1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
3 * incompatible change: the thread-safe (on most platforms) getaddrinfo
4 and getnameinfo sockets functions are used instead of gethostbyaddr
5 and gethostbyname, on platforms where the newer functions are available.
6 As a result, the ALIASES field of HOST-ENT will always be NIL on these
8 * optimization: code using alien values with undeclared types is much faster
9 * bug fix: >= and <= gave wrong results when used with NaNs.
10 * bug fix: the #= and ## reader macros now interact reasonably with
11 funcallable instances.
12 * bug fix: type-checks for function arguments were compiled using the
13 compiler policy settings of the wrong lexical environment (in 1.0.2
15 * bug fix: SHADOW accepts characters as string designators, as required
16 by the spec (thanks to Eric Marsden)
17 * bug fix: fixed GC safety issues when foreign functions are called with
18 non-base strings as arguments
20 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
21 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
22 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
23 produce more efficient code on the x86 if the compiler can determine
24 that (+ INDEX <constant>) does not require a bounds check and FOO
25 has an element type at least 8 bits wide.
26 * bug fix: references to undefined variables are handled the same way
27 in toplevel forms as in the normal compiler
28 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
30 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
31 (thanks to Jon Buller)
32 * improvement: faster compilation times for complex functions
33 * improvement: added readlink support to SB-POSIX (thanks to Richard
36 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
37 * improvement: experimental support for mach exception handling on
38 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
40 * improvement: support for GBK external format.
41 (thanks to Chun Tian (binghe))
42 * improvement: the debugger now displays variables that have been closed
43 over, in code compiled with (DEBUG 2) or higher
44 * improvement: support for executable cores on NetBSD (thanks to
46 * new feature: added a RESTART-FRAME debugger command
47 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
48 be used to provide an implementation for FILE-POSITION on Gray streams
49 (thanks to Eric Marsden)
50 * new feature: add syslog, openlog and closelog support to SB-POSIX
51 (thanks to Richard Kreuter)
52 * optimization: the function call overhead in code compiled with
53 a high DEBUG optimization setting is significantly
54 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
55 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
56 (thanks to Tony Martinez)
57 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
58 evaluated once (reported by Kevin Reid)
59 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
60 works again (thanks to Kilian Sprotte)
61 * bug fix: an error is signaled for tagbodies with duplicate tags
62 (thanks to Stephen Wilson)
63 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
64 * bug fix: Win32 port can now handle foreign code unwinding Lisp
65 stack frames from alien callbacks.
66 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
67 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
68 65k pages (thanks to David Woodhouse)
69 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
71 changes in sbcl-1.0.1 relative to sbcl-1.0:
72 * new platform: FreeBSD/x86-64, including support for threading.
73 * new feature: the compiler stores cross-referencing information
74 abount function calls (who-calls), macroexpansion (who-macroexpands)
75 and special variables (who-binds, who-sets, who-references) for code
76 compiled with (< SPACE 3). This information is available through the
77 sb-introspect contrib.
78 * new feature: users may subclass SEQUENCE, and have instances of
79 these classes interoperate with standard Common Lisp functions if
80 a number of methods are defined. (This feature is experimental
81 and the interface subject to change based on feedback from SBCL
82 users and the general community)
83 * improvement: sb-sprof traces call stacks to an arbitrary depth on
84 x86/x86-64, rather than the previous fixed depth of 8
85 * improvement: another pthread back-end of mutex "pthread-futex".
86 * bug fix: non-ascii command-line arguments are processed correctly
87 (thanks to Yaroslav Kavenchuk)
88 * bug fix: non-required arguments were not passed correctly when a method
89 defined using DEFMETHOD was called from a mop-generated method using
90 CALL-NEXT-METHOD (reported by Pascal Costanza)
91 * bug fix: recursion is now permitted in accessors through
92 SLOT-UNBOUND. (reported by Pascal Costanza)
93 * bug fix: an error was signaled at startup if the HOME environment
94 variable was defined, but had an empty value (reported by Peter Van Eynde)
95 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
96 signal a PROGRAM-ERROR, not a BUG.
97 * bug fix: SB-POSIX:READDIR works when built with large file support.
98 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
100 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
101 for better automatic indentation support. (thanks to Matt Pillsbury)
102 * optimization: loading generic functions no longer takes O(n^2) time,
103 proportional to the amount of methods in the generic function
104 (reported by Todd Sabin and Jeremy Brown)
105 * optimization: the FIND and POSITION family of sequence functions
106 are significantly faster on arrays whose element types have been
108 * improvements to the Windows port:
109 ** Intermittent heap corruption problems have been fixed. (thanks
110 to Alastair Bridgewater)
111 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
113 ** Lisp is able to unwind foreign exception frames from alien
114 callbacks. (thanks to Alastair Bridgewater)
116 changes in sbcl-1.0 relative to sbcl-0.9.18:
117 * improvement: experimental support for threading on FreeBSD/x86.
118 (thanks to NIIMI Satoshi)
119 * improvement: runtime option --dynamic-space-size can be used
120 to set the size of the dynamic space reserved on startup.
121 * improvement: floating point modes in effect are now saved in
122 core, and restored on startup.
123 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
124 startup, not time since first call to GET-INTERNAL-REAL-TIME.
125 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
126 threads are not running after *SAVE-HOOKS* have run.
127 * improvement: writes to CLOS instance slots are type-checked in code
128 compiled with (SAFETY 3)
129 * improvement: floating-point exception handling on FreeBSD (thanks to
131 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
132 (thanks to Zach Beane)
133 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
135 * improvement: added support for the Shift-JIS external format.
136 (contributed by NIIMI Satoshi)
137 * improvement: callbacks are supported on Linux/PPC. (thanks to
139 * bug fix: compiler bug triggered by a (non-standard) VALUES
140 declaration in a LET* was fixed. (reported by Kaersten Poeck)
141 * bug fix: file compiler no longer confuses validated and already
142 dumped structurres. (reported by Kaersten Poeck)
143 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
144 fill-pointer signals a type-error as required. (thanks to
146 * bug fix: disassemly of funcallable instances works.
147 * bug fix: single stepping on PPC.
148 * bug fix: fix thread-safety problems in the type system (generally
149 manifesting as nonsensical errors like "STRING is a bad type specifier
150 for sequences" or "The value 1 is not of type FIXNUM")
151 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
152 (reported by Josip Gracin)
153 * bug fix: an error is signaled for attempts to displace arrays with
154 incompatible element types (thanks to Mario Mommer)
155 * bug fix: more correct handling of wide characters in debug info
156 (bug reported by Attila Lendvai and fixed by Juho Snellman)
157 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
158 and don't cause extra consing
159 * optimization: MAP and MAP-INTO are significantly faster on vectors
160 whose elements types have been declared.
161 * Improvements to SB-SPROF:
162 ** Support for allocation profiling
163 ** Reduced profiling overhead, especially for long profiling runs
164 * Improvements to the Windows port:
165 ** floating point exceptions are now reported correctly.
166 ** stack exhaustion detection works partially.
167 ** more accurate GET-INTERNAL-REAL-TIME.
168 ** better RUN-PROGRAM behaviour in case of errors in the child
170 ** PROBE-FILE now simplifies pathnames correctly.
171 ** DIRECTORY now works correctly with :WILD-INFERIORS.
172 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
173 1st 1970 (although time timezone data will be unreliable).
174 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
176 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
177 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
178 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
180 * bug fix: two potential GC deadlocks affecting threaded builds.
181 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
182 returns T (reported by Anton Kazennikov)
183 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
184 with non-variable places
185 * bug fix: remove a race condition in the setting of
186 funcallable-instance functions, this should make threaded CLOS
187 code more stable against memory faults.
188 * bug fix: corruption of specials when unbinding is interrupted by an
189 asynchronous unwind (reported by Hannu Koivisto)
190 * improvement: the debugger will now also display local variables that
191 are only used once, for code compiled with a DEBUG optimization quality
194 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
195 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
196 * incompatible change: External-format support for FFI calls. The
197 SB-ALIEN:C-STRING no longer implies an ASCII
198 external-format. Instead the string is subject to external-format
199 conversion. Additionally return values of type SB-ALIEN:C-STRING
200 are by default of type (SIMPLE-ARRAY CHARACTER), not
201 SIMPLE-BASE-STRING. For an alien type that behaves like the old
202 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
203 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
204 * incompatible change: SB-EXT package no longer contains the
205 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
206 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
208 * incompatible change: the single-stepper is no longer available
209 on Alpha, Mips and Sparc platforms.
210 * minor incompatible change: the direct superclasses of
211 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
212 not (STANDARD-OBJECT FUNCTION). This makes the
213 class-precedence-lists of GENERIC-FUNCTION and
214 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
216 * minor incompatible change: the default stream external format on
217 non-unicode SBCL is no longer detected from the locale, but is always
219 * new feature: Add a version of evaluator that uses an interpreter instead
220 of the compiler. EVAL still uses the compiler by default, to switch it
221 to use the interpreter, set the value of the variable
222 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
223 * minor incompatible change: the single-stepper REPL has been merged
224 with the normal debugger (see the "Stepping" heading of the debugger help
225 for more details). The debugger command STEP will no longer switch
226 to the single-stepper REPL.
227 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
228 for a type now works.
229 * bug fix: loading of default sysinit file works. (thanks to Leonid
231 * bug fix: better detection of circularities in the file-compiler.
232 (reported by Marco Monteiro)
233 * bug fix: the CL pathname functions now work with files that have
234 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
235 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
236 systems with Unix98 pty semantics.
237 * bug fix: ASDF-INSTALL will now work with bsd tar.
238 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
240 * bug fix: timers expiring in dead threads no longer cause a
241 type-error (reported by Paul "Nonny Mouse").
242 * bug fix: thanks to more lightweight single-stepper instrumentation,
243 code compiled with (DEBUG 3) will compile and execute significantly faster,
244 and will have more accurate type-inferencing than before
245 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
246 whose bindings are modified
247 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
248 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
249 ** replace dummy implementations of CL:MACHINE-INSTANCE and
250 CL:SOFTWARE-VERSION with proper version
252 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
253 * feature: implemented the READER-METHOD-CLASS and
254 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
255 as specified by AMOP.
256 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
258 * optimization: faster LOGCOUNT implementation on x86 and x86-64
259 (thanks to Lutz Euler)
260 * optimization: hashing of general arrays and vectors has been
261 improved. (reported by Any Fingerhut)
262 * enhancement: SB-INTROSPECT is now able to find definitions of
263 profiled functions. (thanks to Troels Henriksen)
264 * enhancement: compiler-macro expansion applies now to FUNCALL forms
265 as well, allowing compiler-macros for SETF-functions to expand.
266 * enhancement: step-instrumentation no longer wraps known
267 single-value functions with multiple-value context, allowing
268 better type inference.
269 * fixed bug #337: use of MAKE-METHOD in method combination now works
270 even in the presence of user-defined method classes. (reported by
271 Bruno Haible and Pascal Costanza)
272 * fixed bug #339(c): if there are applicable methods not part of any
273 long-form method-combination group, call INVALID-METHOD-ERROR.
274 (reported by Bruno Haible)
275 * fixed bug #361: the :FUNCTION initarg in the protocol for
276 initialization of methods can now be used to override
277 internally-produced optimized functions. (reported by Bruno
279 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
280 system-provided lambda expression no longer cause warnings about
281 unbound #:|pv-table| symbols.
282 * bug fix: improved the handling of type declarations and the
283 detection of violations for keyword arguments with non-constant
285 * bug fix: potentially erronous calls to PATHNAME and
286 MERGE-PATHNAMES were being flushed in some cases.
287 (reported by Richard Kreuter)
288 * bug fix: compiled calls to TYPEP were mishandling obsolete
289 instances. (reported by James Bielman and Attila Lendvai)
290 * bug fix: format strings with ~> without matching ~< no longer
291 trigger an AVER, but signal an understandable error instead.
292 (reported by Antonio Martinez)
293 * bug fix: specifying an output-file without a pathname-type for
294 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
295 of input-file instead of "fasl". (reported by Robert Dodier)
296 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
297 (reported by James Y Knight).
298 * bug fix: compiler-macros-function did not consider the environment
299 argument for shadowing by local functions.
300 * bug fix: compiler-macros expansion was inhibited by local INLINE
302 * bug fix: inline expansions of known functions were subject to
303 step-instrumentation in high debug policies, leading to problems
305 * bug fix: compiler failed to differentiate between different CONS
307 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
308 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
309 * bug fix: #368: incorrect use of expressed vs. upgraded array
311 * bug fix: #306a: more precise unions of array types.
312 * thread-safety improvements:
313 ** CONDITION-WAIT could return early on Linux, if the thread was
314 interrupted and subsequently continued with SIGCONT.
315 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
317 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
318 * added support for the ucs-2 external format. (contributed by Ivan
320 * minor incompatible change: pretty printing of objects of type
321 (cons symbol) is, in the default pprint-dispatch-table, now
322 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
324 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
325 called later in a class's lifetime, possibly as late as when the
326 first instance of the class is created. Previously,
327 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
328 class became finalizeable.
329 * fixed bug: FILE-POSITION sometimes returned inconsistent results
330 for multibyte external-format streams. (thanks to "vbzoli")
331 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
332 with :ALLOCATION :CLASS inherited from superclasses of the
334 * fixed bug: anonymous classes can now be created using the :NAME
335 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
336 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
337 * fixed bug: core-files saved with :EXECUTABLE T can again be
338 executed when SBCL_HOME isn't set. (reported by James Knight)
339 * fixed bug: toplevel LOCALLY forms with declarations could
340 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
341 * fixed bug: printing from several different threads using different
342 values of *print-case* could cause invalid output, due to
343 some internal special variables of the printer not being bound
344 thread-locally (reported by Max Mikhanosha)
345 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
346 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
347 (reported by Pascal Costanza)
348 * fixed bug: FINALIZE-INHERITANCE is called from
349 REINITIALIZE-INSTANCE on classes when the class has previously
350 been finalized, as required by AMOP.
351 * minor code generation optimizations:
352 ** better register allocation in CLOS dispatching functions
353 ** overflow detection when coercing signed bytes to fixnums on x86-64
354 ** is now implemented with one IMUL instruction instead of three shifts
355 ** more efficient bit-vector access on x86 and x86-64
356 ** more efficient access to raw structure slots on x86 and x86-64
357 * fixed some bugs revealed by Paul Dietz' test suite:
358 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
361 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
362 * feature: thread support on Solaris/x86, and experimental thread support
364 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
366 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
367 default method for SLOT-UNBOUND.
368 * minor incompatible change: prevent the user from specializing the
369 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
370 somewhat counter to the intent of the protocol, I (CSR) think, and
371 additionally it just doesn't work in SBCL as currently
372 implemented, thanks to optimizations (that are always valid for
373 the other three SLOT-VALUEish functions, but not for the setter).
374 * minor incompatible change: the :SB-LDB feature is now enabled by
375 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
376 the low-level debugger.
377 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
378 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
379 down to an SB-SYS:FD-STREAM, allowing direct communication in
380 more cases. Temporary files and pipes are still used as a fallback
382 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
383 on the non-reentrant C functions localtime(3) and gmtime(3).
384 * bug fix: global optimization policy was not visible in LOCALLY and
386 * bug fix: class objects can be used as specializers in methods.
387 (reported by Pascal Costanza)
388 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
389 with a type error. (reported by blitz_ on #lisp)
390 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
391 failed with a type error. (reported by Pascal Bourguignon)
392 * bug fix: merging pathnames against defaults with :DIRECTORY
393 starting with '(:RELATIVE :BACK) should preserve the :BACK.
394 (reported by James Y Knight)
395 * bug fix: saving large (>2GB) cores on x86-64 now works
396 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
397 constant index and a (simple-array (signed-byte 32)) array
398 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
399 error (patch by Robert J. Macomber)
400 * feature: TIME now displays the amount of run-time spent in GC
401 * bug fix: The debugger now does a better job of respecting
402 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
403 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
404 workaround for bug 403.)
405 * bug fix: floating point exception handling now works on FreeBSD
406 (thanks to NIIMI Satoshi)
407 * fixed some bugs revealed by Paul Dietz' test suite:
408 ** MISC.641: LET-conversion were not supposed to work in late
410 * improvements to the Win32/x86 port:
411 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
412 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
414 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
415 * new feature: source path information is generated for macro-expansion
416 errors for use in IDE's like Slime (thanks to Helmut Eller)
417 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
418 * bug fix: compiler does not loop forever on an invalid type in
420 * improvement: compilation of most CLOS applications is significantly
422 * optimization: added a limited bytecode compiler for simple toplevel
423 forms, speeding up compilation and FASL loading
424 * bug fix: the statistical profiler now properly distinguishes anonymous
427 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
428 * minor incompatible change: in sbcl-0.9.11 (but not earlier
429 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
430 into subclasses of the finalized class. Now user calls to
431 FINALIZE-INHERITANCE finalize just the one class, and calls by the
432 system return before any subclasses are finalized.
433 * minor incompatible change: The reader no longer ignores errors
434 regarding non-existent packages in #+ and #- feature tests.
435 * new feature: command line options --no-sysinit, --no-userinit to
436 inhibit loading the corresponding init files
437 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
438 for sending data through UDP sockets (thanks to François-René Rideau)
439 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
440 error is signalled instead (thanks to François-René Rideau)
441 * minor incompatible change: Error signalling behaviour of lexical
442 operations violating package locks has changed slightly. Refer to
443 documentation on package locks for details.
444 * enhancement: EVAL can process IF-expressions without resorting to the
446 * bug fix: Some operations involving SIN, COS, or TAN weren't being
447 constant-folded properly due to missing out-of-line functions.
448 (reported by Mika Pihlajamäki)
449 * bug fix: LISTEN sometimes returned T even in cases where no data was
450 immediately available from the stream
451 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
452 were reversed. (reported by Levente Mészáros)
453 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
454 when given a vector displaced to an adjustable array. (reported
456 * bug fix: derivation of float boundaries from numbers outside the
457 appropriate float range (reported by John Wiseman)
458 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
460 * bug fix: precomputing cache entries for generic functions with
461 some subclasses of specializers as yet invalid does not attempt to
462 fill a cache line with a negative offset. (reported by Levente
464 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
465 allow condition class objects as arguments to DOCUMENTATION and
466 (SETF DOCUMENTATION); only find and set documentation for
467 structure names for the STRUCTURE doc-type. (suggested by Gary
469 * improvements to the Win32/x86 port:
470 ** (user-homedir-pathname) and default initialization file
471 locations now know about the user's "Documents and Settings"
472 directory (thanks to Yaroslav Kavenchuk)
473 ** run-program is implemented (thanks to Mike Thomas)
474 ** sockets support (thanks to Timothy Ritchey)
475 ** better backtrace support (thanks to Alastair Bridgewater)
476 ** sb-grovel supported
477 ** asdf-install and sb-posix work somewhat
478 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
479 * improvements to the Solaris/x86 port:
480 ** works on Solaris 11/Solaris Express
481 ** floating-point exception handling support
482 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
483 * fixed some bugs revealed by Paul Dietz' test suite:
484 ** REMOVE-METHOD returns its generic function argument even when
485 the method is not one of the generic functions' methods.
486 ** objects declared as MEMBER types can be admissible arguments to
488 ** printing characters should simply be printed by the FORMAT ~:C
490 ** compiler failure when compiling functions with hairy constant
491 defaults for optional parameters.
492 ** compiler produces wrong code when MAYBE-INLINE-expanding a
493 function, which is already optimized.
495 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
496 * new platform: experimental support for SBCL x86/Darwin, including
497 MacOS X 10.4.5 on Intel.
498 * new feature: Unicode character names are now known to the system
499 (through CHAR-NAME and NAME-CHAR).
500 * new feature: the filesystem location of SBCL's core file is
501 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
502 * minor incompatible change: the contrib modules SB-POSIX and
503 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
504 this change is to make it easier to distribute
505 location-independent binaries.
506 * bug fix: as implied by AMOP, standardized classes no longer have
507 slots named by external symbols of public packages. (reported by
509 * optimization: calling functions via constant symbols -- as in
510 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
511 via the function object as in (FUNCALL #'FOO).
512 * optimization: on x86 and x86-64, the calling convention for the
513 case of dealing with an unknown number of values has been altered
514 to be friendlier to the prediction heuristics implemented,
515 particularly with reference to CALL and RET pairing. (thanks to
516 Alastair Bridgewater)
517 * optimization: on x86, the code for access to thread-local symbol
518 values and binding/unbinding of thread-local symbols is smaller.
