of type declarations, i.e.
3740 violations of the Python "declarations are assertions" principle
3741 (thanks to Alexey Dejneka)
3742 * fixed several bugs in PCL's error checking (thanks to Gerd
3744 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3746 * fixed bug in compilation of functions as first class values
3747 (thanks to Antonio Martinez-Shotton)
3748 * The compiler's handling TYPE-ERRORs which it can prove will
3749 inevitably happen at runtime has been cleaned up and corrected
3750 in several ways. (thanks to Alexey Dejneka)
3751 * improved argument type checking for various basic arithmetic
3752 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3753 lavished on them in the past that they can be compiled in many
3754 ways in different special cases
3755 * fixed bug 181: compiler checks validity of user supplied type
3757 * cleaned up code flushing in optimization: Function calls which
3758 should signal errors for safety purposes (e.g. which ANSI says
3759 should signal errors when their arguments are of incorrect type)
3760 are no longer optimized away.
3761 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3762 * incremented fasl file version number, because changes in the
3763 implementation of sequence functions like COERCE caused
3764 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3765 in old inline expansions) to become undefined. (Actually these
3766 changes were later undone, so we might very well be binary
3767 compatible with 0.7.7 after all, but leaving the version number
3768 incremented seemed like the simplest and most conservative
3771 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3772 * An alpha-quality port to the parisc architecture running Linux,
3773 based on the old CMUCL backend, has been made. This, even more so
3774 than the other backends, should be considered still a work in
3775 progress; known problems include that the Linux kernel in 64-bit
3776 mode does not propagate the correct sigcontext structure to
3777 userspace, and consequently SBCL on a parisc64 kernel will not
3779 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3780 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3781 suppressed.) Also now that the compiler is looking at declarations
3782 in the environment, it checks optimization declarations as well,
3783 and suppresses inlining when (> DEBUG SPEED).
3784 * More fixes have been made to treatment of floating point exception
3785 treatment and other Unix signals. In particular, floating point
3786 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3787 * The detection and handling of control stack exhaustion (infinite
3788 or very deeply nested recursion) has changed. Stack exhaustion
3789 detection is now done by write-protecting pages at the OS level
3790 and applies at all optimization settings; when found, a
3791 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3792 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3793 be caught using IGNORE-ERRORS.
3794 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3795 correctly when transforming recursive LABELS functions to LETs.
3796 (thanks to Alexey Dejneka)
3797 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3798 that are names of constants or global variables.
3799 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3800 alien routines with docstrings.
3801 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3802 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3804 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3805 (thanks to Eric Marsden)
3806 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3807 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3808 * Bug fix: The SPARC backend can now compile functions involving
3809 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3810 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3811 to a functions expecting a PATHNAME-DESIGNATOR.
3812 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3813 lambda lists. (thanks to Alexey Dejneka)
3814 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3815 functions so that EXPORTing the name of the class doesn't cause
3816 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3817 to Antonio Martinez for reporting this)
3818 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3819 constant-folded without failing an assertion. (thanks to Einar
3820 Floystad Dorum for reporting this)
3821 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3822 types are now dealt with more correctly by the compiler.
3823 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3824 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3825 behaviour analogous to RENAME-FILE. This puts its behaviour more
3826 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3828 * The fasl file version number has changed again. (because of the
3829 bug fix involving the names of PCL MAKE-INSTANCE functions)
3831 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3832 * bug fix: Floating point exceptions are treated much more
3833 consistently on the x86/Linux and PPC/Linux platforms.
3834 * Array initialization with :INITIAL-ELEMENT is now much faster for
3835 cases when the compiler cannot open code the array creation, but
3836 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3837 array accesses have also seen a speed increase.
3838 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3839 specified. (This was at the root of some bad interactions between
3840 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3842 * bug fix: Internal error arguments for undefined functions are now
3843 computed correctly on the PPC/Linux platform.
3844 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3845 Raymond Toy's patch for CMU CL)
3846 * Support for the Solaris 9 operating environment has been included
3847 (thanks to Daniel Merritt)
3848 * A very ugly but hopefully complete draft of the missing FFI chapter
3849 of the manual has been created by reformatting the corresponding
3850 CMU CL manual chapter into (currently very ugly and incoherent)
3851 DocBook and bringing it up to date for SBCL behavior. Thus, the
3852 manual is now essentially complete, at least by my extreme
3853 once-and-only-once standards, whereby it's acceptable to refer to
3854 the doc strings of SB-EXT functions as the primary documentation.
3855 * The fasl file version number has changed again, due to cleanup of
3856 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3857 is no longer a static symbol.)
3859 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3860 * SBCL now builds with OpenMCL (version 0.12) as the
3861 cross-compilation host; also, more progress has been made toward
3862 bootstrapping under CLISP.
3863 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3865 * bug 158 fixed: The compiler can now deal with integer loop
3866 increments different from 1; fixing this turned out also to fix
3868 * bug 169 fixed: no more bogus warnings about using lexical bindings
3869 despite the presence of perfectly good SPECIAL declarations (thanks
3870 to David Lichteblau)
3871 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3872 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3873 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3875 * bug 180 fixed: Method combination specifications no longer ignore
3876 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3877 * bug fix: Structure type predicate functions now check their argument
3878 count as they should.
3879 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3880 correctly. (thanks to Pierre Mai)
3881 * minor incompatible change: The --noprogrammer option is deprecated
3882 in favor of the new --disable-debugger option, which is very similar.
3883 (The major difference is that it takes effect at a slightly different
3884 time at startup, causing handling of errors in --sysinit and
3885 --userinit files will be affected differently.) The
3886 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3887 been added to allow this functionality to be controlled from ordinary
3888 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3889 maintainers, who might want to run non-interactive scripts to
3890 build SBCL cores which will later be used interactively.)
3891 * minor incompatible change: The LOAD function no longer, when given
3892 a wild pathname to load, loads all files matching that pathname.
3893 Instead, an error of type FILE-ERROR is signalled.
3895 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3896 * bug 147 fixed: The compiler preserves its block link/count
3897 invariants more correctly now so that it doesn't crash. (thanks
3899 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3901 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3902 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3903 no-value-for-this-character as NIL (as specified by ANSI).
3904 * HOST-NAMESTRING on physical pathnames now returns a string that is
3905 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3906 (thanks to Christophe Rhodes)
3907 * The Alpha port handles icache flushing more correctly. (thanks to
3909 * More progress has been made toward bootstrapping under CLISP. (thanks
3910 to Christophe Rhodes)
3911 * The fasl file format has changed again, because dynamic loading
3912 on OpenBSD (which has non-ELF object files) motivated some cleanups
3913 in the way that foreign symbols are transformed and passed around.
