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7 * enhancement: MUTEX-VALUE is to be superseded by MUTEX-OWNER that has a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)
32 changes in sbcl-1.0.24 relative to 1.0.23:
33 * new feature: ARRAY-STORAGE-VECTOR provides access to the underlying data
34 vector of a multidimensional SIMPLE-ARRAY.
35 * new feature: the system now signals a continuable error if standard
36 readtable modification is attempted. (thanks to Tobias Rittweiler)
37 * new feature: DIRECTORY has been extended with a non-standard keyword
38 argument :RESOLVE-SYMLINKS. (thanks to TC-Rucho)
39 * enhancement: SB-BSD-SOCKETS:MAKE-SOCKET-STREAM keyword argument handling
40 has been robustified and documented better. (thanks to Robert Goldman)
41 * optimization: faster generic arithmetic dispatch on x86 and x86-64.
42 * optimization: FORMAT ~D and ~F are now approximately 5% faster.
43 * tradeoff: constant FORMAT control strings are now compiled unless
44 SPACE > SPEED (previously only when SPEED > SPACE.)
45 * bug fix: Red Hat Enterprise 3 mmap randomization workaround. (thanks
47 * bug fix: DEFCLASS and ENSURE-CLASS-USING-CLASS are now expected to
49 * bug fix: lexical type declarations are now correctly reported by
50 SB-CLTL2. (reported by Larry D'Anna)
51 * bug fix: STRING-TO-OCTETS did not handle :START properly when
52 using UTF-8 as external format. (thanks to Luis Oliveira)
53 * bug fix: errors from invalid fill-pointer values to (SETF FILL-POINTER)
54 are signalled correctly. (thanks to Stas Boukarev)
55 * bug fix: SET-MACRO-CHARACTER accepts NIL as the readtable
56 designator. (thanks to Tobias Rittweiler)
57 * bug fix: SET-DISPATCH-MACRO-CHARACTER accepts NIL as the readtable
58 designator, and returns T instead of the function. (thanks to
60 * bug fix: direct superclasses of STANDARD-CLASS and
61 FUNCALLABLE-STANDARD-CLASS now default to STANDARD-OBJECT and
62 FUNCALLABLE-STANDARD-OBJECT as required by AMOP.
63 * bug fix: compiling a call to SLOT-VALUE with a constant slot-name
64 when no class with the named slot yet exists no longer causes a
65 compile-time style-warning.
66 * bug fix: :ALLOCATION :CLASS slots are type-checked properly
67 in safe code. (reported by Didier Verna)
68 * bug fix: #430; nested structure constructors can stack allocate.
69 * bug fix: on some 64-bit platforms dynamic space size was truncated
70 to #xffff0000 bytes. (reported by Benjamin Lambert)
71 * bug fix: setting *READ-SUPPRESS* to T no longer renders the default
72 REPL unusable. (reported by Daniel Herring)
73 * bug fix: return values of READ-SEQUENCE did not take :START into
74 account on file streams, regressions since 1.0.12.22. (reported by
75 Thomas Russ, patch by Paul Khuong)
76 * bug fix: using SET or (SETF SYMBOL-VALUE) to change the value of a
77 method specializer used to confuse permuation vector optimization.
78 * bug fix: system inserted bogus implicit type declarations for local
79 special variables in DEFMETHOD bodies.
80 * bug fix: #354; duplicated frames in backtraces due to
81 non-tail-call-optimized XEPs to functions with return type NIL
83 * bug fix: #357; MAKE-INSTANCE/SHARED-INITIALIZE now
84 initializes structure object slots according to DEFSTRUCT initforms,
85 and DEFSTRUCT forms :INCLUDEind structure classes defined using
86 DEFCLASS :METACLASS STRUCTURE-CLASS now inherit their initforms.
87 (reported by Bruno Haible and Stephen Wilson)
88 * bug fix: #395; fill-pointer output streams used now support
89 element-type BASE-CHAR as well.
90 * bug fix: compiler error when attempting to derive return value of
91 ARRAY-ELEMENT-TYPE when the array type was a union of intersection
93 * bug fix: address-spaces overlapped on OpenBSD. (thanks to Josh
95 * bug fix: Mac OS X binaries should now be portable between Leopard
98 changes in sbcl-1.0.23 relative to 1.0.22:
99 * enhancement: when disassembling method functions, disassembly
100 for the associated fast function is also produced.
101 * enhancement: system stores DEFTYPE lambda-lists, so DESCRIBE can
103 * optimization: printing with *PRINT-PRETTY* true is now more
104 efficient as long as the object being printed doesn't require
105 special handling by the pretty printer.
106 * bug fix: slot symbol-macros from WITH-SLOTS inside DEFMETHOD bodies
107 now interact correctly with type declarations.
108 * partial bug fix: PCL detects infinite recursion during wrapper
109 validation. (thanks to Attila Lendvai)
110 * bug fix: #426; nested function calls are inlined properly.
111 Previously if FOO was an inline function, in calls of the form
112 (FOO (FOO ...)) the outer call was not inlined.
113 * bug fix: long long arguments passed in registers to alien
114 functions on PowerPC Linux and NetBSD are now handled correctly
115 when preceded by SYSTEM-AREA-POINTERs. (reported by Josh Elsasser)
117 changes in sbcl-1.0.22 relative to 1.0.21:
118 * minor incompatible change: LOAD-SHARED-OBJECT no longer by default looks
119 for the shared object in the current directory, but passes the native
120 namestring of the designated pathname to the operation system's shared
121 object loading function as-it.
122 * minor incompatible change: --disable-debugger toplevel option now takes
123 effect before processing of initialization files and --eval or --load
125 * new feature: new commandline argument: --script, which supports
126 shebang lines. See documentation for details. (based on work by
128 * new feature: SAVE-LISP-AND-DIE can save current values of
129 --dynamic-space-size and --control-stack-size in the executable core,
130 causing it to skip normal runtime option processing. See documentation
131 for details. (thanks to Zach Beane)
132 * enhancement: inoccous calls to EVAL or generic functions dispatching
133 on subclasses of eg. STREAM no longer cause compiler notes to appear.
134 * enhancement: the system no longer resignals errors from --load and
135 --eval toplevel arguments as SIMPLE-ERRORS, which caused restarts
136 associated with the original error to be lost. (thanks to Ariel
138 * enhancement: :DONT-SAVE keyword argument has been added to
139 LOAD-SHARED-OBJECT for controlling interaction with
140 SAVE-LISP-AND-DIE, and UNLOAD-SHARED-OBJECT can be used to undo
141 the effects of an earlier LOAD-SHARED-OBJECT call.
142 * bug fix: ADJUST-ARRAY on multidimensional arrays used bogusly give
143 them a fill pointer unless :DISPLACED-TO or :INITIAL-CONTENTS were
144 provided. (reported by Cedric St-Jean)
145 * bug fix: circularity handling in the reader did not treat raw
146 structure slots correctly. (reported by Cedric St-Jean)
147 * bug fix: SERVE-EVENT occasionally signaled an error about bogus
148 file descriptors when there were none.
149 * bug fix: DEFINE-COMPILER-MACRO support of destructuring lambda-lists
150 was broken. (reporteed by Willem Broekema)
151 * bug fix: SB-EXT:NATIVE-NAMESTRING on Windows did not work on
152 pathnames without a directory.
153 * bug fix: COMPILE-FILE did not warn about undefined variable
154 references at toplevel, and LOAD of the resulting fasl did
156 * bug fix: functions with non-required arguments used to end up with
157 (SB-C::&OPTIONAL-DISPATCH ...) as their name.
158 * bug fix: redefining a function with non-required arguments didn't
159 update the system's knowledge about its call signature properly.
160 * bug fix: fixed #431; incompatible alien record type redefinitions
161 are detected and handled. (reported by Neil Haven)
162 * bug fix: using RESTRICT-COMPILER-POLICY with DEBUG 3 could cause
163 PROGV miscompilation. (reported by Matthias Benkard, patch by Juho
165 * bug fix: on x86 and x86-64 backtraces were sometimes truncated
166 after alien stack frames.
167 * bug fix: strings passed to mk{s,d}temp lacked null termination
169 changes in sbcl-1.0.21 relative to 1.0.20:
170 * new feature: the compiler is able to track the effective type of a
171 generic function across method addition and removal.
172 * new feature: SB-EXT:ATOMIC-INCF allows atomic incrementation of
173 appropriately typed structure slots without locking.
174 * new feature: SB-EXT:CALL-WITH-TIMING provides access to timing
175 information like those gathered by TIME using a programming-friendly
177 * new feature: TIME reports time taken even if the form performs a
178 non-local transfer of control.
179 * enhancement: reduced conservativism on GENCGC platforms: on
180 average 45% less pages pinned (measured from SBCL self build).
181 * bug fix: SB-EXT:COMPARE-AND-SWAP on SYMBOL-VALUE can no longer
182 mutate constant symbols or violate declaimed type of the symbol.
183 * bug fix: SB-THREAD:SIGNAL-SEMAPHORE could fail to wakeup threads
184 sleeping on the semaphore in heavily contested semaphores.
185 * bug fix: semaphores and condition variables were not interrupt
187 * bug fix: SB-THREAD:CONDITION-WAIT doesn't allow waits on mutexes
188 owned by other threads anymore.
189 * bug fix: FIND on lists called KEY outside the specified
190 subsequence. (reported by budden)
191 * bug fix: LOG doesn't use single-float intermediate results when
192 given mixed integer and double-float arguments, leading to better
193 precision. (reported by Bob Felts)
194 * bug fix: LOG with base zero returned values of inconsistent type.
196 changes in sbcl-1.0.20 relative to 1.0.19:
197 * minor incompatible change: OPTIMIZE qualities
198 SB-C::STACK-ALLOCATE-DYNAMIC-EXTENT, SB-C::STACK-ALLOCATE-VECTOR,
199 and SB-C::STACK-ALLOCATE-VALUE-CELLS no longer exist. See documentation
200 and SB-EXT:*STACK-ALLOCATE-DYNAMIC-EXTENT* for details.
201 * documentation: some slot access efficiency guidelines have been
202 added to the user manual.
203 * optimization: ASSOC-IF, ASSOC-IF-NOT, MEMBER-IF, MEMBER-IF-NOT,
204 RASSOC, RASSOC-IF, and RASSOC-IF-NOT are now equally efficient
205 as ASSOC and MEMEBER.
206 * optimization: calls to ASSOC, MEMBER, and RASSOC can be transformed
207 to more efficient EQ-comparison versions more often.
208 * optimization: enhanced derivation of DOLIST iteration variable type
210 * optimization: constant folding of simple (LIST ...) forms as DOLIST
212 * optimization: runtime lookup of function definitions can be
213 elided in more cases, eg: (let ((x 'foo)) (funcall foo)).
214 * optimization: compiler is able to derive the return type of
215 (AREF (THE STRING X) Y) as being CHARACTER.
216 * optimization: CLRHASH on empty hash-tables no longer does pointless
217 work. (thanks to Alec Berryman)
218 * optimization: REPLACE deftransforms don't punt when :START1 or
220 * bug fix: compiling DESTRUCTURING-BIND on constant list arguments
221 no longer emits a code deletion note. (reported by Andrew
223 * bug fix: bogus odd-number-of-keywords STYLE-WARNINGs from calls to
224 functions with an odd number of &OPTIONAL arguments, a &REST
225 argument, and one or more &KEY arguments at the call site.
226 * bug fix: STYLE-WARNINGs for functions with both &OPTIONAL and &KEY
227 arguments appeared at call sites as well.
228 * bug fix: fixed #425; CLOSE drops input buffers from streams, so
229 READ-LINE &co can no longer read from them afterwards. (reported
231 * bug fix: fixed #427: unused local aliens no longer cause compiler
232 breakage. (reported by Stelian Ionescu, Andy Hefner and Stanislaw
234 * bug fix: non-local exit from a WITH-ALIEN form no longer causes
235 alien-stack leakage. (reported by Andy Hefner)
236 * bug fix: PROGV signals an error when an attempt to violate declared
237 type of a variable or bind a constant is made.
238 * bug fix: SET signals an error when an attempt to violate declared
239 type of a variable is made.
240 * bug fix: restart computation during the execution of a restart
241 test function no longer causes infinite recursion. (reported by
243 * bug fix: calling SB-COVER:REPORT with a non-directory pathname now
244 signals an error. (thanks to Pierre Mai)
245 * bug fix: EXPORT left symbol unexported in conflict situations.
246 (thanks to Michael Weber)
247 * bug fix: correctly handle name conflicts involving CL:NIL.
248 (thanks to Michael Weber)
249 * bug fix: RESOLVE-CONFLICT restart for name conflicts handles
250 conflicts arising from USEing package with conflicting symbols
251 correctly. (thanks to Michael Weber)
253 changes in sbcl-1.0.19 relative to 1.0.18:
254 * new feature: user-customizable variable SB-EXT:*MUFFLED-WARNINGS*;
255 warnings that go otherwise unhandled will be muffled if they are
256 of the type that's the value of this variable.
257 * optimization: stack allocation is slightly more efficient on x86
259 * bug fix: DEFSTRUCT forms with user-specified :CONSTRUCTOR options,
260 where a raw slot always is initialized using the initform whose
261 type is not know sufficiently well a compile-time are now compiled
262 correctly. (reported by John Morrison)
263 * bug fix: compiler no longer makes erronous assumptions in the
264 presence of non-foldable SATISFIES types.
265 * bug fix: stack analysis missed cleanups of dynamic-extent
266 arguments in non-let-converted calls to local functions.
267 * improvements to the Windows port:
268 ** adjusted address spaces for building on both Win32 and
269 Win64. (thanks for John Connors)
270 * fixed some bugs revealed by Paul Dietz' test suite:
271 ** interval arithmetic during type derivation used inexact integer
272 to single-float coercions.
273 ** arithmetic operations involving large integers and single
274 floats give the same results in compiled and interpreted code.
275 ** deriving the result type of COERCE no longer signals an error
276 if the derived type of the second argument is a MEMBER type
277 containing invalid type specifiers.
278 ** ADJOIN with constant NIL as second argument works correctly.
279 ** FUNCTION-KEYWORDS was called SB-PCL::FUNCTION-KEYWORD-PARAMETERS.
281 changes in sbcl-1.0.18 relative to 1.0.17:
282 * minor incompatible change: SB-SPROF:WITH-PROFILING now by default
283 profiles only the current thread.
284 * minor incompatible change: changes to SYMBOL-VALUE of constants
285 defined with DEFCONSTANT now signal an error.
286 * enhancement: SB-SPROF now has support for wallclock profiling,
287 and is also able to profile specific threads. REPORT output
288 has also additional sorting options.
289 * enhancement: better pretty-printing of DEFPACKAGE forms. (Thanks
291 * optimization: structure allocation has been improved
292 ** constructors created by non-toplevel DEFSTRUCTs are ~40% faster.
293 ** out of line constructors are ~10% faster.
294 ** inline constructors are ~15% faster.
295 ** inline constructors are capable of dynamic extent allocation
296 (generally on x86 and x86-64, in some cases on other platforms
298 * optimization: simple uses of HANDLER-CASE and HANDLER-BIND no
300 * optimization: file compiler is now able to coalesce non-circular
301 lists, non-base strings, and bit-vectors. Additionally, constants
302 are never referenced through SYMBOL-VALUE at runtime.
303 * optimization: code defining methods on PRINT-OBJECT (and other
304 generic functions in the COMMON-LISP package) now loads faster.
305 * bug fix: EAI_NODATA is deprecated since RFC 3493. Stop using it
307 * bug fix: if COMPILE-FILE aborts due to an unwind, the partial
308 fasl is now deleted. (reported by Attila Lendvai)
309 * bug fix: READ-LINE always returned NIL for the last line in files.
310 (reported by Yoshinori Tahara)
311 * bug fix: more accurate disassembly annotations of foreign function
312 calls. (thanks to Andy Hefner)
313 * bug fix: trimming non-simple strings and non-string string
314 designators when the there is nothing to trim works properly.
315 (thanks to James Knight)
316 * new feature: SB-POSIX bindings for mlockall, munlockall, and setsid.
317 (thanks to Travis Cross)
318 * fixed some bugs revealed by Paul Dietz' test suite:
319 ** NIL is a valid function name (regression at 1.0.13.38)
320 ** FILL on lists was missing its return value (regression at 1.0.12.27)
321 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
322 fill pointers properly (regression at 1.0.12.23)
323 ** STRING-TRIM, STRING-LEFT-TRIM, and STRING-RIGHT-TRIM did not respect
324 displacement indices properly (regression at 1.0.12.23)
326 changes in sbcl-1.0.17 relative to 1.0.16:
327 * temporary regression: user code can no longer allocate closure
328 variable storage on stack, due to bug 419 without explicitly
329 requesting it. Please consult sbcl-devel for advice if you need to
330 use this feature in the meanwhile.
331 * new feature: runtime argument --control-stack-size can be used to
332 adjust thread default control stack size.
333 * enhancement: improved TIME output
334 ** all times are reported using the measured accuracy (milliseconds
335 for real and GC times, microseconds for everything else.)
336 ** processor cycle counts on x86 and x86-64.
337 ** interpreted forms are counted for both evaluator modes.
338 ** number of lambdas converted by the compiler is reported.
339 ** CPU percentage report (computed from real and total run time.)
340 ** more comprehensive run time reporting, using a condenced format
341 ** interperted form, lambda, and page fault counts are omitted
343 * optimization: ADJOIN and PUSHNEW are upto ~70% faster in normal
345 * optimization: APPEND is upto ~10% faster in normal SPEED policies.
346 * optimization: two argument forms of LAST are upto ~10% faster
347 in normal SPEED policies.
348 * optimization: NCONC no longer needs to heap cons its &REST list
349 in normal SPEED policies.
350 * bug fix: SB-FLUID build feature no longer breaks the build. (thanks
352 * bug fix: UNION and NUNION work with :TEST-NOT once more,
353 regression since 1.0.9.1. (thanks to Eric Marsden)
354 * bug fix: result of MAKE-ARRAY can be stack allocated - regression
355 since 1.0.15.36. (thanks to Paul Khuong)
356 * bug fix: LAST when always returned the whole list when given a bignum
357 as the second argument.
358 * bug fix: dynamic extent allocation of nested lists and vectors
359 could leak to otherwise accessible parts.
360 * bug fix: invalid optimization of heap-allocated alien variable
362 * bug fix: fasl header checking is less vulnerable to different
363 platform word lengths.
364 * bug fix: more correct assembler syntax for GNU binutils
365 2.18.50.0.4 support. (thanks to Marijn Schouten)
366 * bug fix: fix ECASE warnings from CMUCL-as-xc-host. (reported by
368 * bug fix: the fopcompiler can handle LOCALLY forms (with no
369 declarations) successfully. (reported by Attila Lendvai)
371 changes in sbcl-1.0.16 relative to 1.0.15:
372 * minor incompatible change: revert the changes to sb-posix's error
373 signaling added in 1.0.14.
374 * minor incompatible change: change PROBE-FILE back to returning
375 NIL whenever we can't get a truename, as was the case before 1.0.14.
376 * minor incompatible change: SB-BSD-SOCKETS:NAME-SERVICE-ERROR now
377 inherits from ERROR instead of just CONDITION.
378 * new feature: SB-INTROSPECT can provide source locations for instances
379 as well. (thanks to Tobias Rittweiler)
380 * optimization: binding special variables now generates smaller code
381 on threaded platforms.
382 * optimization: MEMBER and ASSOC are over 50% faster for :TEST #'EQ
383 and cases where no :TEST is given but the compiler can infer that
384 the element to search is of type (OR FIXNUM (NOT NUMBER)).
385 * optimization: better LOGNOT on fixnums.
386 * optimization: modular arithmetic for a particular requested width
387 is implemented using a tagged representation unless a better
388 representation is available.
389 * fixed bug 423: TRULY-THE and *CHECK-CONSISTENCY* interaction.
390 * bug fix: SB-BSD-SOCKETS:MAKE-INET-ADDRESS checks the input string
391 for wellformedness and returns a specialized vector. (reported by
392 Francois-Rene Rideau)
393 * bug fix: FIND-CLASS was not thread-safe. (reported by Attila Lendvai)
394 * bug fix: ~R was broken for vigtillions. (thanks to Luis Oliveira)
395 * bug fix: attempt to obtain *SCHEDULER-LOCK* recursively when
396 unscheduling timer at the same time as another timer fires.
397 * bug fix: don't reschedule timers for dead threads.
398 * bug fix: periodic polling was broken. (thanks to Espen S Johnsen)
399 * bug fix: copying output from RUN-PROGRAM to a stream signalled
400 bogus errors if select() was interrupted.
401 * enhancement: add support for fcntl's struct flock to SB-POSIX.
403 changes in sbcl-1.0.15 relative to sbcl-1.0.14:
404 * enhancement: cleaner backtraces for interactive interrupts, as
405 well as other cases where the interesting frames used to be
406 obscured by interrupt handling frames.
407 * enhancement: untracing a whole package using (UNTRACE "FOO") is
408 now supported, and tracing a whole package using (TRACE "FOO") now
409 traces SETF-functions as well.
410 * enhancement: implement SB-POSIX:MKTEMP and SB-POSIX:MKDTEMP.
411 * SB-DEBUG:PRINT-FRAME-CALL now prints the entry-point kind even
412 when SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS* is NIL.
413 * unreadably printed representation of hash-tables now includes
415 * bug fix: partially fixed #188: type propagation from assignments
416 is now more efficient.
417 * bug fix: fixed #407: (COERCE X 'SINGLE-FLOAT) and (COERCE X
418 'DOUBLE-FLOAT) are not flushable.
419 * bug fix: on x86 and x86-64 pointer based EQ-hashing now uses the
420 full address of the object, and none of the tag bits.
421 * bug fix: readably printing hash-tables now respects other printer
422 control variables. (reported by Cedric St-Jean)
423 * bug fix: compiler gave a bogus STYLE-WARNING for the :SYNCHRONIZED
424 keyword with MAKE-HASH-TABLE.
425 * bug fix: export SB-POSIX:MKSTEMP.
426 * bug fix: SORT was not interrupt safe.
427 * bug fix: XREF accounts for the last node of each basic-block as
429 * bug fix: MAKE-INSTANCE optimizations interacted badly with
430 non-keyword :DEFAULT-INITARGS in the presence of :BEFORE/:AFTER
431 methods on SHARED-INITIALIZE. (thanks to Matt Marjanovic)
432 * bug fix: the CTOR optimization for MAKE-INSTANCE should no longer
433 create obsolete instances in the case of redefinition or
434 obsoletion of a superclass. (thanks to Andy Hefner)
435 * bug fix: Support for the Alpha architecture has been revived; it had
436 suffered somewhat from lack of maintenance since sbcl-1.0.
437 * improvements to the Windows port:
438 ** The system detects the codepage to use at startup, instead of
439 using the value from the saved core. (thanks to Kei Suzuki)
441 changes in sbcl-1.0.14 relative to sbcl-1.0.13:
442 * new feature: SB-EXT:*EXIT-HOOKS* are called when the process exits
443 (see documentation for details.)
444 * revived support for OpenBSD (contributed by Josh Elsasser)
445 * partially fixed bug #108: ROOM no longer suffers from occasional
446 (AVER (SAP= CURRENT END)) failures.
447 * fixed bug #402: proclaimed non-standard declarations in DEFMETHOD
448 bodies no longer cause a WARNING to be signalled. (reported by
450 * bug fix: (TRUNCATE X 0) when X is a bignum now correctly signals
451 DIVISION-BY-ZERO. Similarly for MOD and REM (which suffered due to
452 the bug in TRUNCATE.) (reported by Michael Weber)
453 * bug fix: SB-SPROF:REPORT no longer signals an error if there are
454 no samples. (reported by Andy Hefner)
455 * bug fix: functions compiled using (COMPILE NIL '(LAMBDA ...))
456 no longer appear as (NIL ...) frames in backtraces.
457 * bug fix: RESOLVE-CONFLICT (and the other name conflict machinery)
458 is now actually exported from SB-EXT as documented. (reported by
460 * bug fix: sb-aclrepl now correctly understands how to inspect
461 single-floats on 64-bit platforms where single-floats are not boxed.
462 * bug fix: SB-MOP:CLASS-SLOTS now signals an error if the class has not
463 yet been finalized. (reported by Levente Meszaros)
464 * bug fix: CLOSE :ABORT T behaves more correctly on Windows.
465 * DESCRIBE and (DOCUMENTATION ... 'OPTIMIZE) describe meaning of
466 SBCL-specific optimize qualities.
468 changes in sbcl-1.0.13 relative to sbcl-1.0.12:
469 * minor incompatible change: RUN-PROGRAM now uses execvp(3) to find
470 an executable in the search path, and does so in the child
471 process's PATH. The function FIND-EXECUTABLE-IN-SEARCH-PATH has
472 been removed; it can be found in the CVS history, for somebody who
473 needs that search behavior (see the manual).
474 * minor incompatible change: compiler policy re. weakening type
475 checks has changed: now type checks are weakened only if SAFETY < 2
477 * SB-EXT:NATIVE-NAMESTRING takes a new keyword AS-FILE, forcing
478 unparsing of directory pathnames as files. Analogously,
479 SB-EXT:PARSE-NATIVE-NAMESTRING takes an AS-DIRECTORY, forcing a
480 filename to parse into a directory pathname.
481 * enhancement: implicit generic function creation now signals a
482 specific style-warning IMPLICIT-GENERIC-FUNCTION-WARNING, which
483 users can bind handlers for and muffle around calls to LOAD.
484 * enhancement: RUN-PROGRAM allows unicode arguments and environments
485 to be used (using the default stream external format), and allows
486 non-simple strings to be used. (thanks to Harald Hanche-Olsen)
487 * optimization: COPY-SEQ, FILL, and SUBSEQ are 30-80% faster for
488 strings and vectors whose element-type or simplicity is not fully
489 known at compile-time.
490 * optimization: STRING-TRIM and related functions no longer allocate
491 a new string when no trimming needs to be performed. These functions
492 are also faster than before when the input string has been declared
494 * optimization: READ-SEQUENCE on simple-strings is up to 80% faster.
495 * optimization: READ-LINE is significantly faster for files containing
497 * optimization: non-open coded uses of character comparison operators
498 (e.g. char=) no longer cons when called with more than one parameter
499 on platforms supporting dynamic-extent allocation.
500 * bug fix: READ-SEQUENCE on composite stream wrapping a Gray stream
501 with STREAM-ELEMENT-TYPE (UNSIGNED-BYTE 8) signalled an error.
502 * bug fix: COPY-SEQ on lists did not signal a type-error on improper
504 * bug fix: some sequence functions elided bounds checking when
506 * bug fix: too liberal weakening of union-type checks when SPEED >
508 * bug fix: more bogus fixnum declarations in ROOM implementation
511 changes in sbcl-1.0.12 relative to sbcl-1.0.11:
512 * new feature: MAKE-HASH-TABLE now experimentally accepts a
513 :SYNCHRONIZED argument, which makes the hash-table safe for
514 concurrent accesses (but not iteration.) See also:
515 SB-EXT:WITH-LOCKED-HASH-TABLE, and
516 SB-EXT:HASH-TABLE-SYNCHRONIZED-P.
517 * optimization: CONCATENATE on strings is an order of magnitude faster
518 in code compiled with (> SPEED SPACE).
