1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.4 relative to sbcl-1.0.3:
3 * bug fix: >= and <= gave wrong results when used with NaNs.
4 * bug fix: the #= and ## reader macros now interact reasonably with
7 changes in sbcl-1.0.3 relative to sbcl-1.0.2:
8 * new platform: NetBSD/PPC. (thanks to Aymeric Vincent)
9 * optimization: calls of the form (AREF FOO (+ INDEX <constant>)) now
10 produce more efficient code on the x86 if the compiler can determine
11 that (+ INDEX <constant>) does not require a bounds check and FOO
12 has an element type at least 8 bits wide.
13 * bug fix: references to undefined variables are handled the same way
14 in toplevel forms as in the normal compiler
15 * bug fix: the build scripts again work with non-bash /bin/sh (thanks
17 * bug fix: use "gtar" as the asdf-install *TAR-PROGRAM* on NetBSD
18 (thanks to Jon Buller)
19 * improvement: faster compilation times for complex functions
20 * improvement: added readlink support to SB-POSIX (thanks to Richard
23 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
24 * improvement: experimental support for mach exception handling on
25 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
27 * improvement: support for GBK external format.
28 (thanks to Chun Tian (binghe))
29 * improvement: the debugger now displays variables that have been closed
30 over, in code compiled with (DEBUG 2) or higher
31 * improvement: support for executable cores on NetBSD (thanks to
33 * new feature: added a RESTART-FRAME debugger command
34 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
35 be used to provide an implementation for FILE-POSITION on Gray streams
36 (thanks to Eric Marsden)
37 * new feature: add syslog, openlog and closelog support to SB-POSIX
38 (thanks to Richard Kreuter)
39 * optimization: the function call overhead in code compiled with
40 a high DEBUG optimization setting is significantly
41 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
42 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
43 (thanks to Tony Martinez)
44 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
45 evaluated once (reported by Kevin Reid)
46 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
47 works again (thanks to Kilian Sprotte)
48 * bug fix: an error is signaled for tagbodies with duplicate tags
49 (thanks to Stephen Wilson)
50 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
51 * bug fix: Win32 port can now handle foreign code unwinding Lisp
52 stack frames from alien callbacks.
53 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
54 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
55 65k pages (thanks to David Woodhouse)
56 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
58 changes in sbcl-1.0.1 relative to sbcl-1.0:
59 * new platform: FreeBSD/x86-64, including support for threading.
60 * new feature: the compiler stores cross-referencing information
61 abount function calls (who-calls), macroexpansion (who-macroexpands)
62 and special variables (who-binds, who-sets, who-references) for code
63 compiled with (< SPACE 3). This information is available through the
64 sb-introspect contrib.
65 * new feature: users may subclass SEQUENCE, and have instances of
66 these classes interoperate with standard Common Lisp functions if
67 a number of methods are defined. (This feature is experimental
68 and the interface subject to change based on feedback from SBCL
69 users and the general community)
70 * improvement: sb-sprof traces call stacks to an arbitrary depth on
71 x86/x86-64, rather than the previous fixed depth of 8
72 * improvement: another pthread back-end of mutex "pthread-futex".
73 * bug fix: non-ascii command-line arguments are processed correctly
74 (thanks to Yaroslav Kavenchuk)
75 * bug fix: non-required arguments were not passed correctly when a method
76 defined using DEFMETHOD was called from a mop-generated method using
77 CALL-NEXT-METHOD (reported by Pascal Costanza)
78 * bug fix: recursion is now permitted in accessors through
79 SLOT-UNBOUND. (reported by Pascal Costanza)
80 * bug fix: an error was signaled at startup if the HOME environment
81 variable was defined, but had an empty value (reported by Peter Van Eynde)
82 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
83 signal a PROGRAM-ERROR, not a BUG.
84 * bug fix: SB-POSIX:READDIR works when built with large file support.