3914 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3915 is no longer treated as whitespace by the reader, but instead as
3916 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3917 |A
\7fB|, instead of A as it used to.
3919 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3920 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3921 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3922 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3923 this as of 0.7.1.45, but was left out of the previous news section
3924 (thanks to Dan Barlow)
3925 * SBCL now runs on the Solaris operating system on SPARC architectures
3926 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3927 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3928 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3929 * SPARC backend cleanups, allowing builds of cores optimized for V8
3930 and V9 SPARCS, and also emission of code targeted to a particular
3931 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3933 * SBCL is closer to bootstrapping under CLISP, thanks to various
3934 fixes by Christophe Rhodes.
3935 * The fasl file format has changed again, to allow the compiler's
3936 INFO database to support symbol macros.
3937 * The user manual (in doc/) is formatted into HTML more nicely.
3938 (thanks to coreythomas)
3939 * The system is smarter about SUBTYPEP relationships, especially
3940 those involving NOT types (including types such as ATOM which are
3941 represented internally using NOT types). Thus SUBTYPEP is less
3942 likely to return (VALUES NIL NIL) in general, and in particular
3943 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3945 * The fasl file format has changed again, because the internal
3946 representation of types now includes a new slot to support the new
3947 SUBTYPEP-of-NOT-types logic.
3948 * (not a change in the main branch of SBCL, but a related prototype
3949 which can hopefully be merged into the main branch of SBCL in the
3950 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3951 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3952 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3953 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3954 dispatch macro character combinations. (thanks to Alexey Dejneka)
3955 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3956 that superficially look like logical namestrings correctly.
3957 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3959 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3960 default supplied-p) for &optional and &key arguments. (thanks to
3963 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3964 * incompatible change: The compiler is now less aggressive about
3965 tail call optimization, doing it only when (> SPACE DEBUG) or
3966 (> SPEED DEBUG). (This is an incompatible change because there are
3967 programs which relied on the old CMU-CL-style behavior to optimize
3968 away their unbounded recursion which will now die of stack overflow.)
3969 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3970 for non-GENCGC systems has been increased to 20M (since that
3971 seems much closer to the likely performance optimum for modern
3972 systems than the old 4M value was)
3973 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3974 and *DEBUG-PRINT-LEVEL*
3975 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3976 of CMU CL's support for SPARC, and various endianness and other
3977 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3978 * new syntactic sugar for the Unix command line: --load foo.bar is now
3979 an alternate notation for --eval '(load "foo.bar")'.
3981 ** The system now detects stack overflow and handles it gracefully,
3982 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3983 optimization settings. (This is a good thing in general, and
3984 its introduction in this version should be particularly timely
3985 for anyone whose code fails because of suppression of tail
3987 ** The system now hunts for the C variable "environ" in a more
3988 devious way, to avoid segfaults when the C library version differs
3989 between compile time and run time. (thanks to Christophe Rhodes)
3990 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3991 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3992 ** The compiler no longer issues bogus style warnings for undefined
3993 classes in the same source file as the DEFCLASSes which defined
3994 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3996 ** fixes in CONDITION class precedence list for undefined function
3997 errors (thanks to Alexei Dejneka)
3998 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3999 correctly. (thanks to Dan Barlow)
4000 ** portability fixes aiming at bootstrapping under CLISP (thanks
4001 to Dave McDonald and Christophe Rhodes)
4002 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4003 ** fixes in type translation and and type inference (thanks to
4005 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4006 ** various fixes in the new ports (thanks to Dan Barlow)
4007 * several changes related to debugging:
4008 ** suppression of tail recursion, as noted above
4009 ** stack overflow detection, as noted above
4010 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4011 is now the default. (For some time encapsulation has been more
4012 reliable than the breakpoint-based :ENCAPSULATE NIL
4013 implementation, at least on X86 systems; and I just noticed that
4014 encapsulation also seems closer to the spirit of the ANSI
4017 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4019 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4020 up properly again. (There was a packaging bug in 0.7.0 which
4021 left their definitions in SB-SYS::LOAD-FOREIGN and
4022 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4023 vital for most things which interface to C-level interfaces,
4024 like extensions working with sockets or databases or
4025 Perl-compatible regexes or whatever, and the need to fix
4026 this bug is the main reason that 0.7.1 was released so
4028 ** DEFGENERIC is now choosier about the methods it redefines, so that
4029 reLOADing a previously-LOADed file containing DEFGENERICs does
4030 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4031 cycle now works as it should. (thanks to Alexey Dejneka)
4032 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4033 fixed by Christophe Rhodes. (He actually submitted this patch
4034 months ago, and I delayed until after 0.7.0.)
4035 ** Bug 111 (internal compiler confusion about runtime checks on
4036 FUNCTION types) was fixed by Alexey Dejneka.
4037 * Some internal cleanups (getting rid of variables which aren't
4038 needed now that the byte interpreter is gone) caused the fasl
4039 file format number to change again.
4041 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4042 * major incompatible change: The default fasl file extension, i.e. the
4043 default extension for files produced by COMPILE-FILE, has changed
4044 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4046 ** There are many changes in the implementation of the compiler.
4047 SBCL is now essentially a compiler-only implementation of ANSI
4048 Common Lisp. EVAL still "interprets" a few special cases, but
4049 almost all the interesting cases are handled by creating
4050 a LAMBDA expression, calling COMPILE on it, then calling
4051 FUNCALL on the result.
4052 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4053 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4054 Since the code is newer, there might still be some new bugs
4055 (though not as many as before Martin Atzmueller's fixes:-). But
4056 the new code is substantially simpler and clearer, and hopefully
4057 any remaining bugs will be simpler, less fundamental, and more
4058 fixable then the bugs in the old code.
4059 ** The revised compiler is still a little unsteady on its feet.
4061 *** The debugging information it produces (particularly the names
4062 of FUNCTION objects) is sometimes much less useful than what
4063 the old compiler produced.
4064 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4065 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4066 within a MACROLET) has been temporarily weakened.
4067 ** There are new compiler optimizations for various functions:
4068 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4069 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4070 *** the math functions TRUNCATE, FLOOR, and CEILING
4071 *** the function-of-all-trades COERCE
4072 Mostly these should be transparent, but there's one
4073 potentially-annoying problem (bug 117): when the compiler
4074 inline-expands a function and does type analysis on the result,
4075 it can create control paths which have type mismatches, and
4076 when it can't prove that those control paths aren't taken,
4077 it will issue WARNINGs about the type mismatches. This is
4078 a particular problem in practice for the new sequence functions.