519 * optimization: SUBSEQ is ~50% faster on lists.
520 * bug fix: bug 417 fixed -- source location reporting is now more robust.
521 * bug fix: SUBSEQ on a list will now correctly signal an error if if
522 END is smaller then START.
523 * bug fix: SB-PROFILE will no longer report extra consing for nested
524 calls to profiled functions.
525 * bug fix: ROOM implementation had bogus fixnum declarations which
526 could cause type-errors when calling ROOM on large images.
527 * bug fix: if file compilation is aborted, the partial fasl is now
528 deleted, and COMPILE-FILE returns NIL as the primary value.
529 * bug fix: number of thread safety issues relating to SBCL's internal
530 hash-table usage have been fixed.
531 * bug fix: SB-SYS:WITH-PINNED-OBJECTS could cause garbage values to
532 be returned from its body when the values were being returned
533 using unknown-values return convection and the W-P-O was wrapped
534 inside an UNWIND-PROTECT.
535 * bug fix: sb-posix should now compile again under Windows, enabling
538 changes in sbcl-1.0.11 relative to sbcl-1.0.10:
539 * incompatible change: hash-table accessor functions are no longer
540 automatically protected by locks. Concurrent accesses on the same hash-table
541 from multiple threads can give inconsistent results or even corrupt the
542 hash-table completely. Multi-threaded applications should do their own
543 locking at the correct granularity. In the current implementation it is
544 still safe to have multiple readers access the same table, but it's not
545 guaranteed that this property will be maintained in future releases.
546 * minor incompatible change: (SETF MUTEX-VALUE) is not longer supported,
547 and will signal an error at runtime.
548 * enhancement: SB-THREAD package now exports a semaphore interface.
549 * enhancement: CONS can now stack-allocate on x86 and
550 x86-64. (Earlier LIST and LIST* supported stack-allocation, but
552 * enhancement: nested lists can now be stack allocated on
553 platforms providing stack allocation support.
554 * enhancement: dynamic-extent support has been extended to support
555 cases where there are multiple possible sources for the stack
557 * optimization: RELEASE-MUTEX no longer needs to perform a syscall
558 if the mutex is uncontested on Linux.
559 * bug fix: symbol-macro expansion now uses the *MACROEXPAND-HOOK*
560 as specified by the CLHS. (thanks to Tobias Rittweiler)
561 * bug fix: NaN comparison now works on x86-64.
562 * bug fix: CLOSE :ABORT T on a stream with pending output now
564 * bug fix: instances of non-standard metaclasses using standard
565 instance structure protocol sometimes missed the slot type checks
567 * bug fix: known functions can be inlined in high-debug code.
569 changes in sbcl-1.0.10 relative to sbcl-1.0.9:
570 * minor incompatible change: the MSI installer on Windows no longer
571 associates .lisp and .fasl files with the installed SBCL.
572 * minor incompatible change: :UNIX is no longer present in *FEATURES*
573 on Windows. (thanks to Luis Oliviera)
574 * new feature: SB-CLTL2 contrib module now implements
575 FUNCTION-INFORMATION. (thanks to Larry D'Anna)
576 * optimization: scavenging weak pointers is now more efficient,
577 requiring O(1) instead of O(N) per weak pointer to identify
578 scanvenged vs. unscavenged pointers. (thanks to Paul Khuong)
579 * optimization: typechecking (SETF SLOT-VALUE) is now ~5x faster in
580 method bodies using constant slot-names when the first argument is
581 a specializer parameter for the method.
582 * optimization: SLOT-VALUE &co are now ~50% faster for variable slot
583 names, when the class of the instance is a direct instance
584 STANDARD-CLASS or FUNCALLABLE-STANDARD-CLASS (making them only 3x
585 as slow as the constant slot-name case.)
586 * optimization: member type construction is now O(N) instead
588 * optimization: UNION and NUNION are now O(N+M) for large
589 inputs as long as the :TEST function is one of EQ, EQL, EQUAL, or
591 * enhancement: DEFINE-MODIFY-MACRO lambda-list information is
592 now more readable in environments like Slime which display it.
593 (thanks to Tobias C. Rittweiler)
594 * bug fix: SB-EXT:COMPARE-AND-SWAP was non-atomic unless the compiler
595 was able to infer the correct argument type for the object on which
596 the CAS operation was being performed.
597 * bug fix: copy propagation interfered with parallel assignment
598 semantics in local calls. (reported by Paul Khuong)
599 * bug fix: the signed modular fixnum shift compiled to wrong code on
600 x86 and x86-64. (spotted by a slight modification to some of
603 changes in sbcl-1.0.9 relative to sbcl-1.0.8:
604 * minor incompatible change: SB-SYS:OUTPUT-RAW-BYTES is deprecated.
605 * enhancement: SB-EXT:FINALIZE accepts a :DONT-SAVE keyword argument,
606 indicating the finalizer should be cancelled when SAVE-LISP-AND-DIE
608 * enhancement: default PRINT-OBJECT methods for classes, slot-definitions,
609 and generic function now print the object identity for anonymous
610 instances. (thanks to Attila Lendvai)
611 * enhancement: as an extension to MOP, SBCL now supports SETF of
612 STANDARD-INSTANCE-ACCESS and FUNCALLABLE-STANDARD-INSTANCE-ACCESS.
613 (thanks to Attila Lendvai)
614 * enhancement: sb-cover has an alternate annotation mode which puts
615 the form annotations at the car of the form, instead of annotating
616 it completely. The benefit of this mode is that it shows better
617 which forms were instrumented by the compiler.
618 * bug fix: new compiler transforms for MEMBER and ASSOC were affected
619 by printer control variables. (reported by Dan Corkill)
620 * bug fix: system leaked memory when delayed output was performed by
621 the OS in smaller chunks then expected. (thanks to David Smith)
622 * bug fix: system leaked memory when file streams were not closed
624 * bug fix: large objects written to slow streams that were modified
625 after the write could end up with the modified state written to
626 the underlying file descriptor.
627 * bug fix: multiple threads operating in parallel on the same stream
628 could cause buffer-overflows.
629 * bug fix: source location information is stored correctly
630 (broken since 1.0.6). This bug would generally show up as the
631 Slime debugger highlighting the wrong form when the sldb-show-source
633 * bug fix: Branch forms are again annotated as branches in the sb-cover
635 * bug fix: GCD on MOST-NEGATIVE-FIXNUM no longer causes an infinite loop
636 on x86-64. (reported by Gregory Vanuxem)
637 * bug fix: EQUALP could return wrong results for structures with raw
638 slots (slots with a :TYPE of SINGLE-FLOAT, DOUBLE-FLOAT, or a machine
639 word). (reported by Vjacheslav Fyodorov)
640 * bug fix: sb-sprof sampling didn't work on non-x86oid platforms. (patch
643 changes in sbcl-1.0.8 relative to sbcl-1.0.7:
644 * enhancement: experimental macro SB-EXT:COMPARE-AND-SWAP provides
645 atomic compare-and-swap operations on threaded platforms.
646 * enhancement: experimental function SB-EXT:RESTRICT-COMPILER-POLICY
647 allows assigning a global minimum value to optimization qualities
648 (overriding proclamations and declarations).
649 * enhancement: closed over variables can be stack-allocated on x86
651 * performance bug fix: GETHASH and (SETF GETHASH) are once again
653 * optimization: slot definition lookup is now O(1). This speeds up
654 eg. SLOT-VALUE and (SETF SLOT-VALUE) with variable slot names.
655 * optimization: STRING-TO-OCTETS is now up to 60% faster for UTF-8.
656 * optimization: ASSOC and MEMBER can now be open-coded for all
657 combinations of keyword arguments when second argument is constant
658 and SPEED >= SPACE. In other cases a specialized version is
660 * bug fix: using obsoleted structure instances with TYPEP and
661 generic functions now signals a sensible error.
662 * bug fix: threads waiting on GET-FOREGROUND can be interrupted.
663 (reported by Kristoffer Kvello)
664 * bug fix: backtrace construction is now more careful when making
665 lisp-objects from pointers on the stack, to avoid creating bogus
666 objects that can be seen by the GC.
667 * bug fix: defaulting of values in contexts expecting more than 7
668 variables now works on x86-64. (reported by Christopher Laux)
669 * bug fix: modifications to packages (INTERN, EXPORT, etc) are now
671 * bug fix: (SETF SYMBOL-PLIST) no longer allows assigning a non-list
672 as the property-list of a symbol.
673 * bug fix: DEFMETHOD forms with CALL-NEXT-METHOD in the method body,
674 in EVAL-WHEN forms with both :COMPILE-TOPLEVEL and :LOAD-TOPLEVEL
675 situations requested, are once again file-compileable. (reported
678 changes in sbcl-1.0.7 relative to sbcl-1.0.6:
679 * MOP improvement: support for user-defined subclasses of
680 SB-MOP:SPECIALIZER has been enhanced. The experimental interface
681 function SB-PCL:MAKE-METHOD-SPECIALIZERS-FORM, called as part of
682 the expansion of DEFMETHOD, is responsible for generating a form
683 which creates a list of specializers when evaluated. Additional
684 functions SB-PCL:[UN]PARSE-SPECIALIZER-USING-CLASS provide
685 debugging and introspective support.
686 * minor incompatible change: the (unsupported) spinlock interface
687 has changed: free spinlock now has the value NIL, and a held spinlock
688 has the owning thread as its value.
689 * enhancement: WITHOUT-INTERRUPTS now binds ALLOW-WITH-INTERRUPTS and
690 WITH-LOCAL-INTERRUPTS as local macros. Refer to documentation string
692 * enhancement: name of a socket-stream is now "a socket" instead of
694 * enhancement: SB-POSIX now supports lockf(). (Thanks to Zach Beane.)
695 * enhancement: SB-POSIX now supports getcwd(). (Thanks to Tassilo Horn.)
696 * enhancement: SB-INTROSPECT:FUNCTION-ARGLIST shows nicer argument lists
697 for generic functions. (Thanks to Tobias C. Rittweiler)
698 * optimization: bignum printing speed has been improved by 20-40%
699 (depending on the bignum size.)
700 * bug fix: WITH-MUTEX and WITH-RECURSIVE-LOCK are now interrupt safe
702 * bug fix: the cache used by the CLOS to store precomputed effective
703 methods, slot offsets, and constant return values is now thread and
705 * bug fix: generic function dispatch function updating is now thread
706 and interrupt safe (in the sense that the known issues have been
708 * bug fix: ADD/REMOVE-METHOD is now thread and interrupt safe.
709 * bug fix: interning EQL-specializers is now thread and interrupt safe.
710 * bug fix: asdf systems with dependencies to the SB-POSIX or
711 SB-BSD-SOCKETS contribs can be loaded with :FORCE T.
712 * bug fix: interrupt safety of applicable method computation has been
715 changes in sbcl-1.0.6 relative to sbcl-1.0.5:
716 * new contrib: sb-cover, an experimental code coverage tool, is included
718 * optimization: STRING-TO-OCTETS for unibyte encodings and UTF-8 is
719 significantly faster.
720 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
721 produce more efficient code on x86-64 if the compiler can determine
722 that (+ INDEX <constant>) does not require a bounds check and FOO
723 has an element type at least 8 bits wide.
724 * enhancement: a new, experimental synchronous timeout facility is
725 provided. Refer to SB-SYS:WITH-DEADLINE for details.
726 * enhancement: when a symbol name conflict error arises, the
727 conflicting symbols are always printed with a package prefix.
728 (thanks to Kevin Reid)
729 * enhancement: stepping is now once again supported on the SPARC and
730 MIPS platforms. (It is also now more likely to work on CheneyGC
732 * enhancement: sb-sprof can now also track and report accurate call
734 * bug fixes: the treatment of non-standard subclasses of
735 SB-MOP:SPECIALIZER is more correct.
736 * incompatible change: PURIFY no longer copies the data from the
737 dynamic space into the static and read-only spaces on platforms
738 that use the generational garbage collector
739 * bug fix: GETHASH, (SETF GETHASH), CLRHASH and REMHASH are now
741 * bug fix: GC race condition occasionally resulting in crashes with
742 the error message "SIG_STOP_FOR_GC blocked at a bad place" has been
744 * bug fix: &ENVIRONMENT variables in macro lambda lists can now be
746 * bug fix: DEFSETF lambda lists without &ENVIRONMENT no longer cause
747 a STYLE-WARNING to be signalled (regression from 1.0.4.)
748 * bug fix: an asynchronous interrupt could previously leave the
749 system running with GC inhibited.
750 * bug fix: a DECLARE form evaluated at top-level now causes an error
751 rather than silently (or verbosely) returning NIL.
752 * bug fix: trying to dissassemble functions compiled at high DEBUG could
753 result in a error being signalled due to source form lookup errors.
754 (reported by Peter Graves)
756 changes in sbcl-1.0.5 relative to sbcl-1.0.4:
757 * incompatible change: removed writer methods for host-ent-name,
758 host-ent-addresses -- changing the values did not update the DNS
760 * minor incompatible change: changed experimental JOIN-THREAD interface
761 * documentation: the manual now lists reader and writer methods
762 in class slot documentation sections. (thanks to Richard M Kreuter)
763 * documentation: unwinding from asyncronous events has been
764 documented as unsafe.
765 * documentation: SB-SYS:WITHOUT-GCING has been documented as unsafe
766 in multithreaded application code.
767 * optimization: GET-INTERNAL-REAL-TIME has been optimized on POSIX
768 platforms. (thanks to James Anderson for the optimization hint)
769 * optimization: REPLACE, SUBSEQ, and COPY-SEQ are now optimized in
771 * optimization: STRING-{EQUAL,LESSP,GREATER-P} and their NOT-
772 variants no longer cons.
773 * optimization: Direct calls to CHAR-{EQUAL,LESSP,GREATERP} and
774 their NOT- variants no longer cons.
775 * optimization: EQUAL hash tables no longer use SXHASH for objects
776 of all data types, but instead use an EQL hash for types for which
777 EQUAL is the same as EQL.
778 * optimization: the non-inlined generic versions of AREF and (SETF AREF)
779 are significantly faster.
780 * optimization: new STRING-OUTPUT-STREAM implementation is much
781 faster for large outputs and conses 30% less on average.
782 * enhancement: XREF information is now collected for references made
783 to global variables using SYMBOL-VALUE and a constant argument.
784 * enhancement: SIGINT now causes a specific condition
785 SB-SYS:INTERACTIVE-INTERRUPT to be signalled.
786 * bug fix: FILE-POSITION used to signal an error in some cases where
787 ANSI requires it to return NIL.
788 * bug fix: ADJUST-ARRAY is now interrupt-safe.
789 * bug fix: adding and removing fd-handlers is now interrupt-safe.
790 * bug fix: inlined calls to C now ensure 16byte stack alignment on
792 * bug fix: bad type declaration in the CLOS implementation has
793 been fixed. (reported by James Anderson)
794 * bug fix: incorrect ROOM reporting on x86-64 has been fixed.
795 (thanks to Lutz Euler)
796 * bug fix: DEFSETF now allows &ENVIRONMENT and disallows &AUX as
797 required by the CLHS. (reported by Samium Gromoff)
798 * bug fix: dead unbound variable references now signal an error.
799 (reported by Marco Monteiro)
800 * bug fix: / with an unused value was being deleted in safe code.
801 (thanks to Marco Monteiro and Kevin Reid)
802 * bug fix: number of characters that can be written onto a single
803 line in a file is unlimited.
804 * bug fix: some GC deadlocks caused by asynchronous interrupts have
805 been fixed by inhibiting interrupts for when GC is disabled.
806 * bug fix: some interrupt safety issues with GETHASH, (SETF
807 GETHASH), CLRHASH and REMHASH have been fixed.
808 * bug fix: binding *BREAK-ON-SIGNALS* to a value that is not a type
809 specifier no longer causes infinite recursion.
810 * bug fix: SB-EXT:MUFFLE-CONDITIONS declarations no longer trigger a
811 bogus warning in DEFMETHOD bodies. (reported by Kevin Reid)
812 * bug fix: an &environment argument with the correct variable information
813 is passed to macros that are expanded during byte compilation.
814 (reported by Samium Gromoff)
815 * bug fix: a logic error was causing the hash function for bignums to
816 have excessive amounts of collisions. (reported by Faré Rideau)
817 * bug fix: modifying the contents of an array could change the return
818 value of SXHASH on that array, which is only allowed for strings
819 and bit vectors. (bug introduced in 0.9.16)
820 * improvement: the x86-64/darwin port now passes all tests (except
821 for the debugger tests) but should still be considered
822 experimental until this is fixed.
823 * improvement: a style-warning is signaled for CASE (etc) clauses with
824 duplicate keys. (thanks to Kevin Reid)
825 * improvement: macos/x86 and macos/x86-64 now use mach exception
826 handlers for dealing with illegal instructions (for trapping and
827 error handling) and memory protection violations (for GC).
829 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
830 * new platform: experimental support for x86-64/darwin (MacOS).
831 * incompatible change: the thread-safe (on most platforms) getaddrinfo
832 and getnameinfo sockets functions are used instead of gethostbyaddr
833 and gethostbyname, on platforms where the newer functions are available.
834 As a result, the ALIASES field of HOST-ENT will always be NIL on these
836 * change: runtimes with embedded cores (i.e. saved with :EXECUTABLE T)
837 don't print the startup banner, but behave as if --noinform was passed
838 as a command line argument. (thanks to Kevin Reid)
839 * new experimental feature: added JOIN-THREAD (by NIIMI Satoshi)
840 * optimization: code using alien values with undeclared types is much faster.
841 * optimization: the compiler is now able to open code SEARCH in more cases.
842 * optimization: more compact typechecks on x86-64 (thanks to Lutz Euler)
843 * bug fix: using standardized COMMON-LISP special variables as loop
844 variables no longer signals bogus package lock violations. (reported
846 * bug fix: declaring local loop variables to be of a range-limited type
847 such as (SINGLE-FLOAT 1.0 2.0) no longer causes a compile-time error.
848 (reported by Andras Simon)
849 * bug fix: >= and <= gave wrong results when used with NaNs. (Some NaN
850 bugs remain on x86-64.)
851 * bug fix: the #= and ## reader macros now interact reasonably with
852 funcallable instances.
853 * bug fix: type-checks for function arguments were compiled using the
854 compiler policy settings of the wrong lexical environment (in 1.0.2
856 * bug fix: SHADOW accepts characters as string designators, as required
857 by the spec (thanks to Eric Marsden)
858 * bug fix: fixed GC safety issues when foreign functions are called with
859 non-base strings as arguments
860 * bug fix: more consistent error output for fatal-compiler-warnings, like
862 * bug fix: functions with proper names were showing up as NIL in
863 backtraces (reported by Edi Weitz, regression in 1.0.2)
865 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
866 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
867 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
868 produce more efficient code on the x86 if the compiler can determine
869 that (+ INDEX <constant>) does not require a bounds check and FOO
870 has an element type at least 8 bits wide.
871 * bug fix: references to undefined variables are handled the same way
872 in toplevel forms as in the normal compiler
873 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
875 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
876 (thanks to Jon Buller)
877 * improvement: faster compilation times for complex functions
878 * improvement: added readlink support to SB-POSIX (thanks to Richard
881 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
882 * improvement: experimental support for mach exception handling on
883 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
885 * improvement: support for GBK external format.
886 (thanks to Chun Tian (binghe))
887 * improvement: the debugger now displays variables that have been closed
888 over, in code compiled with (DEBUG 2) or higher
889 * improvement: support for executable cores on NetBSD (thanks to
891 * new feature: added a RESTART-FRAME debugger command
892 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
893 be used to provide an implementation for FILE-POSITION on Gray streams
894 (thanks to Eric Marsden)
895 * new feature: add syslog, openlog and closelog support to SB-POSIX
896 (thanks to Richard Kreuter)
897 * optimization: the function call overhead in code compiled with
898 a high DEBUG optimization setting is significantly
899 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
900 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
901 (thanks to Tony Martinez)
902 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
903 evaluated once (reported by Kevin Reid)
904 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
905 works again (thanks to Kilian Sprotte)
906 * bug fix: an error is signaled for tagbodies with duplicate tags
907 (thanks to Stephen Wilson)
908 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
909 * bug fix: Win32 port can now handle foreign code unwinding Lisp
910 stack frames from alien callbacks.
911 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
912 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
913 65k pages (thanks to David Woodhouse)
914 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
916 changes in sbcl-1.0.1 relative to sbcl-1.0:
917 * new platform: FreeBSD/x86-64, including support for threading.
918 * new feature: the compiler stores cross-referencing information
919 about function calls (who-calls), macroexpansion (who-macroexpands)
920 and special variables (who-binds, who-sets, who-references) for code
921 compiled with (< SPACE 3). This information is available through the
922 sb-introspect contrib.
923 * new feature: users may subclass SEQUENCE, and have instances of
924 these classes interoperate with standard Common Lisp functions if
925 a number of methods are defined. (This feature is experimental
926 and the interface subject to change based on feedback from SBCL
927 users and the general community)
928 * improvement: sb-sprof traces call stacks to an arbitrary depth on
929 x86/x86-64, rather than the previous fixed depth of 8
930 * improvement: another pthread back-end of mutex "pthread-futex".
931 * bug fix: non-ascii command-line arguments are processed correctly
932 (thanks to Yaroslav Kavenchuk)
933 * bug fix: non-required arguments were not passed correctly when a method
934 defined using DEFMETHOD was called from a mop-generated method using
935 CALL-NEXT-METHOD (reported by Pascal Costanza)
936 * bug fix: recursion is now permitted in accessors through
937 SLOT-UNBOUND. (reported by Pascal Costanza)
938 * bug fix: an error was signaled at startup if the HOME environment
939 variable was defined, but had an empty value (reported by Peter Van Eynde)
940 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
941 signal a PROGRAM-ERROR, not a BUG.
942 * bug fix: SB-POSIX:READDIR works when built with large file support.
943 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
945 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
946 for better automatic indentation support. (thanks to Matt Pillsbury)
947 * optimization: loading generic functions no longer takes O(n^2) time,
948 proportional to the amount of methods in the generic function
949 (reported by Todd Sabin and Jeremy Brown)
950 * optimization: the FIND and POSITION family of sequence functions
951 are significantly faster on arrays whose element types have been
953 * improvements to the Windows port:
954 ** Intermittent heap corruption problems have been fixed. (thanks
955 to Alastair Bridgewater)
956 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
958 ** Lisp is able to unwind foreign exception frames from alien
959 callbacks. (thanks to Alastair Bridgewater)
961 changes in sbcl-1.0 relative to sbcl-0.9.18:
962 * improvement: experimental support for threading on FreeBSD/x86.
963 (thanks to NIIMI Satoshi)
964 * improvement: runtime option --dynamic-space-size can be used
965 to set the size of the dynamic space reserved on startup.
966 * improvement: floating point modes in effect are now saved in
967 core, and restored on startup.
968 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
969 startup, not time since first call to GET-INTERNAL-REAL-TIME.
970 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
971 threads are not running after *SAVE-HOOKS* have run.
972 * improvement: writes to CLOS instance slots are type-checked in code
973 compiled with (SAFETY 3)
974 * improvement: floating-point exception handling on FreeBSD (thanks to
976 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
977 (thanks to Zach Beane)
978 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
980 * improvement: added support for the Shift-JIS external format.
981 (contributed by NIIMI Satoshi)
982 * improvement: callbacks are supported on Linux/PPC. (thanks to
984 * bug fix: compiler bug triggered by a (non-standard) VALUES
985 declaration in a LET* was fixed. (reported by Kaersten Poeck)
986 * bug fix: file compiler no longer confuses validated and already
987 dumped structurres. (reported by Kaersten Poeck)
988 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
989 fill-pointer signals a type-error as required. (thanks to
991 * bug fix: disassemly of funcallable instances works.
992 * bug fix: single stepping on PPC.
993 * bug fix: fix thread-safety problems in the type system (generally
994 manifesting as nonsensical errors like "STRING is a bad type specifier
995 for sequences" or "The value 1 is not of type FIXNUM")
996 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
997 (reported by Josip Gracin)
998 * bug fix: an error is signaled for attempts to displace arrays with
999 incompatible element types (thanks to Mario Mommer)
1000 * bug fix: more correct handling of wide characters in debug info
1001 (bug reported by Attila Lendvai and fixed by Juho Snellman)
1002 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
1003 and don't cause extra consing
1004 * optimization: MAP and MAP-INTO are significantly faster on vectors
1005 whose elements types have been declared.
1006 * Improvements to SB-SPROF:
1007 ** Support for allocation profiling
1008 ** Reduced profiling overhead, especially for long profiling runs
1009 * Improvements to the Windows port:
1010 ** floating point exceptions are now reported correctly.
1011 ** stack exhaustion detection works partially.
1012 ** more accurate GET-INTERNAL-REAL-TIME.
1013 ** better RUN-PROGRAM behaviour in case of errors in the child
1015 ** PROBE-FILE now simplifies pathnames correctly.
1016 ** DIRECTORY now works correctly with :WILD-INFERIORS.
1017 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
1018 1st 1970 (although time timezone data will be unreliable).
1019 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
1021 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
1022 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
1023 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
1025 * bug fix: two potential GC deadlocks affecting threaded builds.
1026 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
1027 returns T (reported by Anton Kazennikov)
1028 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
1029 with non-variable places
1030 * bug fix: remove a race condition in the setting of
1031 funcallable-instance functions, this should make threaded CLOS
1032 code more stable against memory faults.
1033 * bug fix: corruption of specials when unbinding is interrupted by an
1034 asynchronous unwind (reported by Hannu Koivisto)
1035 * improvement: the debugger will now also display local variables that
1036 are only used once, for code compiled with a DEBUG optimization quality
1039 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
1040 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
1041 * incompatible change: External-format support for FFI calls. The
1042 SB-ALIEN:C-STRING no longer implies an ASCII
1043 external-format. Instead the string is subject to external-format
1044 conversion. Additionally return values of type SB-ALIEN:C-STRING
1045 are by default of type (SIMPLE-ARRAY CHARACTER), not
1046 SIMPLE-BASE-STRING. For an alien type that behaves like the old
1047 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
1048 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
1049 * incompatible change: SB-EXT package no longer contains the
1050 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
1051 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
1053 * incompatible change: the single-stepper is no longer available
1054 on Alpha, Mips and Sparc platforms.