85 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
87 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
88 for better automatic indentation support. (thanks to Matt Pillsbury)
89 * optimization: loading generic functions no longer takes O(n^2) time,
90 proportional to the amount of methods in the generic function
91 (reported by Todd Sabin and Jeremy Brown)
92 * optimization: the FIND and POSITION family of sequence functions
93 are significantly faster on arrays whose element types have been
95 * improvements to the Windows port:
96 ** Intermittent heap corruption problems have been fixed. (thanks
97 to Alastair Bridgewater)
98 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
100 ** Lisp is able to unwind foreign exception frames from alien
101 callbacks. (thanks to Alastair Bridgewater)
103 changes in sbcl-1.0 relative to sbcl-0.9.18:
104 * improvement: experimental support for threading on FreeBSD/x86.
105 (thanks to NIIMI Satoshi)
106 * improvement: runtime option --dynamic-space-size can be used
107 to set the size of the dynamic space reserved on startup.
108 * improvement: floating point modes in effect are now saved in
109 core, and restored on startup.
110 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
111 startup, not time since first call to GET-INTERNAL-REAL-TIME.
112 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
113 threads are not running after *SAVE-HOOKS* have run.
114 * improvement: writes to CLOS instance slots are type-checked in code
115 compiled with (SAFETY 3)
116 * improvement: floating-point exception handling on FreeBSD (thanks to
118 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
119 (thanks to Zach Beane)
120 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
122 * improvement: added support for the Shift-JIS external format.
123 (contributed by NIIMI Satoshi)
124 * improvement: callbacks are supported on Linux/PPC. (thanks to
126 * bug fix: compiler bug triggered by a (non-standard) VALUES
127 declaration in a LET* was fixed. (reported by Kaersten Poeck)
128 * bug fix: file compiler no longer confuses validated and already
129 dumped structurres. (reported by Kaersten Poeck)
130 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
131 fill-pointer signals a type-error as required. (thanks to
133 * bug fix: disassemly of funcallable instances works.
134 * bug fix: single stepping on PPC.
135 * bug fix: fix thread-safety problems in the type system (generally
136 manifesting as nonsensical errors like "STRING is a bad type specifier
137 for sequences" or "The value 1 is not of type FIXNUM")
138 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
139 (reported by Josip Gracin)
140 * bug fix: an error is signaled for attempts to displace arrays with
141 incompatible element types (thanks to Mario Mommer)
142 * bug fix: more correct handling of wide characters in debug info
143 (bug reported by Attila Lendvai and fixed by Juho Snellman)
144 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
145 and don't cause extra consing
146 * optimization: MAP and MAP-INTO are significantly faster on vectors
147 whose elements types have been declared.
148 * Improvements to SB-SPROF:
149 ** Support for allocation profiling
150 ** Reduced profiling overhead, especially for long profiling runs
151 * Improvements to the Windows port:
152 ** floating point exceptions are now reported correctly.
153 ** stack exhaustion detection works partially.
154 ** more accurate GET-INTERNAL-REAL-TIME.
155 ** better RUN-PROGRAM behaviour in case of errors in the child
157 ** PROBE-FILE now simplifies pathnames correctly.
158 ** DIRECTORY now works correctly with :WILD-INFERIORS.
159 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
160 1st 1970 (although time timezone data will be unreliable).
161 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
163 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
164 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
165 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
167 * bug fix: two potential GC deadlocks affecting threaded builds.
168 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
169 returns T (reported by Anton Kazennikov)
170 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
171 with non-variable places
172 * bug fix: remove a race condition in the setting of
173 funcallable-instance functions, this should make threaded CLOS
174 code more stable against memory faults.
175 * bug fix: corruption of specials when unbinding is interrupted by an
176 asynchronous unwind (reported by Hannu Koivisto)
177 * improvement: the debugger will now also display local variables that
178 are only used once, for code compiled with a DEBUG optimization quality
181 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
182 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
183 * incompatible change: External-format support for FFI calls. The
184 SB-ALIEN:C-STRING no longer implies an ASCII
185 external-format. Instead the string is subject to external-format
186 conversion. Additionally return values of type SB-ALIEN:C-STRING
187 are by default of type (SIMPLE-ARRAY CHARACTER), not
188 SIMPLE-BASE-STRING. For an alien type that behaves like the old
189 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
190 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
191 * incompatible change: SB-EXT package no longer contains the
192 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
193 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
195 * incompatible change: the single-stepper is no longer available
196 on Alpha, Mips and Sparc platforms.