519 (thanks to Alastair Bridgewater)
520 * enhancement: CONSTANTP is now able to determine constantness of
521 more complex forms, including calls to constant-foldable standardized
522 functions and some special forms beyond QUOTE.
523 * fixed bug: occasional GC crashes on Solaris/x86
524 * optimization: x86-64 supports stack allocation of results of simple
525 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
526 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
527 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
528 (thanks to James Knight)
529 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
530 by the standard (thanks to Stephen Compall)
532 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
533 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
534 be used for bundling the runtime and the core file into one
535 executable binary. This feature is not currently supported on all SBCL
536 platforms. (thanks to James Bielman and NIIMI Satoshi)
537 * new feature: a generational or ephemeral garbage collector is now
538 the default on the PowerPC platform (both Linux and Darwin). The
539 old Cheney (stop and copy) collector is a build-time option.
540 (thanks to Cyrus Harmon, after Raymond Toy)
541 * minor incompatible change: the method by which SBCL finds its
542 contributed modules has changed; it no longer relies on symbolic
543 links from an $SBCL_HOME/systems directory, but searches directly
544 in the subdirectories of $SBCL_HOME.
545 * enhancement: the dynamic heap size on the Linux/PPC platform is
546 markedly larger, even using the older Cheney garbage collector.
547 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
548 certain complicated string types. (reported by Gary King)
549 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
550 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
551 * fixed bug: DOCUMENTATION on structures no longer signals an
552 error if no structure type of the right name exists. (reported by
554 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
555 applicable non-standard primary or :AROUND method on
556 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
557 no longer cause unbound variable errors. (reported by Kevin Reid)
558 * fixed bug: metacircle resolution in cases where methods have slots
559 added before the slots from STANDARD-METHOD. (reported by Jean
561 * fixed bug: the Power PC assembly code for calling into lisp
562 sometimes computed its return address wrongly (depending on the
563 15th bit of the address, fixed at link time). This used to
564 manifest itself in a segmentation violation while building PCL.
565 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
566 many others over the years)
567 * fixed bug: ROOM no longer reports silly numbers for stack usage if
568 the stack pointer (treated as a signed integer) is negative.
569 (thanks to Peter van Eynde)
571 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
572 * new platform: experimental support for the Windows operating
573 system has been added. (thanks to Alastair Bridgewater)
574 * fixed several bugs in and robustified the PPC FFI (including
575 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
576 * bug fix: allow non-simple string symbol names (reported by Paul
578 * bug fix: interrupt handling on NetBSD (thanks to Richard M
580 * bug fix: saving a core corrupted callbacks on x86/x86-64
581 * bug fix: closed a loophole in metacircularity detection and
582 grounding in the PCL implementation of CLOS.
583 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
584 (reported by tomppa on #lisp)
585 * bug fix: clear the direction flag on Lisp -> C transitions, as
586 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
587 (reported by Andrej Grozin and Hendrik Maryns)
588 * optimization: major improvements to GC efficiency on GENCGC platforms
589 * optimization: faster implementation of EQUAL
590 * optimization: emit more efficient opcodes for some common
591 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
592 * optimization: several other minor code-generation improvements on x86-64
593 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
595 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
596 * minor incompatible change: (SETF CLASS-NAME) and (SETF
597 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
598 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
599 by ANSI as a generic function, and for consistency (SETF
600 GENERIC-FUNCTION-NAME) is treated likewise.
601 * fixed bug #233.b: make constraint propagation notice when a variable
602 value is changed after it is referenced but before it is used
603 * fixed bug #296: no more arbitrary behaviour from filenames with
604 odd characters as --load arguments. (reported by Adam Warner)
605 * fixed bug #390: :CHARACTER-SET pathname components now work as
606 expected. (reported by Tim Daly Jr)
607 * fixed bug #391: complicated :TYPE intersections in slot
608 definitions no longer cause an error in PCL internals.
609 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
610 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
611 returning the number of octets which would be written to the
612 file-stream. (thanks to Robert J. Macomber)
613 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
614 arguments to the internal stream functions. (thanks to David
616 * bug fix: stack exhaustion is now handled on the x86/SunOS
617 platform. (thanks to Daisuke Homma)
618 * bug fix: more accurate ROOM results on GENCGC platforms
619 * optimization: improved type inference for arithmetic-for
620 index variables in LOOP
621 * optimization: faster floating-point SQRT on x86-64
622 * fixed some bugs revealed by Paul Dietz' test suite:
623 ** DOCUMENTATION returns NIL instead of "" for method combinations
624 that don't have a docstring
626 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
627 * minor incompatible change: (SETF CLASS-NAME) and (SETF
628 GENERIC-FUNCTION-NAME) are no longer generic functions, and
629 therefore are not customizeable by user code (as seems to be at
630 least permitted and maybe required by AMOP). As a consolation,
631 however, the SBCL implementation of these functions now calls
632 REINITIALIZE-INSTANCE as specified by AMOP.
633 * bug fix: slot-definition documentation is propagated as per ANSI
634 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
635 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
636 argument, as do eslotd initialization methods. (from Pascal
637 Costanza's "Closer" project)
638 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
639 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
641 * bug fix: it is now possible to have more than one subclass of
642 STANDARD-GENERIC-FUNCTION without causing stack overflow.
643 (reported by Bruno Haible, Pascal Costanza and others)
644 * bug fix: the dependent update protocol now works for generic
645 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
647 * bug fix: condition-class instances corresponding to
648 DEFINE-CONDITION forms are now created eagerly. (reported by
649 Kalle Olavi Niemitalo on comp.lang.lisp)
650 * bug fix: floating point printing is more accurate in some
651 circumstances. (thanks to Simon Alexander)
652 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
653 merged with *DEFAULT-PATHNAME-DEFAULTS*.
654 * bug fix: callbacks on OS X now preserve stack-alignment.
655 (thanks to Cyrus Harmon)
656 * enhancement: the x86-64 disassembler is much better at
657 disassembling SSE instructions. (thanks to Lutz Euler)
658 * enhancement: CHECK-TYPE on a variable now allows type-inference
659 (thanks to Vincent Arkesteijn)
660 * optimization: improved performance of EUC-JP external format.
661 (thanks to NIIMI Satoshi)
662 * optimization: performance improvements to IO on file streams of
663 :ELEMENT-TYPE CHARACTER
664 * optimization: much faster memory allocation on x86-64
665 * optimization: faster garbage collections (latency and throughput)
667 * optimization: faster bignum subtraction and division on x86 and x86-64
668 (thanks to Lutz Euler)
669 * optimization: more accurate type-inference for loops that use a
670 floating point index variable or a negative step.
672 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
673 * bug fix: add a workaround to SBCL looping infinitely at startup on
674 Linux kernels with apparently buggy implementations of personality().
675 (thanks to Svein Ove Aas)
676 * bug fix: Unicode symbols are correctly printed in LDB backtraces
677 (thanks to David Lichteblau)
678 * bug fix: local bindings shadow global symbol macros for
679 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
680 * optimization: non-open coded uses of numeric comparison operators
681 (e.g. >) no longer cons when called with more than one parameter
682 on platforms supporting dynamic-extent allocation.
683 * enhancement: saving cores with foreign code loaded is now supported
684 on MIPS/Linux in addition to the previously supported platforms.
685 * bug fix: FIND-RESTART now tests for activity, not applicability when given
686 a restart object as identifier. (reported by Helmut Eller for CMUCL)
687 * bug fix: division by zero in sb-sprof when no samples were collected
688 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
689 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
690 (thanks to Alexander Kjeldaas)
691 * new feature: ignoring the return values of destructive functions
692 like NREVERSE signals a compile-time style-warning.
693 (thanks to Kevin Reid)
694 * new feature: the alignment of alien structure fields can be
695 explicitly specified. (thanks to Cyrus Harmon)
696 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
697 and dump core on SIGQUIT
699 ** incompatible change: threads do not inherit values of specials
700 from their parents (see manual)
701 ** bug fix: threads stacks belonging to dead threads are freed by the
702 next exiting thread, no need to gc to collect thread stacks anymore
703 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
704 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
705 ** bug fix: dynamic variable and thread start related gc lossage
706 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
708 * fixed some bugs revealed by Paul Dietz' test suite:
709 ** SUBTYPEP is slightly more accurate on heinously complicated
710 CONS types where some of the members have uncertain (in the
711 NIL, NIL sense) type relationships to each other.
713 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
714 no longer purified unless :PURIFY T is explicitly specified.
715 ** Non-purified cores are significantly smaller than before
717 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
718 * new feature: timers based on Zach Beane's excellent timer package
719 * added support for the following external formats: koi8-u,
720 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
721 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
722 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
723 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
724 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
725 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
726 * incompatible change: a threaded SBCL will no longer revert to
727 non-threaded mode on non-NPTL systems, but refuse to start entirely.
728 * bug fix: interrupts are disabled until startup is complete; no
729 more sigsegvs when receiving a signal to soon
730 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
732 * bug fix: add a workaround for the memory randomization features in
733 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
734 workaround will only be in effect on systems with the proc filesystem
736 * bug fix: printing objects of type HASH-TABLE signals a
737 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
739 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
740 have been running for over 50 days. (reported by Gilbert Baumann)
741 * bug fix: the logic for getting names of functions gets less
742 confused when confronded with alternate-metaclass
743 funcallable-instances. (reported by Cyrus Harmon)
744 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
745 consistent, even on internal alternate-metaclass objects.
746 * bug fix: flush closure information collected by physical
747 environment analysis prepass before the main pass. (bug reported
749 * bug fix: compiler pack phase does not modify a hashtable, which is
750 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
751 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
752 may be expanded. (reported by Alexander <malishev> on c.l.l)
753 * bug fix: no more style-warnings when compiling code using LAST
754 under high-SPEED low-SPACE settings. (reported by David Wallin
756 * fixed some incompatibilities between SBCL's MOP and the MOP
758 ** the METAOBJECT class is now implemented;
759 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
760 STANDARD-OBJECT, as required;
761 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
762 now compatible (as required by VALIDATE-SUPERCLASS); there
763 remains a consistency requirement over the presence or absence
764 of the FUNCTION class in the superclass list of finalized
765 classes; see the manual for more details;
766 ** the :AROUND method for COMPUTE-SLOTS on
767 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
768 requested slot ordering.
770 ** bug fix: parent thread now can be gc'ed even with a live
772 ** bug fix: binding a special with PROGV to no value is not
773 the same as that symbol not having been bound (thanks to
775 ** bug fix: binding specials is thread safe (thanks to
777 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
778 and SB-SPROF do not die with 'no handler for signal XX in
779 interrupt_handle_now(..)' anymore
780 ** bug fix: WITH-TIMEOUT works with multiple running threads
781 * fixed some bugs revealed by Paul Dietz' test suite:
782 ** ENSURE-GENERIC-FUNCTION should take a method class object for
783 the :method-class keyword argument.
785 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
786 * new port: the Solaris operating system on x86 processors is now
787 mostly supported, though some rough edges in the environment
788 remain. (thanks to Daisuke Homma)
789 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
790 space for its heap. (thanks to Thiemo Seufer)
791 * optimization: SBCL on MIPS platforms now supports dynamic-extent
792 closures. (thanks to Thiemo Seufer)
793 * minor incompatible change: eof selects abort in the debugger.
794 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
795 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
797 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
798 deprecated, and will go away in a future revision of SBCL.
799 * minor incompatible change: GC-ON and GC-OFF are no longer
800 implemented with a counter, it does not matter how many times gc
801 is switched on or off
802 * bug fix: discriminating functions for generic function classes
803 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
804 make invalid assumptions about method precedence order. (reported
806 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
807 platforms. (thanks to Thiemo Seufer)
808 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
809 * bug fix: correct pathname printing with printer escaping is on.
810 (thanks to Kevin Reid)
811 * bug fix: complex VOP definitions in "user-space" no longer trigger
812 package locks. (reported by Zach Beane)
813 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
814 for funcallable-instance functions, and is no different from
816 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
817 characters in the current readtable. (reported by Nicholas Neuss)
818 * bug fix: SUBTYPEP on various CONS types returns more a more
819 accurate acknowledgment of its certainty.
820 * optimizations: REMOVE-DUPLICATES now runs in linear time on
821 lists in some cases. This partially fixes bug 384.
822 * flush all standard streams before prompting in the REPL and the
824 * bug fix: signal handling and triggering gc do not conflict
825 directly anymore, in particular a high frequency sb-sprof does
826 not prevent gc from running
827 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
828 approximation for timezone and DST information between the
829 universal time epoch and the smallest negative 32-bit time_t.
830 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
831 year 1899 so long as the encoded time is non-negative.
832 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
833 the correct answer for non-constant rotate amounts when performing
834 an inline 32-bit rotation.
836 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
837 there is only one thread in the session
838 ** bug fix: memory leak for streams created in one thread and
839 written to in another
840 ** bug fix: lockup when compiled with gcc4
841 ** bug fix: race that allows the gc to be triggered when gc is
843 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
844 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
846 ** bug fix: finished threads are gc'ed properly
847 * fixed some bugs revealed by Paul Dietz' test suite:
848 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
849 has arguments with a different set of applicable methods from
850 the orignal arguments.
851 ** The type error thrown by MAP now has a correct expected-type
853 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
854 name a compiled function.
855 ** Three MISC tests where a large bignum was improperly coerced to
856 a float that couldn't represent that bignum during type
857 derivation were fixed.
858 ** SUBTYPEP can now handle types involving the negation of a
859 list-form FUNCTION type.
860 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
861 as (COMPLEX (AND RATIO (NOT FIXNUM))).
862 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
864 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
865 * New feature: Experimental support for bivalent streams: streams
866 opened with :element-type :default now allow character and binary
867 (unsigned-byte 8) I/O
868 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
869 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
871 * bug fix: no more highly sporadic "couldn't check whether ~S is
872 readable" when reading a stream and an interrupt hits in the middle
873 of a select system call
874 * compiler better recognizes complex arrays (reported by Eduardo
876 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
877 for variables on SBCL built with linkage-tables. (reported by Luis
879 * various error reporting improvements.
880 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
881 (Thanks to Thiemo Seufer)
882 * bug fix: nested reader invokations work correctly
883 * bug fix: it is possible to have simultaneous references to foreign
884 code and foreign data with the same name.
886 ** added x86-64 support
887 ** incompatible change: the threading api now works with thread
888 objects instead of thread ids
889 ** bug fix: threads are protected from signals and interruption when
890 starting up or going down
891 ** bug fix: a race where an exiting thread could lose its stack to gc
892 ** bug fix: don't halt on infinite error in threads if possible
893 ** fixed numerous gc deadlocks introduced in the pthread merge
894 ** bug fix: fixed thread safety issues in read and print
895 ** bug fix: debugger doesn't hang on session lock if interrupted at
896 an inappropriate moment
897 ** bug fix: run-program is now thread safe(r)
898 ** bug fix: inner with-recursive-lock no longer releases the mutex
899 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
900 * fixed some bugs revealed by Paul Dietz' test suite:
901 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
902 EXPECTED-TYPE slots filled.
903 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
905 ** MAKE-CONDITION accepts classes as type-designators.
906 ** COMPILE may never return NIL.
907 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
908 range before calling Unix time functions
910 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
911 * numerous signal handling fixes to increase stability
912 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
913 * minor incompatible change: we now correctly canonize default
914 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
915 as per the MOP, rather than the historical (INITARG INITFUNCTION
916 INITFORM). (reported by Bruno Haible)
917 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
918 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
919 for more information.
920 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
921 pathname is a directory pathname.
922 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
923 * dynamic space size on PPC has been increased to 768Mb. (thanks to
925 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
926 :METACLASS argument in addition to a class name. (reported by
927 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
928 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
929 (reported by Bruno Haible; patch from Gerd Moellmann)
930 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
932 * bug fix: more cleanups to the floating point exception handling on
933 x86-64 (thanks to James Knight)
934 * bug fix: the compiler does not try to constant fold calls of
935 COERCE and BIT-* functions when they return freshly constructed
936 objects. (reported by Nikodemus Siivola and Paul Dietz)
937 * optimization: improved the MIPS versions of generic (in the
938 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
939 * optimization: direct conversion of (unsigned-byte 32) to floats on
940 the PowerPC platform.
941 * optimization: structure instances with raw slots now use less
942 memory, and probably show better memory locality. (thanks to
944 * optimization: DYNAMIC-EXTENT declarations for lists and closures
945 are treated as requests for stack allocation on the x86-64,
946 Alpha-32, PPC and SPARC platforms.
947 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
948 the wrong file descriptor; implementation of SOCKET-OPEN-P.
949 (thanks to Tony Martinez)
951 ** gcing a dead thread can no longer lead to lockups
952 ** threads block signals until they are set up properly
953 ** errno is no longer shared by threads
954 ** interrupt-thread restores the eflags register on x86
955 ** fixed some lockups due to gc/thread interaction
956 * fixed some bugs revealed by Paul Dietz' test suite:
957 ** invalid dotted lists no longer raise a read error when
959 ** signal an error if a symbol that names a declaration is used
960 as the name of a type, or vice versa
961 ** allow using the (declare (typespec var*)) abbreviation for
962 (declare (type typespec var*)) with all type specifiers
963 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
964 ** always return NIL from PPRINT-POP when OBJECT is NIL
965 ** don't signal errors when pretty-printing malformed LABELS,
966 FLET or MACROLET forms
967 ** declarations in a DOLIST body are also in scope for the
969 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
971 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
974 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
975 * fixed cross-compiler leakages that prevented building a 32-bit
976 target with a 64-bit host compiler.
977 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
978 opened with :IF-EXISTS :OVERWRITE.
979 * fixed bug 281: error for an invalid qualifier in a short-form method
980 combination method is not signalled until the faulty method is called.
981 * bug fix: iteration variable type inferrer failed to deal with open
982 intervals. (reported by Alan Shields)
983 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
984 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
985 or more runtime options were provided to the sbcl binary.
986 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
987 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
988 * optimization: there's now a fast-path for fixnum arguments in the
989 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
990 * optimization: the code generated on x86-64 is more compact thanks
991 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
992 * bug fix: floating-point exception handling now partly works on
993 x86-64. (Thanks to James Knight)
994 * improvement to the MIPS backend from Thiemo Seufer: C-style
995 64-bit long long arguments and return values to alien functions
997 * fixed some bugs revealed by Paul Dietz' test suite:
998 ** the type-error signalled from WARN has a filled-in DATUM slot.
999 ** the type-error required when a stream is not associated with
1000 a file has the stream as its datum.
1001 ** type-errors on single-floats on x86-64 no longer have
1002 :INVALID-OBJECT as the datum
1003 ** the type-errors signalled for invalid function names now have
1004 a correct expected type
1005 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
1006 for structures defined with a :TYPE. Documentation strings for
1007 typed structures are no longer immediately discarded
1008 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
1009 broadcast streams and synonym streams. FILE-LENGTH now also works
1010 on broadcast streams.
1012 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1013 * incompatible change: the --noprogrammer option, deprecated since
1014 version 0.7.5, has been removed. Please use the equivalent
1015 --disable-debugger option instead.
1016 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1018 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1019 inoperational for a while now) has been completely removed.
1020 * null lexical environments are now printed as #<NULL-LEXENV>,
1021 significantly reducing the amount of clutter in typical backtraces.
1022 * documentation on weak pointers, finalization, and after GC hooks
1023 has been added to the manual.
1024 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1025 as other specialized array types, is much faster. SUBSEQ and
1026 COPY-SEQ on such arrays have also been sped up.
1027 * optimization: EQL is now more efficient when at least other argument
1028 is known to be of type (OR FIXNUM (NOT NUMBER)).
1029 * fixed bug: the runtime is now less vulnerable to changes in the
1030 size of the SBCL object on OS X, and virtual memory is reserved for
1031 all spaces that need to be at a fixed address.
1032 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1033 * fixed bug: finalizers and after GC hooks that cause consing are now
1035 * fixed bug: compiler error messages and summaries are now printed to
1036 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1037 * fixed inference of the upper bound of an iteration variable.
1038 (reported by Rajat Datta).
1039 * fixed bug 211e: calling local functions with duplicated constant
1040 keyword argument no longer causes a bogus style warning about an
1042 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1043 ftype declaration to be disregarded. (reported by Dave Roberts)
1044 * fixed bug 373: caused by erronous compilation of references to alien
1045 variables in the runtime on ppc/darwin.