1055 * minor incompatible change: the direct superclasses of
1056 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
1057 not (STANDARD-OBJECT FUNCTION). This makes the
1058 class-precedence-lists of GENERIC-FUNCTION and
1059 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
1061 * minor incompatible change: the default stream external format on
1062 non-unicode SBCL is no longer detected from the locale, but is always
1064 * new feature: Add a version of evaluator that uses an interpreter instead
1065 of the compiler. EVAL still uses the compiler by default, to switch it
1066 to use the interpreter, set the value of the variable
1067 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
1068 * minor incompatible change: the single-stepper REPL has been merged
1069 with the normal debugger (see the "Stepping" heading of the debugger help
1070 for more details). The debugger command STEP will no longer switch
1071 to the single-stepper REPL.
1072 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
1073 for a type now works.
1074 * bug fix: loading of default sysinit file works. (thanks to Leonid
1076 * bug fix: better detection of circularities in the file-compiler.
1077 (reported by Marco Monteiro)
1078 * bug fix: the CL pathname functions now work with files that have
1079 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
1080 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
1081 systems with Unix98 pty semantics.
1082 * bug fix: ASDF-INSTALL will now work with bsd tar.
1083 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
1085 * bug fix: timers expiring in dead threads no longer cause a
1086 type-error (reported by Paul "Nonny Mouse").
1087 * bug fix: thanks to more lightweight single-stepper instrumentation,
1088 code compiled with (DEBUG 3) will compile and execute significantly faster,
1089 and will have more accurate type-inferencing than before
1090 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
1091 whose bindings are modified
1092 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
1093 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
1094 ** replace dummy implementations of CL:MACHINE-INSTANCE and
1095 CL:SOFTWARE-VERSION with proper version
1097 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
1098 * feature: implemented the READER-METHOD-CLASS and
1099 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
1100 as specified by AMOP.
1101 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
1103 * optimization: faster LOGCOUNT implementation on x86 and x86-64
1104 (thanks to Lutz Euler)
1105 * optimization: hashing of general arrays and vectors has been
1106 improved. (reported by Any Fingerhut)
1107 * enhancement: SB-INTROSPECT is now able to find definitions of
1108 profiled functions. (thanks to Troels Henriksen)
1109 * enhancement: compiler-macro expansion applies now to FUNCALL forms
1110 as well, allowing compiler-macros for SETF-functions to expand.
1111 * enhancement: step-instrumentation no longer wraps known
1112 single-value functions with multiple-value context, allowing
1113 better type inference.
1114 * fixed bug #337: use of MAKE-METHOD in method combination now works
1115 even in the presence of user-defined method classes. (reported by
1116 Bruno Haible and Pascal Costanza)
1117 * fixed bug #339(c): if there are applicable methods not part of any
1118 long-form method-combination group, call INVALID-METHOD-ERROR.
1119 (reported by Bruno Haible)
1120 * fixed bug #361: the :FUNCTION initarg in the protocol for
1121 initialization of methods can now be used to override
1122 internally-produced optimized functions. (reported by Bruno
1124 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
1125 system-provided lambda expression no longer cause warnings about
1126 unbound #:|pv-table| symbols.
1127 * bug fix: improved the handling of type declarations and the
1128 detection of violations for keyword arguments with non-constant
1130 * bug fix: potentially erronous calls to PATHNAME and
1131 MERGE-PATHNAMES were being flushed in some cases.
1132 (reported by Richard Kreuter)
1133 * bug fix: compiled calls to TYPEP were mishandling obsolete
1134 instances. (reported by James Bielman and Attila Lendvai)
1135 * bug fix: format strings with ~> without matching ~< no longer
1136 trigger an AVER, but signal an understandable error instead.
1137 (reported by Antonio Martinez)
1138 * bug fix: specifying an output-file without a pathname-type for
1139 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
1140 of input-file instead of "fasl". (reported by Robert Dodier)
1141 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
1142 (reported by James Y Knight).
1143 * bug fix: compiler-macros-function did not consider the environment
1144 argument for shadowing by local functions.
1145 * bug fix: compiler-macros expansion was inhibited by local INLINE
1147 * bug fix: inline expansions of known functions were subject to
1148 step-instrumentation in high debug policies, leading to problems
1149 with type-inference.
1150 * bug fix: compiler failed to differentiate between different CONS
1151 types in some cases.
1152 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
1153 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
1154 * bug fix: #368: incorrect use of expressed vs. upgraded array
1156 * bug fix: #306a: more precise unions of array types.
1157 * thread-safety improvements:
1158 ** CONDITION-WAIT could return early on Linux, if the thread was
1159 interrupted and subsequently continued with SIGCONT.
1160 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
1162 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
1163 * added support for the ucs-2 external format. (contributed by Ivan
1165 * minor incompatible change: pretty printing of objects of type
1166 (cons symbol) is, in the default pprint-dispatch-table, now
1167 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
1169 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
1170 called later in a class's lifetime, possibly as late as when the
1171 first instance of the class is created. Previously,
1172 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
1173 class became finalizeable.
1174 * fixed bug: FILE-POSITION sometimes returned inconsistent results
1175 for multibyte external-format streams. (thanks to "vbzoli")
1176 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
1177 with :ALLOCATION :CLASS inherited from superclasses of the
1179 * fixed bug: anonymous classes can now be created using the :NAME
1180 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
1181 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
1182 * fixed bug: core-files saved with :EXECUTABLE T can again be
1183 executed when SBCL_HOME isn't set. (reported by James Knight)
1184 * fixed bug: toplevel LOCALLY forms with declarations could
1185 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
1186 * fixed bug: printing from several different threads using different
1187 values of *print-case* could cause invalid output, due to
1188 some internal special variables of the printer not being bound
1189 thread-locally (reported by Max Mikhanosha)
1190 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
1191 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
1192 (reported by Pascal Costanza)
1193 * fixed bug: FINALIZE-INHERITANCE is called from
1194 REINITIALIZE-INSTANCE on classes when the class has previously
1195 been finalized, as required by AMOP.
1196 * minor code generation optimizations:
1197 ** better register allocation in CLOS dispatching functions
1198 ** overflow detection when coercing signed bytes to fixnums on x86-64
1199 ** is now implemented with one IMUL instruction instead of three shifts
1200 ** more efficient bit-vector access on x86 and x86-64
1201 ** more efficient access to raw structure slots on x86 and x86-64
1202 * fixed some bugs revealed by Paul Dietz' test suite:
1203 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
1204 return its argument.
1206 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
1207 * feature: thread support on Solaris/x86, and experimental thread support
1209 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
1211 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
1212 default method for SLOT-UNBOUND.
1213 * minor incompatible change: prevent the user from specializing the
1214 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
1215 somewhat counter to the intent of the protocol, I (CSR) think, and
1216 additionally it just doesn't work in SBCL as currently
1217 implemented, thanks to optimizations (that are always valid for
1218 the other three SLOT-VALUEish functions, but not for the setter).
1219 * minor incompatible change: the :SB-LDB feature is now enabled by
1220 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
1221 the low-level debugger.
1222 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
1223 from SYNONYM-STREAM and TWO-WAY-STREAM provided they can be decomposed
1224 down to an SB-SYS:FD-STREAM, allowing direct communication in
1225 more cases. Temporary files and pipes are still used as a fallback
1227 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
1228 on the non-reentrant C functions localtime(3) and gmtime(3).
1229 * bug fix: global optimization policy was not visible in LOCALLY and
1231 * bug fix: class objects can be used as specializers in methods.
1232 (reported by Pascal Costanza)
1233 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
1234 with a type error. (reported by blitz_ on #lisp)
1235 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
1236 failed with a type error. (reported by Pascal Bourguignon)
1237 * bug fix: merging pathnames against defaults with :DIRECTORY
1238 starting with '(:RELATIVE :BACK) should preserve the :BACK.
1239 (reported by James Y Knight)
1240 * bug fix: saving large (>2GB) cores on x86-64 now works
1241 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
1242 constant index and a (simple-array (signed-byte 32)) array
1243 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
1244 error (patch by Robert J. Macomber)
1245 * feature: TIME now displays the amount of run-time spent in GC
1246 * bug fix: The debugger now does a better job of respecting
1247 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
1248 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
1249 workaround for bug 403.)
1250 * bug fix: floating point exception handling now works on FreeBSD
1251 (thanks to NIIMI Satoshi)
1252 * fixed some bugs revealed by Paul Dietz' test suite:
1253 ** MISC.641: LET-conversion were not supposed to work in late
1255 * improvements to the Win32/x86 port:
1256 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
1257 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
1259 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
1260 * new feature: source path information is generated for macro-expansion
1261 errors for use in IDE's like Slime (thanks to Helmut Eller)
1262 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
1263 * bug fix: compiler does not loop forever on an invalid type in
1265 * improvement: compilation of most CLOS applications is significantly
1267 * optimization: added a limited bytecode compiler for simple toplevel
1268 forms, speeding up compilation and FASL loading
1269 * bug fix: the statistical profiler now properly distinguishes anonymous
1272 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
1273 * minor incompatible change: in sbcl-0.9.11 (but not earlier
1274 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
1275 into subclasses of the finalized class. Now user calls to
1276 FINALIZE-INHERITANCE finalize just the one class, and calls by the
1277 system return before any subclasses are finalized.
1278 * minor incompatible change: The reader no longer ignores errors
1279 regarding non-existent packages in #+ and #- feature tests.
1280 * new feature: command line options --no-sysinit, --no-userinit to
1281 inhibit loading the corresponding init files
1282 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
1283 for sending data through UDP sockets (thanks to François-René Rideau)
1284 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
1285 error is signalled instead (thanks to François-René Rideau)
1286 * minor incompatible change: Error signalling behaviour of lexical
1287 operations violating package locks has changed slightly. Refer to
1288 documentation on package locks for details.
1289 * enhancement: EVAL can process IF-expressions without resorting to the
1291 * bug fix: Some operations involving SIN, COS, or TAN weren't being
1292 constant-folded properly due to missing out-of-line functions.
1293 (reported by Mika Pihlajamäki)
1294 * bug fix: LISTEN sometimes returned T even in cases where no data was
1295 immediately available from the stream
1296 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
1297 were reversed. (reported by Levente Mészáros)
1298 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
1299 when given a vector displaced to an adjustable array. (reported
1301 * bug fix: derivation of float boundaries from numbers outside the
1302 appropriate float range (reported by John Wiseman)
1303 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
1305 * bug fix: precomputing cache entries for generic functions with
1306 some subclasses of specializers as yet invalid does not attempt to
1307 fill a cache line with a negative offset. (reported by Levente
1309 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
1310 allow condition class objects as arguments to DOCUMENTATION and
1311 (SETF DOCUMENTATION); only find and set documentation for
1312 structure names for the STRUCTURE doc-type. (suggested by Gary
1314 * improvements to the Win32/x86 port:
1315 ** (user-homedir-pathname) and default initialization file
1316 locations now know about the user's "Documents and Settings"
1317 directory (thanks to Yaroslav Kavenchuk)
1318 ** run-program is implemented (thanks to Mike Thomas)
1319 ** sockets support (thanks to Timothy Ritchey)
1320 ** better backtrace support (thanks to Alastair Bridgewater)
1321 ** sb-grovel supported
1322 ** asdf-install and sb-posix work somewhat
1323 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
1324 * improvements to the Solaris/x86 port:
1325 ** works on Solaris 11/Solaris Express
1326 ** floating-point exception handling support
1327 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
1328 * fixed some bugs revealed by Paul Dietz' test suite:
1329 ** REMOVE-METHOD returns its generic function argument even when
1330 the method is not one of the generic functions' methods.
1331 ** objects declared as MEMBER types can be admissible arguments to
1332 structure accessors.
1333 ** printing characters should simply be printed by the FORMAT ~:C
1335 ** compiler failure when compiling functions with hairy constant
1336 defaults for optional parameters.
1337 ** compiler produces wrong code when MAYBE-INLINE-expanding a
1338 function, which is already optimized.
1340 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
1341 * new platform: experimental support for SBCL x86/Darwin, including
1342 MacOS X 10.4.5 on Intel.
1343 * new feature: Unicode character names are now known to the system
1344 (through CHAR-NAME and NAME-CHAR).
1345 * new feature: the filesystem location of SBCL's core file is
1346 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
1347 * minor incompatible change: the contrib modules SB-POSIX and
1348 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
1349 this change is to make it easier to distribute
1350 location-independent binaries.
1351 * bug fix: as implied by AMOP, standardized classes no longer have
1352 slots named by external symbols of public packages. (reported by
1354 * optimization: calling functions via constant symbols -- as in
1355 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
1356 via the function object as in (FUNCALL #'FOO).
1357 * optimization: on x86 and x86-64, the calling convention for the
1358 case of dealing with an unknown number of values has been altered
1359 to be friendlier to the prediction heuristics implemented,
1360 particularly with reference to CALL and RET pairing. (thanks to
1361 Alastair Bridgewater)
1362 * optimization: on x86, the code for access to thread-local symbol
1363 values and binding/unbinding of thread-local symbols is smaller.
1364 (thanks to Alastair Bridgewater)
1365 * enhancement: CONSTANTP is now able to determine constantness of
1366 more complex forms, including calls to constant-foldable standardized
1367 functions and some special forms beyond QUOTE.
1368 * fixed bug: occasional GC crashes on Solaris/x86
1369 * optimization: x86-64 supports stack allocation of results of simple
1370 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
1371 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
1372 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
1373 (thanks to James Knight)
1374 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
1375 by the standard (thanks to Stephen Compall)
1377 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
1378 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
1379 be used for bundling the runtime and the core file into one
1380 executable binary. This feature is not currently supported on all SBCL
1381 platforms. (thanks to James Bielman and NIIMI Satoshi)
1382 * new feature: a generational or ephemeral garbage collector is now
1383 the default on the PowerPC platform (both Linux and Darwin). The
1384 old Cheney (stop and copy) collector is a build-time option.
1385 (thanks to Cyrus Harmon, after Raymond Toy)
1386 * minor incompatible change: the method by which SBCL finds its
1387 contributed modules has changed; it no longer relies on symbolic
1388 links from an $SBCL_HOME/systems directory, but searches directly
1389 in the subdirectories of $SBCL_HOME.
1390 * enhancement: the dynamic heap size on the Linux/PPC platform is
1391 markedly larger, even using the older Cheney garbage collector.
1392 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
1393 certain complicated string types. (reported by Gary King)
1394 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
1395 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
1396 * fixed bug: DOCUMENTATION on structures no longer signals an
1397 error if no structure type of the right name exists. (reported by
1399 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
1400 applicable non-standard primary or :AROUND method on
1401 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
1402 no longer cause unbound variable errors. (reported by Kevin Reid)
1403 * fixed bug: metacircle resolution in cases where methods have slots
1404 added before the slots from STANDARD-METHOD. (reported by Jean
1406 * fixed bug: the Power PC assembly code for calling into lisp
1407 sometimes computed its return address wrongly (depending on the
1408 15th bit of the address, fixed at link time). This used to
1409 manifest itself in a segmentation violation while building PCL.
1410 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
1411 many others over the years)
1412 * fixed bug: ROOM no longer reports silly numbers for stack usage if
1413 the stack pointer (treated as a signed integer) is negative.
1414 (thanks to Peter van Eynde)
1416 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
1417 * new platform: experimental support for the Windows operating
1418 system has been added. (thanks to Alastair Bridgewater)
1419 * fixed several bugs in and robustified the PPC FFI (including
1420 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
1421 * bug fix: allow non-simple string symbol names (reported by Paul
1423 * bug fix: interrupt handling on NetBSD (thanks to Richard M
1425 * bug fix: saving a core corrupted callbacks on x86/x86-64
1426 * bug fix: closed a loophole in metacircularity detection and
1427 grounding in the PCL implementation of CLOS.
1428 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
1429 (reported by tomppa on #lisp)
1430 * bug fix: clear the direction flag on Lisp -> C transitions, as
1431 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
1432 (reported by Andrej Grozin and Hendrik Maryns)
1433 * optimization: major improvements to GC efficiency on GENCGC platforms
1434 * optimization: faster implementation of EQUAL
1435 * optimization: emit more efficient opcodes for some common
1436 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
1437 * optimization: several other minor code-generation improvements on x86-64
1438 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
1440 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
1441 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1442 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
1443 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
1444 by ANSI as a generic function, and for consistency (SETF
1445 GENERIC-FUNCTION-NAME) is treated likewise.
1446 * fixed bug #233.b: make constraint propagation notice when a variable
1447 value is changed after it is referenced but before it is used
1448 * fixed bug #296: no more arbitrary behaviour from filenames with
1449 odd characters as --load arguments. (reported by Adam Warner)
1450 * fixed bug #390: :CHARACTER-SET pathname components now work as
1451 expected. (reported by Tim Daly Jr)
1452 * fixed bug #391: complicated :TYPE intersections in slot
1453 definitions no longer cause an error in PCL internals.
1454 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
1455 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
1456 returning the number of octets which would be written to the
1457 file-stream. (thanks to Robert J. Macomber)
1458 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
1459 arguments to the internal stream functions. (thanks to David
1461 * bug fix: stack exhaustion is now handled on the x86/SunOS
1462 platform. (thanks to Daisuke Homma)
1463 * bug fix: more accurate ROOM results on GENCGC platforms
1464 * optimization: improved type inference for arithmetic-for
1465 index variables in LOOP
1466 * optimization: faster floating-point SQRT on x86-64
1467 * fixed some bugs revealed by Paul Dietz' test suite:
1468 ** DOCUMENTATION returns NIL instead of "" for method combinations
1469 that don't have a docstring
1471 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
1472 * minor incompatible change: (SETF CLASS-NAME) and (SETF
1473 GENERIC-FUNCTION-NAME) are no longer generic functions, and
1474 therefore are not customizeable by user code (as seems to be at
1475 least permitted and maybe required by AMOP). As a consolation,
1476 however, the SBCL implementation of these functions now calls
1477 REINITIALIZE-INSTANCE as specified by AMOP.
1478 * bug fix: slot-definition documentation is propagated as per ANSI
1479 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
1480 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
1481 argument, as do eslotd initialization methods. (from Pascal
1482 Costanza's "Closer" project)
1483 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
1484 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
1486 * bug fix: it is now possible to have more than one subclass of
1487 STANDARD-GENERIC-FUNCTION without causing stack overflow.
1488 (reported by Bruno Haible, Pascal Costanza and others)
1489 * bug fix: the dependent update protocol now works for generic
1490 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
1491 and Pascal Costanza)
1492 * bug fix: condition-class instances corresponding to
1493 DEFINE-CONDITION forms are now created eagerly. (reported by
1494 Kalle Olavi Niemitalo on comp.lang.lisp)
1495 * bug fix: floating point printing is more accurate in some
1496 circumstances. (thanks to Simon Alexander)
1497 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
1498 merged with *DEFAULT-PATHNAME-DEFAULTS*.
1499 * bug fix: callbacks on OS X now preserve stack-alignment.
1500 (thanks to Cyrus Harmon)
1501 * enhancement: the x86-64 disassembler is much better at
1502 disassembling SSE instructions. (thanks to Lutz Euler)
1503 * enhancement: CHECK-TYPE on a variable now allows type-inference
1504 (thanks to Vincent Arkesteijn)
1505 * optimization: improved performance of EUC-JP external format.
1506 (thanks to NIIMI Satoshi)
1507 * optimization: performance improvements to IO on file streams of
1508 :ELEMENT-TYPE CHARACTER
1509 * optimization: much faster memory allocation on x86-64
1510 * optimization: faster garbage collections (latency and throughput)
1512 * optimization: faster bignum subtraction and division on x86 and x86-64
1513 (thanks to Lutz Euler)
1514 * optimization: more accurate type-inference for loops that use a
1515 floating point index variable or a negative step.
1517 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
1518 * bug fix: add a workaround to SBCL looping infinitely at startup on
1519 Linux kernels with apparently buggy implementations of personality().
1520 (thanks to Svein Ove Aas)
1521 * bug fix: Unicode symbols are correctly printed in LDB backtraces
1522 (thanks to David Lichteblau)
1523 * bug fix: local bindings shadow global symbol macros for
1524 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
1525 * optimization: non-open coded uses of numeric comparison operators
1526 (e.g. >) no longer cons when called with more than one parameter
1527 on platforms supporting dynamic-extent allocation.
1528 * enhancement: saving cores with foreign code loaded is now supported
1529 on MIPS/Linux in addition to the previously supported platforms.
1530 * bug fix: FIND-RESTART now tests for activity, not applicability when given
1531 a restart object as identifier. (reported by Helmut Eller for CMUCL)
1532 * bug fix: division by zero in sb-sprof when no samples were collected
1533 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
1534 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
1535 (thanks to Alexander Kjeldaas)
1536 * new feature: ignoring the return values of destructive functions
1537 like NREVERSE signals a compile-time style-warning.
1538 (thanks to Kevin Reid)
1539 * new feature: the alignment of alien structure fields can be
1540 explicitly specified. (thanks to Cyrus Harmon)
1541 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
1542 and dump core on SIGQUIT
1544 ** incompatible change: threads do not inherit values of specials
1545 from their parents (see manual)
1546 ** bug fix: threads stacks belonging to dead threads are freed by the
1547 next exiting thread, no need to gc to collect thread stacks anymore
1548 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
1549 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
1550 ** bug fix: dynamic variable and thread start related gc lossage
1551 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
1553 * fixed some bugs revealed by Paul Dietz' test suite:
1554 ** SUBTYPEP is slightly more accurate on heinously complicated
1555 CONS types where some of the members have uncertain (in the
1556 NIL, NIL sense) type relationships to each other.
1558 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
1559 no longer purified unless :PURIFY T is explicitly specified.
1560 ** Non-purified cores are significantly smaller than before
1562 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
1563 * new feature: timers based on Zach Beane's excellent timer package
1564 * added support for the following external formats: koi8-u,
1565 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
1566 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
1567 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
1568 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
1569 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
1570 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
1571 * incompatible change: a threaded SBCL will no longer revert to
1572 non-threaded mode on non-NPTL systems, but refuse to start entirely.
1573 * bug fix: interrupts are disabled until startup is complete; no
1574 more sigsegvs when receiving a signal to soon
1575 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
1577 * bug fix: add a workaround for the memory randomization features in
1578 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
1579 workaround will only be in effect on systems with the proc filesystem
1581 * bug fix: printing objects of type HASH-TABLE signals a
1582 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
1584 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
1585 have been running for over 50 days. (reported by Gilbert Baumann)
1586 * bug fix: the logic for getting names of functions gets less
1587 confused when confronded with alternate-metaclass
1588 funcallable-instances. (reported by Cyrus Harmon)
1589 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
1590 consistent, even on internal alternate-metaclass objects.
1591 * bug fix: flush closure information collected by physical
1592 environment analysis prepass before the main pass. (bug reported
1594 * bug fix: compiler pack phase does not modify a hashtable, which is
1595 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
1596 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
1597 may be expanded. (reported by Alexander <malishev> on c.l.l)
1598 * bug fix: no more style-warnings when compiling code using LAST
1599 under high-SPEED low-SPACE settings. (reported by David Wallin
1601 * fixed some incompatibilities between SBCL's MOP and the MOP
1603 ** the METAOBJECT class is now implemented;
1604 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
1605 STANDARD-OBJECT, as required;
1606 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
1607 now compatible (as required by VALIDATE-SUPERCLASS); there
1608 remains a consistency requirement over the presence or absence
1609 of the FUNCTION class in the superclass list of finalized
1610 classes; see the manual for more details;
1611 ** the :AROUND method for COMPUTE-SLOTS on
1612 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
1613 requested slot ordering.
1615 ** bug fix: parent thread now can be gc'ed even with a live
1617 ** bug fix: binding a special with PROGV to no value is not
1618 the same as that symbol not having been bound (thanks to
1620 ** bug fix: binding specials is thread safe (thanks to
1622 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
1623 and SB-SPROF do not die with 'no handler for signal XX in
1624 interrupt_handle_now(..)' anymore
1625 ** bug fix: WITH-TIMEOUT works with multiple running threads
1626 * fixed some bugs revealed by Paul Dietz' test suite:
1627 ** ENSURE-GENERIC-FUNCTION should take a method class object for
1628 the :method-class keyword argument.
1630 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
1631 * new port: the Solaris operating system on x86 processors is now
1632 mostly supported, though some rough edges in the environment
1633 remain. (thanks to Daisuke Homma)
1634 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
1635 space for its heap. (thanks to Thiemo Seufer)
1636 * optimization: SBCL on MIPS platforms now supports dynamic-extent
1637 closures. (thanks to Thiemo Seufer)
1638 * minor incompatible change: eof selects abort in the debugger.
1639 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
1640 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
1642 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
1643 deprecated, and will go away in a future revision of SBCL.
1644 * minor incompatible change: GC-ON and GC-OFF are no longer
1645 implemented with a counter, it does not matter how many times gc
1646 is switched on or off
1647 * bug fix: discriminating functions for generic function classes
1648 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
1649 make invalid assumptions about method precedence order. (reported
1651 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
1652 platforms. (thanks to Thiemo Seufer)
1653 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
1654 * bug fix: correct pathname printing with printer escaping is on.
1655 (thanks to Kevin Reid)
1656 * bug fix: complex VOP definitions in "user-space" no longer trigger
1657 package locks. (reported by Zach Beane)
1658 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
1659 for funcallable-instance functions, and is no different from
1661 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
1662 characters in the current readtable. (reported by Nicholas Neuss)
1663 * bug fix: SUBTYPEP on various CONS types returns more a more
1664 accurate acknowledgment of its certainty.
1665 * optimizations: REMOVE-DUPLICATES now runs in linear time on
1666 lists in some cases. This partially fixes bug 384.
1667 * flush all standard streams before prompting in the REPL and the
1669 * bug fix: signal handling and triggering gc do not conflict
1670 directly anymore, in particular a high frequency sb-sprof does
1671 not prevent gc from running
1672 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
1673 approximation for timezone and DST information between the
1674 universal time epoch and the smallest negative 32-bit time_t.
1675 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
1676 year 1899 so long as the encoded time is non-negative.
1677 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
1678 the correct answer for non-constant rotate amounts when performing
1679 an inline 32-bit rotation.
1681 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
1682 there is only one thread in the session
1683 ** bug fix: memory leak for streams created in one thread and
1684 written to in another
1685 ** bug fix: lockup when compiled with gcc4
1686 ** bug fix: race that allows the gc to be triggered when gc is
1688 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
1689 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
1691 ** bug fix: finished threads are gc'ed properly
1692 * fixed some bugs revealed by Paul Dietz' test suite:
1693 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
1694 has arguments with a different set of applicable methods from
1695 the orignal arguments.
1696 ** The type error thrown by MAP now has a correct expected-type
1698 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
1699 name a compiled function.
1700 ** Three MISC tests where a large bignum was improperly coerced to
1701 a float that couldn't represent that bignum during type
1702 derivation were fixed.
1703 ** SUBTYPEP can now handle types involving the negation of a
1704 list-form FUNCTION type.
1705 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
1706 as (COMPLEX (AND RATIO (NOT FIXNUM))).