197 * minor incompatible change: the direct superclasses of
198 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
199 not (STANDARD-OBJECT FUNCTION). This makes the
200 class-precedence-lists of GENERIC-FUNCTION and
201 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
203 * minor incompatible change: the default stream external format on
204 non-unicode SBCL is no longer detected from the locale, but is always
206 * new feature: Add a version of evaluator that uses an interpreter instead
207 of the compiler. EVAL still uses the compiler by default, to switch it
208 to use the interpreter, set the value of the variable
209 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
210 * minor incompatible change: the single-stepper REPL has been merged
211 with the normal debugger (see the "Stepping" heading of the debugger help
212 for more details). The debugger command STEP will no longer switch
213 to the single-stepper REPL.
214 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
215 for a type now works.
216 * bug fix: loading of default sysinit file works. (thanks to Leonid
218 * bug fix: better detection of circularities in the file-compiler.
219 (reported by Marco Monteiro)
220 * bug fix: the CL pathname functions now work with files that have
221 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
222 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
223 systems with Unix98 pty semantics.
224 * bug fix: ASDF-INSTALL will now work with bsd tar.
225 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
227 * bug fix: timers expiring in dead threads no longer cause a
228 type-error (reported by Paul "Nonny Mouse").
229 * bug fix: thanks to more lightweight single-stepper instrumentation,
230 code compiled with (DEBUG 3) will compile and execute significantly faster,
231 and will have more accurate type-inferencing than before
232 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
233 whose bindings are modified
234 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
235 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
236 ** replace dummy implementations of CL:MACHINE-INSTANCE and
237 CL:SOFTWARE-VERSION with proper version
239 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
240 * feature: implemented the READER-METHOD-CLASS and
241 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
242 as specified by AMOP.
243 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
245 * optimization: faster LOGCOUNT implementation on x86 and x86-64
246 (thanks to Lutz Euler)
247 * optimization: hashing of general arrays and vectors has been
248 improved. (reported by Any Fingerhut)
249 * enhancement: SB-INTROSPECT is now able to find definitions of
250 profiled functions. (thanks to Troels Henriksen)
251 * enhancement: compiler-macro expansion applies now to FUNCALL forms
252 as well, allowing compiler-macros for SETF-functions to expand.
253 * enhancement: step-instrumentation no longer wraps known
254 single-value functions with multiple-value context, allowing
255 better type inference.
256 * fixed bug #337: use of MAKE-METHOD in method combination now works
257 even in the presence of user-defined method classes. (reported by
258 Bruno Haible and Pascal Costanza)
259 * fixed bug #339(c): if there are applicable methods not part of any
260 long-form method-combination group, call INVALID-METHOD-ERROR.
261 (reported by Bruno Haible)
262 * fixed bug #361: the :FUNCTION initarg in the protocol for
263 initialization of methods can now be used to override
264 internally-produced optimized functions. (reported by Bruno
266 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
267 system-provided lambda expression no longer cause warnings about
268 unbound #:|pv-table| symbols.
269 * bug fix: improved the handling of type declarations and the
270 detection of violations for keyword arguments with non-constant
272 * bug fix: potentially erronous calls to PATHNAME and
273 MERGE-PATHNAMES were being flushed in some cases.
274 (reported by Richard Kreuter)
275 * bug fix: compiled calls to TYPEP were mishandling obsolete
276 instances. (reported by James Bielman and Attila Lendvai)
277 * bug fix: format strings with ~> without matching ~< no longer
278 trigger an AVER, but signal an understandable error instead.
279 (reported by Antonio Martinez)
280 * bug fix: specifying an output-file without a pathname-type for
281 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
282 of input-file instead of "fasl". (reported by Robert Dodier)
283 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
284 (reported by James Y Knight).
285 * bug fix: compiler-macros-function did not consider the environment
286 argument for shadowing by local functions.
287 * bug fix: compiler-macros expansion was inhibited by local INLINE
289 * bug fix: inline expansions of known functions were subject to
290 step-instrumentation in high debug policies, leading to problems
292 * bug fix: compiler failed to differentiate between different CONS
294 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
295 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
296 * bug fix: #368: incorrect use of expressed vs. upgraded array
298 * bug fix: #306a: more precise unions of array types.
299 * thread-safety improvements:
300 ** CONDITION-WAIT could return early on Linux, if the thread was
301 interrupted and subsequently continued with SIGCONT.