1046 * fixed bug 376: CONJUGATE type deriver.
1047 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1048 (reported by Baughn on #lisp)
1049 * a cross-compiler bug on non-x86 platforms has been identified and
1050 fixed. (thanks to Bruno Haible)
1051 * improvements to the MIPS runtime code for increased stability.
1052 (thanks to Thiemo Seufer)
1053 * increased the maximimum compact environment size to allow
1054 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1055 * partial workaround for bug 135: don't name the function we're
1056 calling for hairy cases of EVAL, so as not to accumulate one environment
1057 entry per call to EVAL. (reported by Kevin Reid)
1058 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1059 * optimization: type testing for non-vector arrays should be faster.
1060 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1062 * bug fix: setting 31st element of a bit vector to zero did not work
1064 * bug fix: redefining a class definition which failed due to a
1065 previous accessor / function clash now works (but see BUGS entry
1066 #380 for more problems in this area). (thanks to Zach Beane)
1067 * the long form of DEFINE-METHOD-COMBINATION disables method group
1068 checking when given a single method group with pattern *. (thanks
1070 * on x86 compiler supports stack allocation of results of simple
1071 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1072 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1073 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1074 * fixed some bugs related to Unicode integration:
1075 ** the restarts for recovering from input and output encoding
1076 errors only appear when there is in fact such an error to
1078 * fixed some bugs revealed by Paul Dietz' test suite:
1079 ** MISC.549 and similar: late transformation of unsafe type
1080 assertions into derived types caused unexpected code
1082 ** SCALE-FLOAT type deriver is less wrong.
1083 ** type derivers for EXP, LOG and similar functions compute result
1084 types for complex arguments better.
1085 ** (MISC.563) CONJUGATE type deriver works for very restricted
1087 ** out-of-line type testers for character strings are available.
1088 ** EQUAL compiler transform understands specialness of objects
1090 ** accessing double-floats stored on the stack now works on x86-64.
1091 ** debugger internals could sometimes create invalid lispobjs,
1092 resulting in GC crashes.
1093 ** MISC.548: type check weakening can convert required type into
1095 ** initialization forms for bindings are not in scope of free special
1098 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1099 * incompatible change: thread support for non-NPTL systems has
1100 been removed - locking is buggy and unreliable. A threaded
1101 SBCL build will now warn at startup and refuse to create
1102 new threads, unless futex support is detected in the kernel
1103 * incompatible change: the top level REPL now has only an ABORT
1104 restart associated with it, not TOPLEVEL and ABORT as it used to.
1105 TOP and TOPLEVEL are now available as debugger commands for
1106 returning to the top level.
1107 * incompatible change: forms evaluated in the REPL now use the
1108 global optimization policy.
1109 * incompatible change: user- and system-initialization files are
1110 no longer processed with LOAD, but by READ and EVAL; hence the
1111 global optimization policy, startup package, readtable, etc,
1113 * The .fasl file format number has been incremented because of
1114 various incompatible changes.
1115 * internal entry point details and argument counts no longer appear
1116 in backtraces unless explicitly requested by setting
1117 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1118 * built-in and standard functions no longer have names like "top
1119 level local call to FOO".
1120 * fixed bug 32: functions defined in non-null lexical environments
1121 now have more legible printed representation
1122 * fixed bug 33: functions defined in non-null lexical environemnts
1123 are now more amenable to inspection by INSPECT.
1124 * workaround for bug 354: XEPs no longer appear in backtraces unless
1125 explicitly requested.
1126 * fixed bug: receiving the signal which results from attempting to
1127 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1128 SIGBUS on other *BSDs) on architectures where the C stack is also
1129 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1130 to debugger confusion or wild instability. (reported by Cheuksan
1132 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1133 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1134 (reported by Lutz Euler)
1135 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1136 compiled code. (reported by Paul Dietz)
1137 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1138 specialized parameter to a method to something that is not TYPEP
1139 the specializer is now possible.
1140 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1141 face of package deletion.
1142 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1143 pathname. (reported by David Tolpin on c.l.l)
1144 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1145 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1146 STRING-STREAM and FILE-STREAM, respectively.
1147 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1148 than previously; in addition, its test suite is now run on build.
1149 (thanks to Luke Gorrie)
1150 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1151 * optimization: added a immediate representation for single-floats
1153 * fixed some bugs revealed by Paul Dietz' test suite:
1154 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1155 ** attempting to create a package with a colliding nickname causes
1156 correctable errors to be signalled.
1157 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1158 ** several x86-64 backend bugs related to sign-extension of immediate
1161 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1162 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1163 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1165 * fixed loading of multiply forward-referenced layouts.
1166 (thanks to Cheuksan Wang)
1167 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1168 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1169 (see COMPILE-FILE documentation for details of :PRINT option)
1170 * fixed bugs 19 and 317: fixed-format floating point printing is
1171 more accurate. This also fixes a bug reported by Adam Warner
1172 related to the ~@F format directive.
1173 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1175 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1176 dispatch table if the from-char is a dispatch macro character.
1177 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1178 by the wordsize no longer ignore the last word. (reported by Lutz
1180 * fixed bug in type checking of dynamic-extent variables. (reported
1182 * optimization: sequence traversal functions use their freedom to
1183 coerce function designators to functions.
1184 * optimization: code with many calls to CLOS methods specialized on
1185 CLOS classes has had redundant type checks removed.
1186 * optimization: added declarations to speed up operations that access
1187 the internal character database (for example STRING-UPCASE)
1188 * optimization: comparison operations between floats and sufficiently small
1189 fixnums no longer create extra rationals
1190 * fixed some bugs related to Unicode integration:
1191 ** portions of multibyte characters at the end of buffers for
1192 character-based file input are correctly transferred to the
1193 start of the buffer at the next read.
1194 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1195 passing it through to OPEN.
1196 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1197 argument given to internal calls to OPEN.
1198 ** the built-in comment readers (introduced by character sequences
1199 ";" and "#|") are more forgiving to encoding errors; they will
1200 STYLE-WARN and then attempt to resync the stream at a character
1201 boundary. (thanks to Teemu Kalvas)
1202 * fixed some bugs in the x86-64 port:
1203 ** Negative short int return values from c-calls are sign-extended
1205 ** The stack is aligned to 16-bytes for c-calls, as required by
1206 the ABI. (thanks to Cheuksan Wang)
1207 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1208 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1209 for enums. (thanks to Vincent Arkesteijn)
1210 ** Multiple small optimizations and bugfixes for floating point
1212 * fixed some bugs revealed by Paul Dietz' test suite:
1213 ** Space, Tab, Linefeed, Return and Page have the invalid
1214 secondary constituent character trait.
1215 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1217 ** WITH-INPUT-FROM-STRING should only update the index place on
1219 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1220 works more reliably.
1221 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1222 with one-dimensional arrays with fill pointers.
1223 ** TYPE-OF failed on a complex with an integer realpart and a
1225 ** compiler failure during type inference for the code of form
1226 (IF (EQL X (THE ...)) ...) (MISC.535).
1228 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1229 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1230 platform. The port supports 61-bit fixnums, large memory spaces
1231 and reloading shared object files.
1232 * enhancement: saving cores with foreign code loaded is now
1233 supported on ppc/Darwin in addition to the previously supported
1235 * enhancement: the statistical profiler now walks deeper into the
1236 call stack for more meaningful call-graphs and accrued time
1237 reports (x86/x86-64 only). It also now reports time spent in
1239 * enhancement: it is now possible to trace most individual methods
1240 of a generic function in addition to tracing the generic function
1242 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1244 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1245 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1246 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1247 produces an error. (thanks to Vincent Arkesteijn)
1248 * bug fix: NAMESTRING on pathnames with :WILD components in their
1249 directories works correctly. (thanks to Artem V. Andreev)
1250 * fixed bug 125: compiler preserves identity of closures. (reported
1252 * bug fixed: functions with &REST arguments sometimes failed with
1253 "Undefined function" when compiled with (DEBUG 3). (reported by
1255 * bug fix: overflow during compiling of setting element of a bit
1256 vector with constant index and value. (reported by Timmy Douglas)
1257 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1258 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1259 * on x86 compiler partially supports stack allocation of dynamic-extent
1261 * GO and RETURN-FROM do not check the extent of their exit points
1262 when compiled with SAFETY 0.
1263 * fixed some bugs related to Unicode integration:
1264 ** encoding and decoding errors are now much more robustly
1265 handled; it should now be possible to recover even from invalid
1266 input or output to the terminal. (thanks to Teemu Kalvas)
1267 ** provided a first cut at implementing STRING-TO-OCTETS and
1268 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1269 ** altered the SB-MD5 contributed module slightly, changing the
1270 interface just enough for it to be supportable for builds where
1271 lisp characters are not eight bits.
1272 * fixed some bugs revealed by Paul Dietz' test suite:
1273 ** the FORMATTER-generated functions for ~V[ conditionals require
1274 the correct number of arguments.
1275 ** READ-FROM-STRING returns the mandated second value when applied
1276 to displaced strings.
1277 ** the #\Rubout and #\Backspace characters are treated as invalid
1278 constituent characters by the tokenizer.
1280 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1281 * new feature: reloading changed shared object files with
1282 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1283 * new feature: references to foreign variables and functions
1284 can now be compiled and loaded before the corresponding shared
1285 object file is loaded, as long as the foreign definitions are
1286 available at runtime.
1287 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1288 supported. (thanks to Dan Debertin)
1289 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1290 just keywords. (thanks to Vincent Arkesteijn)
1291 * fixed bug #331: structure-class instances corresponding to
1292 DEFSTRUCT forms are now created eagerly.
1293 * fixed bug #345: backtraces from calls to undefined functions work
1294 on x86 as well. Related bug #61 is now also partially fixed on x86
1295 (backtraces from throws to unknown catch tags.)
1296 * bug fix: lambda-list parsing is now stricter vrt. order and number
1297 of lambda-list keywords.
1298 * bug fix: as specified by AMOP, an error is signalled if a
1299 class-option appears multiple times in a DEFCLASS form. (reported
1301 * bug fix: RANDOM can be compiled when the compiler derives the type
1302 of its numeric argument as a disjoint set of small integers.
1303 (reported by Paul Dietz)
1304 * bug fix: starting a core saved with shared objects loaded when
1305 those objects are not available no longer causes threaded SBCL to
1306 hang. (reported by Sean Ross)
1307 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1309 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1310 argument, which must always be NIL. (reported by Kalle Niemitalo)
1311 * bug fix: printing 1.0d+23 no longer results in an error.
1312 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1313 * bug fix: structure slot setters preserve evaluation order. (thanks
1315 * fixed some bugs related to Unicode integration:
1316 ** RUN-PROGRAM can allow its child to take input from a Lisp
1317 stream. (reported by Stefan Scholl)
1318 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1319 locales. (reported by Ken Causey)
1320 * fixed some bugs revealed by Paul Dietz' test suite:
1321 ** INCF, DECF and REMF evaluate their place form as specified in
1323 ** FORMATTER expands ~{ ~} iteration directives with V or #
1324 parameters correctly.
1325 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1326 consequent uses no arguments correctly.
1327 ** the system has a partial understanding of the (COMPLEX RATIO)
1329 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1332 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1333 * new feature: a build-time option (controlled by the :SB-UNICODE
1334 keyword feature, enabled by default) for building the system with
1335 support for the entire 21-bit character space defined by the
1337 * new feature: the system now has rudimentary external-format
1338 support; the primary user-visible change at this time is that
1339 characters with the high bit set (such as lower-case-e-acute) will
1340 print correctly to a terminal in a UTF-8 environment.
1341 * minor incompatible change: BASE-CHAR no longer names a class;
1342 however, CHARACTER continues to do so, as required by ANSI.
1343 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1344 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1345 has been moved to the SB-EXT package (temporarily re-exported from
1347 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1348 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1350 * fixed bug #351: better error-handlind and reporting for malformed
1352 * fixed bug #350: bignum-printing is now more memory-efficient,
1353 allowing printing of very large bignums, eg. (expt 2 10000000).
1354 (reported by Bruno Haible)
1355 * fixed bug #302: better primitive-type selection for intersection
1357 * fixed bug #308: non-graphic characters now all have names, as
1358 required. (reported by Bruno Haible)
1359 * bug fix: redefining a class with different superclasses now correctly
1360 removes it from the direct-subclasses of its previous superclasses.
1361 (reported by David Morse)
1362 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1363 new value now works. (reported by Bruno Haible)
1364 * bug fix: correct canonicalization of multiple non-standard slot
1365 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1366 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1367 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1368 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1369 now exists, an signals an error.
1370 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1371 during macroexpansion for non-iteration variables as well. (reported
1372 by Bruno Haible for CMUCL)
1373 * bug fix: Cyclic structures and unprintable objects in compiler
1374 messages no longer cause errors. (reported by Bruno Haible)
1375 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1376 and READ-FROM-STRING all now return a primary value of NIL if
1377 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1378 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1379 (reported by Bruno Haible for CMUCL)
1380 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1381 do not designate a condition. (reported by Bruno Haible for
1383 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1384 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1385 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1386 permitting resolution in favour of any of the conflicting symbols.
1387 (reported by Bruno Haible for CMUCL)
1388 * bug fix: EQUAL compiler optimizations is less aggressive on
1389 strings which can potentially compare true despite having distinct
1390 specialized array element types.
1391 * bug fix: unit enumerations can be defined without dividing by
1392 zero. (thanks to Vincent Arkesteijn)
1393 * FORMAT compile-time argument count checking has been enhanced.
1394 (report from Bruno Haible for CMUCL)
1395 * a partial workaround for the bug 262: the compiler does not try to
1396 inline-expand a local function doing RETURN-FROM from a deleted
1397 BLOCK. (thanks to Peter Denno for the bug report and to David
1398 Wragg for the simple test case)
1399 * fixed some bugs revealed by Paul Dietz' test suite:
1400 ** RENAME-PACKAGE allows all package designators as new package
1402 ** constraint propagation and lambda variable substitution are
1403 more cautious in dealing with partially deleted code.
1404 ** compiler sometimes lost reoptimization passes.
1405 ** CERROR, given a condition as condition designator, passes its
1406 remaining arguments to the continue format control without
1408 ** Case-altering FORMAT directives work correctly on non-ASCII
1410 ** The REAL type specifier handles bounds outside the floating
1411 point ranges without signalling FLOATING-POINT-OVERFLOW.
1412 ** Functions with IR1-transformations can create intercomponent
1413 references to global functions.
1414 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1416 ** FORMAT ~R treats a nil value for its first parameter correctly.
1418 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1419 * enhancement: saving cores with foreign code loaded is now
1420 supported on x86/NetBSD and sparc/Linux in addition to the previously
1421 supported platforms.
1422 * bug fix: on some platforms repeated installations caused multiple
1423 copies of HTML documentation to be installed -- should not happen
1424 any more. (reported by Stefan Scholl)
1425 * bug fix: parsing self-recursive alien record types multiple times
1426 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1427 original patch by Helmut Eller for CMUCL)
1428 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1429 (thanks to Richard Kreuter)
1430 * bug fix: defining classes whose accessors are methods on existing
1431 generic functions in other (locked) packages no longer signals
1432 bogus package lock violations. (reported by François-René Rideau)
1433 * bug fix: special variables as DEFMETHOD parameters no longer have
1434 associated bogus type declarations. (reported by David Wragg and
1436 * bug fix: read-write consistency on streams of element-type
1437 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1438 * bug fix: redefiniton of the only method of a generic function with
1439 no DEFGENERIC no longer emits a full WARNING. In addition,
1440 redefinition of generic functions with no DEFGENERIC to an
1441 incompatible lambda list now signals an error. (thanks to Zach
1443 * bug fix: DEFGENERIC now works even when there's a function of the
1444 same name in an enclosing lexical environment. (thanks to Zach
1446 * fixed compiler failure, caused by instrumenting code during
1447 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1448 * optimization: added loop analysis and improved register allocation
1449 * fixed some bugs revealed by Paul Dietz' test suite:
1450 ** POSITION on displaced vectors with non-zero displacement
1451 returns the right answer.
1452 ** (SIMPLE-STRING) is a valid type specifier for sequence
1454 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1456 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1457 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1459 ** The justification version of the FORMAT ~< directive treats
1460 non-zero minpad parameter correctly.
1462 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1463 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1464 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1465 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1466 the supported interface.
1467 * new feature: Single-stepping of code compiled with DEBUG 2 or
1468 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1469 * new feature: saving cores with foreign code loaded is now
1470 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1471 Timothy Moore's work for CMUCL)
1472 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1473 parameters to * instead of NIL if no initform is supplied.
1474 (reported by Johan Bockgaard)
1475 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1476 correctly when FUNCALL appears as the car of the form. Note:
1477 despite this FUNCALL forms are not currently subject to
1478 compiler-macro expansion. (port of Raymond Toy's fix for the
1479 same from CMUCL, reported by Johan Bockgaard)
1480 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1481 (thanks for Teemu Kalvas)
1482 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1483 case instead of terminating the iteration (thanks for Julian
1484 Squires, Sean Champ and Raymond Toy)
1485 * bug fix: incorrect expansion of defgeneric that caused a style
1486 warning. (thanks for Zach Beane)
1487 * bug fix: slot accessor effective method computation works properly
1488 for classes with multiple non-standard applicable methods on
1489 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1490 * on x86 compiler supports stack allocation of results of LIST and
1491 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1492 CMUCL implementation by Gerd Moellmann)
1493 * fixed some bugs revealed by Paul Dietz' test suite:
1494 ** FORMAT strings with both the ~<~:;~> form of the justification
1495 directive and pretty-printing directives cause an error.
1497 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1498 * incompatible change: the internal functions
1499 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1500 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1501 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1502 instead of the old functions.
1503 * new feature: on platforms where "dladdr" is available foreign
1504 function names now appear in backtraces. (based on Helmut Eller's
1506 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1507 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1509 * documentation: ASDF manual is now distributed with SBCL.
1510 * bug fix: SBCL can now load its contributed modules with REQUIRE
1511 even if the system-provided entries have been removed from
1512 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1514 * bug fix: backtraces involving undefined functions or assembly
1515 routines are more informative. (thanks to Brian Downing)
1516 * bug fix: mutually referent alien structures now work correctly.
1517 (reported by Rick Taube)
1518 * bug fix: structures defined by WITH-ALIEN can be referred to
1519 within other definitions in the same WITH-ALIEN.
1520 * bug fix: division operators (MOD, TRUNCATE and the like) with
1521 constant zero divisors and integer dividends no longer generate
1523 * bug fix: provide default methods for INPUT-STREAM-P and
1524 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1525 * bug fix: improve the detection and resolution of MOP metacycles
1526 (where effective method computation depends on the generic function
1527 having its effective method computed). (reported by Bruno Haible)
1528 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1529 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1530 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1531 * bug fix: pathnames with relative directory components can be
1532 represented relative to default pathnames.
1533 * optimization: in taking the GCD of bignums, reduce the two bignums
1534 to approximately the same size (using Euclid's algorithm) before
1535 applying the more sophisticated binary GCD. (thanks to Juho
1537 * optimization: COUNT on bitvectors now operates word-at-a-time.
1538 * optimization: ASH with a positive, but not necessarily constant,
1539 (leftwards) shift, when in a modular context, is compiled to a
1541 * fixed some bugs revealed by Paul Dietz' test suite:
1542 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1543 if the corresponding argument is NIL.
1544 ** FORMAT directives accepting a minpad parameter treat negative
1546 ** literal commas as character parameters to FORMAT directives are
1548 ** literal spaces directly after ~<Newline> directives within a
1549 format-logical-block (~:< ~@:>) do not induce :FILL-style
1550 conditional newlines.
1551 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1552 as specified: it no longer includes conditional newlines).
1553 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1555 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1556 :PER-LINE-PREFIX argument does not evaluate to a string.
1558 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1559 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1560 the manual for details; for now, package locks can be disabled by
1561 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1562 if no major problems are found then it is likely that they will be
1563 compiled in unconditionally.
1564 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1565 unsupported operators on all platforms. To load a shared library
1566 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1567 object file, link it into a shared library outside of SBCL and
1568 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1570 * bug workaround: ROOM T and the GC were not working together
1571 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1572 reporting facility aren't preserved. That reporting has been
1573 disabled, so now until and unless someone figures out how to make it
1574 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1575 * minor incompatible change: as threatened around sbcl-0.8.0, the
1576 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1577 The symbols are also exported from SB-PCL for backwards
1578 compatibility, but more so than before SB-PCL should be treated as
1579 an implementation-internal package.