1707 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
1709 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
1710 * New feature: Experimental support for bivalent streams: streams
1711 opened with :element-type :default now allow character and binary
1712 (unsigned-byte 8) I/O
1713 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
1714 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
1716 * bug fix: no more highly sporadic "couldn't check whether ~S is
1717 readable" when reading a stream and an interrupt hits in the middle
1718 of a select system call
1719 * compiler better recognizes complex arrays (reported by Eduardo
1721 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
1722 for variables on SBCL built with linkage-tables. (reported by Luis
1724 * various error reporting improvements.
1725 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
1726 (Thanks to Thiemo Seufer)
1727 * bug fix: nested reader invokations work correctly
1728 * bug fix: it is possible to have simultaneous references to foreign
1729 code and foreign data with the same name.
1731 ** added x86-64 support
1732 ** incompatible change: the threading api now works with thread
1733 objects instead of thread ids
1734 ** bug fix: threads are protected from signals and interruption when
1735 starting up or going down
1736 ** bug fix: a race where an exiting thread could lose its stack to gc
1737 ** bug fix: don't halt on infinite error in threads if possible
1738 ** fixed numerous gc deadlocks introduced in the pthread merge
1739 ** bug fix: fixed thread safety issues in read and print
1740 ** bug fix: debugger doesn't hang on session lock if interrupted at
1741 an inappropriate moment
1742 ** bug fix: run-program is now thread safe(r)
1743 ** bug fix: inner with-recursive-lock no longer releases the mutex
1744 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
1745 * fixed some bugs revealed by Paul Dietz' test suite:
1746 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
1747 EXPECTED-TYPE slots filled.
1748 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
1750 ** MAKE-CONDITION accepts classes as type-designators.
1751 ** COMPILE may never return NIL.
1752 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
1753 range before calling Unix time functions
1755 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
1756 * numerous signal handling fixes to increase stability
1757 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
1758 * minor incompatible change: we now correctly canonize default
1759 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
1760 as per the MOP, rather than the historical (INITARG INITFUNCTION
1761 INITFORM). (reported by Bruno Haible)
1762 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
1763 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
1764 for more information.
1765 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
1766 pathname is a directory pathname.
1767 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
1768 * dynamic space size on PPC has been increased to 768Mb. (thanks to
1770 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
1771 :METACLASS argument in addition to a class name. (reported by
1772 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
1773 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
1774 (reported by Bruno Haible; patch from Gerd Moellmann)
1775 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
1777 * bug fix: more cleanups to the floating point exception handling on
1778 x86-64 (thanks to James Knight)
1779 * bug fix: the compiler does not try to constant fold calls of
1780 COERCE and BIT-* functions when they return freshly constructed
1781 objects. (reported by Nikodemus Siivola and Paul Dietz)
1782 * optimization: improved the MIPS versions of generic (in the
1783 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
1784 * optimization: direct conversion of (unsigned-byte 32) to floats on
1785 the PowerPC platform.
1786 * optimization: structure instances with raw slots now use less
1787 memory, and probably show better memory locality. (thanks to
1789 * optimization: DYNAMIC-EXTENT declarations for lists and closures
1790 are treated as requests for stack allocation on the x86-64,
1791 Alpha-32, PPC and SPARC platforms.
1792 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
1793 the wrong file descriptor; implementation of SOCKET-OPEN-P.
1794 (thanks to Tony Martinez)
1796 ** gcing a dead thread can no longer lead to lockups
1797 ** threads block signals until they are set up properly
1798 ** errno is no longer shared by threads
1799 ** interrupt-thread restores the eflags register on x86
1800 ** fixed some lockups due to gc/thread interaction
1801 * fixed some bugs revealed by Paul Dietz' test suite:
1802 ** invalid dotted lists no longer raise a read error when
1803 *READ-SUPPRESS* is T
1804 ** signal an error if a symbol that names a declaration is used
1805 as the name of a type, or vice versa
1806 ** allow using the (declare (typespec var*)) abbreviation for
1807 (declare (type typespec var*)) with all type specifiers
1808 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
1809 ** always return NIL from PPRINT-POP when OBJECT is NIL
1810 ** don't signal errors when pretty-printing malformed LABELS,
1811 FLET or MACROLET forms
1812 ** declarations in a DOLIST body are also in scope for the
1814 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
1816 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
1819 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
1820 * fixed cross-compiler leakages that prevented building a 32-bit
1821 target with a 64-bit host compiler.
1822 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
1823 opened with :IF-EXISTS :OVERWRITE.
1824 * fixed bug 281: error for an invalid qualifier in a short-form method
1825 combination method is not signalled until the faulty method is called.
1826 * bug fix: iteration variable type inferrer failed to deal with open
1827 intervals. (reported by Alan Shields)
1828 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
1829 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
1830 or more runtime options were provided to the sbcl binary.
1831 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
1832 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
1833 * optimization: there's now a fast-path for fixnum arguments in the
1834 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
1835 * optimization: the code generated on x86-64 is more compact thanks
1836 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
1837 * bug fix: floating-point exception handling now partly works on
1838 x86-64. (Thanks to James Knight)
1839 * improvement to the MIPS backend from Thiemo Seufer: C-style
1840 64-bit long long arguments and return values to alien functions
1842 * fixed some bugs revealed by Paul Dietz' test suite:
1843 ** the type-error signalled from WARN has a filled-in DATUM slot.
1844 ** the type-error required when a stream is not associated with
1845 a file has the stream as its datum.
1846 ** type-errors on single-floats on x86-64 no longer have
1847 :INVALID-OBJECT as the datum
1848 ** the type-errors signalled for invalid function names now have
1849 a correct expected type
1850 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1851 for structures defined with a :TYPE. Documentation strings for
1852 typed structures are no longer immediately discarded
1853 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1854 broadcast streams and synonym streams. FILE-LENGTH now also works
1855 on broadcast streams.
1857 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1858 * incompatible change: the --noprogrammer option, deprecated since
1859 version 0.7.5, has been removed. Please use the equivalent
1860 --disable-debugger option instead.
1861 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1863 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1864 inoperational for a while now) has been completely removed.
1865 * null lexical environments are now printed as #<NULL-LEXENV>,
1866 significantly reducing the amount of clutter in typical backtraces.
1867 * documentation on weak pointers, finalization, and after GC hooks
1868 has been added to the manual.
1869 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1870 as other specialized array types, is much faster. SUBSEQ and
1871 COPY-SEQ on such arrays have also been sped up.
1872 * optimization: EQL is now more efficient when at least other argument
1873 is known to be of type (OR FIXNUM (NOT NUMBER)).
1874 * fixed bug: the runtime is now less vulnerable to changes in the
1875 size of the SBCL object on OS X, and virtual memory is reserved for
1876 all spaces that need to be at a fixed address.
1877 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1878 * fixed bug: finalizers and after GC hooks that cause consing are now
1880 * fixed bug: compiler error messages and summaries are now printed to
1881 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1882 * fixed inference of the upper bound of an iteration variable.
1883 (reported by Rajat Datta).
1884 * fixed bug 211e: calling local functions with duplicated constant
1885 keyword argument no longer causes a bogus style warning about an
1887 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1888 ftype declaration to be disregarded. (reported by Dave Roberts)
1889 * fixed bug 373: caused by erronous compilation of references to alien
1890 variables in the runtime on ppc/darwin.
1891 * fixed bug 376: CONJUGATE type deriver.
1892 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1893 (reported by Baughn on #lisp)
1894 * a cross-compiler bug on non-x86 platforms has been identified and
1895 fixed. (thanks to Bruno Haible)
1896 * improvements to the MIPS runtime code for increased stability.
1897 (thanks to Thiemo Seufer)
1898 * increased the maximimum compact environment size to allow
1899 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1900 * partial workaround for bug 135: don't name the function we're
1901 calling for hairy cases of EVAL, so as not to accumulate one environment
1902 entry per call to EVAL. (reported by Kevin Reid)
1903 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1904 * optimization: type testing for non-vector arrays should be faster.
1905 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1907 * bug fix: setting 31st element of a bit vector to zero did not work
1909 * bug fix: redefining a class definition which failed due to a
1910 previous accessor / function clash now works (but see BUGS entry
1911 #380 for more problems in this area). (thanks to Zach Beane)
1912 * the long form of DEFINE-METHOD-COMBINATION disables method group
1913 checking when given a single method group with pattern *. (thanks
1915 * on x86 compiler supports stack allocation of results of simple
1916 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1917 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1918 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1919 * fixed some bugs related to Unicode integration:
1920 ** the restarts for recovering from input and output encoding
1921 errors only appear when there is in fact such an error to
1923 * fixed some bugs revealed by Paul Dietz' test suite:
1924 ** MISC.549 and similar: late transformation of unsafe type
1925 assertions into derived types caused unexpected code
1927 ** SCALE-FLOAT type deriver is less wrong.
1928 ** type derivers for EXP, LOG and similar functions compute result
1929 types for complex arguments better.
1930 ** (MISC.563) CONJUGATE type deriver works for very restricted
1932 ** out-of-line type testers for character strings are available.
1933 ** EQUAL compiler transform understands specialness of objects
1935 ** accessing double-floats stored on the stack now works on x86-64.
1936 ** debugger internals could sometimes create invalid lispobjs,
1937 resulting in GC crashes.
1938 ** MISC.548: type check weakening can convert required type into
1940 ** initialization forms for bindings are not in scope of free special
1943 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1944 * incompatible change: thread support for non-NPTL systems has
1945 been removed - locking is buggy and unreliable. A threaded
1946 SBCL build will now warn at startup and refuse to create
1947 new threads, unless futex support is detected in the kernel
1948 * incompatible change: the top level REPL now has only an ABORT
1949 restart associated with it, not TOPLEVEL and ABORT as it used to.
1950 TOP and TOPLEVEL are now available as debugger commands for
1951 returning to the top level.
1952 * incompatible change: forms evaluated in the REPL now use the
1953 global optimization policy.
1954 * incompatible change: user- and system-initialization files are
1955 no longer processed with LOAD, but by READ and EVAL; hence the
1956 global optimization policy, startup package, readtable, etc,
1958 * The .fasl file format number has been incremented because of
1959 various incompatible changes.
1960 * internal entry point details and argument counts no longer appear
1961 in backtraces unless explicitly requested by setting
1962 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1963 * built-in and standard functions no longer have names like "top
1964 level local call to FOO".
1965 * fixed bug 32: functions defined in non-null lexical environments
1966 now have more legible printed representation
1967 * fixed bug 33: functions defined in non-null lexical environemnts
1968 are now more amenable to inspection by INSPECT.
1969 * workaround for bug 354: XEPs no longer appear in backtraces unless
1970 explicitly requested.
1971 * fixed bug: receiving the signal which results from attempting to
1972 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1973 SIGBUS on other *BSDs) on architectures where the C stack is also
1974 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1975 to debugger confusion or wild instability. (reported by Cheuksan
1977 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1978 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1979 (reported by Lutz Euler)
1980 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1981 compiled code. (reported by Paul Dietz)
1982 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1983 specialized parameter to a method to something that is not TYPEP
1984 the specializer is now possible.
1985 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1986 face of package deletion.
1987 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1988 pathname. (reported by David Tolpin on c.l.l)
1989 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1990 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1991 STRING-STREAM and FILE-STREAM, respectively.
1992 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1993 than previously; in addition, its test suite is now run on build.
1994 (thanks to Luke Gorrie)
1995 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1996 * optimization: added a immediate representation for single-floats
1998 * fixed some bugs revealed by Paul Dietz' test suite:
1999 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
2000 ** attempting to create a package with a colliding nickname causes
2001 correctable errors to be signalled.
2002 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
2003 ** several x86-64 backend bugs related to sign-extension of immediate
2006 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
2007 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
2008 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
2010 * fixed loading of multiply forward-referenced layouts.
2011 (thanks to Cheuksan Wang)
2012 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
2013 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
2014 (see COMPILE-FILE documentation for details of :PRINT option)
2015 * fixed bugs 19 and 317: fixed-format floating point printing is
2016 more accurate. This also fixes a bug reported by Adam Warner
2017 related to the ~@F format directive.
2018 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
2020 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
2021 dispatch table if the from-char is a dispatch macro character.
2022 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
2023 by the wordsize no longer ignore the last word. (reported by Lutz
2025 * fixed bug in type checking of dynamic-extent variables. (reported
2027 * optimization: sequence traversal functions use their freedom to
2028 coerce function designators to functions.
2029 * optimization: code with many calls to CLOS methods specialized on
2030 CLOS classes has had redundant type checks removed.
2031 * optimization: added declarations to speed up operations that access
2032 the internal character database (for example STRING-UPCASE)
2033 * optimization: comparison operations between floats and sufficiently small
2034 fixnums no longer create extra rationals
2035 * fixed some bugs related to Unicode integration:
2036 ** portions of multibyte characters at the end of buffers for
2037 character-based file input are correctly transferred to the
2038 start of the buffer at the next read.
2039 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
2040 passing it through to OPEN.
2041 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
2042 argument given to internal calls to OPEN.
2043 ** the built-in comment readers (introduced by character sequences
2044 ";" and "#|") are more forgiving to encoding errors; they will
2045 STYLE-WARN and then attempt to resync the stream at a character
2046 boundary. (thanks to Teemu Kalvas)
2047 * fixed some bugs in the x86-64 port:
2048 ** Negative short int return values from c-calls are sign-extended
2050 ** The stack is aligned to 16-bytes for c-calls, as required by
2051 the ABI. (thanks to Cheuksan Wang)
2052 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
2053 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
2054 for enums. (thanks to Vincent Arkesteijn)
2055 ** Multiple small optimizations and bugfixes for floating point
2057 * fixed some bugs revealed by Paul Dietz' test suite:
2058 ** Space, Tab, Linefeed, Return and Page have the invalid
2059 secondary constituent character trait.
2060 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
2062 ** WITH-INPUT-FROM-STRING should only update the index place on
2064 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
2065 works more reliably.
2066 ** Bit-array operations (BIT-AND and similar) worked incorrectly
2067 with one-dimensional arrays with fill pointers.
2068 ** TYPE-OF failed on a complex with an integer realpart and a
2070 ** compiler failure during type inference for the code of form
2071 (IF (EQL X (THE ...)) ...) (MISC.535).
2073 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
2074 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
2075 platform. The port supports 61-bit fixnums, large memory spaces
2076 and reloading shared object files.
2077 * enhancement: saving cores with foreign code loaded is now
2078 supported on ppc/Darwin in addition to the previously supported
2080 * enhancement: the statistical profiler now walks deeper into the
2081 call stack for more meaningful call-graphs and accrued time
2082 reports (x86/x86-64 only). It also now reports time spent in
2084 * enhancement: it is now possible to trace most individual methods
2085 of a generic function in addition to tracing the generic function
2087 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
2089 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
2090 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
2091 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
2092 produces an error. (thanks to Vincent Arkesteijn)
2093 * bug fix: NAMESTRING on pathnames with :WILD components in their
2094 directories works correctly. (thanks to Artem V. Andreev)
2095 * fixed bug 125: compiler preserves identity of closures. (reported
2097 * bug fixed: functions with &REST arguments sometimes failed with
2098 "Undefined function" when compiled with (DEBUG 3). (reported by
2100 * bug fix: overflow during compiling of setting element of a bit
2101 vector with constant index and value. (reported by Timmy Douglas)
2102 * build fix: fixed the dependence on *LOAD-PATHNAME* and
2103 *COMPILE-FILE-PATHNAME* being absolute pathnames.
2104 * on x86 compiler partially supports stack allocation of dynamic-extent
2106 * GO and RETURN-FROM do not check the extent of their exit points
2107 when compiled with SAFETY 0.
2108 * fixed some bugs related to Unicode integration:
2109 ** encoding and decoding errors are now much more robustly
2110 handled; it should now be possible to recover even from invalid
2111 input or output to the terminal. (thanks to Teemu Kalvas)
2112 ** provided a first cut at implementing STRING-TO-OCTETS and
2113 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
2114 ** altered the SB-MD5 contributed module slightly, changing the
2115 interface just enough for it to be supportable for builds where
2116 lisp characters are not eight bits.
2117 * fixed some bugs revealed by Paul Dietz' test suite:
2118 ** the FORMATTER-generated functions for ~V[ conditionals require
2119 the correct number of arguments.
2120 ** READ-FROM-STRING returns the mandated second value when applied
2121 to displaced strings.
2122 ** the #\Rubout and #\Backspace characters are treated as invalid
2123 constituent characters by the tokenizer.
2125 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
2126 * new feature: reloading changed shared object files with
2127 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
2128 * new feature: references to foreign variables and functions
2129 can now be compiled and loaded before the corresponding shared
2130 object file is loaded, as long as the foreign definitions are
2131 available at runtime.
2132 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
2133 supported. (thanks to Dan Debertin)
2134 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
2135 just keywords. (thanks to Vincent Arkesteijn)
2136 * fixed bug #331: structure-class instances corresponding to
2137 DEFSTRUCT forms are now created eagerly.
2138 * fixed bug #345: backtraces from calls to undefined functions work
2139 on x86 as well. Related bug #61 is now also partially fixed on x86
2140 (backtraces from throws to unknown catch tags.)
2141 * bug fix: lambda-list parsing is now stricter vrt. order and number
2142 of lambda-list keywords.
2143 * bug fix: as specified by AMOP, an error is signalled if a
2144 class-option appears multiple times in a DEFCLASS form. (reported
2146 * bug fix: RANDOM can be compiled when the compiler derives the type
2147 of its numeric argument as a disjoint set of small integers.
2148 (reported by Paul Dietz)
2149 * bug fix: starting a core saved with shared objects loaded when
2150 those objects are not available no longer causes threaded SBCL to
2151 hang. (reported by Sean Ross)
2152 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
2154 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
2155 argument, which must always be NIL. (reported by Kalle Niemitalo)
2156 * bug fix: printing 1.0d+23 no longer results in an error.
2157 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
2158 * bug fix: structure slot setters preserve evaluation order. (thanks
2160 * fixed some bugs related to Unicode integration:
2161 ** RUN-PROGRAM can allow its child to take input from a Lisp
2162 stream. (reported by Stefan Scholl)
2163 ** ASDF-INSTALL successfully downloads .tar.gz files in all
2164 locales. (reported by Ken Causey)
2165 * fixed some bugs revealed by Paul Dietz' test suite:
2166 ** INCF, DECF and REMF evaluate their place form as specified in
2168 ** FORMATTER expands ~{ ~} iteration directives with V or #
2169 parameters correctly.
2170 ** FORMATTER deals with the ~@[ ~] conditional directive where the
2171 consequent uses no arguments correctly.
2172 ** the system has a partial understanding of the (COMPLEX RATIO)
2174 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
2177 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
2178 * new feature: a build-time option (controlled by the :SB-UNICODE
2179 keyword feature, enabled by default) for building the system with
2180 support for the entire 21-bit character space defined by the
2182 * new feature: the system now has rudimentary external-format
2183 support; the primary user-visible change at this time is that
2184 characters with the high bit set (such as lower-case-e-acute) will
2185 print correctly to a terminal in a UTF-8 environment.
2186 * minor incompatible change: BASE-CHAR no longer names a class;
2187 however, CHARACTER continues to do so, as required by ANSI.
2188 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
2189 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
2190 has been moved to the SB-EXT package (temporarily re-exported from
2192 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
2193 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
2195 * fixed bug #351: better error-handlind and reporting for malformed
2197 * fixed bug #350: bignum-printing is now more memory-efficient,
2198 allowing printing of very large bignums, eg. (expt 2 10000000).
2199 (reported by Bruno Haible)
2200 * fixed bug #302: better primitive-type selection for intersection
2202 * fixed bug #308: non-graphic characters now all have names, as
2203 required. (reported by Bruno Haible)
2204 * bug fix: redefining a class with different superclasses now correctly
2205 removes it from the direct-subclasses of its previous superclasses.
2206 (reported by David Morse)
2207 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
2208 new value now works. (reported by Bruno Haible)
2209 * bug fix: correct canonicalization of multiple non-standard slot
2210 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
2211 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
2212 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
2213 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
2214 now exists, an signals an error.
2215 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
2216 during macroexpansion for non-iteration variables as well. (reported
2217 by Bruno Haible for CMUCL)
2218 * bug fix: Cyclic structures and unprintable objects in compiler
2219 messages no longer cause errors. (reported by Bruno Haible)
2220 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
2221 and READ-FROM-STRING all now return a primary value of NIL if
2222 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
2223 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
2224 (reported by Bruno Haible for CMUCL)
2225 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
2226 do not designate a condition. (reported by Bruno Haible for
2228 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
2229 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
2230 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
2231 permitting resolution in favour of any of the conflicting symbols.
2232 (reported by Bruno Haible for CMUCL)
2233 * bug fix: EQUAL compiler optimizations is less aggressive on
2234 strings which can potentially compare true despite having distinct
2235 specialized array element types.
2236 * bug fix: unit enumerations can be defined without dividing by
2237 zero. (thanks to Vincent Arkesteijn)
2238 * FORMAT compile-time argument count checking has been enhanced.
2239 (report from Bruno Haible for CMUCL)
2240 * a partial workaround for the bug 262: the compiler does not try to
2241 inline-expand a local function doing RETURN-FROM from a deleted
2242 BLOCK. (thanks to Peter Denno for the bug report and to David
2243 Wragg for the simple test case)
2244 * fixed some bugs revealed by Paul Dietz' test suite:
2245 ** RENAME-PACKAGE allows all package designators as new package
2247 ** constraint propagation and lambda variable substitution are
2248 more cautious in dealing with partially deleted code.
2249 ** compiler sometimes lost reoptimization passes.
2250 ** CERROR, given a condition as condition designator, passes its
2251 remaining arguments to the continue format control without
2253 ** Case-altering FORMAT directives work correctly on non-ASCII
2255 ** The REAL type specifier handles bounds outside the floating
2256 point ranges without signalling FLOATING-POINT-OVERFLOW.
2257 ** Functions with IR1-transformations can create intercomponent
2258 references to global functions.
2259 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
2261 ** FORMAT ~R treats a nil value for its first parameter correctly.
2263 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
2264 * enhancement: saving cores with foreign code loaded is now
2265 supported on x86/NetBSD and sparc/Linux in addition to the previously
2266 supported platforms.
2267 * bug fix: on some platforms repeated installations caused multiple
2268 copies of HTML documentation to be installed -- should not happen
2269 any more. (reported by Stefan Scholl)
2270 * bug fix: parsing self-recursive alien record types multiple times
2271 no longer causes infinite recursion. (reported by Thomas F. Burdick,
2272 original patch by Helmut Eller for CMUCL)
2273 * bug fix: stack-exhaustion detection works now on NetBSD as well.
2274 (thanks to Richard Kreuter)
2275 * bug fix: defining classes whose accessors are methods on existing
2276 generic functions in other (locked) packages no longer signals
2277 bogus package lock violations. (reported by François-René Rideau)
2278 * bug fix: special variables as DEFMETHOD parameters no longer have
2279 associated bogus type declarations. (reported by David Wragg and
2281 * bug fix: read-write consistency on streams of element-type
2282 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
2283 * bug fix: redefiniton of the only method of a generic function with
2284 no DEFGENERIC no longer emits a full WARNING. In addition,
2285 redefinition of generic functions with no DEFGENERIC to an
2286 incompatible lambda list now signals an error. (thanks to Zach
2288 * bug fix: DEFGENERIC now works even when there's a function of the
2289 same name in an enclosing lexical environment. (thanks to Zach
2291 * fixed compiler failure, caused by instrumenting code during
2292 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
2293 * optimization: added loop analysis and improved register allocation
2294 * fixed some bugs revealed by Paul Dietz' test suite:
2295 ** POSITION on displaced vectors with non-zero displacement
2296 returns the right answer.
2297 ** (SIMPLE-STRING) is a valid type specifier for sequence
2299 ** *PRINT-LEVEL* handling for slotless structures is pedantically
2301 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
2302 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
2304 ** The justification version of the FORMAT ~< directive treats
2305 non-zero minpad parameter correctly.
2307 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
2308 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
2309 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
2310 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
2311 the supported interface.
2312 * new feature: Single-stepping of code compiled with DEBUG 2 or
2313 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
2314 * new feature: saving cores with foreign code loaded is now
2315 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
2316 Timothy Moore's work for CMUCL)
2317 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
2318 parameters to * instead of NIL if no initform is supplied.
2319 (reported by Johan Bockgaard)
2320 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
2321 correctly when FUNCALL appears as the car of the form. Note:
2322 despite this FUNCALL forms are not currently subject to
2323 compiler-macro expansion. (port of Raymond Toy's fix for the
2324 same from CMUCL, reported by Johan Bockgaard)
2325 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
2326 (thanks for Teemu Kalvas)
2327 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
2328 case instead of terminating the iteration (thanks for Julian
2329 Squires, Sean Champ and Raymond Toy)
2330 * bug fix: incorrect expansion of defgeneric that caused a style
2331 warning. (thanks for Zach Beane)
2332 * bug fix: slot accessor effective method computation works properly
2333 for classes with multiple non-standard applicable methods on
2334 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
2335 * on x86 compiler supports stack allocation of results of LIST and
2336 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
2337 CMUCL implementation by Gerd Moellmann)
2338 * fixed some bugs revealed by Paul Dietz' test suite:
2339 ** FORMAT strings with both the ~<~:;~> form of the justification
2340 directive and pretty-printing directives cause an error.
2342 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
2343 * incompatible change: the internal functions
2344 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
2345 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
2346 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
2347 instead of the old functions.
2348 * new feature: on platforms where "dladdr" is available foreign
2349 function names now appear in backtraces. (based on Helmut Eller's
2351 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
2352 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
2354 * documentation: ASDF manual is now distributed with SBCL.
2355 * bug fix: SBCL can now load its contributed modules with REQUIRE
2356 even if the system-provided entries have been removed from
2357 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
2359 * bug fix: backtraces involving undefined functions or assembly
2360 routines are more informative. (thanks to Brian Downing)
2361 * bug fix: mutually referent alien structures now work correctly.
2362 (reported by Rick Taube)
2363 * bug fix: structures defined by WITH-ALIEN can be referred to
2364 within other definitions in the same WITH-ALIEN.
2365 * bug fix: division operators (MOD, TRUNCATE and the like) with
2366 constant zero divisors and integer dividends no longer generate
2368 * bug fix: provide default methods for INPUT-STREAM-P and
2369 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
2370 * bug fix: improve the detection and resolution of MOP metacycles
2371 (where effective method computation depends on the generic function
2372 having its effective method computed). (reported by Bruno Haible)
2373 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
2374 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
2375 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
2376 * bug fix: pathnames with relative directory components can be
2377 represented relative to default pathnames.
2378 * optimization: in taking the GCD of bignums, reduce the two bignums
2379 to approximately the same size (using Euclid's algorithm) before
2380 applying the more sophisticated binary GCD. (thanks to Juho
2382 * optimization: COUNT on bitvectors now operates word-at-a-time.
2383 * optimization: ASH with a positive, but not necessarily constant,
2384 (leftwards) shift, when in a modular context, is compiled to a
2386 * fixed some bugs revealed by Paul Dietz' test suite:
2387 ** FORMAT variable parameters ("~V<char>") are defaulted properly
2388 if the corresponding argument is NIL.