302 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
304 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
305 * added support for the ucs-2 external format. (contributed by Ivan
307 * minor incompatible change: pretty printing of objects of type
308 (cons symbol) is, in the default pprint-dispatch-table, now
309 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
311 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
312 called later in a class's lifetime, possibly as late as when the
313 first instance of the class is created. Previously,
314 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
315 class became finalizeable.
316 * fixed bug: FILE-POSITION sometimes returned inconsistent results
317 for multibyte external-format streams. (thanks to "vbzoli")
318 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
319 with :ALLOCATION :CLASS inherited from superclasses of the
321 * fixed bug: anonymous classes can now be created using the :NAME
322 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
323 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
324 * fixed bug: core-files saved with :EXECUTABLE T can again be
325 executed when SBCL_HOME isn't set. (reported by James Knight)
326 * fixed bug: toplevel LOCALLY forms with declarations could
327 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
328 * fixed bug: printing from several different threads using different
329 values of *print-case* could cause invalid output, due to
330 some internal special variables of the printer not being bound
331 thread-locally (reported by Max Mikhanosha)
332 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
333 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
334 (reported by Pascal Costanza)
335 * fixed bug: FINALIZE-INHERITANCE is called from
336 REINITIALIZE-INSTANCE on classes when the class has previously
337 been finalized, as required by AMOP.
338 * minor code generation optimizations:
339 ** better register allocation in CLOS dispatching functions
340 ** overflow detection when coercing signed bytes to fixnums on x86-64
341 ** is now implemented with one IMUL instruction instead of three shifts
342 ** more efficient bit-vector access on x86 and x86-64
343 ** more efficient access to raw structure slots on x86 and x86-64
344 * fixed some bugs revealed by Paul Dietz' test suite:
345 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
348 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
349 * feature: thread support on Solaris/x86, and experimental thread support
351 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
353 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
354 default method for SLOT-UNBOUND.
355 * minor incompatible change: prevent the user from specializing the
356 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
357 somewhat counter to the intent of the protocol, I (CSR) think, and
358 additionally it just doesn't work in SBCL as currently
359 implemented, thanks to optimizations (that are always valid for
360 the other three SLOT-VALUEish functions, but not for the setter).
361 * minor incompatible change: the :SB-LDB feature is now enabled by
362 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
363 the low-level debugger.
364 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
365 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
366 down to an SB-SYS:FD-STREAM, allowing direct communication in
367 more cases. Temporary files and pipes are still used as a fallback
369 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
370 on the non-reentrant C functions localtime(3) and gmtime(3).
371 * bug fix: global optimization policy was not visible in LOCALLY and
373 * bug fix: class objects can be used as specializers in methods.
374 (reported by Pascal Costanza)
375 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
376 with a type error. (reported by blitz_ on #lisp)
377 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
378 failed with a type error. (reported by Pascal Bourguignon)
379 * bug fix: merging pathnames against defaults with :DIRECTORY
380 starting with '(:RELATIVE :BACK) should preserve the :BACK.
381 (reported by James Y Knight)
382 * bug fix: saving large (>2GB) cores on x86-64 now works
383 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
384 constant index and a (simple-array (signed-byte 32)) array
385 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
386 error (patch by Robert J. Macomber)
387 * feature: TIME now displays the amount of run-time spent in GC
388 * bug fix: The debugger now does a better job of respecting
389 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
390 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
391 workaround for bug 403.)
392 * bug fix: floating point exception handling now works on FreeBSD
393 (thanks to NIIMI Satoshi)
394 * fixed some bugs revealed by Paul Dietz' test suite:
395 ** MISC.641: LET-conversion were not supposed to work in late
397 * improvements to the Win32/x86 port:
398 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
399 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
401 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
402 * new feature: source path information is generated for macro-expansion
403 errors for use in IDE's like Slime (thanks to Helmut Eller)
404 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
405 * bug fix: compiler does not loop forever on an invalid type in
407 * improvement: compilation of most CLOS applications is significantly
409 * optimization: added a limited bytecode compiler for simple toplevel
410 forms, speeding up compilation and FASL loading
411 * bug fix: the statistical profiler now properly distinguishes anonymous
414 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
415 * minor incompatible change: in sbcl-0.9.11 (but not earlier
416 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
417 into subclasses of the finalized class. Now user calls to
418 FINALIZE-INHERITANCE finalize just the one class, and calls by the
419 system return before any subclasses are finalized.