1580 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1582 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1583 It is known as of this release not to work on the Alpha, however.
1584 * fixed bug #167: errors signalled due to illegal syntax in method
1585 bodies are now more legible.
1586 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1587 designators and can hence be used with TYPEP.
1588 * fixed bug #333: CHECK-TYPE now ensures that the type error
1589 signalled, if any, has the right object to be accessed by
1590 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1591 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1592 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1594 * fixed bug #334: programmatic addition of slots using specialized
1595 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1596 and :ALLOCATION :CLASS slots.
1597 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1598 float exponents. (rereported by Peter Seibel)
1599 * fixed a bug: #\Space (and other whitespace characters) are no
1600 longer considered to be macro characters in standard syntax by
1601 GET-MACRO-CHARACTER.
1602 * fixed bug: initialization of condition class metaobjects no longer
1603 causes an instance of the condition to be created. (reported by
1605 * fixed bug: it is now possible to have slots such that
1606 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1607 description is neither :INSTANCE nor :CLASS.
1608 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1609 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1610 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1611 non-standard SLOT-DEFINITION-ALLOCATION values.
1612 * fixed bugs: various race conditions handling exiting threads.
1613 CL-PPCRE's thread torture test now passes.
1614 * fixed arguably-a-bug: GC time proportional to number of threads in
1615 system even when most of them are idle
1616 * optimization: improved performance of BIT and SBIT on bit-vectors.
1617 * .fasl file incompatibility: The fasl file version number has
1618 been incremented because of changes associated with package locks.
1620 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1621 * minor incompatible change: the system no longer provides
1622 optimization hints (conditions which are TYPEP
1623 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1624 mode; these hints are still emitted when higher SPEED optimization
1626 * new contrib module: a sampling profiler (profiling by statistical
1627 sampling, rather than by instrumenting functions) is available as
1628 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1629 * the behaviour of the standard function ED is now customizeable by
1630 third parties through a hook variable: see ED's documentation
1631 string for information on the protocol.
1632 * the compiler no longer emits efficiency notes for (FUNCALL X)
1633 when the type of X is uncertain under default optimization
1635 * fixed bug 276: mutating a binding of a specialized parameter to a
1636 method to something that is not TYPEP the specializer is now
1638 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1639 exhibit the required behaviour on the x86 platform. (thanks to
1640 Peter van Eynde, Eric Marsden and Bruno Haible)
1641 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1642 even for difficult arguments. (reported by Peter Graves)
1643 * fixed bug 141a: the backquote printer now descends quoted
1645 * fixed another bug in backquote printing: no more destructive
1646 modification of the form's list structure. (reported by Brian
1648 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1649 move between its address being taken and the call to
1650 interrupt_thread, fixing a crashing race condition.
1651 * the SB-POSIX contrib implementation has been adjusted so that it
1652 no longer exhibits ridiculously poor performance when constructing
1653 instances corresponding to C structs.
1655 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1656 * minor incompatible change: the sb-grovel contrib now treats C
1657 structures as alien (in the sense of SB-ALIEN) objects rather than
1658 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1659 has implications for memory management of client code
1660 (sb-grovel-returned objects must now be manually managed) and for
1661 type safety (alien objects now have full types).
1662 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1663 used to control emission of compiler diagnostics, rather than the
1664 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1665 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1666 quality should be considered deprecated.
1667 * install.sh now installs the user manual as well
1668 * (not quite a new documentable feature, but worth considering in
1669 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1670 beginnings of a semantically meaningful condition hierarchy is
1671 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1673 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1674 designator as the defaults argument.
1675 * fixed bug: Displaced arrays whose displaced-to array has become
1676 too small now cause ARRAY-DIMENSION to signal an error, providing
1677 for safer bounds-checking. (reported by Bruno Haible)
1678 * fixed bug: DEFCLASS slot definitions with identical :READER and
1679 :WRITER names now signal a reasonable error. (reported by Thomas
1681 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1683 * fixed bug: Invalid slot specification errors now print correctly.
1684 (thanks to Zach Beane)
1685 * fixed bug 320: Shared to local slot value transfers in class
1686 redefinitions now happen corrently from superclasses as
1687 well. (reported by Bruno Haible)
1688 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1690 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1691 improper lists correctly. (reported by Bruno Haible)
1692 * fixed bug 313: source-transform for <fun-name> was erroneously
1693 applied to a call of a value of a variable with name <fun-name>.
1694 (reported by Antonio Menezes Leitao)
1695 * fixed bug 307: The obsolete instance protocol ensures that
1696 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1697 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1698 remove dead unknown-values globs from the middle of the stack.
1699 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1700 easier to resume long computations after using *BREAK-ON-SIGNALS*
1701 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1702 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1703 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1704 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1705 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1706 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1708 * fixed a (fairly theoretical) bug in string printing: if
1709 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1710 does not have array-element-type equal to the most general string
1712 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1713 function name. (thanks to Nikodemus Siivola)
1714 * fixed bug: compile-time format string checker failed on
1715 non-closed ~{. (reported by Thomas F Burdick)
1716 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1717 don't warn on legal loop code involving more than one aggregate
1718 boolean. (thanks to Nikodemus Siivola)
1719 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1720 in keyword position to MAKE-PACKAGE was not being reported as
1722 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1723 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1725 * fixed bugs 280 and 312: the checking for multiple definitions in a
1726 file is less likely to become confused by uses of inline
1728 * fixed bug: the #S reader macro performs the keyword coercion
1729 specified for slot names. (reported by Kalle Niemitalo)
1730 * fixed bug: lambda lists may contain symbols whose names start with
1731 & but are not lambda-list-keywords; their occurrence triggers a
1733 * fixed bug 321: define-method-combination argument lambda lists do
1734 not cause invalid code to be generated when &optional or &aux
1735 variables have default values. (reported by Bruno Haible)
1736 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1737 have CLOS classes; this appears also to have fixed reported
1738 instability in stack exhaustion detection.
1739 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1740 MUFFLE-WARNING when no associated restart is present is now
1742 * optimization: rearranged the expansion of various defining macros
1743 so that each expands into only one top-level form in a
1744 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1746 * optimization: used a previously little-used slot in symbols to
1747 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1749 * fixed some bugs revealed by Paul Dietz' test suite:
1750 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1752 ** Fixed an optimization bug related to inheritance of initforms
1753 from local to shared slots.
1754 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1755 ** CAST optimizer forgot to flush argument derived type.
1756 ** print/read consistency on floats is now orders of magnitude
1757 more likely. (thanks also to Bruno Haible for a similar report
1759 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1760 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1761 for objects of type REAL. Make it so.
1762 ** SXHASH is specified (infelicitously) to respect similarity,
1763 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1764 it so. (thanks to Markus Ziegler)
1765 ** on the Alpha, the compiler succeeds in compiling functions
1766 returning a known number of arguments greater than 63.
1767 ** fixed handling of invalid NIL arguments in keyword position
1769 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1771 ** provided a readably-printable representation for RANDOM-STATE
1773 ** ensured that pathnames, where they have a namestring, always
1774 print using #P"..." syntax.
1776 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1777 * Support for the forthcoming 2.0 version of the NetBSD kernel
1778 running on x86 hardware has been added. (thanks to Perry
1779 E. Metzger most immediately, and others for their past work)
1780 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1781 current release version; SBCL's OpenBSD support had been broken
1782 since about the time of OpenBSD's switch to ELF binary format.)
1783 * [placeholder for DX summary]
1784 ** user code with &REST lists declared dynamic-extent, under high
1785 speed or space and low safety and debug optimization policy.
1786 * The manual has been converted to Texinfo format and the debugger
1787 chapter from the cmucl manual has been added.
1788 * A facility has been added to extract documentation strings from
1789 sbcl and store them as Texinfo-formatted snippets for inclusion in
1790 the manual (via Texinfo's @include directive)
1791 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1792 the test case to Dave Roberts)
1793 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1794 pointers. (reported by Sean Ross)
1795 * bug fix: PROFILE output is printed nicely even for large numerical
1796 values. (thanks to Zach Beane)
1797 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1798 greater than 32 handle EOF correctly.
1799 * bug fix: on X86 an immediate argument of the IMUL instruction is
1800 correctly printed in disassembly. (thanks to Lutz Euler)
1801 * bug fix: class slots in redefined classes preserve their old
1802 values. (thanks to Bruno Haible and Nikodemus Siivola)
1803 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1804 succeeds. (reported by Marco Baringer)
1805 * fixed some bugs revealed by Paul Dietz' test suite:
1806 ** READ-SEQUENCE now works on ECHO-STREAMs.
1807 ** RATIONALIZE works more according to its specification. (thanks
1810 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1811 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1812 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1813 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1814 should matter to you only if you rebind the printer control
1815 variables and then find you want different bindings in the
1816 debugger than in the ordinary execution of your program.)
1817 * The runtime build system has been tweaked to support building
1818 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1819 assembler and linker. (thanks to Nikodemus Siivola)
1820 * Unbound, undefined, undeclared variables now trigger full
1821 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1822 more often programmer error than deliberate exploitation of undefined
1824 * optimization: the hash algorithm for strings has changed to one
1825 that is less vulnerable to spurious collisions. (thanks to Juho
1827 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1828 needless bounds checking. (thanks to Juho Snellman)
1829 * optimization: implemented multiplication as a modular
1830 (UNSIGNED-BYTE 32) operation on the PPC backend.
1831 * fixed some bugs revealed by Paul Dietz' test suite:
1832 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1834 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1835 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1836 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1838 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1840 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1842 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1843 their output stream on EOF from read.
1844 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1845 have been read to end-of-file.
1846 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1848 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1849 description of determination of which consecutive characters
1851 ** Printing the "Space" character with escaping on now yields "#\\ ",
1852 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1853 ** Reading floating-point numbers with *READ-BASE* set to a number
1854 less than 10 works correctly.
1855 ** Reading floating-point numbers with *READ-BASE* set to a number
1856 more than 10 works correctly.
1857 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1858 the readtable currently in effect.
1860 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1861 * minor incompatible change: parsing of namestrings on a physical
1862 (Unix) host has changed; numbers after the final #\. in a
1863 namestring are no longer interpreted as a version field. This is
1864 intented to be largely invisible to the user, except that the
1865 meaning of the namestring "*.*.*" has changed: it now refers to a
1866 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1867 should usually be replaced by
1868 (make-pathname :name :wild :type :wild :version :wild)
1869 with the added benefit that this is more likely to be portable.
1870 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1871 option to OPEN now signals an error if the file being opened
1872 exists; this may have an impact on existing code.
1873 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1874 signals received, on the PowerPC platforms (both Linux and
1875 Darwin). (thanks to Pierre Mai for pointing out the location of
1877 * several fixes on OS X: The system now builds and runs cleanly on
1878 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1879 (thanks to Brian Mastenbrook)
1880 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1881 second-resolution: integer multiples of 1/3600 between -24 and 24.
1882 (thanks to Vincent Arkesteijn)
1883 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1884 type when called with 1 argument; PEEK-CHAR checked type of
1885 PEEK-TYPE only after having read first character from a
1886 stream. (reported by Peter Graves)
1887 * bug fix: the garbage collector now has much better locality
1888 behaviour, and in particular no longer treats all memory as being
1889 exhausted when many small objects point to each other in a deeply
1891 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1892 recognized as being TYPEP their class.
1893 * bug fix: the PUSHNEW documentation string has been corrected.
1894 (thanks to Vincent Arkesteijn)
1895 * bug fix: defaulting of the value for the last of an atypically
1896 large number of multiple values being bound was not being
1897 performed correctly on the Alpha or PPC platforms
1898 * optimization: implemented multiplication as a modular
1899 (UNSIGNED-BYTE 32) operation on the x86 backend.
1900 * optimization: SEARCH on simple-base-strings can now be open-coded.
1901 (see also contrib/compiler-extras.lisp for inspiration for
1902 teaching the compiler about the Boyer-Moore algorithm).
1903 * value, returned by MAX (and MIN) called with several EQUALP, but
1904 not EQL, arguments now does not depend on compiler settings.
1905 * fixed some bugs revealed by Paul Dietz' test suite:
1906 ** in stack analysis liveness information is propagated from
1907 non-local entry points.
1908 ** pathwise CAST removing failed when the CAST node did not start
1910 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1911 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1913 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1914 host is already defined.
1915 ** RENAME-FILE works on streams instead of signalling an internal
1917 ** PEEK-CHAR uses the current readtable when determining whether
1918 or not a character is whitespace.
1919 ** MERGE-PATHNAMES handles the case when the pathname does not
1920 specify a name while the default-pathname specifies a version.
1921 ** Pathnames now stand a better chance of respecting print/read
1923 ** Attempting to use standardized file system operators with a
1924 pathname with invalid :DIRECTORY components signals a
1926 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1928 ** DIRECTORY on logical pathnames is more correct.
1929 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1930 signal a TYPE-ERROR if their argument is not a stream.
1931 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1932 designator argument does not designate a stream.
1933 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1934 examining the synonym.
1935 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1937 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1938 element-type larger than ([UN]SIGNED-BYTE 32).
1940 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1941 * When built with the :SB-FUTEX feature, threaded builds now take
1942 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1943 for faster/more reliable mutex and condition variable support.
1944 * Incompatible change (but one you probably shouldn't have been using
1945 anyway): the interface and code for arbitrating between multiple
1946 threads in the same user session has been redesigned.
1947 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1948 error when called without an explicit environment argument.
1949 (thanks to Nikodemus Siivola)
1950 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1951 confused about their position. (thanks to Adam Warner and Gerd
1953 * bug fix: Pretty printing backquoted forms with unquotations in the
1954 argument list position of various code constructs such as LAMBDA
1955 now works correctly. (reported by Paul Dietz)
1956 * bug fix: Pretty printing unquotations no longer loses all
1957 stream position information.
1958 * optimization: performance of string output streams is now less
1959 poor for multiple small sequence writes.
1960 * optimization: performance of CSUBTYPEP in the presence of complex
1961 expressions involving CONS and NOT many times has been improved.
1962 (reported by Paul Dietz)
1963 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1965 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1966 streams for the REPL, for the debugger, and so on. (thanks to
1968 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1969 * fixed some bugs revealed by Paul Dietz' test suite:
1970 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1971 is ignored if it is too small, rather than propagating through
1972 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1973 ** extremely complex negations of CONS types were not being
1974 sufficiently canonicalized, leading to inconsistencies in
1976 ** VALUES tranformer lost derived type.
1978 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1979 * fixed a bootstrapping bug: the build process no longer assumes
1980 that the various BOOLE-related constants have the same value in
1981 host and target lisps. (noted by Paul Dietz' test suite on an
1982 SBCL binary built from CLISP)
1983 * The system can now be dynamically linked on the MIPS platform,
1984 which enables dynamic loading of foreign code from Lisp. (thanks
1985 to Ralf Baechle for discussions on the MIPS ABI)
1986 * The system now records debugging information for its own source
1987 files in a filesystem-position-independent manner, relative to
1988 the "SYS" logical host.
1989 * fixed a compiler bug: MV-LET convertion did not check references
1990 to the "max args" entry point. (reported by Brian Downing)
1991 * tweaked disassembly notes to be less confident about proclaiming
1992 some instruction as an LRA. (thanks to Brian Downing)
1993 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1994 now each have their own history, command character, and other
1995 characteristics. (thanks to David Lichteblau)
1996 * fixed some bugs revealed by Paul Dietz' test suite:
1997 ** compiler failure in compiling LOGAND expressions including a
1999 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
2000 shift greater than 32.
2001 ** FLUSH-DEST did not mark blocks for type check regeneration.
2002 ** HANDLER-CASE failed to accept declarations in handler clauses
2003 in some circumstances.
2005 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
2006 * New code in contrib/sb-introspect (still probably not entirely
2007 stable yet) provides some support for smart Lisp development
2008 environments like SLIME.
2009 * The conditions signalled for errors occurring when loading .fasl
2010 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
2011 in a way which should help ASDF recover gracefully.
2012 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2013 stuff has been cleaned up. If you code contrib/ stuff, this might
2014 affect you, and you can look at contrib/README, contrib/STANDARDS,
2015 and/or the 0.8.4.27 diff to check.
2016 * In full calls the compiler now does not generate checks for declared
2017 argument types for all arguments.
2018 * various threading fixes
2019 ** and some experimental patches which didn't make it into
2020 the main tree for this release, but which are shipped in
2021 contrib/experimental-thread.patch as a possible fix for some
2022 failures (deadlock, spinning...) in GC-intensive multithreaded
2024 * fixed PPC build problem (source code incompatibility of different
2025 library versions): added offsetof() hackery which attempts to divine
2026 where glibc maintainers put uc_mcontext today
2027 * fixed bug 282: compiler does not trust type assertions while passing
2028 arguments to a full call.
2029 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2030 &OPTIONAL VALUES type parameter.
2031 * fix bug 214: algorithm for noting rejected templates is now more
2032 similar to that of template seletion. (also reported by rydis on
2034 * fixed bug 141b: printing backquoted information readably and prettily
2035 inserts a space where necessary.
2036 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2037 (CHARACTER 10) are now reported as errors, rather than propagated
2038 as unknown types. (reported by piso on #lisp)
2039 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2040 with values NIL and :ERROR. (thanks to Milan Zamazal)
2041 * fixed bug 191c: CLOS now does proper keyword argument checking as
2042 described in CLHS 7.6.5 and 7.6.5.1.
2043 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2044 longer raise an error; further, using a list as a for-as-arithmetic
2045 counter now raises a meaningful error.
2046 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2047 now understood by sequence creation functions such as MAKE-SEQUENCE
2049 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2050 a STRING-INPUT-STREAM.
2051 * compiler enhancement: SIGNUM is now better able to derive the type
2053 * type declarations inside WITH-SLOTS are checked. (reported by
2055 * fixed some bugs revealed by Paul Dietz' test suite:
2056 ** incorrect optimization of TRUNCATE for a positive first
2057 argument and negative second.
2058 ** compiler failure in let-convertion during flushing dead code.
2059 ** compiler failure while deriving type of TRUNCATE on an
2060 interval, containing 0.
2061 ** ASH of a negative bignum by a negative bignum count now returns
2063 ** intersection of CONS types now canonicalizes properly, fixing
2064 inconsistencies in SUBTYPEP.
2066 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2067 * incompatible change: The --disable-debugger command line
2068 option now clobbers the debugger at a more fundamental
2069 level, by redefining #'INVOKE-DEBUGGER instead of by
2070 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2071 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2072 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2073 enter the debugger REPL and then suffer recursive errors
2074 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2075 while under the new system, BREAK in --disable-debugger mode
2076 terminates the system just as an unhandled error would.
2077 * fixed compiler performance when processing loops with a step >1;
2078 * bug fix: DOCUMENTATION now retrieves generic function
2079 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2080 support has been systematized, and now supports the methods
2081 specified by ANSI, along with a default method and a method for
2082 slot documentation. (reported by Nathan Froyd)
2083 * bug fix: effective methods associated with a generic function are
2084 no longer cached over a change of that generic function's method
2085 combination. (reported by Andreas Fuchs)
2086 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2087 names a non-existent directory. (thanks to Andreas Fuchs)
2088 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2089 the PPC platform, lead to stack corruption; this has been fixed.
2090 (reported by Rainer Joswig)
2091 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2092 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2093 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2094 arithmetic in general, is now much more reliable on the PPC
2096 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2097 platform now returns the right answer.
2098 * optimization: restored some effective method precomputation in
2099 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2100 precomputation is now tunable.
2101 * optimization: compiler-internal data structure use has been
2102 reviewed, and changes have been made that should improve the
2103 performance of the compiler by about 20%.
2104 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2105 simple-base-strings and simple-bit-vectors is improved.
2106 * optimization: the optimization of 32-bit logical and arithmetic
2107 functions introduced in version 0.8.3 on the x86 has been
2108 implemented on the mips, ppc and sparc platforms; an
2109 implementation of the same facility, but for 64-bit arithmetic,
2110 has been added for the alpha.
2111 * microoptimization: the compiler is better able to make use of the
2112 x86 LEA instruction for multiplication by constants.
2113 * bug fix: in some situations compiler did not report usage of
2114 generic arithmetic in (SPEED 3) policy.
2115 * bug 145b fix: compiler used wrong type specifier while converting
2116 MEMBER-types to numeric.
2117 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2119 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2120 index argument now works properly on the Alpha platform.