2389 ** FORMAT directives accepting a minpad parameter treat negative
2391 ** literal commas as character parameters to FORMAT directives are
2393 ** literal spaces directly after ~<Newline> directives within a
2394 format-logical-block (~:< ~@:>) do not induce :FILL-style
2395 conditional newlines.
2396 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
2397 as specified: it no longer includes conditional newlines).
2398 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
2400 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
2401 :PER-LINE-PREFIX argument does not evaluate to a string.
2403 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
2404 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
2405 the manual for details; for now, package locks can be disabled by
2406 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
2407 if no major problems are found then it is likely that they will be
2408 compiled in unconditionally.
2409 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
2410 unsupported operators on all platforms. To load a shared library
2411 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
2412 object file, link it into a shared library outside of SBCL and
2413 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
2415 * bug workaround: ROOM T and the GC were not working together
2416 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
2417 reporting facility aren't preserved. That reporting has been
2418 disabled, so now until and unless someone figures out how to make it
2419 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
2420 * minor incompatible change: as threatened around sbcl-0.8.0, the
2421 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
2422 The symbols are also exported from SB-PCL for backwards
2423 compatibility, but more so than before SB-PCL should be treated as
2424 an implementation-internal package.
2425 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
2427 * the SB-SPROF contrib now works on (most) non-x86 architectures.
2428 It is known as of this release not to work on the Alpha, however.
2429 * fixed bug #167: errors signalled due to illegal syntax in method
2430 bodies are now more legible.
2431 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
2432 designators and can hence be used with TYPEP.
2433 * fixed bug #333: CHECK-TYPE now ensures that the type error
2434 signalled, if any, has the right object to be accessed by
2435 TYPE-ERROR-DATUM. (reported by Tony Martinez)
2436 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
2437 5.1.2.3 for multiple-value place subforms. (reported by Kalle
2439 * fixed bug #334: programmatic addition of slots using specialized
2440 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
2441 and :ALLOCATION :CLASS slots.
2442 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
2443 float exponents. (rereported by Peter Seibel)
2444 * fixed a bug: #\Space (and other whitespace characters) are no
2445 longer considered to be macro characters in standard syntax by
2446 GET-MACRO-CHARACTER.
2447 * fixed bug: initialization of condition class metaobjects no longer
2448 causes an instance of the condition to be created. (reported by
2450 * fixed bug: it is now possible to have slots such that
2451 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
2452 description is neither :INSTANCE nor :CLASS.
2453 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
2454 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
2455 SLOT-VALUE-USING-CLASS), and no longer causes errors with
2456 non-standard SLOT-DEFINITION-ALLOCATION values.
2457 * fixed bugs: various race conditions handling exiting threads.
2458 CL-PPCRE's thread torture test now passes.
2459 * fixed arguably-a-bug: GC time proportional to number of threads in
2460 system even when most of them are idle
2461 * optimization: improved performance of BIT and SBIT on bit-vectors.
2462 * .fasl file incompatibility: The fasl file version number has
2463 been incremented because of changes associated with package locks.
2465 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
2466 * minor incompatible change: the system no longer provides
2467 optimization hints (conditions which are TYPEP
2468 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
2469 mode; these hints are still emitted when higher SPEED optimization
2471 * new contrib module: a sampling profiler (profiling by statistical
2472 sampling, rather than by instrumenting functions) is available as
2473 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
2474 * the behaviour of the standard function ED is now customizeable by
2475 third parties through a hook variable: see ED's documentation
2476 string for information on the protocol.
2477 * the compiler no longer emits efficiency notes for (FUNCALL X)
2478 when the type of X is uncertain under default optimization
2480 * fixed bug 276: mutating a binding of a specialized parameter to a
2481 method to something that is not TYPEP the specializer is now
2483 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
2484 exhibit the required behaviour on the x86 platform. (thanks to
2485 Peter van Eynde, Eric Marsden and Bruno Haible)
2486 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
2487 even for difficult arguments. (reported by Peter Graves)
2488 * fixed bug 141a: the backquote printer now descends quoted
2490 * fixed another bug in backquote printing: no more destructive
2491 modification of the form's list structure. (reported by Brian
2493 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
2494 move between its address being taken and the call to
2495 interrupt_thread, fixing a crashing race condition.
2496 * the SB-POSIX contrib implementation has been adjusted so that it
2497 no longer exhibits ridiculously poor performance when constructing
2498 instances corresponding to C structs.
2500 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
2501 * minor incompatible change: the sb-grovel contrib now treats C
2502 structures as alien (in the sense of SB-ALIEN) objects rather than
2503 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
2504 has implications for memory management of client code
2505 (sb-grovel-returned objects must now be manually managed) and for
2506 type safety (alien objects now have full types).
2507 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
2508 used to control emission of compiler diagnostics, rather than the
2509 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
2510 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
2511 quality should be considered deprecated.
2512 * install.sh now installs the user manual as well
2513 * (not quite a new documentable feature, but worth considering in
2514 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
2515 beginnings of a semantically meaningful condition hierarchy is
2516 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
2518 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
2519 designator as the defaults argument.
2520 * fixed bug: Displaced arrays whose displaced-to array has become
2521 too small now cause ARRAY-DIMENSION to signal an error, providing
2522 for safer bounds-checking. (reported by Bruno Haible)
2523 * fixed bug: DEFCLASS slot definitions with identical :READER and
2524 :WRITER names now signal a reasonable error. (reported by Thomas
2526 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
2528 * fixed bug: Invalid slot specification errors now print correctly.
2529 (thanks to Zach Beane)
2530 * fixed bug 320: Shared to local slot value transfers in class
2531 redefinitions now happen corrently from superclasses as
2532 well. (reported by Bruno Haible)
2533 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
2535 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
2536 improper lists correctly. (reported by Bruno Haible)
2537 * fixed bug 313: source-transform for <fun-name> was erroneously
2538 applied to a call of a value of a variable with name <fun-name>.
2539 (reported by Antonio Menezes Leitao)
2540 * fixed bug 307: The obsolete instance protocol ensures that
2541 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
2542 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
2543 remove dead unknown-values globs from the middle of the stack.
2544 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
2545 easier to resume long computations after using *BREAK-ON-SIGNALS*
2546 to diagnose and fix failures. (thanks to Nikodemus Siivola)
2547 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
2548 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
2549 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
2550 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
2551 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
2553 * fixed a (fairly theoretical) bug in string printing: if
2554 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
2555 does not have array-element-type equal to the most general string
2557 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
2558 function name. (thanks to Nikodemus Siivola)
2559 * fixed bug: compile-time format string checker failed on
2560 non-closed ~{. (reported by Thomas F Burdick)
2561 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
2562 don't warn on legal loop code involving more than one aggregate
2563 boolean. (thanks to Nikodemus Siivola)
2564 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
2565 in keyword position to MAKE-PACKAGE was not being reported as
2567 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
2568 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
2570 * fixed bugs 280 and 312: the checking for multiple definitions in a
2571 file is less likely to become confused by uses of inline
2573 * fixed bug: the #S reader macro performs the keyword coercion
2574 specified for slot names. (reported by Kalle Niemitalo)
2575 * fixed bug: lambda lists may contain symbols whose names start with
2576 & but are not lambda-list-keywords; their occurrence triggers a
2578 * fixed bug 321: define-method-combination argument lambda lists do
2579 not cause invalid code to be generated when &optional or &aux
2580 variables have default values. (reported by Bruno Haible)
2581 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
2582 have CLOS classes; this appears also to have fixed reported
2583 instability in stack exhaustion detection.
2584 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
2585 MUFFLE-WARNING when no associated restart is present is now
2587 * optimization: rearranged the expansion of various defining macros
2588 so that each expands into only one top-level form in a
2589 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
2591 * optimization: used a previously little-used slot in symbols to
2592 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
2594 * fixed some bugs revealed by Paul Dietz' test suite:
2595 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
2597 ** Fixed an optimization bug related to inheritance of initforms
2598 from local to shared slots.
2599 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
2600 ** CAST optimizer forgot to flush argument derived type.
2601 ** print/read consistency on floats is now orders of magnitude
2602 more likely. (thanks also to Bruno Haible for a similar report
2604 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
2605 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
2606 for objects of type REAL. Make it so.
2607 ** SXHASH is specified (infelicitously) to respect similarity,
2608 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
2609 it so. (thanks to Markus Ziegler)
2610 ** on the Alpha, the compiler succeeds in compiling functions
2611 returning a known number of arguments greater than 63.
2612 ** fixed handling of invalid NIL arguments in keyword position
2614 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
2616 ** provided a readably-printable representation for RANDOM-STATE
2618 ** ensured that pathnames, where they have a namestring, always
2619 print using #P"..." syntax.
2621 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
2622 * Support for the forthcoming 2.0 version of the NetBSD kernel
2623 running on x86 hardware has been added. (thanks to Perry
2624 E. Metzger most immediately, and others for their past work)
2625 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
2626 current release version; SBCL's OpenBSD support had been broken
2627 since about the time of OpenBSD's switch to ELF binary format.)
2628 * [placeholder for DX summary]
2629 ** user code with &REST lists declared dynamic-extent, under high
2630 speed or space and low safety and debug optimization policy.
2631 * The manual has been converted to Texinfo format and the debugger
2632 chapter from the cmucl manual has been added.
2633 * A facility has been added to extract documentation strings from
2634 sbcl and store them as Texinfo-formatted snippets for inclusion in
2635 the manual (via Texinfo's @include directive)
2636 * bug fix: compiler emitted division in optimized DEREF. (thanks for
2637 the test case to Dave Roberts)
2638 * bug fix: multidimensional simple arrays loaded from FASLs had fill
2639 pointers. (reported by Sean Ross)
2640 * bug fix: PROFILE output is printed nicely even for large numerical
2641 values. (thanks to Zach Beane)
2642 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
2643 greater than 32 handle EOF correctly.
2644 * bug fix: on X86 an immediate argument of the IMUL instruction is
2645 correctly printed in disassembly. (thanks to Lutz Euler)
2646 * bug fix: class slots in redefined classes preserve their old
2647 values. (thanks to Bruno Haible and Nikodemus Siivola)
2648 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
2649 succeeds. (reported by Marco Baringer)
2650 * fixed some bugs revealed by Paul Dietz' test suite:
2651 ** READ-SEQUENCE now works on ECHO-STREAMs.
2652 ** RATIONALIZE works more according to its specification. (thanks
2655 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
2656 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
2657 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
2658 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
2659 should matter to you only if you rebind the printer control
2660 variables and then find you want different bindings in the
2661 debugger than in the ordinary execution of your program.)
2662 * The runtime build system has been tweaked to support building
2663 (on SPARC/SunOS) using a C compiler which invokes Sun's own
2664 assembler and linker. (thanks to Nikodemus Siivola)
2665 * Unbound, undefined, undeclared variables now trigger full
2666 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
2667 more often programmer error than deliberate exploitation of undefined
2669 * optimization: the hash algorithm for strings has changed to one
2670 that is less vulnerable to spurious collisions. (thanks to Juho
2672 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
2673 needless bounds checking. (thanks to Juho Snellman)
2674 * optimization: implemented multiplication as a modular
2675 (UNSIGNED-BYTE 32) operation on the PPC backend.
2676 * fixed some bugs revealed by Paul Dietz' test suite:
2677 ** ADJUST-ARRAY now copies the datum in a zero rank array if
2679 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
2680 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
2681 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
2683 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
2685 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
2687 ** ECHO-STREAMs no longer attempt to echo the end of file value to
2688 their output stream on EOF from read.
2689 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
2690 have been read to end-of-file.
2691 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
2693 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
2694 description of determination of which consecutive characters
2696 ** Printing the "Space" character with escaping on now yields "#\\ ",
2697 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
2698 ** Reading floating-point numbers with *READ-BASE* set to a number
2699 less than 10 works correctly.
2700 ** Reading floating-point numbers with *READ-BASE* set to a number
2701 more than 10 works correctly.
2702 ** Printing with *PRINT-READABLY* targets the standard readtable, not
2703 the readtable currently in effect.
2705 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
2706 * minor incompatible change: parsing of namestrings on a physical
2707 (Unix) host has changed; numbers after the final #\. in a
2708 namestring are no longer interpreted as a version field. This is
2709 intented to be largely invisible to the user, except that the
2710 meaning of the namestring "*.*.*" has changed: it now refers to a
2711 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
2712 should usually be replaced by
2713 (make-pathname :name :wild :type :wild :version :wild)
2714 with the added benefit that this is more likely to be portable.
2715 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
2716 option to OPEN now signals an error if the file being opened
2717 exists; this may have an impact on existing code.
2718 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
2719 signals received, on the PowerPC platforms (both Linux and
2720 Darwin). (thanks to Pierre Mai for pointing out the location of
2722 * several fixes on OS X: The system now builds and runs cleanly on
2723 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
2724 (thanks to Brian Mastenbrook)
2725 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
2726 second-resolution: integer multiples of 1/3600 between -24 and 24.
2727 (thanks to Vincent Arkesteijn)
2728 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
2729 type when called with 1 argument; PEEK-CHAR checked type of
2730 PEEK-TYPE only after having read first character from a
2731 stream. (reported by Peter Graves)
2732 * bug fix: the garbage collector now has much better locality
2733 behaviour, and in particular no longer treats all memory as being
2734 exhausted when many small objects point to each other in a deeply
2736 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
2737 recognized as being TYPEP their class.
2738 * bug fix: the PUSHNEW documentation string has been corrected.
2739 (thanks to Vincent Arkesteijn)
2740 * bug fix: defaulting of the value for the last of an atypically
2741 large number of multiple values being bound was not being
2742 performed correctly on the Alpha or PPC platforms
2743 * optimization: implemented multiplication as a modular
2744 (UNSIGNED-BYTE 32) operation on the x86 backend.
2745 * optimization: SEARCH on simple-base-strings can now be open-coded.
2746 (see also contrib/compiler-extras.lisp for inspiration for
2747 teaching the compiler about the Boyer-Moore algorithm).
2748 * value, returned by MAX (and MIN) called with several EQUALP, but
2749 not EQL, arguments now does not depend on compiler settings.
2750 * fixed some bugs revealed by Paul Dietz' test suite:
2751 ** in stack analysis liveness information is propagated from
2752 non-local entry points.
2753 ** pathwise CAST removing failed when the CAST node did not start
2755 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
2756 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
2758 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
2759 host is already defined.
2760 ** RENAME-FILE works on streams instead of signalling an internal
2762 ** PEEK-CHAR uses the current readtable when determining whether
2763 or not a character is whitespace.
2764 ** MERGE-PATHNAMES handles the case when the pathname does not
2765 specify a name while the default-pathname specifies a version.
2766 ** Pathnames now stand a better chance of respecting print/read
2768 ** Attempting to use standardized file system operators with a
2769 pathname with invalid :DIRECTORY components signals a
2771 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
2773 ** DIRECTORY on logical pathnames is more correct.
2774 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
2775 signal a TYPE-ERROR if their argument is not a stream.
2776 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
2777 designator argument does not designate a stream.
2778 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
2779 examining the synonym.
2780 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
2782 ** OPEN and WITH-OPEN-STREAM allow opening streams with
2783 element-type larger than ([UN]SIGNED-BYTE 32).
2785 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
2786 * When built with the :SB-FUTEX feature, threaded builds now take
2787 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
2788 for faster/more reliable mutex and condition variable support.
2789 * Incompatible change (but one you probably shouldn't have been using
2790 anyway): the interface and code for arbitrating between multiple
2791 threads in the same user session has been redesigned.
2792 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
2793 error when called without an explicit environment argument.
2794 (thanks to Nikodemus Siivola)
2795 * bug fix: buffered :DIRECTION :IO streams are less likely to become
2796 confused about their position. (thanks to Adam Warner and Gerd
2798 * bug fix: Pretty printing backquoted forms with unquotations in the
2799 argument list position of various code constructs such as LAMBDA
2800 now works correctly. (reported by Paul Dietz)
2801 * bug fix: Pretty printing unquotations no longer loses all
2802 stream position information.
2803 * optimization: performance of string output streams is now less
2804 poor for multiple small sequence writes.
2805 * optimization: performance of CSUBTYPEP in the presence of complex
2806 expressions involving CONS and NOT many times has been improved.
2807 (reported by Paul Dietz)
2808 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
2810 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
2811 streams for the REPL, for the debugger, and so on. (thanks to
2813 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
2814 * fixed some bugs revealed by Paul Dietz' test suite:
2815 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
2816 is ignored if it is too small, rather than propagating through
2817 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
2818 ** extremely complex negations of CONS types were not being
2819 sufficiently canonicalized, leading to inconsistencies in
2821 ** VALUES tranformer lost derived type.
2823 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
2824 * fixed a bootstrapping bug: the build process no longer assumes
2825 that the various BOOLE-related constants have the same value in
2826 host and target lisps. (noted by Paul Dietz' test suite on an
2827 SBCL binary built from CLISP)
2828 * The system can now be dynamically linked on the MIPS platform,
2829 which enables dynamic loading of foreign code from Lisp. (thanks
2830 to Ralf Baechle for discussions on the MIPS ABI)
2831 * The system now records debugging information for its own source
2832 files in a filesystem-position-independent manner, relative to
2833 the "SYS" logical host.
2834 * fixed a compiler bug: MV-LET convertion did not check references
2835 to the "max args" entry point. (reported by Brian Downing)
2836 * tweaked disassembly notes to be less confident about proclaiming
2837 some instruction as an LRA. (thanks to Brian Downing)
2838 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
2839 now each have their own history, command character, and other
2840 characteristics. (thanks to David Lichteblau)
2841 * fixed some bugs revealed by Paul Dietz' test suite:
2842 ** compiler failure in compiling LOGAND expressions including a
2844 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2845 shift greater than 32.
2846 ** FLUSH-DEST did not mark blocks for type check regeneration.
2847 ** HANDLER-CASE failed to accept declarations in handler clauses
2848 in some circumstances.
2850 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2851 * New code in contrib/sb-introspect (still probably not entirely
2852 stable yet) provides some support for smart Lisp development
2853 environments like SLIME.
2854 * The conditions signalled for errors occurring when loading .fasl
2855 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2856 in a way which should help ASDF recover gracefully.
2857 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2858 stuff has been cleaned up. If you code contrib/ stuff, this might
2859 affect you, and you can look at contrib/README, contrib/STANDARDS,
2860 and/or the 0.8.4.27 diff to check.
2861 * In full calls the compiler now does not generate checks for declared
2862 argument types for all arguments.
2863 * various threading fixes
2864 ** and some experimental patches which didn't make it into
2865 the main tree for this release, but which are shipped in
2866 contrib/experimental-thread.patch as a possible fix for some
2867 failures (deadlock, spinning...) in GC-intensive multithreaded
2869 * fixed PPC build problem (source code incompatibility of different
2870 library versions): added offsetof() hackery which attempts to divine
2871 where glibc maintainers put uc_mcontext today
2872 * fixed bug 282: compiler does not trust type assertions while passing
2873 arguments to a full call.
2874 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2875 &OPTIONAL VALUES type parameter.
2876 * fix bug 214: algorithm for noting rejected templates is now more
2877 similar to that of template seletion. (also reported by rydis on
2879 * fixed bug 141b: printing backquoted information readably and prettily
2880 inserts a space where necessary.
2881 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2882 (CHARACTER 10) are now reported as errors, rather than propagated
2883 as unknown types. (reported by piso on #lisp)
2884 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2885 with values NIL and :ERROR. (thanks to Milan Zamazal)
2886 * fixed bug 191c: CLOS now does proper keyword argument checking as
2887 described in CLHS 7.6.5 and 7.6.5.1.
2888 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2889 longer raise an error; further, using a list as a for-as-arithmetic
2890 counter now raises a meaningful error.
2891 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2892 now understood by sequence creation functions such as MAKE-SEQUENCE
2894 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2895 a STRING-INPUT-STREAM.
2896 * compiler enhancement: SIGNUM is now better able to derive the type
2898 * type declarations inside WITH-SLOTS are checked. (reported by
2900 * fixed some bugs revealed by Paul Dietz' test suite:
2901 ** incorrect optimization of TRUNCATE for a positive first
2902 argument and negative second.
2903 ** compiler failure in let-convertion during flushing dead code.
2904 ** compiler failure while deriving type of TRUNCATE on an
2905 interval, containing 0.
2906 ** ASH of a negative bignum by a negative bignum count now returns
2908 ** intersection of CONS types now canonicalizes properly, fixing
2909 inconsistencies in SUBTYPEP.
2911 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2912 * incompatible change: The --disable-debugger command line
2913 option now clobbers the debugger at a more fundamental
2914 level, by redefining #'INVOKE-DEBUGGER instead of by
2915 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2916 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2917 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2918 enter the debugger REPL and then suffer recursive errors
2919 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2920 while under the new system, BREAK in --disable-debugger mode
2921 terminates the system just as an unhandled error would.
2922 * fixed compiler performance when processing loops with a step >1;
2923 * bug fix: DOCUMENTATION now retrieves generic function
2924 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2925 support has been systematized, and now supports the methods
2926 specified by ANSI, along with a default method and a method for
2927 slot documentation. (reported by Nathan Froyd)
2928 * bug fix: effective methods associated with a generic function are
2929 no longer cached over a change of that generic function's method
2930 combination. (reported by Andreas Fuchs)
2931 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2932 names a non-existent directory. (thanks to Andreas Fuchs)
2933 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2934 the PPC platform, lead to stack corruption; this has been fixed.
2935 (reported by Rainer Joswig)
2936 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2937 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2938 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2939 arithmetic in general, is now much more reliable on the PPC
2941 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2942 platform now returns the right answer.
2943 * optimization: restored some effective method precomputation in
2944 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2945 precomputation is now tunable.
2946 * optimization: compiler-internal data structure use has been
2947 reviewed, and changes have been made that should improve the
2948 performance of the compiler by about 20%.
2949 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2950 simple-base-strings and simple-bit-vectors is improved.
2951 * optimization: the optimization of 32-bit logical and arithmetic
2952 functions introduced in version 0.8.3 on the x86 has been
2953 implemented on the mips, ppc and sparc platforms; an
2954 implementation of the same facility, but for 64-bit arithmetic,
2955 has been added for the alpha.
2956 * microoptimization: the compiler is better able to make use of the
2957 x86 LEA instruction for multiplication by constants.
2958 * bug fix: in some situations compiler did not report usage of
2959 generic arithmetic in (SPEED 3) policy.
2960 * bug 145b fix: compiler used wrong type specifier while converting
2961 MEMBER-types to numeric.
2962 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2964 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2965 index argument now works properly on the Alpha platform.
2966 * bug fix: floating point exception treatment on the Alpha platform
2968 * bug fix: FILE-POSITION works much better on string input and
2969 output streams. (thanks to Nikodemus Siivola)
2970 * bug fix: many threading/garbage collection symptoms sorted.
2971 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2972 might be pseudo-atomic.
2973 * internal change: Stopping for GC is now done with signals not
2974 ptrace. GC is now done in whichever thread wanted it, instead of
2976 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2978 * bug fix: result form in DO is not contained in the implicit
2980 * incompatible change: ICR structure is changed; the value part of
2981 CONTINUATION is now called LVAR; corresponding functions are
2982 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2983 * added type deriver for ISQRT (thanks to Robert E. Brown).
2984 * bug fix: better support for loading from the command line when an
2985 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2987 * fixed some bugs revealed by Paul Dietz' test suite:
2988 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2989 ** ROUND and FROUND now give the right answer when given very
2990 small float arguments.
2991 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2993 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2994 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2995 ** LCM should return a non-negative integer.
2996 ** PARSE-INTEGER returned the index of a terminator instead of the
2997 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2998 ** PARSE-INTEGER returned an incorrect index being applied to a
3000 ** LCM with two arguments of 0 returns 0 rather than signalling
3002 ** unsigned addition of a 32-bit constant with the high bit set no
3003 longer causes an internal compiler error on the x86.
3004 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
3005 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
3006 with negative last argument.
3007 ** byte specifiers with zero size and position no longer cause
3008 an error during type derivation.
3009 ** bignum multiplication on the Alpha platform now returns the
3011 * porting: The system now builds on SuSE AMD64, although it still
3012 generates a 32-bit binary.
3013 * .fasl file incompatibility: The fasl file version number has
3014 been incremented (because of the changes to internal compiler
3015 data structures referred to above).
3017 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
3018 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
3019 more accurately, on the Darwin kernel running on PowerPC hardware.
3020 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
3021 * Compiler code deletion notes now signal a condition of type
3022 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
3023 an associated MUFFLE-WARNING restart.
3024 * The compiler now performs limited argument count validation of
3025 constant format strings in FORMAT, and where appropriate in ERROR,
3026 CERROR and WARN. (thanks to Gerd Moellmann)
3027 * New ASDF-INSTALL contrib can be used for automatic download and
3028 installation of third-party Lisp code from CCLAN or other sites
3030 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
3031 which forces another thread to execute a function supplied by the
3033 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
3034 now accept and act upon their :ELEMENT-TYPE keyword argument.
3035 (reported by Martin Atzmueller, Edi Weitz)
3036 * bug fix: FILE-POSITION now accepts position designators up to
3037 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
3038 is the greater. (thanks to Patrik Nordebo)
3039 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
3040 Kalle Olavi Niemitalo)
3041 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
3042 printable. (reported by Eric Marsden)
3043 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
3044 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
3045 * x86 bug fix in control stack exhaustion checking: now shows backtrace
3046 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
3047 (thanks to Stig Sandoe)
3048 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
3049 UNREAD-CHAR now correctly includes the unread character in the
3050 target sequence. (thanks to Gerd Moellmann)
3051 * bug fix in threaded builds: the system can now be suspended and
3052 resumed by shell job control with minimal disruption.
3053 * bug fixes in times and timezones >2038AD
3054 * better handling of "where is GNU make?" problem in build scripts
3055 (thanks to Nikodemus Siivola)
3056 * new optimization: inside a named function any reference to a
3057 function with the same name is considered to be a self-reference;
3058 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
3059 optimization quality.
3060 * new optimization on x86: logical functions and + now have
3061 optimized (UNSIGNED-BYTE 32) versions, which are automatically
3062 used when the result is truncated to 32 bits.
3063 * VALUES declaration is partially enabled.
3064 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
3065 * fixed some bugs revealed by Paul Dietz' test suite:
3066 ** The system now obeys the constraint imposed by
3067 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
3068 types form a lattice under type intersection.
3069 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
3070 ** ASSOC now ignores NIL elements in an alist.
3071 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
3072 and (1+ MOST-POSITIVE-FIXNUM) answers.
3073 ** The addition of a method with invalid qualifiers to a generic
3074 function does not cause an error to be signalled immediately;
3075 a warning is signalled, and the error is generated only on
3076 calling the generic function.
3077 * changed .fasl file version number, in part to add type codes for
3078 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
3079 obscure ANSI requirements
3081 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
3082 * fixed bug 148: failure to inline-expand a local function left
3083 garbage, confusing the compiler.