420 * minor incompatible change: The reader no longer ignores errors
421 regarding non-existent packages in #+ and #- feature tests.
422 * new feature: command line options --no-sysinit, --no-userinit to
423 inhibit loading the corresponding init files
424 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
425 for sending data through UDP sockets (thanks to François-René Rideau)
426 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
427 error is signalled instead (thanks to François-René Rideau)
428 * minor incompatible change: Error signalling behaviour of lexical
429 operations violating package locks has changed slightly. Refer to
430 documentation on package locks for details.
431 * enhancement: EVAL can process IF-expressions without resorting to the
433 * bug fix: Some operations involving SIN, COS, or TAN weren't being
434 constant-folded properly due to missing out-of-line functions.
435 (reported by Mika Pihlajamäki)
436 * bug fix: LISTEN sometimes returned T even in cases where no data was
437 immediately available from the stream
438 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
439 were reversed. (reported by Levente Mészáros)
440 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
441 when given a vector displaced to an adjustable array. (reported
443 * bug fix: derivation of float boundaries from numbers outside the
444 appropriate float range (reported by John Wiseman)
445 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
447 * bug fix: precomputing cache entries for generic functions with
448 some subclasses of specializers as yet invalid does not attempt to
449 fill a cache line with a negative offset. (reported by Levente
451 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
452 allow condition class objects as arguments to DOCUMENTATION and
453 (SETF DOCUMENTATION); only find and set documentation for
454 structure names for the STRUCTURE doc-type. (suggested by Gary
456 * improvements to the Win32/x86 port:
457 ** (user-homedir-pathname) and default initialization file
458 locations now know about the user's "Documents and Settings"
459 directory (thanks to Yaroslav Kavenchuk)
460 ** run-program is implemented (thanks to Mike Thomas)
461 ** sockets support (thanks to Timothy Ritchey)
462 ** better backtrace support (thanks to Alastair Bridgewater)
463 ** sb-grovel supported
464 ** asdf-install and sb-posix work somewhat
465 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
466 * improvements to the Solaris/x86 port:
467 ** works on Solaris 11/Solaris Express
468 ** floating-point exception handling support
469 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
470 * fixed some bugs revealed by Paul Dietz' test suite:
471 ** REMOVE-METHOD returns its generic function argument even when
472 the method is not one of the generic functions' methods.
473 ** objects declared as MEMBER types can be admissible arguments to
475 ** printing characters should simply be printed by the FORMAT ~:C
477 ** compiler failure when compiling functions with hairy constant
478 defaults for optional parameters.
479 ** compiler produces wrong code when MAYBE-INLINE-expanding a
480 function, which is already optimized.
482 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
483 * new platform: experimental support for SBCL x86/Darwin, including
484 MacOS X 10.4.5 on Intel.
485 * new feature: Unicode character names are now known to the system
486 (through CHAR-NAME and NAME-CHAR).
487 * new feature: the filesystem location of SBCL's core file is
488 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
489 * minor incompatible change: the contrib modules SB-POSIX and
490 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
491 this change is to make it easier to distribute
492 location-independent binaries.
493 * bug fix: as implied by AMOP, standardized classes no longer have
494 slots named by external symbols of public packages. (reported by
496 * optimization: calling functions via constant symbols -- as in
497 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
498 via the function object as in (FUNCALL #'FOO).
499 * optimization: on x86 and x86-64, the calling convention for the
500 case of dealing with an unknown number of values has been altered
501 to be friendlier to the prediction heuristics implemented,
502 particularly with reference to CALL and RET pairing. (thanks to
503 Alastair Bridgewater)
504 * optimization: on x86, the code for access to thread-local symbol
505 values and binding/unbinding of thread-local symbols is smaller.
506 (thanks to Alastair Bridgewater)
507 * enhancement: CONSTANTP is now able to determine constantness of
508 more complex forms, including calls to constant-foldable standardized
509 functions and some special forms beyond QUOTE.