4079 It's not clear how this should be fixed, and for now, a
4080 workaround is given in the entry for 117 in the BUGS file.
4081 ** (Because of the interaction between the two previous items --
4082 occasional inlining problems and new inline expansions -- some
4083 of the new sequence function optimizations won't really kick in
4084 completely until debugging information, and then inlining, are
4085 straightened out in some future version.)
4086 * minor incompatible changes:
4087 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4088 DIRECTORY no longer implicitly promotes NIL slots of its
4089 pathname argument to :WILD. In particular, when you ask for the
4090 contents of a directory (which you used to be able to do without
4091 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4092 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4093 ** changes in behavior that ANSI explicitly defines to be
4094 implementation dependent:
4095 *** The new compiler-only implementation still conforms with ANSI,
4096 but acts a little different than before. Besides the obvious
4097 changes in performance tradeoffs (that the cost per form passed
4098 to EVAL has gone up, and the cost per form executed by EVAL
4099 has gone down), the behavior of the system changes a little
4100 because there are no longer any interpreted function objects.
4101 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4102 e.g. doing COMPILE on the output of interactive DEFUN is
4104 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4106 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4107 has changed from (:CL) to NIL.
4108 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4109 e.g. #\Newline and #\Tab, don't have names specified in the
4110 ANSI Common Lisp standard, is now based on their ASCII symbolic
4111 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4112 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4113 are no longer used for output.
4114 ** changes in internal implementation constants:
4115 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4116 4 million. (If your application spends a lot of time GCing and
4117 you have a lot of RAM, you might want to experiment with
4118 increasing it even more.)
4119 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4120 However, almost all old code should still continue to work without
4121 immediate update, as SB-C-CALL is now a (deprecated) nickname
4123 ** Old operator names in the style DEF-FOO are now deprecated in
4124 favor of new corresponding names DEFINE-FOO, for consistency with
4125 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4126 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4127 internal symbols, but a few supported extensions like
4128 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4129 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4130 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4131 etc. as you get deeper into recursive calls to the debugger
4132 command loop, instead of the old "5]", "5]]", "5]]]"
4133 sequence. (I was motivated to do this when squabbles between
4134 ILISP and SBCL left me very deeply nested in the debugger. In the
4135 short term, this change will probably provoke more ILISP/SBCL
4136 squabbles, but hopefully it will be an improvement in the long run.)
4137 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4138 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4139 function, since declaiming return types involving aliens is
4140 (1) annoyingly messy to do by hand and (2) vital to efficient
4141 compilation of code which calls such functions.
4142 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4143 longer reexported by the SB-EXT package. They're solely useful
4144 for alien code, so it seems more logical that you should get
4145 them from the SB-ALIEN package, not in SB-EXT.
4146 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4147 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4148 features. Instead, the code that they used to control is always
4149 built into the system.
4150 * many other bug fixes
4151 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4152 advantage of the new EVAL-WHEN stuff and to clean them up in
4153 general, and they are now more ANSI-compliant in a number of
4154 ways. Martin Atzmueller is responsible for a lot of this.
4155 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4157 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4158 *** correct ERROR type for various file operations
4159 *** some fixes for Lisp streams
4160 *** DEFMETHOD syntax checking
4161 *** changing old weird representation of debug information as
4162 strings (which, among their other deficiencies, don't transform
4163 correctly when you rename packages, and don't change their
4164 print representation when you change things like *PACKAGE*
4165 and *PRINT-LENGTH*) to symbols and lists of symbols
4166 He also made several improvements and fixed several bugs in DESCRIBE.
4167 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4169 *** misbehavior of WRITE-STRING/WRITE-LINE
4170 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4171 and several other LOOP problems as well
4172 *** DIRECTORY when similar filenames are present
4173 *** DEFGENERIC with :METHOD options
4174 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4175 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4176 *** argument ordering in FIND with :TEST option
4177 *** mishandled package designator argument in APROPOS-LIST
4178 *** various problems in the backquote readmacro
4179 *** a bug in APROPOS
4180 *** probably some others that I'm not describing very well here,
4181 since the CVS log documents them by reference to sbcl-devel
4182 messages, and the SourceForge archives aren't working well.:-(
4183 ** Dan Barlow improved the Alpha port (and is making progress on the
4184 PPC port, for those of you who think different).
4185 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4186 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4187 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4188 continues to work on the SPARC port (for those of you in a position
4189 to look down upon our little PC-compatible boxes from a great height).
4190 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4191 and uses that copy, rather than the current dynamic value, when
4192 it's trying to decide whether to truncate output. Thus e.g.
4193 (let ((*print-lines* 50))
4194 (pprint-logical-block (stream nil)
4196 (let ((*print-lines* 8))
4197 (print (aref possiblybigthings i) stream)))))
4198 should now truncate the logical block only at 50 lines, instead of
4199 often truncating it at 8 lines, as it did before.
4200 * The doc/cmucl/ directory, containing old CMU CL documentation
4201 from the time of the fork, is no longer part of the base system.
4202 SourceForge has shut down its anonymous FTP service, and with it
4203 my original plan for distributing the old CMU CL documentation
4204 there. For now, if you need these files you can download an old
4205 SBCL source release and extract them from it.
4206 * The fasl file version number changed again, for dozens of reasons,
4207 some of which are apparent above.
4209 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4210 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4211 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4212 patch, so that DISASSEMBLE output is much nicer.
4213 * The code in the SB-PROFILE package now seems reasonably stable.
4214 I still haven't decided what the final interface should look like
4215 (I'd like PROFILE to interact cleanly with TRACE, since both
4216 facilities use function encapsulation) but if you have a need
4217 for profiling now, you can probably use it successfully with
4218 the current CMU-CL-style interface.
4219 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4220 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4221 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4222 link now returns the dangling link itself, and for similar
4223 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4224 link itself. (In these cases the old code signalled an error and
4225 looped endlessly, respectively.) Thus, DIRECTORY now works even
4226 in the presence of dangling and cyclic symbolic links.
4227 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4228 is now a supported extension again, since the consensus on
4229 sbcl-devel was that it can be useful for ordinary development
4230 work, not just for debugging SBCL itself.
4231 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4232 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4233 that currently-defined functions might be redefined later with
4234 different return types.
4235 * Hash tables can be printed readably, as inspired by CMU CL code
4236 of Eric Marsden and SBCL code of Martin Atzmueller.
4237 * better error handling in CLOS method combination, thanks to
4238 Martin Atzmueller porting Pierre Mai's CMU CL patches
4239 * more overflow fixes for >16Mbyte I/O buffers
4240 * A bug in READ has been fixed, so that now a single Ctrl-D
4241 character suffices to cause end-of-file on character streams.