3084 * fixed bugs 3cd: structure slot readers perform type check if the
3085 slot can have an invalid value (i.e. it is either not initialized
3086 or can be written with a less specific slot writer).
3087 * bug fix: the compiler now traps array references to elements off
3088 the end of an array; previously, the bounds checking in some
3089 circumstances could go off-by-one.
3090 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
3092 * type declarations for array element types now obey the description
3093 on the CLHS page "Declaration TYPE", as per discussions on
3094 sbcl-help around 2003-05-08. This means that a declaration
3095 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
3096 declaration, all references to BAR will be asserted or assumed
3097 (with THE, so dependent on compiler policy) to involve objects of
3098 type FOO. Note that no such declaration is implied in
3099 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
3100 * declared types of functions from the "Conditions"
3101 chapter. (reported by Paul Dietz)
3102 * bug fix: CERROR accepts a function as its first argument.
3103 * bug fix: NTH an NTHCDR accept a bignum as index
3104 arguments. (reported by Adam Warner)
3105 * optimization: character compare routines now optimize comparing
3106 against a constant character. (reported by Gilbert Baumann)
3107 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
3108 argument now works properly on the MIPS platform.
3109 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
3110 * repeated evaluation of the same DEFSTRUCT, a slot of which is
3111 declared to have a functional type, does not cause an error
3113 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
3114 many :DEFAULT-INITARGS. (reported by Istvan Marko)
3115 * fixed bug: if last continuation of a deleted block has a
3116 destination, this destination should be deleted too. (reported by
3118 * fixed a bug in the bootstrap process: the host compiler's values
3119 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
3120 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
3121 test case from Patrik Nordebo)
3122 * improved the ability of the disassembler on the PPC platform to
3123 provide helpful disassembly notes.
3124 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
3125 the class in more cases than previously.
3126 * bug fix: FILE-POSITION now understands :START and :END for
3127 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
3128 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
3129 8) data. (thanks to David Lichteblau)
3130 * bug fix: it is possible to add a method to a generic function
3131 without lambda list.
3132 * bug fix: reader failed to signal END-OF-FILE inside an
3133 object representation. (reported by Nikodemus Siivola)
3134 * fixed some bugs revealed by Paul Dietz' test suite:
3135 ** LAST and [N]BUTLAST should accept a bignum.
3136 ** condition slot accessors are methods.
3137 ** (VECTOR NIL) is a subtype of STRING.
3139 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
3140 * minor incompatible change: some nonsensical specialized lambda
3141 lists (used in DEFMETHOD) which were previously ignored now signal
3143 * minor incompatible change: the system is now aware of the types of
3144 variables in the COMMON-LISP package, and will signal errors for
3145 most violations of these type constraints (where previously they
3146 were silently accepted).
3147 * minor incompatible change: COMPILE-FILE now uses the freedom
3148 afforded (ANSI 3.2.2.3) to use derived function types for
3149 functions defined in the same file. This also permits the system
3150 to warn on static type mismatches and function
3151 redefinition. (Currently it does not work with high DEBUG level.)
3152 * minor incompatible change: VALUES declaration is disabled.
3153 * When issuing notes, the compiler now signals a condition of type
3154 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
3155 restart for use in user handlers. It is expected that the
3156 COMPILER-NOTE condition will eventually become a condition
3157 supertype to a hierarchy of note types, which will then be
3158 handleable in a similar fashion. However, at the moment, no such
3159 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
3160 but it's an implementation detail, not a classification for the
3162 * Changes in type checking closed the following bugs:
3163 ** type checking of unused values (192b, 194d, 203);
3164 ** template selection based on unsafe type assertions (192c, 236);
3165 ** type checking in branches (194bc).
3166 * A short form of VALUES type specifier has ANSI meaning (it has
3167 increased the number of situations when SBCL cannot perform type
3169 * fixed bug in DEFSTRUCT: once again, naming structure slots with
3170 keywords or constants is permissible.
3171 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
3172 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
3173 classes. (thanks to Antonio Martinez)
3174 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
3175 outputs FRESH-LINE or TERPRI, and no longer converts its stream
3176 argument to a pretty-print stream. Instead, it leaves any such
3177 operations to DESCRIBE-OBJECT methods.
3178 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
3180 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
3181 lambda lists. (thanks to David Lichteblau)
3182 * bug fix: Defining a generic function with a :METHOD-CLASS being a
3183 subclass of STANDARD-METHOD no longer causes stack exhaustion.
3184 (thanks to Gerd Moellmann)
3185 * fixed bug 246: increased compilation speed of long
3186 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
3188 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
3190 * DEFCONSTANT now throws a condition of type
3191 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
3192 constant to a non-EQL value; CONTINUE and ABORT restarts
3193 respectively change and preserve the value.
3194 * fixed bug 63: The code walker, part of the implementation of CLOS,
3195 is now better at handling symbol macros.
3196 * bug fix: There is no longer an internal implementation type named
3197 CL:LENGTH. (reported by Raymond Toy)
3198 * bug fix: In macro-like defining macros/special operators the
3199 implicit block does not enclose the lambda list.
3200 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
3201 enclosed in parentheses) are not suitable as type specifiers, and
3202 their use properly signals an error now.
3203 * bug fix: An argument count mismatch for a type specifier in code
3204 being compiled no longer causes an unhandled error at compile
3205 time, but signals a compile-time warning.
3206 * fixed simple vector readable printing
3207 * bug fix: DESCRIBE takes more care over whether the class
3208 precedence list slot of a class is bound before accessing it.
3209 (reported by Markus Krummenacker)
3210 * bug fix: FORMATTER can successfully compile pretty-printer format
3211 strings which use variants of the ~* directive inside.
3212 * bug fix: SEARCH now applies its TEST predicate to the elements of
3213 the arguments in the correct order. (thanks to Wolfhard Buss)
3214 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
3215 in type inference. (thanks to Robert E. Brown)
3216 * bug fix: Reading in symbols with an explicit package name of ""
3217 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
3218 * fixed some bugs revealed by Paul Dietz' test suite:
3219 ** NIL is now allowed as a structure slot name.
3220 ** Arbitrary numbers, not just REALs, are allowed in certain
3221 circumstances in LOOP for-as-arithmetic clauses.
3222 ** Multiple class redefinitions before slot access no longer
3223 causes a type error.
3224 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
3225 association between the name and a class.
3226 ** Generic functions with non-standard method-combination and over
3227 five methods all of which return constants no longer return NIL
3228 after the first few invocations. (thanks to Gerd Moellmann)
3229 ** CALL-NEXT-METHOD with no arguments now passes the original
3230 values of the arguments, even in the presence of assignment.
3231 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
3233 ** DISASSEMBLE works with closures and funcallable instances.
3234 ** ADD-METHOD now returns the generic function, not the new method.
3235 ** FIND-METHOD signals an error if the lengths of the specializers
3236 is incompatible with the generic function, even if the ERRORP
3238 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
3239 which its argument is a member.
3240 ** DEFCLASS only redefines the class named by its class-name
3241 argument if that name is the proper name of the class;
3242 otherwise, it creates a new class.
3243 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
3244 of the UNBOUND-SLOT condition to the name of the slot.
3245 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
3246 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
3247 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
3248 SLOT-MAKUNBOUND in the specified fashion.
3250 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
3251 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
3252 cross-compilation host. As a consequence, we can now bootstrap our
3253 way up to SBCL starting with a bare gcc toolchain and human-readable
3254 source code (first the source to CLISP, then the source to SBCL).
3255 * A contributed module containing a partial implementation of the
3256 simple-streams interface has been included. (thanks to Rudi
3258 * A contributed module implementing the RFC1321 Message Digest
3259 Algorithm, known as MD5, has been included.
3260 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
3261 no longer has any effect, as the code controlled by this feature
3262 has been deleted. (As far as we know, no-one has ever built using
3263 this feature, and its semantics were confused in any case).
3264 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
3265 work on conditions (as required by the ANSI specification),
3266 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
3267 expected behaviour on conditions. Users should note, however,
3268 that such behaviour is not required by the ANSI specification,
3269 and so use of this behaviour may render their code unportable.
3270 * fixed some bugs revealed by Paul Dietz' test suite:
3271 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
3273 ** &ENVIRONMENT parameter in macro lambda list is bound first.
3274 ** SXHASH on condition objects no longer returns NIL.
3275 ** :ALLOCATION :CLASS slots are better treated; their values are
3276 updated on class redefinition, and initforms inherited from
3277 superclasses are applied.
3278 ** REMOVE-METHOD returns its generic function argument even when
3279 no method was removed.
3280 ** SHARED-INITIALIZE now initializes the values of the requested
3281 slots, including those with :ALLOCATION :CLASS.
3282 ** ALLOCATE-INSTANCE now works on structure classes defined via
3283 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
3285 ** SLOT-EXISTS-P now works on conditions, as well as structures
3287 ** MAKE-LOAD-FORM now has the required methods on
3288 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
3289 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
3290 keyword, and now implements the SLOT-NAMES argument.
3291 ** methods with &OPTIONAL arguments no longer allow too many
3292 arguments to be passed in the call without error.
3293 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
3294 option is consistent with the required arguments of the generic
3295 function lambda list.
3296 * bug fix: REQUIRE accepts a string designator. (Thanks to
3298 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
3299 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
3300 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
3301 * bug fix: The compiler now checks for duplicated variables in macro
3303 * bug fix: SETQ on globals returns the correct value.
3304 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
3305 not a COMPILER-ERROR (followed by some other strange error on
3306 choosing the CONTINUE restart).
3307 * bug fix: make.sh and friends are now more consistent in the way that
3308 they look for GNU "make".
3310 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
3311 * experimental native threads support (on x86 Linux >=2.4 only).
3312 This is not compiled in by default: you need to add :SB-THREAD to
3313 the target features. See the "Beyond ANSI" chapter of the manual
3315 * fix for longstanding nonANSIism: The old distinction between
3316 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
3317 The return value from CL:FIND-CLASS is now a CLOS class, and
3318 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
3319 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
3320 * An interface to the MetaObject Protocol, as described in Kiczales,
3321 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
3322 MIT Press, 1991, is available from the SB-MOP package.
3323 * incompatible change: the SB-PCL package should now be considered
3324 a private implementation detail, and no longer a semi-private MOP
3326 * minor incompatible change: due to rearrangement for threads, the
3327 control stack and binding stack are now allocated at arbitrary
3328 addresses instead of being hardcoded per-port. Users affected by
3329 this probably have to be doing advanced things with shared
3330 libraries, and will know who they are.
3331 * minor incompatible change: Previously, all --eval forms used were
3332 processed with READ before any of them were processed with EVAL.
3333 Now each --eval form is processed with both READ and EVAL before
3334 the next --eval form is processed. (Thus package operations like
3335 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
3336 work as the user might reasonably expect.)
3337 * minor incompatible change: *STANDARD-INPUT* is now only an
3338 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
3340 * minor incompatible change: Y-OR-N-P is now character-oriented,
3341 not line oriented. Also, YES-OR-NO-P now works without errors.
3342 (thanks to Antonio Martinez)
3343 * sb-aclrepl module improvements: an integrated inspector, added
3344 repl features, and a bug fix to :trace command.
3345 * Known functions, which cannot be open coded by the backend, are
3346 considered to be able to check types of their arguments. (fixing
3347 a bug report by Nathan J. Froyd)
3348 * fixed a bug in computing method discriminating functions: It is
3349 now possible to define methods specialized on classes which have
3350 forward-referenced superclasses. (thanks to Gerd Moellmann)
3351 * fixed evaluation order in optional entries (reported by Gilbert
3353 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
3354 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
3355 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
3356 required-by-AMOP NAME argument, as well as CLASS and
3357 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
3358 * fixed bug 20: DEFMETHOD can define methods using names that are
3359 not the proper names of classes to designate class specializers.
3360 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
3361 Unix file descriptors, instead of blowing up. (thanks to Antonio
3363 * Garbage collection refactoring: user-visible change is that a
3364 call to the GC function during WITHOUT-GCING will not do garbage
3365 collection until the end of the WITHOUT-GCING. If you were doing
3366 this you were probably losing anyway.
3367 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
3368 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
3369 TYPEP the latter but not the former.
3370 * The compiler issues a full WARNING for calls to undefined functions
3371 with names from the CL package.
3372 * MAP-INTO for a vector destination is open coded. (reported by
3373 Brian Downing on c.l.l)
3374 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
3375 documentation string.
3376 * fixed some bugs revealed by Paul Dietz' test suite:
3377 ** COPY-ALIST now signals an error if its argument is a dotted
3379 ** Condition slots are now accessed more correctly in the presence
3380 of multiple initargs for a given slot.
3381 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
3382 exclude restarts of the same name associated with a different
3384 ** DEFCLASS of forward-referenced classes with another
3385 forward-referenced class in the superclasses list no longer
3387 ** Condition slots are now initialized once each, not multiple
3388 times. (thanks to Gerd Moellmann)
3389 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
3391 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
3392 ** NSET-EXCLUSIVE-OR does not return extra elements when its
3393 arguments contain duplicated elements.
3394 ** RESTART-CASE understands local macros.
3395 ** RESTART-CASE associates exactly its own restarts with a condition.
3396 ** ENDP in safe mode checks its argument to be of type LIST.
3397 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
3398 in question is unbound.
3399 ** Optimized MAKE-INSTANCE functions no longer cause internal
3400 assertion failures in the presence of duplicate initargs.
3401 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
3402 ** Side-effectful :DEFAULT-INITARGS have their side-effects
3403 propagated even in the ctor optimized implementation of
3405 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
3407 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
3408 * a better implementation of SXHASH on (simple) bit vectors,
3409 measured both in execution speed and in distribution of results
3410 over the positive fixnums, has been installed. Likewise, a better
3411 implementation of EQUAL for simple bit vectors is now available.
3412 * fixed CEILING optimization for a divisor of form 2^k.
3413 * fixed bug 240 (emitting extra style warnings "using the lexical
3414 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
3415 by Antonio Martinez)
3416 * fixed SXHASH, giving different results for NIL depending on type
3417 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
3418 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
3419 variable. (found by Rolf Wester)
3420 * SBCL does not ignore type declarations for special
3421 variables. (reported by rif on c.l.l 2003-03-05)
3422 * some bug fixes in contrib/sb-aclrepl/
3423 * fixed some bugs revealed by Paul Dietz' test suite:
3424 ** a bug in the CONS type specifier, whereby the CAR and CDR
3425 types got intertwined, has been fixed;
3426 ** the type system is now able to reason about the interaction
3427 between INTEGER and RATIO types more completely;
3428 ** APPEND, [N]REVERSE and NRECONC check that those their
3429 arguments, which must be proper lists, are really so;
3430 ** An array specialized to be unable to hold elements has been
3431 implemented, as required -- yes, really -- by ANSI;
3432 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
3433 on malformed property lists;
3435 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
3436 * incompatible packaging change: in line with Unix convention,
3437 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
3438 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
3439 * REQUIRE and PROVIDE are now optionally capable of doing something
3440 useful. See the documentation string for REQUIRE.
3441 * infrastructure for a managed SBCL contrib system: contributed
3442 modules in this release include:
3443 ** the ASDF system definition facility;
3444 ** an interface to the BSD Sockets API;
3445 ** an ACL-like convenience interface to the repl;
3446 (thanks to Kevin Rosenberg)
3447 ** an implementation of ROTATE-BYTE, with efficient implementation
3449 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
3450 gives a closer approximation to the right answer than previously.
3451 (thanks to Raymond Toy)
3452 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
3453 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
3454 optional environment arguments, as required by ANSI.
3455 * fixed bugs in other functions taking environment objects, allowing
3456 calls with an explicit NIL environment argument to be compiled
3458 * fixed bug 228: primary return values from
3459 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
3460 COMPILE or FUNCTION.
3461 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
3462 :TYPE will no longer signal errors on innocuous objects.
3463 * fixed bug 231b: SETQ is better at respecting type declarations in
3464 the lexical environment.
3465 * fixed a bug in DEFCLASS: classes named by symbols with no or
3466 unprintable packages can now be defined.
3467 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
3468 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
3469 * fixed compiler failure related to checking types of functions.
3470 (reported by Robert E. Brown)
3471 * the compiler is now much more consistent in its error-checking
3472 treatment of bounding index arguments to sequence functions: in
3473 (SAFETY 3) code, errors will be signalled in almost all cases if
3474 invalid sequence bounding indices are passed to functions defined
3475 by ANSI to operate on sequences.
3476 * fixed a bug in the build procedure: documentation of SBCL-specific
3477 packages is now preserved and available in the final Lisp image.
3478 * lifted FDEFINITION lookup out of loops in the implementation of
3479 many list operations. (thanks to Robert E. Brown)
3480 * fixed a bug in the reader: the #n# reader macro now works for
3481 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
3482 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
3483 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
3484 E. Brown and Tony Martinez respectively)
3485 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
3486 SLOT-BOUNDP: in particular, it is now able to optimize them much
3487 better, and is now not vulnerable to having packages renamed.
3488 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
3489 always called when a slot is not present in an instance. (thanks
3491 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
3492 longer signal an error on unreading a character following EOF on
3493 the previous constituent stream. (thanks to Tony Martinez)
3494 * fixed some bugs revealed by Paul Dietz' test suite:
3495 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
3496 not just nonnegative fixnums;
3497 ** the logical bit-array operators such as BIT-AND now accept an
3498 explicit NIL for their "opt-arg" argument (to indicate a
3499 freshly-consed result bit-array);
3500 ** ELT now signals an error on an invalid sequence index in safe
3502 ** the type system is now cleverer about negations of numeric
3503 types, and consequently understands the BIGNUM and RATIO types
3505 ** the type system is now cleverer about the interaction between
3506 INTEGER and RATIO types: while bugs still remain, many more
3507 cases are accurately computed;
3508 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
3509 if it is in the last clause;
3510 ** CONSTANTLY now correctly returns a side-effect-free function in
3512 ** DECLARE is no longer treated as a special-operator; in
3513 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
3514 * incremented fasl file version number due to the change in the
3515 DEFSTRUCT-SLOT-DESCRIPTION structure.
3517 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
3518 * minor incompatible change: code processed by the "interpreter" or
3519 EVAL now has a compilation optimization policy of (DEBUG 2)
3520 (changed from (DEBUG 1)) to improve debuggability of interactive
3521 development, and to allow the use of the debug RETURN command in
3523 * an experimental implementation of the RETURN command for the
3524 debugger has been included. (thanks to Frederik Kuivinen)
3525 * fixed bug 62: constraints were not propagated into a loop.
3526 * fixed bug in embedded calls of SORT (reported and investigated by
3528 * fixed some bugs revealed by Paul Dietz' test suite:
3529 ** printing and reading of arrays with some dimensions having
3530 length 0 (thanks to Gerd Moellmann);
3531 ** BOA constructor with &AUX argument without a default value does
3532 not cause a type error;
3533 ** CONSTANTP now returns true for all self-evaluating objects.
3535 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
3536 * fixed bug 127: DEFSTRUCT now does not clobber old structure
3537 accessors that are related by inheritance, as specified in the
3538 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
3539 Valtteri Vuorikoski)
3540 * The compiler is now able to inline functions that were defined in
3541 a complex lexical environment (e.g. inside a MACROLET).
3542 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
3544 * Support for the upcoming FreeBSD-5.0 release has been included.
3545 (thanks to Dag-Erling Smorgrav)
3546 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
3547 effect when it is not in a toplevel context.
3548 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
3549 stand a better chance of being correct. (thanks to Gerd
3551 * fixed bug in COERCE, which now signals an error on coercing a
3552 rational to a bounded real type which excludes the expected
3554 * The compiler is now able to derive types more accurately from the
3555 COERCE and COMPILE functions.
3556 * fixed bug 223: functional binding is considered to be constant
3557 only for symbols in the CL package.
3558 * fixed bug 231: SETQ did not check the type of a variable being set
3559 (reported by Robert E. Brown)
3560 * A new optimization for MAKE-INSTANCE has been included, fixing
3561 various bugs (including relating to :ALLOCATION :CLASS slots and
3562 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
3564 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
3565 ** As required by ANSI, LOOP now disallows anonymous collection
3566 clauses such as COLLECT I in conjunction with aggregate boolean
3567 clauses such as THEREIS (= I 1);
3568 ** LOOP now signals an error when any variable is reused in the
3569 same loop (including the potentially useful construct analogous
3570 to WITH A = 1 WITH A = (1+ A);
3571 ** IT is only a special loop symbol within the first clause of a
3572 conditional loop clause;
3573 ** LOOP with a typed iteration variable over a hashtable now
3574 signals a type error iff it should.
3575 * fixed some other bugs revealed by Paul Dietz' test suite:
3576 ** FILE-STREAM now names the class previously known as FD-STREAM;
3577 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
3578 argument) no longer signals an error;
3579 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
3580 of the slot symbol, rather than using the current package
3581 ((:CONC-NAME "") continues to intern the slot's name in the
3583 * incremented fasl file version number, because of the incompatible
3584 change to the DEFSTRUCT-DESCRIPTION structure, and again because
3585 of the new implementation of DEFINE-COMPILER-MACRO.
3587 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
3588 * Support for building SBCL for MIPS platforms running in
3589 little-endian mode has now been checked in, and basic
3590 functionality on said platforms verified.
3591 * minor incompatible change: PCL now records the pathname of a file
3592 in which methods and the like are defined, rather than its
3594 * minor incompatible change: TRUENAME now considers the truename of
3595 a file naming a directory to be the pathname with :DIRECTORY
3596 component indicating that directory.
3597 * minor incompatible change: a NAMED clause in the extended form of
3598 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
3599 reason for the previous behaviour is unclear.
3600 * more systematization and improvement of CLOS and MOP conformance
3601 in PCL (thanks to Gerd Moellman and Pierre Mai):
3602 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
3604 ** DEFINE-METHOD-COMBINATION no longer signals an error for
3605 primary methods with no specializers;
3606 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
3608 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
3609 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
3610 FORWARD-REFERENCED-CLASSes; error reporting on
3611 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
3613 ** SXHASH on CLOS instances now uses a slot internal to the
3614 instance to return different numbers on distinct instances,
3615 while preserving the same return value through invocations of
3617 ** DEFMETHOD signals errors when methods with longer incongruent
3618 lambda lists are added to generic functions;
3619 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
3620 CLASS, as specified in AMOP;
3621 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
3622 on the order returned by the primary method for classes of
3623 class STANDARD-CLASS;
3624 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
3625 * fixed some bugs shown by Paul Dietz' test suite:
3626 ** DOLIST puts its body in TAGBODY;
3627 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
3629 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
3630 value producing form;
3631 ** if more variables are given to PROGV than values, extra
3632 variables are bound and made to have no value;
3633 ** NSUBSTITUTE on list arguments gets the right answer with
3635 ** ELT signals an error of type TYPE-ERROR when the index argument
3636 is not a valid sequence index;
3637 ** LOOP signals (at macroexpansion time) an error of type
3638 PROGRAM-ERROR when duplicate variable names are found;
3639 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
3640 ** FUNCALL of special-operators now cause an error of type
3642 ** PSETQ now works as required in the presence of side-effecting
3643 symbol-macro places;
3644 ** NCONC accepts any object as its last argument;
3645 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
3647 ** loop-for-as-package does not require a package to be explicitely
3649 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
3651 * fixed bug 166: compiler preserves "there is a way to go"
3652 invariant when deleting code.
3653 * fixed bug 172: macro lambda lists with required arguments after
3654 &REST arguments now cause an error to be signalled. (thanks to
3656 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
3657 bug. (thanks to Matthew Danish)
3658 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
3660 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
3661 arguments are lexically rebound. (thanks to Gerd Moellmann and
3663 * fixed bug 194: error messages are now more informative when there
3664 is no primary method applicable in a call to a generic
3665 function. (thanks to Gerd Moellmann)
3666 * fixed bug in command line argument checking (thanks to Julian
3668 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
3669 (thanks to Matthew Danish)
3670 * incremented fasl file version number, because of the
3671 SXHASH-related changes in the layout of CLOS data structures
3673 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
3674 * minor incompatible change: The runtime (the Unix executable named
3675 "sbcl") is now much pickier about the .core files it will load.
3676 Essentially it now requires .core files to descend from the same
3677 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
3678 as the runtime does. (The intent is to prevent the crashes which
3679 can occur, and which can even be reported as mysterious failures,
3680 when people patch the sources or change the build parameters
3681 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
3682 sbcl and .core files.)
3683 * fixed bug: VALUES-LIST is no longer optimized away.
3684 * fixed bug 142: The FFI conversion of C string values to Lisp
3685 string values no longer conses excessively. (thanks to Nathan
3686 Froyd porting Raymond Toy's fix to CMU CL)
3687 * began to systematize and improve MOP conformance in PCL (thanks to
3688 Nathan Froyd, Gerd Moellman and Pierre Mai):
3689 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
3691 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
3692 ** FINALIZE-INHERITANCE is now called on class finalization;
3693 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
3694 argument precedence order.
3695 * fixed bug 202: The compiler no longer fails on functions whose
3696 derived types contradict their declared type.
3697 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
3698 so it can be non-toplevel.
3699 * The fasl file version number has changed (because of the new
3700 implementation of DEFMACRO).
3701 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
3702 safe code, that any length requirement by their type-specifier
3703 argument is valid. The exceptions to this are described in bug
3705 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
3706 functions now check the types of their inputs as required by ANSI.
3707 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
3708 introduced symbol is DECLAREd to be SPECIAL.
3709 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
3710 * fixed bug: :COUNT argument to sequence functions may be negative
3711 * fixed bug: body of DO-SYMBOLS may contain declarations
3712 * fixed bug: PUSHNEW now evaluates its arguments from left to right
3713 (reported by Paul F. Dietz, fixed by Gerd Moellman)
3714 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
3715 symbol macro only once
3716 * fixed printing of call frame when argument list is unavailable
3717 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
3718 * compiler no longer signals WARNING on unknown keyword
3721 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
3722 * A beta-quality port to the mips architecture running Linux,
3723 based on the old CMUCL backend, has been made. It has been tested
3724 on a big-endian kernel, and works sufficiently well to be able to
3725 rebuild itself; it has not been tested in little-endian mode.
3726 * fixed an inconsistency between gencgc.c and purify.c which made
3727 dumping/loading .core files unreliable
3728 * fixed bug 120a: The compiler now deals correctly with IFs where
3729 the consequent is the same as the alternative, instead of
3730 misderiving the return type. (thanks to Alexey Dejneka)
3731 * fixed bug 113: Logical pathnames are now dumpable (the logical
3732 host is resolved at load-time, throwing an error if it is not
3734 * fixed bug 174: FORMAT's error message is slightly clearer when a
3735 non-printing character is used in a format directive.
3736 * fixed several bugs in compiler checking of type declarations, i.e.