510 * fixed bug: occasional GC crashes on Solaris/x86
511 * optimization: x86-64 supports stack allocation of results of simple
512 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
513 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
514 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
515 (thanks to James Knight)
516 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
517 by the standard (thanks to Stephen Compall)
519 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
520 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
521 be used for bundling the runtime and the core file into one
522 executable binary. This feature is not currently supported on all SBCL
523 platforms. (thanks to James Bielman and NIIMI Satoshi)
524 * new feature: a generational or ephemeral garbage collector is now
525 the default on the PowerPC platform (both Linux and Darwin). The
526 old Cheney (stop and copy) collector is a build-time option.
527 (thanks to Cyrus Harmon, after Raymond Toy)
528 * minor incompatible change: the method by which SBCL finds its
529 contributed modules has changed; it no longer relies on symbolic
530 links from an $SBCL_HOME/systems directory, but searches directly
531 in the subdirectories of $SBCL_HOME.
532 * enhancement: the dynamic heap size on the Linux/PPC platform is
533 markedly larger, even using the older Cheney garbage collector.
534 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
535 certain complicated string types. (reported by Gary King)
536 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
537 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
538 * fixed bug: DOCUMENTATION on structures no longer signals an
539 error if no structure type of the right name exists. (reported by
541 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
542 applicable non-standard primary or :AROUND method on
543 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
544 no longer cause unbound variable errors. (reported by Kevin Reid)
545 * fixed bug: metacircle resolution in cases where methods have slots
546 added before the slots from STANDARD-METHOD. (reported by Jean
548 * fixed bug: the Power PC assembly code for calling into lisp
549 sometimes computed its return address wrongly (depending on the
550 15th bit of the address, fixed at link time). This used to
551 manifest itself in a segmentation violation while building PCL.
552 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
553 many others over the years)
554 * fixed bug: ROOM no longer reports silly numbers for stack usage if
555 the stack pointer (treated as a signed integer) is negative.
556 (thanks to Peter van Eynde)
558 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
559 * new platform: experimental support for the Windows operating
560 system has been added. (thanks to Alastair Bridgewater)
561 * fixed several bugs in and robustified the PPC FFI (including
562 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
563 * bug fix: allow non-simple string symbol names (reported by Paul
565 * bug fix: interrupt handling on NetBSD (thanks to Richard M
567 * bug fix: saving a core corrupted callbacks on x86/x86-64
568 * bug fix: closed a loophole in metacircularity detection and
569 grounding in the PCL implementation of CLOS.
570 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
571 (reported by tomppa on #lisp)
572 * bug fix: clear the direction flag on Lisp -> C transitions, as
573 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
574 (reported by Andrej Grozin and Hendrik Maryns)
575 * optimization: major improvements to GC efficiency on GENCGC platforms
576 * optimization: faster implementation of EQUAL
577 * optimization: emit more efficient opcodes for some common
578 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
579 * optimization: several other minor code-generation improvements on x86-64
580 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
582 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
583 * minor incompatible change: (SETF CLASS-NAME) and (SETF
584 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
585 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
586 by ANSI as a generic function, and for consistency (SETF
587 GENERIC-FUNCTION-NAME) is treated likewise.
588 * fixed bug #233.b: make constraint propagation notice when a variable
589 value is changed after it is referenced but before it is used
590 * fixed bug #296: no more arbitrary behaviour from filenames with
591 odd characters as --load arguments. (reported by Adam Warner)
592 * fixed bug #390: :CHARACTER-SET pathname components now work as
593 expected. (reported by Tim Daly Jr)
594 * fixed bug #391: complicated :TYPE intersections in slot
595 definitions no longer cause an error in PCL internals.