4242 In particular, now you only need one Ctrl-D at the command
4243 line (not two) to exit SBCL.
4244 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4246 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4247 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4248 does the right thing.
4249 * The implementation of some type tests, especially for CONDITION
4250 types, is now tidier and maybe faster, due to CMU CL code
4251 originally by Douglas Crosher, ported by Martin Atzmueller.
4252 * Some math functions have been fixed, and there are new
4253 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4254 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4255 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4256 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4257 likely to be built into sbcl-0.7.0.)
4258 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4259 CL, ported to SBCL by Martin Atzmueller)
4260 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4261 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4262 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4263 * There's a new slam.sh hack to shorten the edit/compile/debug
4264 cycle for low-level changes to SBCL itself, and a new
4265 :SB-AFTER-XC-CORE target feature to control the generation of
4266 the after-xc.core file needed by slam.sh.
4267 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4268 COMPILE-FILE is no longer supported, so that now every function
4269 gets an entry point, so that block compilation looks a little
4270 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4271 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4272 returns the number of bytes consed since the system started,
4273 rather than the number consed since the first time the function
4274 was called. (The new definition parallels ANSI functions like
4275 CL:GET-INTERNAL-RUN-TIME.)
4276 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4277 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4278 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4279 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4280 * Fasl file version numbers are now independent of the target CPU,
4281 since historically most system changes which required version
4282 number changes have affected all CPUs equally. Similarly,
4283 the byte fasl file version is now equal to the ordinary
4286 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4287 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4288 is no longer recognized. I apologize for this, because it was
4289 listed in SB-EXT as a supported extension, but I found that
4290 its existing behavior was poorly specified, as well as incorrectly
4291 specified, and it looked like too much of a mess to straighten it
4292 out. I have enough on my hands trying to get ANSI stuff to work..
4293 * many patches ported from CMU CL by Martin Atzmueller, with
4294 half a dozen bug fixes in pretty-printing and the debugger, and
4295 half a dozen others elsewhere
4296 * fixed bug 13: Floating point infinities are now supported again.
4297 They might still be a little bit flaky, but thanks to bug reports
4298 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4299 as flaky as they were.
4300 * The --noprogrammer command line option is now supported. (Its
4301 behavior is slightly different in detail from what the old man
4302 page claimed it would do, but it's still appropriate under the
4303 same circumstances that the man page talks about.)
4304 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4305 are now supported, and enabled by default. Thus, the compiler can
4306 handle many floating point and complex operations much less
4307 inefficiently. (Thus e.g. you can implement a complex FFT
4309 * The compiler now detects type mismatches between DECLAIM FTYPE
4310 and DEFUN better, and implements CHECK-TYPE more correctly, and
4311 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4312 thanks to patches from Martin Atzmueller.
4313 * various fixes to make the cross-compiler more portable to
4314 ANSI-conforming-but-different cross-compilation hosts (notably
4315 Lispworks for Windows, following bug reports from Arthur Lemmens)
4316 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4317 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4318 fixed thanks to Pierre Mai's CMU CL patches.
4319 * fixes to keep the system from overflowing internal counters when
4320 it tries to use i/o buffers larger than 16M bytes
4321 * fixed bug 45a: Various internal functions required to support
4322 complex special functions have been merged from CMU CL sources.
4323 (When I was first setting up SBCL, I misunderstood a compile-time
4324 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4325 * improved support for type intersection and union, fixing bug 12
4326 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4327 more obscure bugs as well
4328 * some steps toward byte-compiling non-performance-critical
4329 parts of the system, courtesy of patches from Martin Atzmueller
4330 * Christophe Rhodes has made some debian packages of sbcl at
4331 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4332 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4333 stable, but are nonetheless usable. When he's ready, I'd be happy
4334 to add them to the SourceForge "File Releases" section. (And if
4335 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4336 * new fasl file format version number (because of changes in
4337 internal representation of (OR ..) types to accommodate the new
4338 support for (AND ..) types, among other things)
4340 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4341 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4343 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4345 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4346 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4347 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4348 * DESCRIBE now works on structure objects again.
4349 * Most function call argument type mismatches are now handled as
4350 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4351 know whether the function will be redefined before the call is
4352 executed. (The compiler could flag local calls with full WARNINGs,
4353 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4354 it doesn't keep track of enough information to know whether calls
4355 are local in this sense.)
4356 * Compiler output is now more verbose, with messages truncated
4357 later than before. (There should be some supported way for users
4358 to override the default verbosity, but I haven't decided how to
4359 provide it yet, so this behavior is still controlled by the internal
4360 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4361 src/compiler/ir1util.lisp.)
4362 * Fasl file format version numbers have increased again, because
4363 support for the Gray streams extension changes the layout of the
4364 system's STREAM objects.
4365 * The Gray subclassable streams extension now works, thanks to a
4366 patch from Martin Atzmueller.
4367 * The full LOAD-FOREIGN extension (not just the primitive
4368 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4369 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4370 but like most other programs, it defaults to copying the Unix
4371 environment from the original process instead of starting the
4372 new process in an empty environment.
4373 * Extensions which manipulate the Unix environment now support
4374 an :ENVIRONMENT keyword option which doesn't smash case or
4375 do other bad things. The CMU-CL-style :ENV option is retained
4376 for porting convenience.
4377 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4378 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4380 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4382 * A patch from Martin Atzmueller seems to have solved the SIGINT
4383 problem, and as far as we know, signal-handling now works cleanly.
4384 (If you find any new bugs, please report them!)
4385 * The system no longer defaults Lisp source file names to types
4386 ".l", ".cl", or ".lsp", but only to ".lisp".
4387 * The compiler no longer uses special default file extensions for
4388 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4389 seems to expect a single default extension for all compiled code,
4390 and there's no compelling reason to try to stretch the standard
4391 to allow two different extensions.) Instead, byte-compiled files
4392 default to the same extension as native-compiled files.
4393 * Fasl file format version numbers have increased again, because
4394 a rearrangement of internal implementation packages made some
4395 dumped symbols in old fasl files unreadable in new cores.
4396 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4397 many fewer weird special cases.
4398 * Bug #17 (differing COMPILE-FILE behavior between logical and
4399 physical pathnames) has been fixed, and some related misbehavior too,
4400 thanks to a patch from Martin Atzmueller.
4401 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4402 by Tim Moore, ported to SBCL by Martin Atzmueller.
4403 * Martin Atzmueller fixed several filesystem-related problems,
4404 including bug #36, in part by porting CMU CL patches, which were
4405 written in part by Paul Werkowski.
4406 * More compiler warnings in src/runtime/ are gone, thanks to
4407 more patches from Martin Atzmueller.