2121 * bug fix: floating point exception treatment on the Alpha platform
2123 * bug fix: FILE-POSITION works much better on string input and
2124 output streams. (thanks to Nikodemus Siivola)
2125 * bug fix: many threading/garbage collection symptoms sorted.
2126 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2127 might be pseudo-atomic.
2128 * internal change: Stopping for GC is now done with signals not
2129 ptrace. GC is now done in whichever thread wanted it, instead of
2131 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2133 * bug fix: result form in DO is not contained in the implicit
2135 * incompatible change: ICR structure is changed; the value part of
2136 CONTINUATION is now called LVAR; corresponding functions are
2137 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2138 * added type deriver for ISQRT (thanks to Robert E. Brown).
2139 * bug fix: better support for loading from the command line when an
2140 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2142 * fixed some bugs revealed by Paul Dietz' test suite:
2143 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2144 ** ROUND and FROUND now give the right answer when given very
2145 small float arguments.
2146 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2148 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2149 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2150 ** LCM should return a non-negative integer.
2151 ** PARSE-INTEGER returned the index of a terminator instead of the
2152 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2153 ** PARSE-INTEGER returned an incorrect index being applied to a
2155 ** LCM with two arguments of 0 returns 0 rather than signalling
2157 ** unsigned addition of a 32-bit constant with the high bit set no
2158 longer causes an internal compiler error on the x86.
2159 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2160 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2161 with negative last argument.
2162 ** byte specifiers with zero size and position no longer cause
2163 an error during type derivation.
2164 ** bignum multiplication on the Alpha platform now returns the
2166 * porting: The system now builds on SuSE AMD64, although it still
2167 generates a 32-bit binary.
2168 * .fasl file incompatibility: The fasl file version number has
2169 been incremented (because of the changes to internal compiler
2170 data structures referred to above).
2172 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2173 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2174 more accurately, on the Darwin kernel running on PowerPC hardware.
2175 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2176 * Compiler code deletion notes now signal a condition of type
2177 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2178 an associated MUFFLE-WARNING restart.
2179 * The compiler now performs limited argument count validation of
2180 constant format strings in FORMAT, and where appropriate in ERROR,
2181 CERROR and WARN. (thanks to Gerd Moellmann)
2182 * New ASDF-INSTALL contrib can be used for automatic download and
2183 installation of third-party Lisp code from CCLAN or other sites
2185 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2186 which forces another thread to execute a function supplied by the
2188 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2189 now accept and act upon their :ELEMENT-TYPE keyword argument.
2190 (reported by Martin Atzmueller, Edi Weitz)
2191 * bug fix: FILE-POSITION now accepts position designators up to
2192 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2193 is the greater. (thanks to Patrik Nordebo)
2194 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2195 Kalle Olavi Niemitalo)
2196 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2197 printable. (reported by Eric Marsden)
2198 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2199 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2200 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2201 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2202 (thanks to Stig Sandoe)
2203 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2204 UNREAD-CHAR now correctly includes the unread character in the
2205 target sequence. (thanks to Gerd Moellmann)
2206 * bug fix in threaded builds: the system can now be suspended and
2207 resumed by shell job control with minimal disruption.
2208 * bug fixes in times and timezones >2038AD
2209 * better handling of "where is GNU make?" problem in build scripts
2210 (thanks to Nikodemus Siivola)
2211 * new optimization: inside a named function any reference to a
2212 function with the same name is considered to be a self-reference;
2213 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2214 optimization quality.
2215 * new optimization on x86: logical functions and + now have
2216 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2217 used when the result is truncated to 32 bits.
2218 * VALUES declaration is partially enabled.
2219 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2220 * fixed some bugs revealed by Paul Dietz' test suite:
2221 ** The system now obeys the constraint imposed by
2222 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2223 types form a lattice under type intersection.
2224 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2225 ** ASSOC now ignores NIL elements in an alist.
2226 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2227 and (1+ MOST-POSITIVE-FIXNUM) answers.
2228 ** The addition of a method with invalid qualifiers to a generic
2229 function does not cause an error to be signalled immediately;
2230 a warning is signalled, and the error is generated only on
2231 calling the generic function.
2232 * changed .fasl file version number, in part to add type codes for
2233 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2234 obscure ANSI requirements
2236 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2237 * fixed bug 148: failure to inline-expand a local function left
2238 garbage, confusing the compiler.
2239 * fixed bugs 3cd: structure slot readers perform type check if the
2240 slot can have an invalid value (i.e. it is either not initialized
2241 or can be written with a less specific slot writer).
2242 * bug fix: the compiler now traps array references to elements off
2243 the end of an array; previously, the bounds checking in some
2244 circumstances could go off-by-one.
2245 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2247 * type declarations for array element types now obey the description
2248 on the CLHS page "Declaration TYPE", as per discussions on
2249 sbcl-help around 2003-05-08. This means that a declaration
2250 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2251 declaration, all references to BAR will be asserted or assumed
2252 (with THE, so dependent on compiler policy) to involve objects of
2253 type FOO. Note that no such declaration is implied in
2254 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2255 * declared types of functions from the "Conditions"
2256 chapter. (reported by Paul Dietz)
2257 * bug fix: CERROR accepts a function as its first argument.
2258 * bug fix: NTH an NTHCDR accept a bignum as index
2259 arguments. (reported by Adam Warner)
2260 * optimization: character compare routines now optimize comparing
2261 against a constant character. (reported by Gilbert Baumann)
2262 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2263 argument now works properly on the MIPS platform.
2264 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2265 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2266 declared to have a functional type, does not cause an error
2268 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2269 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2270 * fixed bug: if last continuation of a deleted block has a
2271 destination, this destination should be deleted too. (reported by
2273 * fixed a bug in the bootstrap process: the host compiler's values
2274 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2275 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2276 test case from Patrik Nordebo)
2277 * improved the ability of the disassembler on the PPC platform to
2278 provide helpful disassembly notes.
2279 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2280 the class in more cases than previously.
2281 * bug fix: FILE-POSITION now understands :START and :END for
2282 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2283 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2284 8) data. (thanks to David Lichteblau)
2285 * bug fix: it is possible to add a method to a generic function
2286 without lambda list.
2287 * bug fix: reader failed to signal END-OF-FILE inside an
2288 object representation. (reported by Nikodemus Siivola)
2289 * fixed some bugs revealed by Paul Dietz' test suite:
2290 ** LAST and [N]BUTLAST should accept a bignum.
2291 ** condition slot accessors are methods.
2292 ** (VECTOR NIL) is a subtype of STRING.
2294 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2295 * minor incompatible change: some nonsensical specialized lambda
2296 lists (used in DEFMETHOD) which were previously ignored now signal
2298 * minor incompatible change: the system is now aware of the types of
2299 variables in the COMMON-LISP package, and will signal errors for
2300 most violations of these type constraints (where previously they
2301 were silently accepted).
2302 * minor incompatible change: COMPILE-FILE now uses the freedom
2303 afforded (ANSI 3.2.2.3) to use derived function types for
2304 functions defined in the same file. This also permits the system
2305 to warn on static type mismatches and function
2306 redefinition. (Currently it does not work with high DEBUG level.)
2307 * minor incompatible change: VALUES declaration is disabled.
2308 * When issuing notes, the compiler now signals a condition of type
2309 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2310 restart for use in user handlers. It is expected that the
2311 COMPILER-NOTE condition will eventually become a condition
2312 supertype to a hierarchy of note types, which will then be
2313 handleable in a similar fashion. However, at the moment, no such
2314 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2315 but it's an implementation detail, not a classification for the
2317 * Changes in type checking closed the following bugs:
2318 ** type checking of unused values (192b, 194d, 203);
2319 ** template selection based on unsafe type assertions (192c, 236);
2320 ** type checking in branches (194bc).
2321 * A short form of VALUES type specifier has ANSI meaning (it has
2322 increased the number of situations when SBCL cannot perform type
2324 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2325 keywords or constants is permissible.
2326 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2327 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2328 classes. (thanks to Antonio Martinez)
2329 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2330 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2331 argument to a pretty-print stream. Instead, it leaves any such
2332 operations to DESCRIBE-OBJECT methods.
2333 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2335 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2336 lambda lists. (thanks to David Lichteblau)
2337 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2338 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2339 (thanks to Gerd Moellmann)
2340 * fixed bug 246: increased compilation speed of long
2341 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2343 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2345 * DEFCONSTANT now throws a condition of type
2346 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2347 constant to a non-EQL value; CONTINUE and ABORT restarts
2348 respectively change and preserve the value.
2349 * fixed bug 63: The code walker, part of the implementation of CLOS,
2350 is now better at handling symbol macros.
2351 * bug fix: There is no longer an internal implementation type named
2352 CL:LENGTH. (reported by Raymond Toy)
2353 * bug fix: In macro-like defining macros/special operators the
2354 implicit block does not enclose the lambda list.
2355 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2356 enclosed in parentheses) are not suitable as type specifiers, and
2357 their use properly signals an error now.
2358 * bug fix: An argument count mismatch for a type specifier in code
2359 being compiled no longer causes an unhandled error at compile
2360 time, but signals a compile-time warning.
2361 * fixed simple vector readable printing
2362 * bug fix: DESCRIBE takes more care over whether the class
2363 precedence list slot of a class is bound before accessing it.
2364 (reported by Markus Krummenacker)
2365 * bug fix: FORMATTER can successfully compile pretty-printer format
2366 strings which use variants of the ~* directive inside.
2367 * bug fix: SEARCH now applies its TEST predicate to the elements of
2368 the arguments in the correct order. (thanks to Wolfhard Buss)
2369 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2370 in type inference. (thanks to Robert E. Brown)
2371 * bug fix: Reading in symbols with an explicit package name of ""
2372 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2373 * fixed some bugs revealed by Paul Dietz' test suite:
2374 ** NIL is now allowed as a structure slot name.
2375 ** Arbitrary numbers, not just REALs, are allowed in certain
2376 circumstances in LOOP for-as-arithmetic clauses.
2377 ** Multiple class redefinitions before slot access no longer
2378 causes a type error.
2379 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2380 association between the name and a class.
2381 ** Generic functions with non-standard method-combination and over
2382 five methods all of which return constants no longer return NIL
2383 after the first few invocations. (thanks to Gerd Moellmann)
2384 ** CALL-NEXT-METHOD with no arguments now passes the original
2385 values of the arguments, even in the presence of assignment.
2386 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2388 ** DISASSEMBLE works with closures and funcallable instances.
2389 ** ADD-METHOD now returns the generic function, not the new method.
2390 ** FIND-METHOD signals an error if the lengths of the specializers
2391 is incompatible with the generic function, even if the ERRORP
2393 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2394 which its argument is a member.
2395 ** DEFCLASS only redefines the class named by its class-name
2396 argument if that name is the proper name of the class;
2397 otherwise, it creates a new class.
2398 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2399 of the UNBOUND-SLOT condition to the name of the slot.
2400 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2401 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2402 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2403 SLOT-MAKUNBOUND in the specified fashion.
2405 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2406 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2407 cross-compilation host. As a consequence, we can now bootstrap our
2408 way up to SBCL starting with a bare gcc toolchain and human-readable
2409 source code (first the source to CLISP, then the source to SBCL).
2410 * A contributed module containing a partial implementation of the
2411 simple-streams interface has been included. (thanks to Rudi
2413 * A contributed module implementing the RFC1321 Message Digest
2414 Algorithm, known as MD5, has been included.
2415 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2416 no longer has any effect, as the code controlled by this feature
2417 has been deleted. (As far as we know, no-one has ever built using
2418 this feature, and its semantics were confused in any case).
2419 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2420 work on conditions (as required by the ANSI specification),
2421 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2422 expected behaviour on conditions. Users should note, however,
2423 that such behaviour is not required by the ANSI specification,
2424 and so use of this behaviour may render their code unportable.
2425 * fixed some bugs revealed by Paul Dietz' test suite:
2426 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2428 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2429 ** SXHASH on condition objects no longer returns NIL.
2430 ** :ALLOCATION :CLASS slots are better treated; their values are
2431 updated on class redefinition, and initforms inherited from
2432 superclasses are applied.
2433 ** REMOVE-METHOD returns its generic function argument even when
2434 no method was removed.
2435 ** SHARED-INITIALIZE now initializes the values of the requested
2436 slots, including those with :ALLOCATION :CLASS.
2437 ** ALLOCATE-INSTANCE now works on structure classes defined via
2438 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2440 ** SLOT-EXISTS-P now works on conditions, as well as structures
2442 ** MAKE-LOAD-FORM now has the required methods on
2443 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2444 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2445 keyword, and now implements the SLOT-NAMES argument.
2446 ** methods with &OPTIONAL arguments no longer allow too many
2447 arguments to be passed in the call without error.
2448 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2449 option is consistent with the required arguments of the generic
2450 function lambda list.
2451 * bug fix: REQUIRE accepts a string designator. (Thanks to
2453 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2454 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2455 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2456 * bug fix: The compiler now checks for duplicated variables in macro
2458 * bug fix: SETQ on globals returns the correct value.
2459 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2460 not a COMPILER-ERROR (followed by some other strange error on
2461 choosing the CONTINUE restart).
2462 * bug fix: make.sh and friends are now more consistent in the way that
2463 they look for GNU "make".
2465 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2466 * experimental native threads support (on x86 Linux >=2.4 only).
2467 This is not compiled in by default: you need to add :SB-THREAD to
2468 the target features. See the "Beyond ANSI" chapter of the manual
2470 * fix for longstanding nonANSIism: The old distinction between
2471 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2472 The return value from CL:FIND-CLASS is now a CLOS class, and
2473 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2474 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2475 * An interface to the MetaObject Protocol, as described in Kiczales,
2476 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2477 MIT Press, 1991, is available from the SB-MOP package.
2478 * incompatible change: the SB-PCL package should now be considered
2479 a private implementation detail, and no longer a semi-private MOP
2481 * minor incompatible change: due to rearrangement for threads, the
2482 control stack and binding stack are now allocated at arbitrary
2483 addresses instead of being hardcoded per-port. Users affected by
2484 this probably have to be doing advanced things with shared
2485 libraries, and will know who they are.
2486 * minor incompatible change: Previously, all --eval forms used were
2487 processed with READ before any of them were processed with EVAL.
2488 Now each --eval form is processed with both READ and EVAL before
2489 the next --eval form is processed. (Thus package operations like
2490 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2491 work as the user might reasonably expect.)
2492 * minor incompatible change: *STANDARD-INPUT* is now only an
2493 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2495 * minor incompatible change: Y-OR-N-P is now character-oriented,
2496 not line oriented. Also, YES-OR-NO-P now works without errors.
2497 (thanks to Antonio Martinez)
2498 * sb-aclrepl module improvements: an integrated inspector, added
2499 repl features, and a bug fix to :trace command.
2500 * Known functions, which cannot be open coded by the backend, are
2501 considered to be able to check types of their arguments. (fixing
2502 a bug report by Nathan J. Froyd)
2503 * fixed a bug in computing method discriminating functions: It is
2504 now possible to define methods specialized on classes which have
2505 forward-referenced superclasses. (thanks to Gerd Moellmann)
2506 * fixed evaluation order in optional entries (reported by Gilbert
2508 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2509 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2510 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2511 required-by-AMOP NAME argument, as well as CLASS and
2512 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2513 * fixed bug 20: DEFMETHOD can define methods using names that are
2514 not the proper names of classes to designate class specializers.
2515 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2516 Unix file descriptors, instead of blowing up. (thanks to Antonio
2518 * Garbage collection refactoring: user-visible change is that a
2519 call to the GC function during WITHOUT-GCING will not do garbage
2520 collection until the end of the WITHOUT-GCING. If you were doing
2521 this you were probably losing anyway.
2522 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2523 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2524 TYPEP the latter but not the former.
2525 * The compiler issues a full WARNING for calls to undefined functions
2526 with names from the CL package.
2527 * MAP-INTO for a vector destination is open coded. (reported by
2528 Brian Downing on c.l.l)
2529 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2530 documentation string.
2531 * fixed some bugs revealed by Paul Dietz' test suite:
2532 ** COPY-ALIST now signals an error if its argument is a dotted
2534 ** Condition slots are now accessed more correctly in the presence
2535 of multiple initargs for a given slot.
2536 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2537 exclude restarts of the same name associated with a different
2539 ** DEFCLASS of forward-referenced classes with another
2540 forward-referenced class in the superclasses list no longer
2542 ** Condition slots are now initialized once each, not multiple
2543 times. (thanks to Gerd Moellmann)
2544 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2546 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2547 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2548 arguments contain duplicated elements.
2549 ** RESTART-CASE understands local macros.
2550 ** RESTART-CASE associates exactly its own restarts with a condition.
2551 ** ENDP in safe mode checks its argument to be of type LIST.
2552 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2553 in question is unbound.
2554 ** Optimized MAKE-INSTANCE functions no longer cause internal
2555 assertion failures in the presence of duplicate initargs.
2556 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2557 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2558 propagated even in the ctor optimized implementation of
2560 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2562 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2563 * a better implementation of SXHASH on (simple) bit vectors,
2564 measured both in execution speed and in distribution of results
2565 over the positive fixnums, has been installed. Likewise, a better
2566 implementation of EQUAL for simple bit vectors is now available.
2567 * fixed CEILING optimization for a divisor of form 2^k.
2568 * fixed bug 240 (emitting extra style warnings "using the lexical
2569 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2570 by Antonio Martinez)
2571 * fixed SXHASH, giving different results for NIL depending on type
2572 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2573 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2574 variable. (found by Rolf Wester)
2575 * SBCL does not ignore type declarations for special
2576 variables. (reported by rif on c.l.l 2003-03-05)
2577 * some bug fixes in contrib/sb-aclrepl/
2578 * fixed some bugs revealed by Paul Dietz' test suite:
2579 ** a bug in the CONS type specifier, whereby the CAR and CDR
2580 types got intertwined, has been fixed;
2581 ** the type system is now able to reason about the interaction
2582 between INTEGER and RATIO types more completely;
2583 ** APPEND, [N]REVERSE and NRECONC check that those their
2584 arguments, which must be proper lists, are really so;
2585 ** An array specialized to be unable to hold elements has been
2586 implemented, as required -- yes, really -- by ANSI;
2587 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2588 on malformed property lists;
2590 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2591 * incompatible packaging change: in line with Unix convention,
2592 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2593 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2594 * REQUIRE and PROVIDE are now optionally capable of doing something
2595 useful. See the documentation string for REQUIRE.
2596 * infrastructure for a managed SBCL contrib system: contributed
2597 modules in this release include:
2598 ** the ASDF system definition facility;
2599 ** an interface to the BSD Sockets API;
2600 ** an ACL-like convenience interface to the repl;
2601 (thanks to Kevin Rosenberg)
2602 ** an implementation of ROTATE-BYTE, with efficient implementation
2604 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2605 gives a closer approximation to the right answer than previously.
2606 (thanks to Raymond Toy)
2607 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2608 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2609 optional environment arguments, as required by ANSI.
2610 * fixed bugs in other functions taking environment objects, allowing
2611 calls with an explicit NIL environment argument to be compiled
2613 * fixed bug 228: primary return values from
2614 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2615 COMPILE or FUNCTION.
2616 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2617 :TYPE will no longer signal errors on innocuous objects.
2618 * fixed bug 231b: SETQ is better at respecting type declarations in
2619 the lexical environment.
2620 * fixed a bug in DEFCLASS: classes named by symbols with no or
2621 unprintable packages can now be defined.
2622 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2623 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2624 * fixed compiler failure related to checking types of functions.
2625 (reported by Robert E. Brown)
2626 * the compiler is now much more consistent in its error-checking
2627 treatment of bounding index arguments to sequence functions: in
2628 (SAFETY 3) code, errors will be signalled in almost all cases if
2629 invalid sequence bounding indices are passed to functions defined
2630 by ANSI to operate on sequences.
2631 * fixed a bug in the build procedure: documentation of SBCL-specific
2632 packages is now preserved and available in the final Lisp image.
2633 * lifted FDEFINITION lookup out of loops in the implementation of
2634 many list operations. (thanks to Robert E. Brown)
2635 * fixed a bug in the reader: the #n# reader macro now works for
2636 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2637 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2638 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2639 E. Brown and Tony Martinez respectively)
2640 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2641 SLOT-BOUNDP: in particular, it is now able to optimize them much
2642 better, and is now not vulnerable to having packages renamed.