596 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
597 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
598 returning the number of octets which would be written to the
599 file-stream. (thanks to Robert J. Macomber)
600 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
601 arguments to the internal stream functions. (thanks to David
603 * bug fix: stack exhaustion is now handled on the x86/SunOS
604 platform. (thanks to Daisuke Homma)
605 * bug fix: more accurate ROOM results on GENCGC platforms
606 * optimization: improved type inference for arithmetic-for
607 index variables in LOOP
608 * optimization: faster floating-point SQRT on x86-64
609 * fixed some bugs revealed by Paul Dietz' test suite:
610 ** DOCUMENTATION returns NIL instead of "" for method combinations
611 that don't have a docstring
613 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
614 * minor incompatible change: (SETF CLASS-NAME) and (SETF
615 GENERIC-FUNCTION-NAME) are no longer generic functions, and
616 therefore are not customizeable by user code (as seems to be at
617 least permitted and maybe required by AMOP). As a consolation,
618 however, the SBCL implementation of these functions now calls
619 REINITIALIZE-INSTANCE as specified by AMOP.
620 * bug fix: slot-definition documentation is propagated as per ANSI
621 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
622 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
623 argument, as do eslotd initialization methods. (from Pascal
624 Costanza's "Closer" project)
625 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
626 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
628 * bug fix: it is now possible to have more than one subclass of
629 STANDARD-GENERIC-FUNCTION without causing stack overflow.
630 (reported by Bruno Haible, Pascal Costanza and others)
631 * bug fix: the dependent update protocol now works for generic
632 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
634 * bug fix: condition-class instances corresponding to
635 DEFINE-CONDITION forms are now created eagerly. (reported by
636 Kalle Olavi Niemitalo on comp.lang.lisp)
637 * bug fix: floating point printing is more accurate in some
638 circumstances. (thanks to Simon Alexander)
639 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
640 merged with *DEFAULT-PATHNAME-DEFAULTS*.
641 * bug fix: callbacks on OS X now preserve stack-alignment.
642 (thanks to Cyrus Harmon)
643 * enhancement: the x86-64 disassembler is much better at
644 disassembling SSE instructions. (thanks to Lutz Euler)
645 * enhancement: CHECK-TYPE on a variable now allows type-inference
646 (thanks to Vincent Arkesteijn)
647 * optimization: improved performance of EUC-JP external format.
648 (thanks to NIIMI Satoshi)
649 * optimization: performance improvements to IO on file streams of
650 :ELEMENT-TYPE CHARACTER
651 * optimization: much faster memory allocation on x86-64
652 * optimization: faster garbage collections (latency and throughput)
654 * optimization: faster bignum subtraction and division on x86 and x86-64
655 (thanks to Lutz Euler)
656 * optimization: more accurate type-inference for loops that use a
657 floating point index variable or a negative step.
659 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
660 * bug fix: add a workaround to SBCL looping infinitely at startup on
661 Linux kernels with apparently buggy implementations of personality().
662 (thanks to Svein Ove Aas)
663 * bug fix: Unicode symbols are correctly printed in LDB backtraces
664 (thanks to David Lichteblau)
665 * bug fix: local bindings shadow global symbol macros for
666 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
667 * optimization: non-open coded uses of numeric comparison operators
668 (e.g. >) no longer cons when called with more than one parameter
669 on platforms supporting dynamic-extent allocation.
670 * enhancement: saving cores with foreign code loaded is now supported
671 on MIPS/Linux in addition to the previously supported platforms.
672 * bug fix: FIND-RESTART now tests for activity, not applicability when given
673 a restart object as identifier. (reported by Helmut Eller for CMUCL)
674 * bug fix: division by zero in sb-sprof when no samples were collected
675 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
676 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
677 (thanks to Alexander Kjeldaas)
678 * new feature: ignoring the return values of destructive functions
679 like NREVERSE signals a compile-time style-warning.
680 (thanks to Kevin Reid)
681 * new feature: the alignment of alien structure fields can be
682 explicitly specified. (thanks to Cyrus Harmon)
683 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
684 and dump core on SIGQUIT
686 ** incompatible change: threads do not inherit values of specials
687 from their parents (see manual)
688 ** bug fix: threads stacks belonging to dead threads are freed by the
689 next exiting thread, no need to gc to collect thread stacks anymore
690 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
691 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
692 ** bug fix: dynamic variable and thread start related gc lossage
693 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
695 * fixed some bugs revealed by Paul Dietz' test suite:
696 ** SUBTYPEP is slightly more accurate on heinously complicated
697 CONS types where some of the members have uncertain (in the
698 NIL, NIL sense) type relationships to each other.
700 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
701 no longer purified unless :PURIFY T is explicitly specified.