3737 violations of the Python "declarations are assertions" principle
3738 (thanks to Alexey Dejneka)
3739 * fixed several bugs in PCL's error checking (thanks to Gerd
3741 * fixed bug: printing of FILE-ERROR (thanks to Antonio
3743 * fixed bug in compilation of functions as first class values
3744 (thanks to Antonio Martinez-Shotton)
3745 * The compiler's handling TYPE-ERRORs which it can prove will
3746 inevitably happen at runtime has been cleaned up and corrected
3747 in several ways. (thanks to Alexey Dejneka)
3748 * improved argument type checking for various basic arithmetic
3749 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
3750 lavished on them in the past that they can be compiled in many
3751 ways in different special cases
3752 * fixed bug 181: compiler checks validity of user supplied type
3754 * cleaned up code flushing in optimization: Function calls which
3755 should signal errors for safety purposes (e.g. which ANSI says
3756 should signal errors when their arguments are of incorrect type)
3757 are no longer optimized away.
3758 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
3759 * incremented fasl file version number, because changes in the
3760 implementation of sequence functions like COERCE caused
3761 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
3762 in old inline expansions) to become undefined. (Actually these
3763 changes were later undone, so we might very well be binary
3764 compatible with 0.7.7 after all, but leaving the version number
3765 incremented seemed like the simplest and most conservative
3768 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
3769 * An alpha-quality port to the parisc architecture running Linux,
3770 based on the old CMUCL backend, has been made. This, even more so
3771 than the other backends, should be considered still a work in
3772 progress; known problems include that the Linux kernel in 64-bit
3773 mode does not propagate the correct sigcontext structure to
3774 userspace, and consequently SBCL on a parisc64 kernel will not
3776 * fixed bug 189: The compiler now respects NOTINLINE declarations for
3777 functions declared in FLET and LABELS. (I.e. "LET conversion" is
3778 suppressed.) Also now that the compiler is looking at declarations
3779 in the environment, it checks optimization declarations as well,
3780 and suppresses inlining when (> DEBUG SPEED).
3781 * More fixes have been made to treatment of floating point exception
3782 treatment and other Unix signals. In particular, floating point
3783 exceptions no longer cause Bus errors on the SPARC/Linux platform.
3784 * The detection and handling of control stack exhaustion (infinite
3785 or very deeply nested recursion) has changed. Stack exhaustion
3786 detection is now done by write-protecting pages at the OS level
3787 and applies at all optimization settings; when found, a
3788 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
3789 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
3790 be caught using IGNORE-ERRORS.
3791 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
3792 correctly when transforming recursive LABELS functions to LETs.
3793 (thanks to Alexey Dejneka)
3794 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
3795 that are names of constants or global variables.
3796 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
3797 alien routines with docstrings.
3798 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
3799 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
3801 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
3802 (thanks to Eric Marsden)
3803 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
3804 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
3805 * Bug fix: The SPARC backend can now compile functions involving
3806 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
3807 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
3808 to a functions expecting a PATHNAME-DESIGNATOR.
3809 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
3810 lambda lists. (thanks to Alexey Dejneka)
3811 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
3812 functions so that EXPORTing the name of the class doesn't cause
3813 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
3814 to Antonio Martinez for reporting this)
3815 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
3816 constant-folded without failing an assertion. (thanks to Einar
3817 Floystad Dorum for reporting this)
3818 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
3819 types are now dealt with more correctly by the compiler.
3820 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
3821 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
3822 behaviour analogous to RENAME-FILE. This puts its behaviour more
3823 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
3825 * The fasl file version number has changed again. (because of the
3826 bug fix involving the names of PCL MAKE-INSTANCE functions)
3828 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
3829 * bug fix: Floating point exceptions are treated much more
3830 consistently on the x86/Linux and PPC/Linux platforms.
3831 * Array initialization with :INITIAL-ELEMENT is now much faster for
3832 cases when the compiler cannot open code the array creation, but
3833 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
3834 array accesses have also seen a speed increase.
3835 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
3836 specified. (This was at the root of some bad interactions between
3837 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
3839 * bug fix: Internal error arguments for undefined functions are now
3840 computed correctly on the PPC/Linux platform.
3841 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
3842 Raymond Toy's patch for CMU CL)
3843 * Support for the Solaris 9 operating environment has been included
3844 (thanks to Daniel Merritt)
3845 * A very ugly but hopefully complete draft of the missing FFI chapter
3846 of the manual has been created by reformatting the corresponding
3847 CMU CL manual chapter into (currently very ugly and incoherent)
3848 DocBook and bringing it up to date for SBCL behavior. Thus, the
3849 manual is now essentially complete, at least by my extreme
3850 once-and-only-once standards, whereby it's acceptable to refer to
3851 the doc strings of SB-EXT functions as the primary documentation.
3852 * The fasl file version number has changed again, due to cleanup of
3853 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3854 is no longer a static symbol.)
3856 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3857 * SBCL now builds with OpenMCL (version 0.12) as the
3858 cross-compilation host; also, more progress has been made toward
3859 bootstrapping under CLISP.
3860 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3862 * bug 158 fixed: The compiler can now deal with integer loop
3863 increments different from 1; fixing this turned out also to fix
3865 * bug 169 fixed: no more bogus warnings about using lexical bindings
3866 despite the presence of perfectly good SPECIAL declarations (thanks
3867 to David Lichteblau)
3868 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3869 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3870 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3872 * bug 180 fixed: Method combination specifications no longer ignore
3873 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3874 * bug fix: Structure type predicate functions now check their argument
3875 count as they should.
3876 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3877 correctly. (thanks to Pierre Mai)
3878 * minor incompatible change: The --noprogrammer option is deprecated
3879 in favor of the new --disable-debugger option, which is very similar.
3880 (The major difference is that it takes effect at a slightly different
3881 time at startup, causing handling of errors in --sysinit and
3882 --userinit files will be affected differently.) The
3883 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3884 been added to allow this functionality to be controlled from ordinary
3885 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3886 maintainers, who might want to run non-interactive scripts to
3887 build SBCL cores which will later be used interactively.)
3888 * minor incompatible change: The LOAD function no longer, when given
3889 a wild pathname to load, loads all files matching that pathname.
3890 Instead, an error of type FILE-ERROR is signalled.
3892 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3893 * bug 147 fixed: The compiler preserves its block link/count
3894 invariants more correctly now so that it doesn't crash. (thanks
3896 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3898 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3899 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3900 no-value-for-this-character as NIL (as specified by ANSI).
3901 * HOST-NAMESTRING on physical pathnames now returns a string that is
3902 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3903 (thanks to Christophe Rhodes)
3904 * The Alpha port handles icache flushing more correctly. (thanks to
3906 * More progress has been made toward bootstrapping under CLISP. (thanks
3907 to Christophe Rhodes)
3908 * The fasl file format has changed again, because dynamic loading
3909 on OpenBSD (which has non-ELF object files) motivated some cleanups
3910 in the way that foreign symbols are transformed and passed around.
3911 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3912 is no longer treated as whitespace by the reader, but instead as
3913 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3914 |A
\7fB|, instead of A as it used to.
3916 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3917 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3918 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3919 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3920 this as of 0.7.1.45, but was left out of the previous news section
3921 (thanks to Dan Barlow)
3922 * SBCL now runs on the Solaris operating system on SPARC architectures
3923 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3924 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3925 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3926 * SPARC backend cleanups, allowing builds of cores optimized for V8
3927 and V9 SPARCS, and also emission of code targeted to a particular
3928 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3930 * SBCL is closer to bootstrapping under CLISP, thanks to various
3931 fixes by Christophe Rhodes.
3932 * The fasl file format has changed again, to allow the compiler's
3933 INFO database to support symbol macros.
3934 * The user manual (in doc/) is formatted into HTML more nicely.
3935 (thanks to coreythomas)
3936 * The system is smarter about SUBTYPEP relationships, especially
3937 those involving NOT types (including types such as ATOM which are
3938 represented internally using NOT types). Thus SUBTYPEP is less
3939 likely to return (VALUES NIL NIL) in general, and in particular
3940 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3942 * The fasl file format has changed again, because the internal
3943 representation of types now includes a new slot to support the new
3944 SUBTYPEP-of-NOT-types logic.
3945 * (not a change in the main branch of SBCL, but a related prototype
3946 which can hopefully be merged into the main branch of SBCL in the
3947 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3948 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3949 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3950 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3951 dispatch macro character combinations. (thanks to Alexey Dejneka)
3952 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3953 that superficially look like logical namestrings correctly.
3954 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3956 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3957 default supplied-p) for &optional and &key arguments. (thanks to
3960 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3961 * incompatible change: The compiler is now less aggressive about
3962 tail call optimization, doing it only when (> SPACE DEBUG) or
3963 (> SPEED DEBUG). (This is an incompatible change because there are
3964 programs which relied on the old CMU-CL-style behavior to optimize
3965 away their unbounded recursion which will now die of stack overflow.)
3966 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3967 for non-GENCGC systems has been increased to 20M (since that
3968 seems much closer to the likely performance optimum for modern
3969 systems than the old 4M value was)
3970 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3971 and *DEBUG-PRINT-LEVEL*
3972 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3973 of CMU CL's support for SPARC, and various endianness and other
3974 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3975 * new syntactic sugar for the Unix command line: --load foo.bar is now
3976 an alternate notation for --eval '(load "foo.bar")'.
3978 ** The system now detects stack overflow and handles it gracefully,
3979 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3980 optimization settings. (This is a good thing in general, and
3981 its introduction in this version should be particularly timely
3982 for anyone whose code fails because of suppression of tail
3984 ** The system now hunts for the C variable "environ" in a more
3985 devious way, to avoid segfaults when the C library version differs
3986 between compile time and run time. (thanks to Christophe Rhodes)
3987 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3988 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3989 ** The compiler no longer issues bogus style warnings for undefined
3990 classes in the same source file as the DEFCLASSes which defined
3991 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3993 ** fixes in CONDITION class precedence list for undefined function
3994 errors (thanks to Alexei Dejneka)
3995 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3996 correctly. (thanks to Dan Barlow)
3997 ** portability fixes aiming at bootstrapping under CLISP (thanks
3998 to Dave McDonald and Christophe Rhodes)
3999 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
4000 ** fixes in type translation and and type inference (thanks to
4002 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
4003 ** various fixes in the new ports (thanks to Dan Barlow)
4004 * several changes related to debugging:
4005 ** suppression of tail recursion, as noted above
4006 ** stack overflow detection, as noted above
4007 ** The default implementation of TRACE has changed. :ENCAPSULATE T
4008 is now the default. (For some time encapsulation has been more
4009 reliable than the breakpoint-based :ENCAPSULATE NIL
4010 implementation, at least on X86 systems; and I just noticed that
4011 encapsulation also seems closer to the spirit of the ANSI
4014 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
4016 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
4017 up properly again. (There was a packaging bug in 0.7.0 which
4018 left their definitions in SB-SYS::LOAD-FOREIGN and
4019 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
4020 vital for most things which interface to C-level interfaces,
4021 like extensions working with sockets or databases or
4022 Perl-compatible regexes or whatever, and the need to fix
4023 this bug is the main reason that 0.7.1 was released so
4025 ** DEFGENERIC is now choosier about the methods it redefines, so that
4026 reLOADing a previously-LOADed file containing DEFGENERICs does
4027 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
4028 cycle now works as it should. (thanks to Alexey Dejneka)
4029 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
4030 fixed by Christophe Rhodes. (He actually submitted this patch
4031 months ago, and I delayed until after 0.7.0.)
4032 ** Bug 111 (internal compiler confusion about runtime checks on
4033 FUNCTION types) was fixed by Alexey Dejneka.
4034 * Some internal cleanups (getting rid of variables which aren't
4035 needed now that the byte interpreter is gone) caused the fasl
4036 file format number to change again.
4038 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
4039 * major incompatible change: The default fasl file extension, i.e. the
4040 default extension for files produced by COMPILE-FILE, has changed
4041 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
4043 ** There are many changes in the implementation of the compiler.
4044 SBCL is now essentially a compiler-only implementation of ANSI
4045 Common Lisp. EVAL still "interprets" a few special cases, but
4046 almost all the interesting cases are handled by creating
4047 a LAMBDA expression, calling COMPILE on it, then calling
4048 FUNCALL on the result.
4049 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
4050 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
4051 Since the code is newer, there might still be some new bugs
4052 (though not as many as before Martin Atzmueller's fixes:-). But
4053 the new code is substantially simpler and clearer, and hopefully
4054 any remaining bugs will be simpler, less fundamental, and more
4055 fixable then the bugs in the old code.
4056 ** The revised compiler is still a little unsteady on its feet.
4058 *** The debugging information it produces (particularly the names
4059 of FUNCTION objects) is sometimes much less useful than what
4060 the old compiler produced.
4061 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
4062 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
4063 within a MACROLET) has been temporarily weakened.
4064 ** There are new compiler optimizations for various functions:
4065 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
4066 FIND-IF-NOT, POSITION-IF-NOT, and FILL
4067 *** the math functions TRUNCATE, FLOOR, and CEILING
4068 *** the function-of-all-trades COERCE
4069 Mostly these should be transparent, but there's one
4070 potentially-annoying problem (bug 117): when the compiler
4071 inline-expands a function and does type analysis on the result,
4072 it can create control paths which have type mismatches, and
4073 when it can't prove that those control paths aren't taken,
4074 it will issue WARNINGs about the type mismatches. This is
4075 a particular problem in practice for the new sequence functions.
4076 It's not clear how this should be fixed, and for now, a
4077 workaround is given in the entry for 117 in the BUGS file.
4078 ** (Because of the interaction between the two previous items --
4079 occasional inlining problems and new inline expansions -- some
4080 of the new sequence function optimizations won't really kick in
4081 completely until debugging information, and then inlining, are
4082 straightened out in some future version.)
4083 * minor incompatible changes:
4084 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
4085 DIRECTORY no longer implicitly promotes NIL slots of its
4086 pathname argument to :WILD. In particular, when you ask for the
4087 contents of a directory (which you used to be able to do without
4088 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
4089 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
4090 ** changes in behavior that ANSI explicitly defines to be
4091 implementation dependent:
4092 *** The new compiler-only implementation still conforms with ANSI,
4093 but acts a little different than before. Besides the obvious
4094 changes in performance tradeoffs (that the cost per form passed
4095 to EVAL has gone up, and the cost per form executed by EVAL
4096 has gone down), the behavior of the system changes a little
4097 because there are no longer any interpreted function objects.
4098 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
4099 e.g. doing COMPILE on the output of interactive DEFUN is
4101 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
4103 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
4104 has changed from (:CL) to NIL.
4105 *** The CHAR-NAME of unprintable ASCII characters which, unlike
4106 e.g. #\Newline and #\Tab, don't have names specified in the
4107 ANSI Common Lisp standard, is now based on their ASCII symbolic
4108 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
4109 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
4110 are no longer used for output.
4111 ** changes in internal implementation constants:
4112 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
4113 4 million. (If your application spends a lot of time GCing and
4114 you have a lot of RAM, you might want to experiment with
4115 increasing it even more.)
4116 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
4117 However, almost all old code should still continue to work without
4118 immediate update, as SB-C-CALL is now a (deprecated) nickname
4120 ** Old operator names in the style DEF-FOO are now deprecated in
4121 favor of new corresponding names DEFINE-FOO, for consistency with
4122 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
4123 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
4124 internal symbols, but a few supported extensions like
4125 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
4126 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
4127 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
4128 etc. as you get deeper into recursive calls to the debugger
4129 command loop, instead of the old "5]", "5]]", "5]]]"
4130 sequence. (I was motivated to do this when squabbles between
4131 ILISP and SBCL left me very deeply nested in the debugger. In the
4132 short term, this change will probably provoke more ILISP/SBCL
4133 squabbles, but hopefully it will be an improvement in the long run.)
4134 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
4135 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
4136 function, since declaiming return types involving aliens is
4137 (1) annoyingly messy to do by hand and (2) vital to efficient
4138 compilation of code which calls such functions.
4139 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
4140 longer reexported by the SB-EXT package. They're solely useful
4141 for alien code, so it seems more logical that you should get
4142 them from the SB-ALIEN package, not in SB-EXT.
4143 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
4144 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
4145 features. Instead, the code that they used to control is always
4146 built into the system.
4147 * many other bug fixes
4148 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
4149 advantage of the new EVAL-WHEN stuff and to clean them up in
4150 general, and they are now more ANSI-compliant in a number of
4151 ways. Martin Atzmueller is responsible for a lot of this.
4152 ** Besides the cleanups discussed above, Martin Atzmueller fixed
4154 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
4155 *** correct ERROR type for various file operations
4156 *** some fixes for Lisp streams
4157 *** DEFMETHOD syntax checking
4158 *** changing old weird representation of debug information as
4159 strings (which, among their other deficiencies, don't transform
4160 correctly when you rename packages, and don't change their
4161 print representation when you change things like *PACKAGE*
4162 and *PRINT-LENGTH*) to symbols and lists of symbols
4163 He also made several improvements and fixed several bugs in DESCRIBE.
4164 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
4166 *** misbehavior of WRITE-STRING/WRITE-LINE
4167 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
4168 and several other LOOP problems as well
4169 *** DIRECTORY when similar filenames are present
4170 *** DEFGENERIC with :METHOD options
4171 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
4172 *** bug in the optimization of ARRAY-ELEMENT-TYPE
4173 *** argument ordering in FIND with :TEST option
4174 *** mishandled package designator argument in APROPOS-LIST
4175 *** various problems in the backquote readmacro
4176 *** a bug in APROPOS
4177 *** probably some others that I'm not describing very well here,
4178 since the CVS log documents them by reference to sbcl-devel
4179 messages, and the SourceForge archives aren't working well.:-(
4180 ** Dan Barlow improved the Alpha port (and is making progress on the
4181 PPC port, for those of you who think different).
4182 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
4183 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
4184 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
4185 continues to work on the SPARC port (for those of you in a position
4186 to look down upon our little PC-compatible boxes from a great height).
4187 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
4188 and uses that copy, rather than the current dynamic value, when
4189 it's trying to decide whether to truncate output. Thus e.g.
4190 (let ((*print-lines* 50))
4191 (pprint-logical-block (stream nil)
4193 (let ((*print-lines* 8))
4194 (print (aref possiblybigthings i) stream)))))
4195 should now truncate the logical block only at 50 lines, instead of
4196 often truncating it at 8 lines, as it did before.
4197 * The doc/cmucl/ directory, containing old CMU CL documentation
4198 from the time of the fork, is no longer part of the base system.
4199 SourceForge has shut down its anonymous FTP service, and with it
4200 my original plan for distributing the old CMU CL documentation
4201 there. For now, if you need these files you can download an old
4202 SBCL source release and extract them from it.
4203 * The fasl file version number changed again, for dozens of reasons,
4204 some of which are apparent above.
4206 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
4207 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
4208 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
4209 patch, so that DISASSEMBLE output is much nicer.
4210 * The code in the SB-PROFILE package now seems reasonably stable.
4211 I still haven't decided what the final interface should look like
4212 (I'd like PROFILE to interact cleanly with TRACE, since both
4213 facilities use function encapsulation) but if you have a need
4214 for profiling now, you can probably use it successfully with
4215 the current CMU-CL-style interface.
4216 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
4217 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
4218 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
4219 link now returns the dangling link itself, and for similar
4220 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
4221 link itself. (In these cases the old code signalled an error and
4222 looped endlessly, respectively.) Thus, DIRECTORY now works even
4223 in the presence of dangling and cyclic symbolic links.
4224 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
4225 is now a supported extension again, since the consensus on
4226 sbcl-devel was that it can be useful for ordinary development
4227 work, not just for debugging SBCL itself.
4228 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
4229 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
4230 that currently-defined functions might be redefined later with
4231 different return types.
4232 * Hash tables can be printed readably, as inspired by CMU CL code
4233 of Eric Marsden and SBCL code of Martin Atzmueller.
4234 * better error handling in CLOS method combination, thanks to
4235 Martin Atzmueller porting Pierre Mai's CMU CL patches
4236 * more overflow fixes for >16Mbyte I/O buffers
4237 * A bug in READ has been fixed, so that now a single Ctrl-D
4238 character suffices to cause end-of-file on character streams.
4239 In particular, now you only need one Ctrl-D at the command
4240 line (not two) to exit SBCL.
4241 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
4243 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
4244 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
4245 does the right thing.
4246 * The implementation of some type tests, especially for CONDITION
4247 types, is now tidier and maybe faster, due to CMU CL code
4248 originally by Douglas Crosher, ported by Martin Atzmueller.
4249 * Some math functions have been fixed, and there are new
4250 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
4251 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
4252 * (There are also some new optimizers in contrib/*-extras.lisp. Those
4253 aren't built into sbcl-0.6.13, but are a sneak preview of what's
4254 likely to be built into sbcl-0.7.0.)
4255 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
4256 CL, ported to SBCL by Martin Atzmueller)
4257 * DESCRIBE now gives more information in some cases. (Pierre Mai's
4258 patch to CMU CL, ported to SBCL by Martin Atzmueller)
4259 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
4260 * There's a new slam.sh hack to shorten the edit/compile/debug
4261 cycle for low-level changes to SBCL itself, and a new
4262 :SB-AFTER-XC-CORE target feature to control the generation of
4263 the after-xc.core file needed by slam.sh.
4264 * minor incompatible change: The ENTRY-POINTS &KEY argument to
4265 COMPILE-FILE is no longer supported, so that now every function
4266 gets an entry point, so that block compilation looks a little
4267 more like the plain vanilla ANSI section 3.2.2.3 scheme.
4268 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
4269 returns the number of bytes consed since the system started,
4270 rather than the number consed since the first time the function
4271 was called. (The new definition parallels ANSI functions like
4272 CL:GET-INTERNAL-RUN-TIME.)
4273 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
4274 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
4275 supported. Now DIRECTORY always does the abstract Common-Lisp-y
4276 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
4277 * Fasl file version numbers are now independent of the target CPU,
4278 since historically most system changes which required version
4279 number changes have affected all CPUs equally. Similarly,
4280 the byte fasl file version is now equal to the ordinary
4283 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
4284 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
4285 is no longer recognized. I apologize for this, because it was
4286 listed in SB-EXT as a supported extension, but I found that
4287 its existing behavior was poorly specified, as well as incorrectly
4288 specified, and it looked like too much of a mess to straighten it
4289 out. I have enough on my hands trying to get ANSI stuff to work..
4290 * many patches ported from CMU CL by Martin Atzmueller, with
4291 half a dozen bug fixes in pretty-printing and the debugger, and
4292 half a dozen others elsewhere
4293 * fixed bug 13: Floating point infinities are now supported again.
4294 They might still be a little bit flaky, but thanks to bug reports
4295 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
4296 as flaky as they were.
4297 * The --noprogrammer command line option is now supported. (Its
4298 behavior is slightly different in detail from what the old man
4299 page claimed it would do, but it's still appropriate under the
4300 same circumstances that the man page talks about.)
4301 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
4302 are now supported, and enabled by default. Thus, the compiler can
4303 handle many floating point and complex operations much less
4304 inefficiently. (Thus e.g. you can implement a complex FFT
4306 * The compiler now detects type mismatches between DECLAIM FTYPE
4307 and DEFUN better, and implements CHECK-TYPE more correctly, and
4308 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
4309 thanks to patches from Martin Atzmueller.
4310 * various fixes to make the cross-compiler more portable to
4311 ANSI-conforming-but-different cross-compilation hosts (notably
4312 Lispworks for Windows, following bug reports from Arthur Lemmens)
4313 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
4314 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
4315 fixed thanks to Pierre Mai's CMU CL patches.
4316 * fixes to keep the system from overflowing internal counters when
4317 it tries to use i/o buffers larger than 16M bytes
4318 * fixed bug 45a: Various internal functions required to support
4319 complex special functions have been merged from CMU CL sources.
4320 (When I was first setting up SBCL, I misunderstood a compile-time
4321 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
4322 * improved support for type intersection and union, fixing bug 12
4323 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
4324 more obscure bugs as well
4325 * some steps toward byte-compiling non-performance-critical
4326 parts of the system, courtesy of patches from Martin Atzmueller
4327 * Christophe Rhodes has made some debian packages of sbcl at
4328 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
4329 From his sbcl-devel e-mail of 2001-04-08 they're not completely
4330 stable, but are nonetheless usable. When he's ready, I'd be happy
4331 to add them to the SourceForge "File Releases" section. (And if
4332 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
4333 * new fasl file format version number (because of changes in
4334 internal representation of (OR ..) types to accommodate the new
4335 support for (AND ..) types, among other things)
4337 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
4338 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
4340 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
4342 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
4343 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
4344 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
4345 * DESCRIBE now works on structure objects again.
4346 * Most function call argument type mismatches are now handled as
4347 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
4348 know whether the function will be redefined before the call is
4349 executed. (The compiler could flag local calls with full WARNINGs,
4350 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
4351 it doesn't keep track of enough information to know whether calls
4352 are local in this sense.)
4353 * Compiler output is now more verbose, with messages truncated
4354 later than before. (There should be some supported way for users
4355 to override the default verbosity, but I haven't decided how to
4356 provide it yet, so this behavior is still controlled by the internal
4357 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
4358 src/compiler/ir1util.lisp.)
4359 * Fasl file format version numbers have increased again, because
4360 support for the Gray streams extension changes the layout of the
4361 system's STREAM objects.
4362 * The Gray subclassable streams extension now works, thanks to a
4363 patch from Martin Atzmueller.
4364 * The full LOAD-FOREIGN extension (not just the primitive
4365 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
4366 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
4367 but like most other programs, it defaults to copying the Unix
4368 environment from the original process instead of starting the
4369 new process in an empty environment.
4370 * Extensions which manipulate the Unix environment now support
4371 an :ENVIRONMENT keyword option which doesn't smash case or
4372 do other bad things. The CMU-CL-style :ENV option is retained
4373 for porting convenience.
4374 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
4375 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
4377 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
4379 * A patch from Martin Atzmueller seems to have solved the SIGINT
4380 problem, and as far as we know, signal-handling now works cleanly.
4381 (If you find any new bugs, please report them!)
4382 * The system no longer defaults Lisp source file names to types
4383 ".l", ".cl", or ".lsp", but only to ".lisp".
4384 * The compiler no longer uses special default file extensions for
4385 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
4386 seems to expect a single default extension for all compiled code,
4387 and there's no compelling reason to try to stretch the standard
4388 to allow two different extensions.) Instead, byte-compiled files
4389 default to the same extension as native-compiled files.
4390 * Fasl file format version numbers have increased again, because
4391 a rearrangement of internal implementation packages made some
4392 dumped symbols in old fasl files unreadable in new cores.
4393 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
4394 many fewer weird special cases.
4395 * Bug #17 (differing COMPILE-FILE behavior between logical and
4396 physical pathnames) has been fixed, and some related misbehavior too,
4397 thanks to a patch from Martin Atzmueller.
4398 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
4399 by Tim Moore, ported to SBCL by Martin Atzmueller.
4400 * Martin Atzmueller fixed several filesystem-related problems,
4401 including bug #36, in part by porting CMU CL patches, which were
4402 written in part by Paul Werkowski.