2643 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2644 always called when a slot is not present in an instance. (thanks
2646 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2647 longer signal an error on unreading a character following EOF on
2648 the previous constituent stream. (thanks to Tony Martinez)
2649 * fixed some bugs revealed by Paul Dietz' test suite:
2650 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2651 not just nonnegative fixnums;
2652 ** the logical bit-array operators such as BIT-AND now accept an
2653 explicit NIL for their "opt-arg" argument (to indicate a
2654 freshly-consed result bit-array);
2655 ** ELT now signals an error on an invalid sequence index in safe
2657 ** the type system is now cleverer about negations of numeric
2658 types, and consequently understands the BIGNUM and RATIO types
2660 ** the type system is now cleverer about the interaction between
2661 INTEGER and RATIO types: while bugs still remain, many more
2662 cases are accurately computed;
2663 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2664 if it is in the last clause;
2665 ** CONSTANTLY now correctly returns a side-effect-free function in
2667 ** DECLARE is no longer treated as a special-operator; in
2668 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2669 * incremented fasl file version number due to the change in the
2670 DEFSTRUCT-SLOT-DESCRIPTION structure.
2672 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2673 * minor incompatible change: code processed by the "interpreter" or
2674 EVAL now has a compilation optimization policy of (DEBUG 2)
2675 (changed from (DEBUG 1)) to improve debuggability of interactive
2676 development, and to allow the use of the debug RETURN command in
2678 * an experimental implementation of the RETURN command for the
2679 debugger has been included. (thanks to Frederik Kuivinen)
2680 * fixed bug 62: constraints were not propagated into a loop.
2681 * fixed bug in embedded calls of SORT (reported and investigated by
2683 * fixed some bugs revealed by Paul Dietz' test suite:
2684 ** printing and reading of arrays with some dimensions having
2685 length 0 (thanks to Gerd Moellmann);
2686 ** BOA constructor with &AUX argument without a default value does
2687 not cause a type error;
2688 ** CONSTANTP now returns true for all self-evaluating objects.
2690 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2691 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2692 accessors that are related by inheritance, as specified in the
2693 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2694 Valtteri Vuorikoski)
2695 * The compiler is now able to inline functions that were defined in
2696 a complex lexical environment (e.g. inside a MACROLET).
2697 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2699 * Support for the upcoming FreeBSD-5.0 release has been included.
2700 (thanks to Dag-Erling Smorgrav)
2701 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2702 effect when it is not in a toplevel context.
2703 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2704 stand a better chance of being correct. (thanks to Gerd
2706 * fixed bug in COERCE, which now signals an error on coercing a
2707 rational to a bounded real type which excludes the expected
2709 * The compiler is now able to derive types more accurately from the
2710 COERCE and COMPILE functions.
2711 * fixed bug 223: functional binding is considered to be constant
2712 only for symbols in the CL package.
2713 * fixed bug 231: SETQ did not check the type of a variable being set
2714 (reported by Robert E. Brown)
2715 * A new optimization for MAKE-INSTANCE has been included, fixing
2716 various bugs (including relating to :ALLOCATION :CLASS slots and
2717 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2719 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2720 ** As required by ANSI, LOOP now disallows anonymous collection
2721 clauses such as COLLECT I in conjunction with aggregate boolean
2722 clauses such as THEREIS (= I 1);
2723 ** LOOP now signals an error when any variable is reused in the
2724 same loop (including the potentially useful construct analogous
2725 to WITH A = 1 WITH A = (1+ A);
2726 ** IT is only a special loop symbol within the first clause of a
2727 conditional loop clause;
2728 ** LOOP with a typed iteration variable over a hashtable now
2729 signals a type error iff it should.
2730 * fixed some other bugs revealed by Paul Dietz' test suite:
2731 ** FILE-STREAM now names the class previously known as FD-STREAM;
2732 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2733 argument) no longer signals an error;
2734 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2735 of the slot symbol, rather than using the current package
2736 ((:CONC-NAME "") continues to intern the slot's name in the
2738 * incremented fasl file version number, because of the incompatible
2739 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2740 of the new implementation of DEFINE-COMPILER-MACRO.
2742 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2743 * Support for building SBCL for MIPS platforms running in
2744 little-endian mode has now been checked in, and basic
2745 functionality on said platforms verified.
2746 * minor incompatible change: PCL now records the pathname of a file
2747 in which methods and the like are defined, rather than its
2749 * minor incompatible change: TRUENAME now considers the truename of
2750 a file naming a directory to be the pathname with :DIRECTORY
2751 component indicating that directory.
2752 * minor incompatible change: a NAMED clause in the extended form of
2753 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2754 reason for the previous behaviour is unclear.
2755 * more systematization and improvement of CLOS and MOP conformance
2756 in PCL (thanks to Gerd Moellman and Pierre Mai):
2757 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2759 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2760 primary methods with no specializers;
2761 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2763 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2764 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2765 FORWARD-REFERENCED-CLASSes; error reporting on
2766 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2768 ** SXHASH on CLOS instances now uses a slot internal to the
2769 instance to return different numbers on distinct instances,
2770 while preserving the same return value through invocations of
2772 ** DEFMETHOD signals errors when methods with longer incongruent
2773 lambda lists are added to generic functions;
2774 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2775 CLASS, as specified in AMOP;
2776 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2777 on the order returned by the primary method for classes of
2778 class STANDARD-CLASS;
2779 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2780 * fixed some bugs shown by Paul Dietz' test suite:
2781 ** DOLIST puts its body in TAGBODY;
2782 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2784 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2785 value producing form;
2786 ** if more variables are given to PROGV than values, extra
2787 variables are bound and made to have no value;
2788 ** NSUBSTITUTE on list arguments gets the right answer with
2790 ** ELT signals an error of type TYPE-ERROR when the index argument
2791 is not a valid sequence index;
2792 ** LOOP signals (at macroexpansion time) an error of type
2793 PROGRAM-ERROR when duplicate variable names are found;
2794 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2795 ** FUNCALL of special-operators now cause an error of type
2797 ** PSETQ now works as required in the presence of side-effecting
2798 symbol-macro places;
2799 ** NCONC accepts any object as its last argument;
2800 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2802 ** loop-for-as-package does not require a package to be explicitely
2804 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2806 * fixed bug 166: compiler preserves "there is a way to go"
2807 invariant when deleting code.
2808 * fixed bug 172: macro lambda lists with required arguments after
2809 &REST arguments now cause an error to be signalled. (thanks to
2811 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2812 bug. (thanks to Matthew Danish)
2813 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2815 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2816 arguments are lexically rebound. (thanks to Gerd Moellmann and
2818 * fixed bug 194: error messages are now more informative when there
2819 is no primary method applicable in a call to a generic
2820 function. (thanks to Gerd Moellmann)
2821 * fixed bug in command line argument checking (thanks to Julian
2823 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2824 (thanks to Matthew Danish)
2825 * incremented fasl file version number, because of the
2826 SXHASH-related changes in the layout of CLOS data structures
2828 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2829 * minor incompatible change: The runtime (the Unix executable named
2830 "sbcl") is now much pickier about the .core files it will load.
2831 Essentially it now requires .core files to descend from the same
2832 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2833 as the runtime does. (The intent is to prevent the crashes which
2834 can occur, and which can even be reported as mysterious failures,
2835 when people patch the sources or change the build parameters
2836 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2837 sbcl and .core files.)
2838 * fixed bug: VALUES-LIST is no longer optimized away.
2839 * fixed bug 142: The FFI conversion of C string values to Lisp
2840 string values no longer conses excessively. (thanks to Nathan
2841 Froyd porting Raymond Toy's fix to CMU CL)
2842 * began to systematize and improve MOP conformance in PCL (thanks to
2843 Nathan Froyd, Gerd Moellman and Pierre Mai):
2844 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2846 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2847 ** FINALIZE-INHERITANCE is now called on class finalization;
2848 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2849 argument precedence order.
2850 * fixed bug 202: The compiler no longer fails on functions whose
2851 derived types contradict their declared type.
2852 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2853 so it can be non-toplevel.
2854 * The fasl file version number has changed (because of the new
2855 implementation of DEFMACRO).
2856 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2857 safe code, that any length requirement by their type-specifier
2858 argument is valid. The exceptions to this are described in bug
2860 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2861 functions now check the types of their inputs as required by ANSI.
2862 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2863 introduced symbol is DECLAREd to be SPECIAL.
2864 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2865 * fixed bug: :COUNT argument to sequence functions may be negative
2866 * fixed bug: body of DO-SYMBOLS may contain declarations
2867 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2868 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2869 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2870 symbol macro only once
2871 * fixed printing of call frame when argument list is unavailable
2872 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2873 * compiler no longer signals WARNING on unknown keyword
2876 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2877 * A beta-quality port to the mips architecture running Linux,
2878 based on the old CMUCL backend, has been made. It has been tested
2879 on a big-endian kernel, and works sufficiently well to be able to
2880 rebuild itself; it has not been tested in little-endian mode.
2881 * fixed an inconsistency between gencgc.c and purify.c which made
2882 dumping/loading .core files unreliable
2883 * fixed bug 120a: The compiler now deals correctly with IFs where
2884 the consequent is the same as the alternative, instead of
2885 misderiving the return type. (thanks to Alexey Dejneka)
2886 * fixed bug 113: Logical pathnames are now dumpable (the logical
2887 host is resolved at load-time, throwing an error if it is not
2889 * fixed bug 174: FORMAT's error message is slightly clearer when a
2890 non-printing character is used in a format directive.
2891 * fixed several bugs in compiler checking of type declarations, i.e.
2892 violations of the Python "declarations are assertions" principle
2893 (thanks to Alexey Dejneka)
2894 * fixed several bugs in PCL's error checking (thanks to Gerd
2896 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2898 * fixed bug in compilation of functions as first class values
2899 (thanks to Antonio Martinez-Shotton)
2900 * The compiler's handling TYPE-ERRORs which it can prove will
2901 inevitably happen at runtime has been cleaned up and corrected
2902 in several ways. (thanks to Alexey Dejneka)
2903 * improved argument type checking for various basic arithmetic
2904 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2905 lavished on them in the past that they can be compiled in many
2906 ways in different special cases
2907 * fixed bug 181: compiler checks validity of user supplied type
2909 * cleaned up code flushing in optimization: Function calls which
2910 should signal errors for safety purposes (e.g. which ANSI says
2911 should signal errors when their arguments are of incorrect type)
2912 are no longer optimized away.
2913 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2914 * incremented fasl file version number, because changes in the
2915 implementation of sequence functions like COERCE caused
2916 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2917 in old inline expansions) to become undefined. (Actually these
2918 changes were later undone, so we might very well be binary
2919 compatible with 0.7.7 after all, but leaving the version number
2920 incremented seemed like the simplest and most conservative
2923 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2924 * An alpha-quality port to the parisc architecture running Linux,
2925 based on the old CMUCL backend, has been made. This, even more so
2926 than the other backends, should be considered still a work in
2927 progress; known problems include that the Linux kernel in 64-bit
2928 mode does not propagate the correct sigcontext structure to
2929 userspace, and consequently SBCL on a parisc64 kernel will not
2931 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2932 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2933 suppressed.) Also now that the compiler is looking at declarations
2934 in the environment, it checks optimization declarations as well,
2935 and suppresses inlining when (> DEBUG SPEED).
2936 * More fixes have been made to treatment of floating point exception
2937 treatment and other Unix signals. In particular, floating point
2938 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2939 * The detection and handling of control stack exhaustion (infinite
2940 or very deeply nested recursion) has changed. Stack exhaustion
2941 detection is now done by write-protecting pages at the OS level
2942 and applies at all optimization settings; when found, a
2943 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2944 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2945 be caught using IGNORE-ERRORS.
2946 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2947 correctly when transforming recursive LABELS functions to LETs.
2948 (thanks to Alexey Dejneka)
2949 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2950 that are names of constants or global variables.
2951 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2952 alien routines with docstrings.
2953 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2954 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2956 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2957 (thanks to Eric Marsden)
2958 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2959 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2960 * Bug fix: The SPARC backend can now compile functions involving
2961 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2962 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2963 to a functions expecting a PATHNAME-DESIGNATOR.
2964 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2965 lambda lists. (thanks to Alexey Dejneka)
2966 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2967 functions so that EXPORTing the name of the class doesn't cause
2968 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2969 to Antonio Martinez for reporting this)
2970 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2971 constant-folded without failing an assertion. (thanks to Einar
2972 Floystad Dorum for reporting this)
2973 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2974 types are now dealt with more correctly by the compiler.
2975 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2976 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2977 behaviour analogous to RENAME-FILE. This puts its behaviour more
2978 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2980 * The fasl file version number has changed again. (because of the
2981 bug fix involving the names of PCL MAKE-INSTANCE functions)
2983 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2984 * bug fix: Floating point exceptions are treated much more
2985 consistently on the x86/Linux and PPC/Linux platforms.
2986 * Array initialization with :INITIAL-ELEMENT is now much faster for
2987 cases when the compiler cannot open code the array creation, but
2988 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2989 array accesses have also seen a speed increase.
2990 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2991 specified. (This was at the root of some bad interactions between
2992 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2994 * bug fix: Internal error arguments for undefined functions are now
2995 computed correctly on the PPC/Linux platform.
2996 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2997 Raymond Toy's patch for CMU CL)
2998 * Support for the Solaris 9 operating environment has been included
2999 (thanks to Daniel Merritt)
3000 * A very ugly but hopefully complete draft of the missing FFI chapter
3001 of the manual has been created by reformatting the corresponding
3002 CMU CL manual chapter into (currently very ugly and incoherent)
3003 DocBook and bringing it up to date for SBCL behavior. Thus, the
3004 manual is now essentially complete, at least by my extreme
3005 once-and-only-once standards, whereby it's acceptable to refer to
3006 the doc strings of SB-EXT functions as the primary documentation.
3007 * The fasl file version number has changed again, due to cleanup of
3008 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
3009 is no longer a static symbol.)
3011 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
3012 * SBCL now builds with OpenMCL (version 0.12) as the
3013 cross-compilation host; also, more progress has been made toward
3014 bootstrapping under CLISP.
3015 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3017 * bug 158 fixed: The compiler can now deal with integer loop
3018 increments different from 1; fixing this turned out also to fix
3020 * bug 169 fixed: no more bogus warnings about using lexical bindings
3021 despite the presence of perfectly good SPECIAL declarations (thanks
3022 to David Lichteblau)
3023 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3024 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3025 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3027 * bug 180 fixed: Method combination specifications no longer ignore
3028 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3029 * bug fix: Structure type predicate functions now check their argument
3030 count as they should.
3031 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3032 correctly. (thanks to Pierre Mai)
3033 * minor incompatible change: The --noprogrammer option is deprecated
3034 in favor of the new --disable-debugger option, which is very similar.
3035 (The major difference is that it takes effect at a slightly different
3036 time at startup, causing handling of errors in --sysinit and
3037 --userinit files will be affected differently.) The
3038 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3039 been added to allow this functionality to be controlled from ordinary
3040 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3041 maintainers, who might want to run non-interactive scripts to
3042 build SBCL cores which will later be used interactively.)
3043 * minor incompatible change: The LOAD function no longer, when given
3044 a wild pathname to load, loads all files matching that pathname.
3045 Instead, an error of type FILE-ERROR is signalled.
3047 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3048 * bug 147 fixed: The compiler preserves its block link/count
3049 invariants more correctly now so that it doesn't crash. (thanks
3051 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3053 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3054 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3055 no-value-for-this-character as NIL (as specified by ANSI).
3056 * HOST-NAMESTRING on physical pathnames now returns a string that is
3057 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3058 (thanks to Christophe Rhodes)
3059 * The Alpha port handles icache flushing more correctly. (thanks to
3061 * More progress has been made toward bootstrapping under CLISP. (thanks
3062 to Christophe Rhodes)
3063 * The fasl file format has changed again, because dynamic loading
3064 on OpenBSD (which has non-ELF object files) motivated some cleanups
3065 in the way that foreign symbols are transformed and passed around.
3066 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3067 is no longer treated as whitespace by the reader, but instead as
3068 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3069 |A
\7fB|, instead of A as it used to.
3071 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3072 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3073 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3074 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3075 this as of 0.7.1.45, but was left out of the previous news section
3076 (thanks to Dan Barlow)
3077 * SBCL now runs on the Solaris operating system on SPARC architectures
3078 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3079 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3080 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3081 * SPARC backend cleanups, allowing builds of cores optimized for V8
3082 and V9 SPARCS, and also emission of code targeted to a particular
3083 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3085 * SBCL is closer to bootstrapping under CLISP, thanks to various
3086 fixes by Christophe Rhodes.
3087 * The fasl file format has changed again, to allow the compiler's
3088 INFO database to support symbol macros.
3089 * The user manual (in doc/) is formatted into HTML more nicely.
3090 (thanks to coreythomas)
3091 * The system is smarter about SUBTYPEP relationships, especially
3092 those involving NOT types (including types such as ATOM which are
3093 represented internally using NOT types). Thus SUBTYPEP is less
3094 likely to return (VALUES NIL NIL) in general, and in particular
3095 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3097 * The fasl file format has changed again, because the internal
3098 representation of types now includes a new slot to support the new
3099 SUBTYPEP-of-NOT-types logic.
3100 * (not a change in the main branch of SBCL, but a related prototype
3101 which can hopefully be merged into the main branch of SBCL in the
3102 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3103 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3104 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3105 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3106 dispatch macro character combinations. (thanks to Alexey Dejneka)
3107 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3108 that superficially look like logical namestrings correctly.
3109 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3111 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3112 default supplied-p) for &optional and &key arguments. (thanks to
3115 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3116 * incompatible change: The compiler is now less aggressive about
3117 tail call optimization, doing it only when (> SPACE DEBUG) or
3118 (> SPEED DEBUG). (This is an incompatible change because there are
3119 programs which relied on the old CMU-CL-style behavior to optimize
3120 away their unbounded recursion which will now die of stack overflow.)
3121 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3122 for non-GENCGC systems has been increased to 20M (since that
3123 seems much closer to the likely performance optimum for modern
3124 systems than the old 4M value was)
3125 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3126 and *DEBUG-PRINT-LEVEL*
3127 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3128 of CMU CL's support for SPARC, and various endianness and other
3129 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3130 * new syntactic sugar for the Unix command line: --load foo.bar is now
3131 an alternate notation for --eval '(load "foo.bar")'.
3133 ** The system now detects stack overflow and handles it gracefully,
3134 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3135 optimization settings. (This is a good thing in general, and
3136 its introduction in this version should be particularly timely
3137 for anyone whose code fails because of suppression of tail
3139 ** The system now hunts for the C variable "environ" in a more
3140 devious way, to avoid segfaults when the C library version differs
3141 between compile time and run time. (thanks to Christophe Rhodes)
3142 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3143 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3144 ** The compiler no longer issues bogus style warnings for undefined
3145 classes in the same source file as the DEFCLASSes which defined
3146 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3148 ** fixes in CONDITION class precedence list for undefined function
3149 errors (thanks to Alexei Dejneka)
3150 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3151 correctly. (thanks to Dan Barlow)
3152 ** portability fixes aiming at bootstrapping under CLISP (thanks
3153 to Dave McDonald and Christophe Rhodes)
3154 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3155 ** fixes in type translation and and type inference (thanks to
3157 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3158 ** various fixes in the new ports (thanks to Dan Barlow)
3159 * several changes related to debugging:
3160 ** suppression of tail recursion, as noted above
3161 ** stack overflow detection, as noted above
3162 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3163 is now the default. (For some time encapsulation has been more
3164 reliable than the breakpoint-based :ENCAPSULATE NIL
3165 implementation, at least on X86 systems; and I just noticed that
3166 encapsulation also seems closer to the spirit of the ANSI
3169 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3171 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3172 up properly again. (There was a packaging bug in 0.7.0 which
3173 left their definitions in SB-SYS::LOAD-FOREIGN and
3174 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3175 vital for most things which interface to C-level interfaces,
3176 like extensions working with sockets or databases or
3177 Perl-compatible regexes or whatever, and the need to fix
3178 this bug is the main reason that 0.7.1 was released so
3180 ** DEFGENERIC is now choosier about the methods it redefines, so that
3181 reLOADing a previously-LOADed file containing DEFGENERICs does
3182 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3183 cycle now works as it should. (thanks to Alexey Dejneka)
3184 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3185 fixed by Christophe Rhodes. (He actually submitted this patch
3186 months ago, and I delayed until after 0.7.0.)