4408 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4411 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4413 * DESCRIBE now works on CONDITION objects.
4414 * The debugger now handles errors which arise when trying to print
4415 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4417 * The build system now uses an additional file, customize-target-features.lisp,
4418 to allow local modifications to the target *FEATURES* list. (The point of
4419 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4420 debugging features enabled, without having to worry about propagating it
4421 into everyone's system when I do a "cvs update".) When no
4422 customize-target-features.lisp file exists, the target *FEATURES* list
4423 should be constructed the same way as before.
4424 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4425 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4426 warn about behavior which is undefined under the ANSI spec. Note: This
4427 is specified by ANSI, but it's not very popular with programmers.
4428 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4429 macro in the SBCL sources for an example of a workaround which you
4430 might use to make portable ANSI-standard code which does what you want.
4431 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4432 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4433 when it's used as a non-toplevel form. (This is required in order
4434 to implement the DEFCONSTANT-EQX macro.)
4435 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4436 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4437 funny things, due to bugs in the implementation of EVAL-WHEN
4438 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4439 non-toplevel DEFCONSTANTs are uncommon.)
4440 * The core file version number and fasl file version number have been
4441 incremented, because the old noncompliant DEFCONSTANT behavior involved
4442 calling functions which no longer exist, and because I also took the
4443 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4444 * fixed bug 1 (error handling before read-eval-print loop starts), and
4445 redid debugger restarts and related debugger commands somewhat while
4447 ** The QUIT debugger command is gone, since it did something
4448 rather different than the SB-EXT:QUIT command, and since it never
4449 worked properly outside the main toplevel read/eval/print loop.
4450 Invoking the new TOPLEVEL restart provides the same functionality.
4451 ** The GO debugger command is also gone, since you can just invoke
4452 the CONTINUE restart directly instead.
4453 ** The TOP debugger command is also gone, since it's redundant with the
4454 FRAME 0 command, and since it interfered with abbreviations for the
4456 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4458 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4459 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4460 some porting work by Martin Atzmueller)
4461 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4462 with a boatload of patches which he ported to SBCL. Now that those
4464 ** The system tries to make sure that its low-priority messages
4465 are prefixed by semicolons, to help people who like to use
4466 syntax highlighting in their ILISP buffer. (This patch
4467 was originally due to Raymond Toy.)
4468 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4469 patches originally written by Raymond Toy.
4470 ** The compiler understands coercion between single-value and
4471 multiple-VALUES type expressions better, getting rid of some very
4472 weird behavior, thanks to patches originally by Robert MacLachlan
4473 and Douglas Crosher.
4474 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4475 lambda lists, thanks to a patch originally by Pierre Mai.
4476 ** The system no longer bogusly warns about "abbreviated type
4478 ** The compiler gets less confused by inlining and RETURN-FROM,
4479 thanks to some patches originally by Tim Moore.
4480 ** The system no longer hangs when dumping circular lists to fasl
4481 files, thanks to a patch originally from Douglas Crosher.
4482 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4483 undefined function error.
4484 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4485 in structure slot initforms) for now, documented workaround instead:-|
4486 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4487 * fixed bug 5: added stubs for various Gray stream functions called
4488 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4489 installed, at least appropriate type errors are generated
4490 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4491 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4492 consistently in DEFMETHOD forms.
4493 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4494 it doesn't seem to affect SBCL after all
4495 * The C runtime system now builds with better optimization and many
4496 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4498 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4500 * The system is now under CVS at SourceForge (instead of the
4501 CVS repository on my home machine).
4502 * The new signal handling code has been tweaked to treat register
4503 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4504 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4505 downstream, e.g. in debug-int.lisp, has implicit dependencies
4506 on the unsignedness of integer representation of machine words,
4507 and that caused the system to bomb out with infinite regress
4508 when trying to recover from type errors involving signed values,
4509 e.g. (BUTLAST '(1 2 3) -1).)
4510 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4511 (This was one of the bugs Peter Van Eynde reported back in July.)
4512 * The system now uses code inspired by Colin Walters' O(N)
4513 implementation of MAP (from the cmucl-imp@cons.org mailing
4514 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4515 inline the MAP operation, and there is more than one
4516 sequence argument to the MAP call (so that it can't just
4517 do ETYPECASE once and for all based on the type of the
4518 single sequence argument). (The old non-inline implementation
4519 of the general M-argument sequence-of-length-N case required
4520 O(M*N*N) time when any of the sequence arguments were LISTs.)
4521 * The QUIT :UNIX-CODE keyword argument has been renamed to
4522 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4523 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4525 * Raymond Wiker's patches to port dynamic loading from Linux to
4526 FreeBSD have been added.
4527 * The BUGS file is now more nearly up to date, thanks in large part
4528 to Martin Atzmueller's review of it.
4529 * The debugger now flushes standard output streams before it begins
4530 its output ("debugger invoked" and so forth).
4531 * The core version number and fasl file version number have both
4532 been incremented, because of incompatible changes in the layout
4534 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4535 used to not be called for a saved Lisp image.)
4536 * Martin Atzmueller's version of a patch to fix a compiler crash,
4537 as posted on sbcl-devel 13 September 2000, has been installed.
4538 * Instead of installing Martin Atzmueller's patch for the
4539 compiler transform for SUBSEQ, I deleted the compiler transform,
4540 and transforms for some similar consing operations.
4541 * A bug in signal handling which kept TRACE from working on OpenBSD
4543 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4544 away by constant folding
4545 * The system now defines its address space constants in one place
4546 (in the Lisp sources), and propagates them automatically elsewhere
4547 (through GENESIS and the sbcl.h file). Therefore, patching the
4548 address map is less unnecessarily tedious and error-prone. The
4549 Lisp names of address space constants have also been systematized.
4550 * CVS tags like dollar-Header-dollar have been removed from
4551 the sources, because they have never saved me trouble and
4552 they've been source of trouble working with patches and other
4553 diff-related operations.
4554 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4555 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4557 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4559 * The system has been ported to OpenBSD.
4560 * The system now compiles with a simple "sh make.sh" on the systems
4561 that it's supported on. I.e., now you no longer need to tweak
4562 text in the target-features.lisp-expr and symlinks in src/runtime/
4563 by hand, the make.sh takes care of it for you.
4564 * The system is no longer so grossly inefficient when compiling code
4565 involving vectors implemented as general (not simple) vectors (VECTOR T),
4566 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4567 which dares to use the various sequence functions on non-simple
4568 vectors, takes less of a performance hit.
4569 * There is now a primitive type predicate VECTOR-T-P
4570 to test for the (VECTOR T) type, so that e.g.