4403 * More compiler warnings in src/runtime/ are gone, thanks to
4404 more patches from Martin Atzmueller.
4405 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
4408 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
4410 * DESCRIBE now works on CONDITION objects.
4411 * The debugger now handles errors which arise when trying to print
4412 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
4414 * The build system now uses an additional file, customize-target-features.lisp,
4415 to allow local modifications to the target *FEATURES* list. (The point of
4416 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
4417 debugging features enabled, without having to worry about propagating it
4418 into everyone's system when I do a "cvs update".) When no
4419 customize-target-features.lisp file exists, the target *FEATURES* list
4420 should be constructed the same way as before.
4421 * fixed bugs in DEFCONSTANT ANSI-compatibility:
4422 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
4423 warn about behavior which is undefined under the ANSI spec. Note: This
4424 is specified by ANSI, but it's not very popular with programmers.
4425 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
4426 macro in the SBCL sources for an example of a workaround which you
4427 might use to make portable ANSI-standard code which does what you want.
4428 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
4429 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
4430 when it's used as a non-toplevel form. (This is required in order
4431 to implement the DEFCONSTANT-EQX macro.)
4432 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
4433 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
4434 funny things, due to bugs in the implementation of EVAL-WHEN
4435 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
4436 non-toplevel DEFCONSTANTs are uncommon.)
4437 * The core file version number and fasl file version number have been
4438 incremented, because the old noncompliant DEFCONSTANT behavior involved
4439 calling functions which no longer exist, and because I also took the
4440 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
4441 * fixed bug 1 (error handling before read-eval-print loop starts), and
4442 redid debugger restarts and related debugger commands somewhat while
4444 ** The QUIT debugger command is gone, since it did something
4445 rather different than the SB-EXT:QUIT command, and since it never
4446 worked properly outside the main toplevel read/eval/print loop.
4447 Invoking the new TOPLEVEL restart provides the same functionality.
4448 ** The GO debugger command is also gone, since you can just invoke
4449 the CONTINUE restart directly instead.
4450 ** The TOP debugger command is also gone, since it's redundant with the
4451 FRAME 0 command, and since it interfered with abbreviations for the
4453 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
4455 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
4456 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
4457 some porting work by Martin Atzmueller)
4458 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
4459 with a boatload of patches which he ported to SBCL. Now that those
4461 ** The system tries to make sure that its low-priority messages
4462 are prefixed by semicolons, to help people who like to use
4463 syntax highlighting in their ILISP buffer. (This patch
4464 was originally due to Raymond Toy.)
4465 ** The system now optimizes INTEGER-LENGTH better, thanks to more
4466 patches originally written by Raymond Toy.
4467 ** The compiler understands coercion between single-value and
4468 multiple-VALUES type expressions better, getting rid of some very
4469 weird behavior, thanks to patches originally by Robert MacLachlan
4470 and Douglas Crosher.
4471 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
4472 lambda lists, thanks to a patch originally by Pierre Mai.
4473 ** The system no longer bogusly warns about "abbreviated type
4475 ** The compiler gets less confused by inlining and RETURN-FROM,
4476 thanks to some patches originally by Tim Moore.
4477 ** The system no longer hangs when dumping circular lists to fasl
4478 files, thanks to a patch originally from Douglas Crosher.
4479 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
4480 undefined function error.
4481 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
4482 in structure slot initforms) for now, documented workaround instead:-|
4483 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
4484 * fixed bug 5: added stubs for various Gray stream functions called
4485 in the not-a-CL:STREAM case, so that even when Gray streams aren't
4486 installed, at least appropriate type errors are generated
4487 * fixed bug 8: better reporting of various PROGRAM-ERRORs
4488 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
4489 consistently in DEFMETHOD forms.
4490 * removed bug 21 from BUGS, since Martin Atzmueller points out that
4491 it doesn't seem to affect SBCL after all
4492 * The C runtime system now builds with better optimization and many
4493 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
4495 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
4497 * The system is now under CVS at SourceForge (instead of the
4498 CVS repository on my home machine).
4499 * The new signal handling code has been tweaked to treat register
4500 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
4501 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
4502 downstream, e.g. in debug-int.lisp, has implicit dependencies
4503 on the unsignedness of integer representation of machine words,
4504 and that caused the system to bomb out with infinite regress
4505 when trying to recover from type errors involving signed values,
4506 e.g. (BUTLAST '(1 2 3) -1).)
4507 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
4508 (This was one of the bugs Peter Van Eynde reported back in July.)
4509 * The system now uses code inspired by Colin Walters' O(N)
4510 implementation of MAP (from the cmucl-imp@cons.org mailing
4511 list, 2 September 2000) when it can't use a DEFTRANSFORM to
4512 inline the MAP operation, and there is more than one
4513 sequence argument to the MAP call (so that it can't just
4514 do ETYPECASE once and for all based on the type of the
4515 single sequence argument). (The old non-inline implementation
4516 of the general M-argument sequence-of-length-N case required
4517 O(M*N*N) time when any of the sequence arguments were LISTs.)
4518 * The QUIT :UNIX-CODE keyword argument has been renamed to
4519 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
4520 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
4522 * Raymond Wiker's patches to port dynamic loading from Linux to
4523 FreeBSD have been added.
4524 * The BUGS file is now more nearly up to date, thanks in large part
4525 to Martin Atzmueller's review of it.
4526 * The debugger now flushes standard output streams before it begins
4527 its output ("debugger invoked" and so forth).
4528 * The core version number and fasl file version number have both
4529 been incremented, because of incompatible changes in the layout
4531 * FINISH-OUTPUT is now called more consistently on QUIT. (It
4532 used to not be called for a saved Lisp image.)
4533 * Martin Atzmueller's version of a patch to fix a compiler crash,
4534 as posted on sbcl-devel 13 September 2000, has been installed.
4535 * Instead of installing Martin Atzmueller's patch for the
4536 compiler transform for SUBSEQ, I deleted the compiler transform,
4537 and transforms for some similar consing operations.
4538 * A bug in signal handling which kept TRACE from working on OpenBSD
4540 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
4541 away by constant folding
4542 * The system now defines its address space constants in one place
4543 (in the Lisp sources), and propagates them automatically elsewhere
4544 (through GENESIS and the sbcl.h file). Therefore, patching the
4545 address map is less unnecessarily tedious and error-prone. The
4546 Lisp names of address space constants have also been systematized.
4547 * CVS tags like dollar-Header-dollar have been removed from
4548 the sources, because they have never saved me trouble and
4549 they've been source of trouble working with patches and other
4550 diff-related operations.
4551 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
4552 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
4554 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
4556 * The system has been ported to OpenBSD.
4557 * The system now compiles with a simple "sh make.sh" on the systems
4558 that it's supported on. I.e., now you no longer need to tweak
4559 text in the target-features.lisp-expr and symlinks in src/runtime/
4560 by hand, the make.sh takes care of it for you.
4561 * The system is no longer so grossly inefficient when compiling code
4562 involving vectors implemented as general (not simple) vectors (VECTOR T),
4563 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
4564 which dares to use the various sequence functions on non-simple
4565 vectors, takes less of a performance hit.
4566 * There is now a primitive type predicate VECTOR-T-P
4567 to test for the (VECTOR T) type, so that e.g.
4568 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
4569 can now be compiled with some semblance of efficiency. (The old code
4570 turned the type declaration into a full call to %TYPEP at runtime!)
4571 * AREF on (VECTOR T) is still not fast, since it's still compiled
4572 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
4573 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
4574 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
4575 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
4576 without making full calls to SUBTYPEP at runtime.
4577 (Some analogous efficiency issues for non-simple vectors specialized to
4578 element types other than T, or for non-simple multidimensional arrays,
4579 have not been addressed. They could almost certainly be handled the
4580 same way if anyone is motivated to do so.)
4581 * The changes in array handling break binary compatibility, so
4582 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
4583 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
4584 instead of (VALUES T T).
4585 * By following the instructions that Dan Barlow posted to sbcl-devel
4586 on 2 July 2000, I was able to enable primitive dynamic object
4587 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
4588 functionality is not implemented (since it calls ld to resolve
4589 library references automatically, requiring RUN-PROGRAM for its
4590 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
4591 try to resolve library references) is now supported.
4592 * The system now flushes the standard output streams when it terminates,
4593 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
4594 them at several other probably-convenient times, e.g. in each pass of
4595 the toplevel read-eval-print loop, and after evaluating a form given
4596 as an "--eval" command-line option. (These changes were motivated by a
4597 discussion of stream flushing issues on cmucl-imp in August 2000.)
4598 * The source transform for TYPEP of array types no longer assumes
4599 that an array whose element type is a not-yet-defined type
4600 is implemented as an array of T, but instead punts, so that the
4601 type will be interpreted at runtime.
4602 * There is now some support for cross-compiling in make.sh: each of
4603 the phases of make.sh has its own script. (This should be transparent
4604 to people doing ordinary, non-cross-compile builds.)
4605 * Since my laptop doesn't have hundreds of megabytes of memory like
4606 my desktop machine, I became more motivated to do some items on
4607 my to-do list in order to reduce the size of the system a little:
4608 ** Arrange for various needed-only-at-cold-init things to be
4609 uninterned after cold init. To support this, those things have
4610 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
4611 symbols with such names are now uninterned after cold init).
4612 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
4613 fasl files for cold load.
4614 ** Remove the old compiler structure pooling code (which used to
4615 be conditional on the target feature :SB-ALLOC) completely.
4616 ** Redo the representation of some data in cold init to be more compact.
4617 (I also looked into supporting byte compiled code at bootstrap time,
4618 which would probably reduce the size of the system a lot, but that
4619 looked too complicated, so I punted for now.)
4620 * The maximum signal nesting depth in the src/runtime/ support code has
4621 been reduced from 4096 to 256. (I don't know any reason for the very
4622 large old value. If the new smaller value turns out to break something,
4623 I'll probably just bump it back up.)
4624 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
4626 * Many, many bugs reported by Peter Van Eynde have been added to
4627 the BUGS list; some have even been fixed.
4628 * While enabling dynamic object file loading, I tried to make the
4629 code easier to understand, renaming various functions and variables
4630 with less ambiguous names, and changing some function calling
4631 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
4632 * While trying to figure out how to do the OpenBSD port, I tried to
4633 clean up some of the code in src/runtime/. In particular, I dropped
4634 support for non-POSIX signal handling, added various comments,
4635 tweaked the code to reduce the number of compilation warnings, and
4636 renamed some files to increase consistency.
4637 * To support the new automatic configuration functionality in make.sh,
4638 the source file target-features.lisp-expr has been replaced with the
4639 source file base-target-features.lisp-expr and the machine-generated
4640 file local-target-features.lisp-expr.
4641 * fixed a stupid quoting error in make.sh so that using CMU CL
4642 "lisp -batch" as cross-compilation host works again
4644 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
4646 * DESCRIBE no longer tries to call itself recursively to describe
4647 bound/fbound values, so that it no longer fails on symbols which are
4648 bound to themselves (like keywords, T, and NIL).
4649 * DESCRIBE now works on generic functions.
4650 * The printer now prints less-screwed-up representations of closures
4651 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
4652 * A private symbol is used instead of the :EMPTY keyword previously
4653 used to mark empty slots in hash tables. Thus
4654 (DEFVAR *HT* (MAKE-HASH-TABLE))
4655 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
4656 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
4657 now does what ANSI says that it should. (You can still get
4658 similar noncompliant behavior if bang on the hash table
4659 implementation with all the symbols you get back from
4660 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
4661 This breaks binary compatibility, since tests for equality to
4662 :EMPTY are wired into things like the macroexpansion of
4663 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
4665 * There's now a minimal placeholder implementation for CL:STEP,
4666 as required by ANSI.
4667 * An obscure bug in the interaction of the normal compiler, the byte
4668 compiler, inlining, and structure predicates has been patched
4669 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
4670 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
4672 * Missing ordinary arguments in a macro call are now detected even
4673 when the macro lambda list contains &KEY or &REST.
4674 * The debugger no longer complains about encountering the top of the
4675 stack when you type "FRAME 0" to explicitly instruct it to go to
4676 the top of the stack. And it now prints the frame you request even
4677 if it's the current frame (instead of saying "You are here.").
4678 * As specified by ANSI, the system now always prints keywords
4679 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
4681 * The default initial SIZE of HASH-TABLEs is now smaller.
4682 * Type information from CLOS class dispatch is now propagated
4683 into DEFMETHOD bodies, so that e.g.
4684 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4686 is now basically equivalent to
4687 (DEFMETHOD FOO ((X SINGLE-FLOAT))
4688 (DECLARE (TYPE SINGLE-FLOAT X))
4690 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
4691 operation, without having to do run-time type dispatch.
4692 * The macroexpansion of DEFMETHOD has been tweaked so that it has
4693 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
4694 * Since I don't seem to be making big file reorganizations very often
4695 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
4696 is overflowing my ability to conveniently back them up), I've finally
4697 checked the system into CVS. (The CVS repository is on my home system,
4698 not at SourceForge -- putting it on SourceForge might come later.)
4699 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
4700 high-level GC-NOTIFY-FOO functions send their output. (There's
4701 still very little control of where low-level verbose GC functions
4702 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
4703 less than it used to -- the GC-NOTIFY-FOO functions are now under
4704 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
4705 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
4706 in only one place in the source code, and propagates it automatically
4707 everywhere that it's needed. Thus e.g. when I bump the version from
4708 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
4709 * The C source files now include boilerplate legalese and documentation
4710 at the head of each file (just as the Lisp source files already did).
4711 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
4712 to his page will be replaced with a link to his new CLiki service.
4714 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
4716 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
4717 * The build process now looks for GNU make under the default name "gmake",
4718 instead of "make" as it used to. If GNU make is not available as "gmake"
4719 on your system, you can change this default behavior by setting the
4720 GNUMAKE environment variable.
4721 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
4722 can build without error under CMU CL.
4724 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
4726 * There is now a partial SBCL user manual (with some new text and some
4727 text cribbed from the CMU CL manual).
4728 * The beginnings of a profiler have been added (starting with the
4729 CMU CL profiler and simplifying and cleaning up). Eventually the
4730 main interface should be through the TRACE macro, but for now,
4731 it's still accessed through vaguely CMU-CL-style functions and macros
4732 exported from the package SB-PROFILE.
4733 * Some problems left over from porting CMU CL to the new
4734 cross-compilation bootstrap process have been cleaned up:
4735 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
4736 instead of SB!XC:DEFMACRO, compounded by an oversight on my
4737 part when getting rid of the compiler *BACKEND* stuff.)
4738 ** The value of *NULL-TYPE* was screwed up, because it was
4739 being initialized before the type system knew the final
4740 definition of the 'NULL type. This screwed up several key
4741 optimizations in the compiler, causing inefficiency in all sorts
4742 of places. (I found it because I wanted to understand why
4743 GET-INTERNAL-RUN-TIME was consing.)
4744 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
4745 PROCLAIM FTYPE information. Unfortunately this broke binary
4746 compatibility again, since now the forms output by DEFGENERIC
4747 to refer to functions which didn't exist in 0.6.3.
4748 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
4749 can use the new (as of 0.6.3) transform for SOME into MAP into
4751 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
4752 values to (INTEGER 0 1000000), so that the time code will no longer
4753 occasionally get blown up by Linux returning 1000000 microseconds
4754 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
4755 its output conform to the ANSI spec. (Alas, this makes its output
4756 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
4757 * A full call to MAP NIL with a single sequence argument no longer conses.
4758 * fixes to problems pointed out by Martin Atzmueller:
4759 * The manual page no longer talks about multiprocessing as though
4760 it were currently supported.
4761 * The ILISP support patches have been removed from the distribution,
4762 because as of version 5.10.1, ILISP now supports SBCL without us
4763 having to maintain patches.
4764 * added a modified version of Raymond Toy's recent CMU CL patch for
4765 EQUALP comparison of HASH-TABLE
4767 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
4769 * The system still can't cross-compile itself with
4770 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
4771 entails), but at least it can compile more of itself that way
4772 than it used to be able to, and various buglets which were uncovered
4773 by trying to cross-compile itself that way have now been fixed.
4774 * This release breaks binary compatibility again. This time
4775 at least I've incremented the FASL file format version to 2, so that the
4776 problem can be detected reliably instead of just causing weird errors.
4777 * various new style warnings:
4778 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
4779 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
4780 ** using the lexical binding of a variable named in the *FOO* style
4781 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
4782 as specified by ANSI.
4783 * *RANDOM-STATE* is no longer automatically initialized from
4784 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
4785 default behavior of the system is to repeat its behavior every time
4786 it's run. If you'd like to change this behavior, you can always
4787 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
4788 old convention there was no comparably easy way to get the system to
4789 repeat its behavior every time it was run.
4790 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
4791 FTYPE declarations has been removed, in favor of their ANSI
4792 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
4793 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
4794 the types of their sequence arguments can be determined at compile time.
4795 This is done through a new open code expansion for MAP which eliminates
4796 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
4797 argument types can be determined at compile time.
4798 * The optimizer now transforms COERCE into an identity operation when it
4799 can prove that the coerced object is already of the correct type. (This
4800 can be a win for machine generated code, including the output of other
4801 optimization transforms, such as the MAP transform above.)
4802 * Credit information has been moved from source file headers into CREDITS.
4803 * Source file headers have been made more standard.
4804 * The CASE macro now compiles without complaining even when it has
4807 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
4809 * (Note that the way that the PCL macroexpansions were rewritten
4810 to accommodate the change in DEFGENERIC below breaks binary
4811 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
4812 not run under sbcl-0.6.2. Once we get out of alpha releases,
4813 i.e. hit release 1.0.0, we'll probably try to maintain binary
4814 compatibility between maintenance releases, e.g. between sbcl-1.4.3
4815 and sbcl-1.4.4. Until then, however, it might be fairly common
4816 for maintenance releases to break binary compatibility.)
4817 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
4819 * The reporting of unhandled signals has been changed to print some
4820 explanatory text as well as the report form. (Previously only
4821 the report form was printed.)
4822 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
4823 it defines, so that the compiler no longer issues undefined function
4824 warnings for compiled-but-not-yet-loaded generic functions.
4825 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
4826 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
4827 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
4828 section of the ANSI spec are supported.
4829 * The "" nickname for the "KEYWORD" package has been removed.
4830 The reader still handles symbol tokens which begin with a package marker
4831 as keywords, but it doesn't expose its mechanism for doing so in the
4832 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
4833 * The system now issues STYLE-WARNINGs for contradictory TYPE
4834 proclamations. (Warnings for contradictory FTYPE proclamations would
4835 be nice too, but those can't be done usefully unless the type system
4836 is made smarter about FUNCTION types.)
4837 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
4838 systematized, so that "*target-*.lisp is supposed to exist only on the
4839 target and imply that there's a related file which exists on the
4840 host, and *host-*.lisp is supposed to exist only on the host and imply
4841 that there's a related file which exists on the target. This involves a
4842 lot of renaming. Hopefully the acute confusion caused by the renaming
4843 will be justified by the reduction in chronic confusion..
4844 ** runtime-type.lisp -> early-target-type.lisp
4845 ** target-type.lisp -> late-target-type.lisp
4846 ** early-host-format.lisp -> early-format.lisp
4847 ** late-host-format.lisp -> late-format.lisp
4848 ** host-error.lisp -> misc-error.lisp
4849 ** early-error.lisp -> early-target-error.lisp
4850 ** late-error.lisp -> late-target-error.lisp
4851 ** host-defboot.lisp -> early-defboot.lisp
4852 ** code/misc.lisp -> code/target-misc.lisp
4853 ** code/host-misc.lisp -> code/misc.lisp
4854 ** code/numbers.lisp -> code/target-numbers.lisp
4855 ** code/early-numbers.lisp -> numbers.lisp
4856 ** early-host-type.lisp -> early-type.lisp
4857 ** late-host-type.lisp -> late-type.lisp
4858 ** host-typep.lisp -> typep.lisp
4859 ** load.lisp -> target-load.lisp
4860 ** host-load.lisp -> load.lisp
4861 ** host-disassem.lisp -> disassem.lisp
4862 ** host-insts.lisp -> insts.lisp
4863 ** byte-comp.lisp -> target-byte-comp.lisp
4864 ** host-byte-comp.lisp -> byte-comp.lisp
4865 ** host-signal.lisp -> signal.lisp
4866 ** host-defstruct.lisp -> defstruct.lisp
4867 ** late-target-type.lisp -> deftypes-for-target.lisp
4868 Furthermore, several other previously target-only files foo.lisp (e.g.
4869 hash-table.lisp and random.lisp) have been split into a target-and-host
4870 foo.lisp file and a target-only target-foo.lisp file, with their key type
4871 definitions in the target-and-host part, so that the cross-compiler will
4872 know more about target types.
4873 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4874 gone away. In their place are various *BACKEND-FOO* variables
4875 corresponding to the slots of the old structure.
4876 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4877 into the target SBCL has been fixed.
4878 * The chill.lisp system for loading cold code into a running SBCL
4880 * Support for the CMU CL "scavenger hook" extension has been removed.
4881 (It was undocumented and unused in the CMU CL sources that SBCL was
4882 derived from, and stale in sbcl-0.6.1.)
4883 * Various errors in the cross-compiler type system were detected
4884 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4885 (enabling various consistency checks). Many of them were fixed,
4886 but some hard problems remain, so the compiler is back to
4887 running without *TYPE-SYSTEM-INITIALIZED* for now.
4888 * As part of the cross-compiler type system cleanup, I implemented
4889 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4890 * I have started adding UNCROSS calls throughout the type system
4891 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4892 on cross-compiler input files will be able to go away, and various
4894 * CONSTANTP now returns true for quoted forms (as explicitly required
4897 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4899 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4900 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4901 (MAKE-LIST -1). In the longer term, it ought to become true
4902 that declarations are assertions even at SAFETY 1. For now, it's not
4903 quite true even at SAFETY 3, but it's at least more nearly true..
4904 (Note that this change seems to increases the size of the system by
4905 O(5%) and to decrease the speed of the compiler by 20% or more.)
4906 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4907 for the problem of printing dozens of lines of distracting information
4908 about low-level system machinery as part of the top stack frame
4909 on entry to the debugger when an undefined function was called.
4910 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4911 is not reset to COMMON-LISP-USER.
4912 * Compilation of stuff related to dyncount.lisp has been made conditional
4913 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4914 smaller. The various dyncount-related symbols have been moved into
4915 a new "SB-DYNCOUNT" package.
4916 * tty-inspect.lisp has been renamed to inspect.lisp.
4917 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4918 feature has gone away. (When we eventually port to other flavors of
4919 libc and/or Unix, we'll try to make the differences between flavors
4920 invisible at the user level.)
4921 * Various other *FEATURES* tags, and/or their associated conditionals,
4922 have been removed if obsolescent, or given better documentation, or
4923 sometimes given more-mnemonic names.
4925 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4927 * tidied up "make.sh" script
4928 * tidied up system directory structure
4929 * better "clean.sh" behavior
4930 * added doc/FOR-CMUCL-DEVELOPERS
4931 * many many small tweaks to output format, e.g. removing possibly-confusing
4932 trailing #\. character in DESCRIBE-INSTANCE
4933 * (EQUALP #\A 'A) no longer signals an error.
4934 * new hashing code, including EQUALP hashing
4935 * tidied up Lisp initialization and toplevel
4936 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4937 * command line argument processing
4938 * added POSIX-GETENV function to deal with Unix-ish environment variables
4939 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4940 terminating SBCL on EOF
4941 * non-verbose GC by default
4942 * There is no more "sbcl" shell script; the sbcl file is now the C
4943 runtime executable (just like CMU CL).
4944 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4946 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4947 symbol and package information as Lisp native symbol and package objects
4948 instead of strings naming symbols and strings naming packages. This way,
4949 whenever packages are renamed (as in warm init), debug information is
4950 transformed along with everything else.
4951 * tweaked the optimization policy declarations which control the building
4952 of SBCL itself. Now, among other things, the system no longer saves
4953 source location debugging information. (This helps two problems at once
4954 by reducing SBCL size and by keeping SBCL from trying to look for its
4955 sources -- which may not exist -- when reporting errors.)
4956 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4957 debugging and testing purposes
4958 * cleaned up printing, making the printer call PRINT-OBJECT for
4959 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4960 methods, giving nearly-ANSI behavior
4961 * converted almost all special variables to use *FOO* naming convention
4962 * deleted PARSE-TIME functionality, since it can be done portably
4963 * moved some files out of cold init into warm init
4964 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4966 * regularized formatting of source files
4967 * added an install.sh script
4968 * fixed ridiculous memory usage of cross-compiler by making
4969 compiler/alloc.lisp not try to do pooling unless it can hook
4970 itself into the GC of the cross-compilation host. Now the system
4971 builds nicely on my old laptop.
4972 * added :SB-ALLOC in target-features.lisp-expr
4973 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4974 was not implemented)
4975 * re-did condition handling and note reporting in the compiler. Notes
4976 are no longer handled by signalling conditions. Style warnings
4977 and warnings are handled more correctly and reported in such a way
4978 that it's easy to find one or the other in your output (so that you
4979 can e.g. figure out which of many problems caused COMPILE-FILE to
4981 * changed the severity of several compiler warnings from full WARNING
4982 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4983 compiler note reporting so that it doesn't use the condition system
4984 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4985 COMPILE-FILE command)
4986 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4987 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4988 no longer supported, since their implementation was deeply intertwingled
4989 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4990 * removed bogus "support" for compiler macros named (SETF FOO), and
4991 removed the compiler macro for SETF INFO (but only after making a fool
4992 of myself on the cmucl-imp mailing list by posting a bogus patch for
4993 DEFINE-COMPILER-MACRO..)
4994 * Compiled files containing forms which have side effects on the Lisp
4995 reader (such as DEFPACKAGE forms) are now handled more correctly.
4996 (Compiler queuing of top level lambdas has been suppressed by setting
4997 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4998 * deleted various currently-unused source files, e.g. gengc.lisp. They
4999 may be added back at some point e.g. when porting to other architectures,
5000 but until they are it's distracting to distribute them and to try to
5002 * deleted "UNCROSS couldn't recurse through.." style warnings, since
5003 there were so many of them they're just distractions, and UNCROSS is
5004 known to be able to handle the current sources
5005 * moved PROFILE functionality into TRACE, so that it will be clear
5006 how the wrapping and unwrapping of functions when you profile them
5007 interacts with the wrapping and unwrapping of functions when you
5008 trace them. (Actually, the functionality isn't there yet, but at least
5009 the interface specification is there. Hopefully, the functionality will
5010 arrive with some future maintenance release.)
5011 * removed host-oops.lisp
5012 * changed signature of QUIT function to allow UNIX-CODE argument
5013 * fixed READ-SEQUENCE bug
5014 * tweaked verbose GC output so that it looks more like the progress
5015 output that ANSI specifies for functions like LOAD
5016 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
5017 * added <http://sbcl.sourceforge.com> to the banner information printed by