3187 ** Bug 111 (internal compiler confusion about runtime checks on
3188 FUNCTION types) was fixed by Alexey Dejneka.
3189 * Some internal cleanups (getting rid of variables which aren't
3190 needed now that the byte interpreter is gone) caused the fasl
3191 file format number to change again.
3193 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3194 * major incompatible change: The default fasl file extension, i.e. the
3195 default extension for files produced by COMPILE-FILE, has changed
3196 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3198 ** There are many changes in the implementation of the compiler.
3199 SBCL is now essentially a compiler-only implementation of ANSI
3200 Common Lisp. EVAL still "interprets" a few special cases, but
3201 almost all the interesting cases are handled by creating
3202 a LAMBDA expression, calling COMPILE on it, then calling
3203 FUNCALL on the result.
3204 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3205 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3206 Since the code is newer, there might still be some new bugs
3207 (though not as many as before Martin Atzmueller's fixes:-). But
3208 the new code is substantially simpler and clearer, and hopefully
3209 any remaining bugs will be simpler, less fundamental, and more
3210 fixable then the bugs in the old code.
3211 ** The revised compiler is still a little unsteady on its feet.
3213 *** The debugging information it produces (particularly the names
3214 of FUNCTION objects) is sometimes much less useful than what
3215 the old compiler produced.
3216 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3217 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3218 within a MACROLET) has been temporarily weakened.
3219 ** There are new compiler optimizations for various functions:
3220 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3221 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3222 *** the math functions TRUNCATE, FLOOR, and CEILING
3223 *** the function-of-all-trades COERCE
3224 Mostly these should be transparent, but there's one
3225 potentially-annoying problem (bug 117): when the compiler
3226 inline-expands a function and does type analysis on the result,
3227 it can create control paths which have type mismatches, and
3228 when it can't prove that those control paths aren't taken,
3229 it will issue WARNINGs about the type mismatches. This is
3230 a particular problem in practice for the new sequence functions.
3231 It's not clear how this should be fixed, and for now, a
3232 workaround is given in the entry for 117 in the BUGS file.
3233 ** (Because of the interaction between the two previous items --
3234 occasional inlining problems and new inline expansions -- some
3235 of the new sequence function optimizations won't really kick in
3236 completely until debugging information, and then inlining, are
3237 straightened out in some future version.)
3238 * minor incompatible changes:
3239 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3240 DIRECTORY no longer implicitly promotes NIL slots of its
3241 pathname argument to :WILD. In particular, when you ask for the
3242 contents of a directory (which you used to be able to do without
3243 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3244 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3245 ** changes in behavior that ANSI explicitly defines to be
3246 implementation dependent:
3247 *** The new compiler-only implementation still conforms with ANSI,
3248 but acts a little different than before. Besides the obvious
3249 changes in performance tradeoffs (that the cost per form passed
3250 to EVAL has gone up, and the cost per form executed by EVAL
3251 has gone down), the behavior of the system changes a little
3252 because there are no longer any interpreted function objects.
3253 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3254 e.g. doing COMPILE on the output of interactive DEFUN is
3256 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3258 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3259 has changed from (:CL) to NIL.
3260 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3261 e.g. #\Newline and #\Tab, don't have names specified in the
3262 ANSI Common Lisp standard, is now based on their ASCII symbolic
3263 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3264 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3265 are no longer used for output.
3266 ** changes in internal implementation constants:
3267 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3268 4 million. (If your application spends a lot of time GCing and
3269 you have a lot of RAM, you might want to experiment with
3270 increasing it even more.)
3271 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3272 However, almost all old code should still continue to work without
3273 immediate update, as SB-C-CALL is now a (deprecated) nickname
3275 ** Old operator names in the style DEF-FOO are now deprecated in
3276 favor of new corresponding names DEFINE-FOO, for consistency with
3277 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3278 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3279 internal symbols, but a few supported extensions like
3280 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3281 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3282 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3283 etc. as you get deeper into recursive calls to the debugger
3284 command loop, instead of the old "5]", "5]]", "5]]]"
3285 sequence. (I was motivated to do this when squabbles between
3286 ILISP and SBCL left me very deeply nested in the debugger. In the
3287 short term, this change will probably provoke more ILISP/SBCL
3288 squabbles, but hopefully it will be an improvement in the long run.)
3289 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3290 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3291 function, since declaiming return types involving aliens is
3292 (1) annoyingly messy to do by hand and (2) vital to efficient
3293 compilation of code which calls such functions.
3294 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3295 longer reexported by the SB-EXT package. They're solely useful
3296 for alien code, so it seems more logical that you should get
3297 them from the SB-ALIEN package, not in SB-EXT.
3298 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3299 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3300 features. Instead, the code that they used to control is always
3301 built into the system.
3302 * many other bug fixes
3303 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3304 advantage of the new EVAL-WHEN stuff and to clean them up in
3305 general, and they are now more ANSI-compliant in a number of
3306 ways. Martin Atzmueller is responsible for a lot of this.
3307 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3309 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3310 *** correct ERROR type for various file operations
3311 *** some fixes for Lisp streams
3312 *** DEFMETHOD syntax checking
3313 *** changing old weird representation of debug information as
3314 strings (which, among their other deficiencies, don't transform
3315 correctly when you rename packages, and don't change their
3316 print representation when you change things like *PACKAGE*
3317 and *PRINT-LENGTH*) to symbols and lists of symbols
3318 He also made several improvements and fixed several bugs in DESCRIBE.
3319 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3321 *** misbehavior of WRITE-STRING/WRITE-LINE
3322 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3323 and several other LOOP problems as well
3324 *** DIRECTORY when similar filenames are present
3325 *** DEFGENERIC with :METHOD options
3326 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3327 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3328 *** argument ordering in FIND with :TEST option
3329 *** mishandled package designator argument in APROPOS-LIST
3330 *** various problems in the backquote readmacro
3331 *** a bug in APROPOS
3332 *** probably some others that I'm not describing very well here,
3333 since the CVS log documents them by reference to sbcl-devel
3334 messages, and the SourceForge archives aren't working well.:-(
3335 ** Dan Barlow improved the Alpha port (and is making progress on the
3336 PPC port, for those of you who think different).
3337 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3338 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3339 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3340 continues to work on the SPARC port (for those of you in a position
3341 to look down upon our little PC-compatible boxes from a great height).
3342 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3343 and uses that copy, rather than the current dynamic value, when
3344 it's trying to decide whether to truncate output. Thus e.g.
3345 (let ((*print-lines* 50))
3346 (pprint-logical-block (stream nil)
3348 (let ((*print-lines* 8))
3349 (print (aref possiblybigthings i) stream)))))
3350 should now truncate the logical block only at 50 lines, instead of
3351 often truncating it at 8 lines, as it did before.
3352 * The doc/cmucl/ directory, containing old CMU CL documentation
3353 from the time of the fork, is no longer part of the base system.
3354 SourceForge has shut down its anonymous FTP service, and with it
3355 my original plan for distributing the old CMU CL documentation
3356 there. For now, if you need these files you can download an old
3357 SBCL source release and extract them from it.
3358 * The fasl file version number changed again, for dozens of reasons,
3359 some of which are apparent above.
3361 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3362 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3363 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3364 patch, so that DISASSEMBLE output is much nicer.
3365 * The code in the SB-PROFILE package now seems reasonably stable.
3366 I still haven't decided what the final interface should look like
3367 (I'd like PROFILE to interact cleanly with TRACE, since both
3368 facilities use function encapsulation) but if you have a need
3369 for profiling now, you can probably use it successfully with
3370 the current CMU-CL-style interface.
3371 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3372 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3373 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3374 link now returns the dangling link itself, and for similar
3375 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3376 link itself. (In these cases the old code signalled an error and
3377 looped endlessly, respectively.) Thus, DIRECTORY now works even
3378 in the presence of dangling and cyclic symbolic links.
3379 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3380 is now a supported extension again, since the consensus on
3381 sbcl-devel was that it can be useful for ordinary development
3382 work, not just for debugging SBCL itself.
3383 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3384 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3385 that currently-defined functions might be redefined later with
3386 different return types.
3387 * Hash tables can be printed readably, as inspired by CMU CL code
3388 of Eric Marsden and SBCL code of Martin Atzmueller.
3389 * better error handling in CLOS method combination, thanks to
3390 Martin Atzmueller porting Pierre Mai's CMU CL patches
3391 * more overflow fixes for >16Mbyte I/O buffers
3392 * A bug in READ has been fixed, so that now a single Ctrl-D
3393 character suffices to cause end-of-file on character streams.
3394 In particular, now you only need one Ctrl-D at the command
3395 line (not two) to exit SBCL.
3396 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3398 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3399 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3400 does the right thing.
3401 * The implementation of some type tests, especially for CONDITION
3402 types, is now tidier and maybe faster, due to CMU CL code
3403 originally by Douglas Crosher, ported by Martin Atzmueller.
3404 * Some math functions have been fixed, and there are new
3405 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3406 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3407 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3408 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3409 likely to be built into sbcl-0.7.0.)
3410 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3411 CL, ported to SBCL by Martin Atzmueller)
3412 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3413 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3414 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3415 * There's a new slam.sh hack to shorten the edit/compile/debug
3416 cycle for low-level changes to SBCL itself, and a new
3417 :SB-AFTER-XC-CORE target feature to control the generation of
3418 the after-xc.core file needed by slam.sh.
3419 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3420 COMPILE-FILE is no longer supported, so that now every function
3421 gets an entry point, so that block compilation looks a little
3422 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3423 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3424 returns the number of bytes consed since the system started,
3425 rather than the number consed since the first time the function
3426 was called. (The new definition parallels ANSI functions like
3427 CL:GET-INTERNAL-RUN-TIME.)
3428 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3429 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3430 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3431 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3432 * Fasl file version numbers are now independent of the target CPU,
3433 since historically most system changes which required version
3434 number changes have affected all CPUs equally. Similarly,
3435 the byte fasl file version is now equal to the ordinary
3438 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3439 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3440 is no longer recognized. I apologize for this, because it was
3441 listed in SB-EXT as a supported extension, but I found that
3442 its existing behavior was poorly specified, as well as incorrectly
3443 specified, and it looked like too much of a mess to straighten it
3444 out. I have enough on my hands trying to get ANSI stuff to work..
3445 * many patches ported from CMU CL by Martin Atzmueller, with
3446 half a dozen bug fixes in pretty-printing and the debugger, and
3447 half a dozen others elsewhere
3448 * fixed bug 13: Floating point infinities are now supported again.
3449 They might still be a little bit flaky, but thanks to bug reports
3450 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3451 as flaky as they were.
3452 * The --noprogrammer command line option is now supported. (Its
3453 behavior is slightly different in detail from what the old man
3454 page claimed it would do, but it's still appropriate under the
3455 same circumstances that the man page talks about.)
3456 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3457 are now supported, and enabled by default. Thus, the compiler can
3458 handle many floating point and complex operations much less
3459 inefficiently. (Thus e.g. you can implement a complex FFT
3461 * The compiler now detects type mismatches between DECLAIM FTYPE
3462 and DEFUN better, and implements CHECK-TYPE more correctly, and
3463 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3464 thanks to patches from Martin Atzmueller.
3465 * various fixes to make the cross-compiler more portable to
3466 ANSI-conforming-but-different cross-compilation hosts (notably
3467 Lispworks for Windows, following bug reports from Arthur Lemmens)
3468 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3469 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3470 fixed thanks to Pierre Mai's CMU CL patches.
3471 * fixes to keep the system from overflowing internal counters when
3472 it tries to use i/o buffers larger than 16M bytes
3473 * fixed bug 45a: Various internal functions required to support
3474 complex special functions have been merged from CMU CL sources.
3475 (When I was first setting up SBCL, I misunderstood a compile-time
3476 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3477 * improved support for type intersection and union, fixing bug 12
3478 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3479 more obscure bugs as well
3480 * some steps toward byte-compiling non-performance-critical
3481 parts of the system, courtesy of patches from Martin Atzmueller
3482 * Christophe Rhodes has made some debian packages of sbcl at
3483 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3484 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3485 stable, but are nonetheless usable. When he's ready, I'd be happy
3486 to add them to the SourceForge "File Releases" section. (And if
3487 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3488 * new fasl file format version number (because of changes in
3489 internal representation of (OR ..) types to accommodate the new
3490 support for (AND ..) types, among other things)
3492 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3493 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3495 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3497 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3498 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3499 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3500 * DESCRIBE now works on structure objects again.
3501 * Most function call argument type mismatches are now handled as
3502 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3503 know whether the function will be redefined before the call is
3504 executed. (The compiler could flag local calls with full WARNINGs,
3505 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3506 it doesn't keep track of enough information to know whether calls
3507 are local in this sense.)
3508 * Compiler output is now more verbose, with messages truncated
3509 later than before. (There should be some supported way for users
3510 to override the default verbosity, but I haven't decided how to
3511 provide it yet, so this behavior is still controlled by the internal
3512 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3513 src/compiler/ir1util.lisp.)
3514 * Fasl file format version numbers have increased again, because
3515 support for the Gray streams extension changes the layout of the
3516 system's STREAM objects.
3517 * The Gray subclassable streams extension now works, thanks to a
3518 patch from Martin Atzmueller.
3519 * The full LOAD-FOREIGN extension (not just the primitive
3520 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3521 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3522 but like most other programs, it defaults to copying the Unix
3523 environment from the original process instead of starting the
3524 new process in an empty environment.
3525 * Extensions which manipulate the Unix environment now support
3526 an :ENVIRONMENT keyword option which doesn't smash case or
3527 do other bad things. The CMU-CL-style :ENV option is retained
3528 for porting convenience.
3529 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3530 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3532 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3534 * A patch from Martin Atzmueller seems to have solved the SIGINT
3535 problem, and as far as we know, signal-handling now works cleanly.
3536 (If you find any new bugs, please report them!)
3537 * The system no longer defaults Lisp source file names to types
3538 ".l", ".cl", or ".lsp", but only to ".lisp".
3539 * The compiler no longer uses special default file extensions for
3540 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3541 seems to expect a single default extension for all compiled code,
3542 and there's no compelling reason to try to stretch the standard
3543 to allow two different extensions.) Instead, byte-compiled files
3544 default to the same extension as native-compiled files.
3545 * Fasl file format version numbers have increased again, because
3546 a rearrangement of internal implementation packages made some
3547 dumped symbols in old fasl files unreadable in new cores.
3548 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3549 many fewer weird special cases.
3550 * Bug #17 (differing COMPILE-FILE behavior between logical and
3551 physical pathnames) has been fixed, and some related misbehavior too,
3552 thanks to a patch from Martin Atzmueller.
3553 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3554 by Tim Moore, ported to SBCL by Martin Atzmueller.
3555 * Martin Atzmueller fixed several filesystem-related problems,
3556 including bug #36, in part by porting CMU CL patches, which were
3557 written in part by Paul Werkowski.
3558 * More compiler warnings in src/runtime/ are gone, thanks to
3559 more patches from Martin Atzmueller.
3560 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3563 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3565 * DESCRIBE now works on CONDITION objects.
3566 * The debugger now handles errors which arise when trying to print
3567 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3569 * The build system now uses an additional file, customize-target-features.lisp,
3570 to allow local modifications to the target *FEATURES* list. (The point of
3571 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3572 debugging features enabled, without having to worry about propagating it
3573 into everyone's system when I do a "cvs update".) When no
3574 customize-target-features.lisp file exists, the target *FEATURES* list
3575 should be constructed the same way as before.
3576 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3577 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3578 warn about behavior which is undefined under the ANSI spec. Note: This
3579 is specified by ANSI, but it's not very popular with programmers.
3580 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3581 macro in the SBCL sources for an example of a workaround which you
3582 might use to make portable ANSI-standard code which does what you want.
3583 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3584 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3585 when it's used as a non-toplevel form. (This is required in order
3586 to implement the DEFCONSTANT-EQX macro.)
3587 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3588 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3589 funny things, due to bugs in the implementation of EVAL-WHEN
3590 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3591 non-toplevel DEFCONSTANTs are uncommon.)
3592 * The core file version number and fasl file version number have been
3593 incremented, because the old noncompliant DEFCONSTANT behavior involved
3594 calling functions which no longer exist, and because I also took the
3595 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3596 * fixed bug 1 (error handling before read-eval-print loop starts), and
3597 redid debugger restarts and related debugger commands somewhat while
3599 ** The QUIT debugger command is gone, since it did something
3600 rather different than the SB-EXT:QUIT command, and since it never
3601 worked properly outside the main toplevel read/eval/print loop.
3602 Invoking the new TOPLEVEL restart provides the same functionality.
3603 ** The GO debugger command is also gone, since you can just invoke
3604 the CONTINUE restart directly instead.
3605 ** The TOP debugger command is also gone, since it's redundant with the
3606 FRAME 0 command, and since it interfered with abbreviations for the
3608 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3610 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3611 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3612 some porting work by Martin Atzmueller)
3613 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3614 with a boatload of patches which he ported to SBCL. Now that those
3616 ** The system tries to make sure that its low-priority messages
3617 are prefixed by semicolons, to help people who like to use
3618 syntax highlighting in their ILISP buffer. (This patch
3619 was originally due to Raymond Toy.)
3620 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3621 patches originally written by Raymond Toy.
3622 ** The compiler understands coercion between single-value and
3623 multiple-VALUES type expressions better, getting rid of some very
3624 weird behavior, thanks to patches originally by Robert MacLachlan
3625 and Douglas Crosher.
3626 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3627 lambda lists, thanks to a patch originally by Pierre Mai.
3628 ** The system no longer bogusly warns about "abbreviated type
3630 ** The compiler gets less confused by inlining and RETURN-FROM,
3631 thanks to some patches originally by Tim Moore.
3632 ** The system no longer hangs when dumping circular lists to fasl
3633 files, thanks to a patch originally from Douglas Crosher.
3634 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3635 undefined function error.
3636 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3637 in structure slot initforms) for now, documented workaround instead:-|
3638 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3639 * fixed bug 5: added stubs for various Gray stream functions called
3640 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3641 installed, at least appropriate type errors are generated
3642 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3643 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3644 consistently in DEFMETHOD forms.
3645 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3646 it doesn't seem to affect SBCL after all
3647 * The C runtime system now builds with better optimization and many
3648 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3650 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3652 * The system is now under CVS at SourceForge (instead of the
3653 CVS repository on my home machine).
3654 * The new signal handling code has been tweaked to treat register
3655 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3656 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3657 downstream, e.g. in debug-int.lisp, has implicit dependencies
3658 on the unsignedness of integer representation of machine words,
3659 and that caused the system to bomb out with infinite regress
3660 when trying to recover from type errors involving signed values,
3661 e.g. (BUTLAST '(1 2 3) -1).)
3662 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3663 (This was one of the bugs Peter Van Eynde reported back in July.)
3664 * The system now uses code inspired by Colin Walters' O(N)
3665 implementation of MAP (from the cmucl-imp@cons.org mailing
3666 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3667 inline the MAP operation, and there is more than one
3668 sequence argument to the MAP call (so that it can't just
3669 do ETYPECASE once and for all based on the type of the
3670 single sequence argument). (The old non-inline implementation
3671 of the general M-argument sequence-of-length-N case required
3672 O(M*N*N) time when any of the sequence arguments were LISTs.)
3673 * The QUIT :UNIX-CODE keyword argument has been renamed to
3674 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3675 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3677 * Raymond Wiker's patches to port dynamic loading from Linux to
3678 FreeBSD have been added.
3679 * The BUGS file is now more nearly up to date, thanks in large part
3680 to Martin Atzmueller's review of it.
3681 * The debugger now flushes standard output streams before it begins
3682 its output ("debugger invoked" and so forth).
3683 * The core version number and fasl file version number have both
3684 been incremented, because of incompatible changes in the layout
3686 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3687 used to not be called for a saved Lisp image.)
3688 * Martin Atzmueller's version of a patch to fix a compiler crash,
3689 as posted on sbcl-devel 13 September 2000, has been installed.
3690 * Instead of installing Martin Atzmueller's patch for the
3691 compiler transform for SUBSEQ, I deleted the compiler transform,
3692 and transforms for some similar consing operations.