4571 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4572 can now be compiled with some semblance of efficiency. (The old code
4573 turned the type declaration into a full call to %TYPEP at runtime!)
4574 * AREF on (VECTOR T) is still not fast, since it's still compiled
4575 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4576 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4577 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4578 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4579 without making full calls to SUBTYPEP at runtime.
4580 (Some analogous efficiency issues for non-simple vectors specialized to
4581 element types other than T, or for non-simple multidimensional arrays,
4582 have not been addressed. They could almost certainly be handled the
4583 same way if anyone is motivated to do so.)
4584 * The changes in array handling break binary compatibility, so
4585 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4586 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4587 instead of (VALUES T T).
4588 * By following the instructions that Dan Barlow posted to sbcl-devel
4589 on 2 July 2000, I was able to enable primitive dynamic object
4590 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4591 functionality is not implemented (since it calls ld to resolve
4592 library references automatically, requiring RUN-PROGRAM for its
4593 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4594 try to resolve library references) is now supported.
4595 * The system now flushes the standard output streams when it terminates,
4596 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4597 them at several other probably-convenient times, e.g. in each pass of
4598 the toplevel read-eval-print loop, and after evaluating a form given
4599 as an "--eval" command-line option. (These changes were motivated by a
4600 discussion of stream flushing issues on cmucl-imp in August 2000.)
4601 * The source transform for TYPEP of array types no longer assumes
4602 that an array whose element type is a not-yet-defined type
4603 is implemented as an array of T, but instead punts, so that the
4604 type will be interpreted at runtime.
4605 * There is now some support for cross-compiling in make.sh: each of
4606 the phases of make.sh has its own script. (This should be transparent
4607 to people doing ordinary, non-cross-compile builds.)
4608 * Since my laptop doesn't have hundreds of megabytes of memory like
4609 my desktop machine, I became more motivated to do some items on
4610 my to-do list in order to reduce the size of the system a little:
4611 ** Arrange for various needed-only-at-cold-init things to be
4612 uninterned after cold init. To support this, those things have
4613 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4614 symbols with such names are now uninterned after cold init).
4615 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4616 fasl files for cold load.
4617 ** Remove the old compiler structure pooling code (which used to
4618 be conditional on the target feature :SB-ALLOC) completely.
4619 ** Redo the representation of some data in cold init to be more compact.
4620 (I also looked into supporting byte compiled code at bootstrap time,
4621 which would probably reduce the size of the system a lot, but that
4622 looked too complicated, so I punted for now.)
4623 * The maximum signal nesting depth in the src/runtime/ support code has
4624 been reduced from 4096 to 256. (I don't know any reason for the very
4625 large old value. If the new smaller value turns out to break something,
4626 I'll probably just bump it back up.)
4627 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4629 * Many, many bugs reported by Peter Van Eynde have been added to
4630 the BUGS list; some have even been fixed.
4631 * While enabling dynamic object file loading, I tried to make the
4632 code easier to understand, renaming various functions and variables
4633 with less ambiguous names, and changing some function calling
4634 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4635 * While trying to figure out how to do the OpenBSD port, I tried to
4636 clean up some of the code in src/runtime/. In particular, I dropped
4637 support for non-POSIX signal handling, added various comments,
4638 tweaked the code to reduce the number of compilation warnings, and
4639 renamed some files to increase consistency.
4640 * To support the new automatic configuration functionality in make.sh,
4641 the source file target-features.lisp-expr has been replaced with the
4642 source file base-target-features.lisp-expr and the machine-generated
4643 file local-target-features.lisp-expr.
4644 * fixed a stupid quoting error in make.sh so that using CMU CL
4645 "lisp -batch" as cross-compilation host works again
4647 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4649 * DESCRIBE no longer tries to call itself recursively to describe
4650 bound/fbound values, so that it no longer fails on symbols which are
4651 bound to themselves (like keywords, T, and NIL).
4652 * DESCRIBE now works on generic functions.
4653 * The printer now prints less-screwed-up representations of closures
4654 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4655 * A private symbol is used instead of the :EMPTY keyword previously
4656 used to mark empty slots in hash tables. Thus
4657 (DEFVAR *HT* (MAKE-HASH-TABLE))
4658 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4659 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4660 now does what ANSI says that it should. (You can still get
4661 similar noncompliant behavior if bang on the hash table
4662 implementation with all the symbols you get back from
4663 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4664 This breaks binary compatibility, since tests for equality to
4665 :EMPTY are wired into things like the macroexpansion of
4666 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4668 * There's now a minimal placeholder implementation for CL:STEP,
4669 as required by ANSI.
4670 * An obscure bug in the interaction of the normal compiler, the byte
4671 compiler, inlining, and structure predicates has been patched
4672 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4673 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4675 * Missing ordinary arguments in a macro call are now detected even
4676 when the macro lambda list contains &KEY or &REST.
4677 * The debugger no longer complains about encountering the top of the
4678 stack when you type "FRAME 0" to explicitly instruct it to go to
4679 the top of the stack. And it now prints the frame you request even
4680 if it's the current frame (instead of saying "You are here.").
4681 * As specified by ANSI, the system now always prints keywords
4682 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4684 * The default initial SIZE of HASH-TABLEs is now smaller.
4685 * Type information from CLOS class dispatch is now propagated
4686 into DEFMETHOD bodies, so that e.g.
4687 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4689 is now basically equivalent to
4690 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4691 (DECLARE (TYPE SINGLE-FLOAT X))
4693 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4694 operation, without having to do run-time type dispatch.
4695 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4696 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4697 * Since I don't seem to be making big file reorganizations very often
4698 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4699 is overflowing my ability to conveniently back them up), I've finally
4700 checked the system into CVS. (The CVS repository is on my home system,
4701 not at SourceForge -- putting it on SourceForge might come later.)
4702 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4703 high-level GC-NOTIFY-FOO functions send their output. (There's
4704 still very little control of where low-level verbose GC functions
4705 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4706 less than it used to -- the GC-NOTIFY-FOO functions are now under
4707 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4708 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4709 in only one place in the source code, and propagates it automatically
4710 everywhere that it's needed. Thus e.g. when I bump the version from
4711 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4712 * The C source files now include boilerplate legalese and documentation
4713 at the head of each file (just as the Lisp source files already did).
4714 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4715 to his page will be replaced with a link to his new CLiki service.
4717 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4719 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4720 * The build process now looks for GNU make under the default name "gmake",
4721 instead of "make" as it used to. If GNU make is not available as "gmake"
4722 on your system, you can change this default behavior by setting the
4723 GNUMAKE environment variable.
4724 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4725 can build without error under CMU CL.
4727 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4729 * There is now a partial SBCL user manual (with some new text and some
4730 text cribbed from the CMU CL manual).
4731 * The beginnings of a profiler have been added (starting with the
4732 CMU CL profiler and simplifying and cleaning up). Eventually the
4733 main interface should be through the TRACE macro, but for now,
4734 it's still accessed through vaguely CMU-CL-style functions and macros
4735 exported from the package SB-PROFILE.
4736 * Some problems left over from porting CMU CL to the new
4737 cross-compilation bootstrap process have been cleaned up:
4738 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4739 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4740 part when getting rid of the compiler *BACKEND* stuff.)
4741 ** The value of *NULL-TYPE* was screwed up, because it was
4742 being initialized before the type system knew the final
4743 definition of the 'NULL type. This screwed up several key
4744 optimizations in the compiler, causing inefficiency in all sorts
4745 of places. (I found it because I wanted to understand why
4746 GET-INTERNAL-RUN-TIME was consing.)
4747 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4748 PROCLAIM FTYPE information. Unfortunately this broke binary
4749 compatibility again, since now the forms output by DEFGENERIC
4750 to refer to functions which didn't exist in 0.6.3.
4751 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4752 can use the new (as of 0.6.3) transform for SOME into MAP into
4754 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4755 values to (INTEGER 0 1000000), so that the time code will no longer
4756 occasionally get blown up by Linux returning 1000000 microseconds
4757 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4758 its output conform to the ANSI spec. (Alas, this makes its output
4759 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4760 * A full call to MAP NIL with a single sequence argument no longer conses.
4761 * fixes to problems pointed out by Martin Atzmueller:
4762 * The manual page no longer talks about multiprocessing as though
4763 it were currently supported.
4764 * The ILISP support patches have been removed from the distribution,
4765 because as of version 5.10.1, ILISP now supports SBCL without us
4766 having to maintain patches.
4767 * added a modified version of Raymond Toy's recent CMU CL patch for
4768 EQUALP comparison of HASH-TABLE
4770 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4772 * The system still can't cross-compile itself with
4773 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4774 entails), but at least it can compile more of itself that way
4775 than it used to be able to, and various buglets which were uncovered
4776 by trying to cross-compile itself that way have now been fixed.
4777 * This release breaks binary compatibility again. This time
4778 at least I've incremented the FASL file format version to 2, so that the
4779 problem can be detected reliably instead of just causing weird errors.
4780 * various new style warnings:
4781 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4782 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4783 ** using the lexical binding of a variable named in the *FOO* style
4784 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4785 as specified by ANSI.
4786 * *RANDOM-STATE* is no longer automatically initialized from
4787 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4788 default behavior of the system is to repeat its behavior every time
4789 it's run. If you'd like to change this behavior, you can always
4790 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4791 old convention there was no comparably easy way to get the system to
4792 repeat its behavior every time it was run.
4793 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4794 FTYPE declarations has been removed, in favor of their ANSI
4795 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4796 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4797 the types of their sequence arguments can be determined at compile time.
4798 This is done through a new open code expansion for MAP which eliminates
4799 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4800 argument types can be determined at compile time.
4801 * The optimizer now transforms COERCE into an identity operation when it
4802 can prove that the coerced object is already of the correct type. (This
4803 can be a win for machine generated code, including the output of other
4804 optimization transforms, such as the MAP transform above.)
4805 * Credit information has been moved from source file headers into CREDITS.
4806 * Source file headers have been made more standard.
4807 * The CASE macro now compiles without complaining even when it has
4810 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4812 * (Note that the way that the PCL macroexpansions were rewritten
4813 to accommodate the change in DEFGENERIC below breaks binary
4814 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4815 not run under sbcl-0.6.2. Once we get out of alpha releases,
4816 i.e. hit release 1.0.0, we'll probably try to maintain binary
4817 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4818 and sbcl-1.4.4. Until then, however, it might be fairly common
4819 for maintenance releases to break binary compatibility.)
4820 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4822 * The reporting of unhandled signals has been changed to print some
4823 explanatory text as well as the report form. (Previously only
4824 the report form was printed.)
4825 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4826 it defines, so that the compiler no longer issues undefined function
4827 warnings for compiled-but-not-yet-loaded generic functions.
4828 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4829 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4830 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4831 section of the ANSI spec are supported.
4832 * The "" nickname for the "KEYWORD" package has been removed.
4833 The reader still handles symbol tokens which begin with a package marker
4834 as keywords, but it doesn't expose its mechanism for doing so in the
4835 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4836 * The system now issues STYLE-WARNINGs for contradictory TYPE
4837 proclamations. (Warnings for contradictory FTYPE proclamations would
4838 be nice too, but those can't be done usefully unless the type system
4839 is made smarter about FUNCTION types.)
4840 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4841 systematized, so that "*target-*.lisp is supposed to exist only on the
4842 target and imply that there's a related file which exists on the
4843 host, and *host-*.lisp is supposed to exist only on the host and imply
4844 that there's a related file which exists on the target. This involves a
4845 lot of renaming. Hopefully the acute confusion caused by the renaming
4846 will be justified by the reduction in chronic confusion..
4847 ** runtime-type.lisp -> early-target-type.lisp
4848 ** target-type.lisp -> late-target-type.lisp
4849 ** early-host-format.lisp -> early-format.lisp
4850 ** late-host-format.lisp -> late-format.lisp
4851 ** host-error.lisp -> misc-error.lisp
4852 ** early-error.lisp -> early-target-error.lisp
4853 ** late-error.lisp -> late-target-error.lisp
4854 ** host-defboot.lisp -> early-defboot.lisp
4855 ** code/misc.lisp -> code/target-misc.lisp
4856 ** code/host-misc.lisp -> code/misc.lisp
4857 ** code/numbers.lisp -> code/target-numbers.lisp
4858 ** code/early-numbers.lisp -> numbers.lisp
4859 ** early-host-type.lisp -> early-type.lisp
4860 ** late-host-type.lisp -> late-type.lisp
4861 ** host-typep.lisp -> typep.lisp
4862 ** load.lisp -> target-load.lisp
4863 ** host-load.lisp -> load.lisp
4864 ** host-disassem.lisp -> disassem.lisp
4865 ** host-insts.lisp -> insts.lisp
4866 ** byte-comp.lisp -> target-byte-comp.lisp
4867 ** host-byte-comp.lisp -> byte-comp.lisp
4868 ** host-signal.lisp -> signal.lisp
4869 ** host-defstruct.lisp -> defstruct.lisp
4870 ** late-target-type.lisp -> deftypes-for-target.lisp
4871 Furthermore, several other previously target-only files foo.lisp (e.g.
4872 hash-table.lisp and random.lisp) have been split into a target-and-host
4873 foo.lisp file and a target-only target-foo.lisp file, with their key type
4874 definitions in the target-and-host part, so that the cross-compiler will
4875 know more about target types.