3693 * A bug in signal handling which kept TRACE from working on OpenBSD
3695 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3696 away by constant folding
3697 * The system now defines its address space constants in one place
3698 (in the Lisp sources), and propagates them automatically elsewhere
3699 (through GENESIS and the sbcl.h file). Therefore, patching the
3700 address map is less unnecessarily tedious and error-prone. The
3701 Lisp names of address space constants have also been systematized.
3702 * CVS tags like dollar-Header-dollar have been removed from
3703 the sources, because they have never saved me trouble and
3704 they've been source of trouble working with patches and other
3705 diff-related operations.
3706 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3707 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3709 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3711 * The system has been ported to OpenBSD.
3712 * The system now compiles with a simple "sh make.sh" on the systems
3713 that it's supported on. I.e., now you no longer need to tweak
3714 text in the target-features.lisp-expr and symlinks in src/runtime/
3715 by hand, the make.sh takes care of it for you.
3716 * The system is no longer so grossly inefficient when compiling code
3717 involving vectors implemented as general (not simple) vectors (VECTOR T),
3718 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3719 which dares to use the various sequence functions on non-simple
3720 vectors, takes less of a performance hit.
3721 * There is now a primitive type predicate VECTOR-T-P
3722 to test for the (VECTOR T) type, so that e.g.
3723 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3724 can now be compiled with some semblance of efficiency. (The old code
3725 turned the type declaration into a full call to %TYPEP at runtime!)
3726 * AREF on (VECTOR T) is still not fast, since it's still compiled
3727 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3728 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3729 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3730 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3731 without making full calls to SUBTYPEP at runtime.
3732 (Some analogous efficiency issues for non-simple vectors specialized to
3733 element types other than T, or for non-simple multidimensional arrays,
3734 have not been addressed. They could almost certainly be handled the
3735 same way if anyone is motivated to do so.)
3736 * The changes in array handling break binary compatibility, so
3737 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3738 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3739 instead of (VALUES T T).
3740 * By following the instructions that Dan Barlow posted to sbcl-devel
3741 on 2 July 2000, I was able to enable primitive dynamic object
3742 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3743 functionality is not implemented (since it calls ld to resolve
3744 library references automatically, requiring RUN-PROGRAM for its
3745 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3746 try to resolve library references) is now supported.
3747 * The system now flushes the standard output streams when it terminates,
3748 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3749 them at several other probably-convenient times, e.g. in each pass of
3750 the toplevel read-eval-print loop, and after evaluating a form given
3751 as an "--eval" command-line option. (These changes were motivated by a
3752 discussion of stream flushing issues on cmucl-imp in August 2000.)
3753 * The source transform for TYPEP of array types no longer assumes
3754 that an array whose element type is a not-yet-defined type
3755 is implemented as an array of T, but instead punts, so that the
3756 type will be interpreted at runtime.
3757 * There is now some support for cross-compiling in make.sh: each of
3758 the phases of make.sh has its own script. (This should be transparent
3759 to people doing ordinary, non-cross-compile builds.)
3760 * Since my laptop doesn't have hundreds of megabytes of memory like
3761 my desktop machine, I became more motivated to do some items on
3762 my to-do list in order to reduce the size of the system a little:
3763 ** Arrange for various needed-only-at-cold-init things to be
3764 uninterned after cold init. To support this, those things have
3765 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3766 symbols with such names are now uninterned after cold init).
3767 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3768 fasl files for cold load.
3769 ** Remove the old compiler structure pooling code (which used to
3770 be conditional on the target feature :SB-ALLOC) completely.
3771 ** Redo the representation of some data in cold init to be more compact.
3772 (I also looked into supporting byte compiled code at bootstrap time,
3773 which would probably reduce the size of the system a lot, but that
3774 looked too complicated, so I punted for now.)
3775 * The maximum signal nesting depth in the src/runtime/ support code has
3776 been reduced from 4096 to 256. (I don't know any reason for the very
3777 large old value. If the new smaller value turns out to break something,
3778 I'll probably just bump it back up.)
3779 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3781 * Many, many bugs reported by Peter Van Eynde have been added to
3782 the BUGS list; some have even been fixed.
3783 * While enabling dynamic object file loading, I tried to make the
3784 code easier to understand, renaming various functions and variables
3785 with less ambiguous names, and changing some function calling
3786 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3787 * While trying to figure out how to do the OpenBSD port, I tried to
3788 clean up some of the code in src/runtime/. In particular, I dropped
3789 support for non-POSIX signal handling, added various comments,
3790 tweaked the code to reduce the number of compilation warnings, and
3791 renamed some files to increase consistency.
3792 * To support the new automatic configuration functionality in make.sh,
3793 the source file target-features.lisp-expr has been replaced with the
3794 source file base-target-features.lisp-expr and the machine-generated
3795 file local-target-features.lisp-expr.
3796 * fixed a stupid quoting error in make.sh so that using CMU CL
3797 "lisp -batch" as cross-compilation host works again
3799 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3801 * DESCRIBE no longer tries to call itself recursively to describe
3802 bound/fbound values, so that it no longer fails on symbols which are
3803 bound to themselves (like keywords, T, and NIL).
3804 * DESCRIBE now works on generic functions.
3805 * The printer now prints less-screwed-up representations of closures
3806 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3807 * A private symbol is used instead of the :EMPTY keyword previously
3808 used to mark empty slots in hash tables. Thus
3809 (DEFVAR *HT* (MAKE-HASH-TABLE))
3810 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3811 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3812 now does what ANSI says that it should. (You can still get
3813 similar noncompliant behavior if bang on the hash table
3814 implementation with all the symbols you get back from
3815 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3816 This breaks binary compatibility, since tests for equality to
3817 :EMPTY are wired into things like the macroexpansion of
3818 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3820 * There's now a minimal placeholder implementation for CL:STEP,
3821 as required by ANSI.
3822 * An obscure bug in the interaction of the normal compiler, the byte
3823 compiler, inlining, and structure predicates has been patched
3824 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3825 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3827 * Missing ordinary arguments in a macro call are now detected even
3828 when the macro lambda list contains &KEY or &REST.
3829 * The debugger no longer complains about encountering the top of the
3830 stack when you type "FRAME 0" to explicitly instruct it to go to
3831 the top of the stack. And it now prints the frame you request even
3832 if it's the current frame (instead of saying "You are here.").
3833 * As specified by ANSI, the system now always prints keywords
3834 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3836 * The default initial SIZE of HASH-TABLEs is now smaller.
3837 * Type information from CLOS class dispatch is now propagated
3838 into DEFMETHOD bodies, so that e.g.
3839 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3841 is now basically equivalent to
3842 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3843 (DECLARE (TYPE SINGLE-FLOAT X))
3845 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3846 operation, without having to do run-time type dispatch.
3847 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3848 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3849 * Since I don't seem to be making big file reorganizations very often
3850 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3851 is overflowing my ability to conveniently back them up), I've finally
3852 checked the system into CVS. (The CVS repository is on my home system,
3853 not at SourceForge -- putting it on SourceForge might come later.)
3854 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3855 high-level GC-NOTIFY-FOO functions send their output. (There's
3856 still very little control of where low-level verbose GC functions
3857 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3858 less than it used to -- the GC-NOTIFY-FOO functions are now under
3859 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3860 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3861 in only one place in the source code, and propagates it automatically
3862 everywhere that it's needed. Thus e.g. when I bump the version from
3863 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3864 * The C source files now include boilerplate legalese and documentation
3865 at the head of each file (just as the Lisp source files already did).
3866 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3867 to his page will be replaced with a link to his new CLiki service.
3869 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3871 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3872 * The build process now looks for GNU make under the default name "gmake",
3873 instead of "make" as it used to. If GNU make is not available as "gmake"
3874 on your system, you can change this default behavior by setting the
3875 GNUMAKE environment variable.
3876 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3877 can build without error under CMU CL.
3879 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3881 * There is now a partial SBCL user manual (with some new text and some
3882 text cribbed from the CMU CL manual).
3883 * The beginnings of a profiler have been added (starting with the
3884 CMU CL profiler and simplifying and cleaning up). Eventually the
3885 main interface should be through the TRACE macro, but for now,
3886 it's still accessed through vaguely CMU-CL-style functions and macros
3887 exported from the package SB-PROFILE.
3888 * Some problems left over from porting CMU CL to the new
3889 cross-compilation bootstrap process have been cleaned up:
3890 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3891 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3892 part when getting rid of the compiler *BACKEND* stuff.)
3893 ** The value of *NULL-TYPE* was screwed up, because it was
3894 being initialized before the type system knew the final
3895 definition of the 'NULL type. This screwed up several key
3896 optimizations in the compiler, causing inefficiency in all sorts
3897 of places. (I found it because I wanted to understand why
3898 GET-INTERNAL-RUN-TIME was consing.)
3899 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3900 PROCLAIM FTYPE information. Unfortunately this broke binary
3901 compatibility again, since now the forms output by DEFGENERIC
3902 to refer to functions which didn't exist in 0.6.3.
3903 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3904 can use the new (as of 0.6.3) transform for SOME into MAP into
3906 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3907 values to (INTEGER 0 1000000), so that the time code will no longer
3908 occasionally get blown up by Linux returning 1000000 microseconds
3909 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3910 its output conform to the ANSI spec. (Alas, this makes its output
3911 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3912 * A full call to MAP NIL with a single sequence argument no longer conses.
3913 * fixes to problems pointed out by Martin Atzmueller:
3914 * The manual page no longer talks about multiprocessing as though
3915 it were currently supported.
3916 * The ILISP support patches have been removed from the distribution,
3917 because as of version 5.10.1, ILISP now supports SBCL without us
3918 having to maintain patches.
3919 * added a modified version of Raymond Toy's recent CMU CL patch for
3920 EQUALP comparison of HASH-TABLE
3922 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3924 * The system still can't cross-compile itself with
3925 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3926 entails), but at least it can compile more of itself that way
3927 than it used to be able to, and various buglets which were uncovered
3928 by trying to cross-compile itself that way have now been fixed.
3929 * This release breaks binary compatibility again. This time
3930 at least I've incremented the FASL file format version to 2, so that the
3931 problem can be detected reliably instead of just causing weird errors.
3932 * various new style warnings:
3933 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3934 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3935 ** using the lexical binding of a variable named in the *FOO* style
3936 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3937 as specified by ANSI.
3938 * *RANDOM-STATE* is no longer automatically initialized from
3939 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3940 default behavior of the system is to repeat its behavior every time
3941 it's run. If you'd like to change this behavior, you can always
3942 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3943 old convention there was no comparably easy way to get the system to
3944 repeat its behavior every time it was run.
3945 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3946 FTYPE declarations has been removed, in favor of their ANSI
3947 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3948 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3949 the types of their sequence arguments can be determined at compile time.
3950 This is done through a new open code expansion for MAP which eliminates
3951 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3952 argument types can be determined at compile time.
3953 * The optimizer now transforms COERCE into an identity operation when it
3954 can prove that the coerced object is already of the correct type. (This
3955 can be a win for machine generated code, including the output of other
3956 optimization transforms, such as the MAP transform above.)
3957 * Credit information has been moved from source file headers into CREDITS.
3958 * Source file headers have been made more standard.
3959 * The CASE macro now compiles without complaining even when it has
3962 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3964 * (Note that the way that the PCL macroexpansions were rewritten
3965 to accommodate the change in DEFGENERIC below breaks binary
3966 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3967 not run under sbcl-0.6.2. Once we get out of alpha releases,
3968 i.e. hit release 1.0.0, we'll probably try to maintain binary
3969 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3970 and sbcl-1.4.4. Until then, however, it might be fairly common
3971 for maintenance releases to break binary compatibility.)
3972 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3974 * The reporting of unhandled signals has been changed to print some
3975 explanatory text as well as the report form. (Previously only
3976 the report form was printed.)
3977 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3978 it defines, so that the compiler no longer issues undefined function
3979 warnings for compiled-but-not-yet-loaded generic functions.
3980 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3981 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3982 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3983 section of the ANSI spec are supported.
3984 * The "" nickname for the "KEYWORD" package has been removed.
3985 The reader still handles symbol tokens which begin with a package marker
3986 as keywords, but it doesn't expose its mechanism for doing so in the
3987 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3988 * The system now issues STYLE-WARNINGs for contradictory TYPE
3989 proclamations. (Warnings for contradictory FTYPE proclamations would
3990 be nice too, but those can't be done usefully unless the type system
3991 is made smarter about FUNCTION types.)
3992 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3993 systematized, so that "*target-*.lisp is supposed to exist only on the
3994 target and imply that there's a related file which exists on the
3995 host, and *host-*.lisp is supposed to exist only on the host and imply
3996 that there's a related file which exists on the target. This involves a
3997 lot of renaming. Hopefully the acute confusion caused by the renaming
3998 will be justified by the reduction in chronic confusion..
3999 ** runtime-type.lisp -> early-target-type.lisp
4000 ** target-type.lisp -> late-target-type.lisp
4001 ** early-host-format.lisp -> early-format.lisp
4002 ** late-host-format.lisp -> late-format.lisp
4003 ** host-error.lisp -> misc-error.lisp
4004 ** early-error.lisp -> early-target-error.lisp
4005 ** late-error.lisp -> late-target-error.lisp
4006 ** host-defboot.lisp -> early-defboot.lisp
4007 ** code/misc.lisp -> code/target-misc.lisp
4008 ** code/host-misc.lisp -> code/misc.lisp
4009 ** code/numbers.lisp -> code/target-numbers.lisp
4010 ** code/early-numbers.lisp -> numbers.lisp
4011 ** early-host-type.lisp -> early-type.lisp
4012 ** late-host-type.lisp -> late-type.lisp
4013 ** host-typep.lisp -> typep.lisp
4014 ** load.lisp -> target-load.lisp
4015 ** host-load.lisp -> load.lisp
4016 ** host-disassem.lisp -> disassem.lisp
4017 ** host-insts.lisp -> insts.lisp
4018 ** byte-comp.lisp -> target-byte-comp.lisp
4019 ** host-byte-comp.lisp -> byte-comp.lisp
4020 ** host-signal.lisp -> signal.lisp
4021 ** host-defstruct.lisp -> defstruct.lisp
4022 ** late-target-type.lisp -> deftypes-for-target.lisp
4023 Furthermore, several other previously target-only files foo.lisp (e.g.
4024 hash-table.lisp and random.lisp) have been split into a target-and-host
4025 foo.lisp file and a target-only target-foo.lisp file, with their key type
4026 definitions in the target-and-host part, so that the cross-compiler will
4027 know more about target types.
4028 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4029 gone away. In their place are various *BACKEND-FOO* variables
4030 corresponding to the slots of the old structure.
4031 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4032 into the target SBCL has been fixed.
4033 * The chill.lisp system for loading cold code into a running SBCL
4035 * Support for the CMU CL "scavenger hook" extension has been removed.
4036 (It was undocumented and unused in the CMU CL sources that SBCL was
4037 derived from, and stale in sbcl-0.6.1.)
4038 * Various errors in the cross-compiler type system were detected
4039 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4040 (enabling various consistency checks). Many of them were fixed,
4041 but some hard problems remain, so the compiler is back to
4042 running without *TYPE-SYSTEM-INITIALIZED* for now.
4043 * As part of the cross-compiler type system cleanup, I implemented
4044 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4045 * I have started adding UNCROSS calls throughout the type system
4046 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4047 on cross-compiler input files will be able to go away, and various
4049 * CONSTANTP now returns true for quoted forms (as explicitly required
4052 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4054 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4055 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4056 (MAKE-LIST -1). In the longer term, it ought to become true
4057 that declarations are assertions even at SAFETY 1. For now, it's not
4058 quite true even at SAFETY 3, but it's at least more nearly true..
4059 (Note that this change seems to increases the size of the system by
4060 O(5%) and to decrease the speed of the compiler by 20% or more.)
4061 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4062 for the problem of printing dozens of lines of distracting information
4063 about low-level system machinery as part of the top stack frame
4064 on entry to the debugger when an undefined function was called.
4065 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4066 is not reset to COMMON-LISP-USER.
4067 * Compilation of stuff related to dyncount.lisp has been made conditional
4068 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4069 smaller. The various dyncount-related symbols have been moved into
4070 a new "SB-DYNCOUNT" package.
4071 * tty-inspect.lisp has been renamed to inspect.lisp.
4072 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4073 feature has gone away. (When we eventually port to other flavors of
4074 libc and/or Unix, we'll try to make the differences between flavors
4075 invisible at the user level.)
4076 * Various other *FEATURES* tags, and/or their associated conditionals,
4077 have been removed if obsolescent, or given better documentation, or
4078 sometimes given more-mnemonic names.
4080 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4082 * tidied up "make.sh" script
4083 * tidied up system directory structure
4084 * better "clean.sh" behavior
4085 * added doc/FOR-CMUCL-DEVELOPERS
4086 * many many small tweaks to output format, e.g. removing possibly-confusing
4087 trailing #\. character in DESCRIBE-INSTANCE
4088 * (EQUALP #\A 'A) no longer signals an error.
4089 * new hashing code, including EQUALP hashing
4090 * tidied up Lisp initialization and toplevel
4091 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4092 * command line argument processing
4093 * added POSIX-GETENV function to deal with Unix-ish environment variables
4094 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4095 terminating SBCL on EOF
4096 * non-verbose GC by default
4097 * There is no more "sbcl" shell script; the sbcl file is now the C
4098 runtime executable (just like CMU CL).
4099 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4101 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4102 symbol and package information as Lisp native symbol and package objects
4103 instead of strings naming symbols and strings naming packages. This way,
4104 whenever packages are renamed (as in warm init), debug information is
4105 transformed along with everything else.
4106 * tweaked the optimization policy declarations which control the building
4107 of SBCL itself. Now, among other things, the system no longer saves
4108 source location debugging information. (This helps two problems at once
4109 by reducing SBCL size and by keeping SBCL from trying to look for its
4110 sources -- which may not exist -- when reporting errors.)
4111 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4112 debugging and testing purposes
4113 * cleaned up printing, making the printer call PRINT-OBJECT for
4114 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4115 methods, giving nearly-ANSI behavior
4116 * converted almost all special variables to use *FOO* naming convention
4117 * deleted PARSE-TIME functionality, since it can be done portably
4118 * moved some files out of cold init into warm init
4119 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4121 * regularized formatting of source files
4122 * added an install.sh script
4123 * fixed ridiculous memory usage of cross-compiler by making
4124 compiler/alloc.lisp not try to do pooling unless it can hook
4125 itself into the GC of the cross-compilation host. Now the system
4126 builds nicely on my old laptop.
4127 * added :SB-ALLOC in target-features.lisp-expr
4128 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4129 was not implemented)
4130 * re-did condition handling and note reporting in the compiler. Notes
4131 are no longer handled by signalling conditions. Style warnings
4132 and warnings are handled more correctly and reported in such a way
4133 that it's easy to find one or the other in your output (so that you
4134 can e.g. figure out which of many problems caused COMPILE-FILE to
4136 * changed the severity of several compiler warnings from full WARNING
4137 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4138 compiler note reporting so that it doesn't use the condition system
4139 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4140 COMPILE-FILE command)
4141 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4142 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4143 no longer supported, since their implementation was deeply intertwingled
4144 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4145 * removed bogus "support" for compiler macros named (SETF FOO), and
4146 removed the compiler macro for SETF INFO (but only after making a fool
4147 of myself on the cmucl-imp mailing list by posting a bogus patch for
4148 DEFINE-COMPILER-MACRO..)
4149 * Compiled files containing forms which have side effects on the Lisp
4150 reader (such as DEFPACKAGE forms) are now handled more correctly.
4151 (Compiler queuing of top level lambdas has been suppressed by setting
4152 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4153 * deleted various currently-unused source files, e.g. gengc.lisp. They
4154 may be added back at some point e.g. when porting to other architectures,
4155 but until they are it's distracting to distribute them and to try to
4157 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4158 there were so many of them they're just distractions, and UNCROSS is
4159 known to be able to handle the current sources
4160 * moved PROFILE functionality into TRACE, so that it will be clear
4161 how the wrapping and unwrapping of functions when you profile them
4162 interacts with the wrapping and unwrapping of functions when you
4163 trace them. (Actually, the functionality isn't there yet, but at least
4164 the interface specification is there. Hopefully, the functionality will
4165 arrive with some future maintenance release.)
4166 * removed host-oops.lisp
4167 * changed signature of QUIT function to allow UNIX-CODE argument
4168 * fixed READ-SEQUENCE bug
4169 * tweaked verbose GC output so that it looks more like the progress
4170 output that ANSI specifies for functions like LOAD
4171 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4172 * added <http://sbcl.sourceforge.com> to the banner information printed by