4876 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4877 gone away. In their place are various *BACKEND-FOO* variables
4878 corresponding to the slots of the old structure.
4879 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4880 into the target SBCL has been fixed.
4881 * The chill.lisp system for loading cold code into a running SBCL
4883 * Support for the CMU CL "scavenger hook" extension has been removed.
4884 (It was undocumented and unused in the CMU CL sources that SBCL was
4885 derived from, and stale in sbcl-0.6.1.)
4886 * Various errors in the cross-compiler type system were detected
4887 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4888 (enabling various consistency checks). Many of them were fixed,
4889 but some hard problems remain, so the compiler is back to
4890 running without *TYPE-SYSTEM-INITIALIZED* for now.
4891 * As part of the cross-compiler type system cleanup, I implemented
4892 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4893 * I have started adding UNCROSS calls throughout the type system
4894 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4895 on cross-compiler input files will be able to go away, and various
4897 * CONSTANTP now returns true for quoted forms (as explicitly required
4900 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4902 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4903 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4904 (MAKE-LIST -1). In the longer term, it ought to become true
4905 that declarations are assertions even at SAFETY 1. For now, it's not
4906 quite true even at SAFETY 3, but it's at least more nearly true..
4907 (Note that this change seems to increases the size of the system by
4908 O(5%) and to decrease the speed of the compiler by 20% or more.)
4909 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4910 for the problem of printing dozens of lines of distracting information
4911 about low-level system machinery as part of the top stack frame
4912 on entry to the debugger when an undefined function was called.
4913 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4914 is not reset to COMMON-LISP-USER.
4915 * Compilation of stuff related to dyncount.lisp has been made conditional
4916 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4917 smaller. The various dyncount-related symbols have been moved into
4918 a new "SB-DYNCOUNT" package.
4919 * tty-inspect.lisp has been renamed to inspect.lisp.
4920 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4921 feature has gone away. (When we eventually port to other flavors of
4922 libc and/or Unix, we'll try to make the differences between flavors
4923 invisible at the user level.)
4924 * Various other *FEATURES* tags, and/or their associated conditionals,
4925 have been removed if obsolescent, or given better documentation, or
4926 sometimes given more-mnemonic names.
4928 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4930 * tidied up "make.sh" script
4931 * tidied up system directory structure
4932 * better "clean.sh" behavior
4933 * added doc/FOR-CMUCL-DEVELOPERS
4934 * many many small tweaks to output format, e.g. removing possibly-confusing
4935 trailing #\. character in DESCRIBE-INSTANCE
4936 * (EQUALP #\A 'A) no longer signals an error.
4937 * new hashing code, including EQUALP hashing
4938 * tidied up Lisp initialization and toplevel
4939 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4940 * command line argument processing
4941 * added POSIX-GETENV function to deal with Unix-ish environment variables
4942 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4943 terminating SBCL on EOF
4944 * non-verbose GC by default
4945 * There is no more "sbcl" shell script; the sbcl file is now the C
4946 runtime executable (just like CMU CL).
4947 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4949 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4950 symbol and package information as Lisp native symbol and package objects
4951 instead of strings naming symbols and strings naming packages. This way,
4952 whenever packages are renamed (as in warm init), debug information is
4953 transformed along with everything else.
4954 * tweaked the optimization policy declarations which control the building
4955 of SBCL itself. Now, among other things, the system no longer saves
4956 source location debugging information. (This helps two problems at once
4957 by reducing SBCL size and by keeping SBCL from trying to look for its
4958 sources -- which may not exist -- when reporting errors.)
4959 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4960 debugging and testing purposes
4961 * cleaned up printing, making the printer call PRINT-OBJECT for
4962 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4963 methods, giving nearly-ANSI behavior
4964 * converted almost all special variables to use *FOO* naming convention
4965 * deleted PARSE-TIME functionality, since it can be done portably
4966 * moved some files out of cold init into warm init
4967 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4969 * regularized formatting of source files
4970 * added an install.sh script
4971 * fixed ridiculous memory usage of cross-compiler by making
4972 compiler/alloc.lisp not try to do pooling unless it can hook
4973 itself into the GC of the cross-compilation host. Now the system
4974 builds nicely on my old laptop.
4975 * added :SB-ALLOC in target-features.lisp-expr
4976 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4977 was not implemented)
4978 * re-did condition handling and note reporting in the compiler. Notes
4979 are no longer handled by signalling conditions. Style warnings
4980 and warnings are handled more correctly and reported in such a way
4981 that it's easy to find one or the other in your output (so that you
4982 can e.g. figure out which of many problems caused COMPILE-FILE to
4984 * changed the severity of several compiler warnings from full WARNING
4985 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4986 compiler note reporting so that it doesn't use the condition system
4987 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4988 COMPILE-FILE command)
4989 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4990 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4991 no longer supported, since their implementation was deeply intertwingled
4992 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4993 * removed bogus "support" for compiler macros named (SETF FOO), and
4994 removed the compiler macro for SETF INFO (but only after making a fool
4995 of myself on the cmucl-imp mailing list by posting a bogus patch for
4996 DEFINE-COMPILER-MACRO..)
4997 * Compiled files containing forms which have side effects on the Lisp
4998 reader (such as DEFPACKAGE forms) are now handled more correctly.
4999 (Compiler queuing of top level lambdas has been suppressed by setting
5000 *TOP-LEVEL-LAMBDA-MAX* to 0. )
5001 * deleted various currently-unused source files, e.g. gengc.lisp. They
5002 may be added back at some point e.g. when porting to other architectures,
5003 but until they are it's distracting to distribute them and to try to
5005 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5006 there were so many of them they're just distractions, and UNCROSS is
5007 known to be able to handle the current sources
5008 * moved PROFILE functionality into TRACE, so that it will be clear
5009 how the wrapping and unwrapping of functions when you profile them
5010 interacts with the wrapping and unwrapping of functions when you
5011 trace them. (Actually, the functionality isn't there yet, but at least
5012 the interface specification is there. Hopefully, the functionality will
5013 arrive with some future maintenance release.)
5014 * removed host-oops.lisp
5015 * changed signature of QUIT function to allow UNIX-CODE argument
5016 * fixed READ-SEQUENCE bug
5017 * tweaked verbose GC output so that it looks more like the progress
5018 output that ANSI specifies for functions like LOAD
5019 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5020 * added <http://sbcl.sourceforge.com> to the banner information printed by