702 ** Non-purified cores are significantly smaller than before
704 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
705 * new feature: timers based on Zach Beane's excellent timer package
706 * added support for the following external formats: koi8-u,
707 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
708 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
709 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
710 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
711 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
712 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
713 * incompatible change: a threaded SBCL will no longer revert to
714 non-threaded mode on non-NPTL systems, but refuse to start entirely.
715 * bug fix: interrupts are disabled until startup is complete; no
716 more sigsegvs when receiving a signal to soon
717 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
719 * bug fix: add a workaround for the memory randomization features in
720 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
721 workaround will only be in effect on systems with the proc filesystem
723 * bug fix: printing objects of type HASH-TABLE signals a
724 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
726 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
727 have been running for over 50 days. (reported by Gilbert Baumann)
728 * bug fix: the logic for getting names of functions gets less
729 confused when confronded with alternate-metaclass
730 funcallable-instances. (reported by Cyrus Harmon)
731 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
732 consistent, even on internal alternate-metaclass objects.
733 * bug fix: flush closure information collected by physical
734 environment analysis prepass before the main pass. (bug reported
736 * bug fix: compiler pack phase does not modify a hashtable, which is
737 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
738 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
739 may be expanded. (reported by Alexander <malishev> on c.l.l)
740 * bug fix: no more style-warnings when compiling code using LAST
741 under high-SPEED low-SPACE settings. (reported by David Wallin
743 * fixed some incompatibilities between SBCL's MOP and the MOP
745 ** the METAOBJECT class is now implemented;
746 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
747 STANDARD-OBJECT, as required;
748 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
749 now compatible (as required by VALIDATE-SUPERCLASS); there
750 remains a consistency requirement over the presence or absence
751 of the FUNCTION class in the superclass list of finalized
752 classes; see the manual for more details;
753 ** the :AROUND method for COMPUTE-SLOTS on
754 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
755 requested slot ordering.
757 ** bug fix: parent thread now can be gc'ed even with a live
759 ** bug fix: binding a special with PROGV to no value is not
760 the same as that symbol not having been bound (thanks to
762 ** bug fix: binding specials is thread safe (thanks to
764 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
765 and SB-SPROF do not die with 'no handler for signal XX in
766 interrupt_handle_now(..)' anymore
767 ** bug fix: WITH-TIMEOUT works with multiple running threads
768 * fixed some bugs revealed by Paul Dietz' test suite:
769 ** ENSURE-GENERIC-FUNCTION should take a method class object for
770 the :method-class keyword argument.
772 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
773 * new port: the Solaris operating system on x86 processors is now
774 mostly supported, though some rough edges in the environment
775 remain. (thanks to Daisuke Homma)
776 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
777 space for its heap. (thanks to Thiemo Seufer)
778 * optimization: SBCL on MIPS platforms now supports dynamic-extent
779 closures. (thanks to Thiemo Seufer)
780 * minor incompatible change: eof selects abort in the debugger.
781 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
782 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
784 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
785 deprecated, and will go away in a future revision of SBCL.
786 * minor incompatible change: GC-ON and GC-OFF are no longer
787 implemented with a counter, it does not matter how many times gc
788 is switched on or off
789 * bug fix: discriminating functions for generic function classes
790 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
791 make invalid assumptions about method precedence order. (reported
793 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
794 platforms. (thanks to Thiemo Seufer)
795 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
796 * bug fix: correct pathname printing with printer escaping is on.
797 (thanks to Kevin Reid)
798 * bug fix: complex VOP definitions in "user-space" no longer trigger
799 package locks. (reported by Zach Beane)
800 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
801 for funcallable-instance functions, and is no different from
803 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
804 characters in the current readtable. (reported by Nicholas Neuss)
805 * bug fix: SUBTYPEP on various CONS types returns more a more
806 accurate acknowledgment of its certainty.
807 * optimizations: REMOVE-DUPLICATES now runs in linear time on
808 lists in some cases. This partially fixes bug 384.
809 * flush all standard streams before prompting in the REPL and the
811 * bug fix: signal handling and triggering gc do not conflict
812 directly anymore, in particular a high frequency sb-sprof does
813 not prevent gc from running
814 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
815 approximation for timezone and DST information between the
816 universal time epoch and the smallest negative 32-bit time_t.
817 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
818 year 1899 so long as the encoded time is non-negative.
819 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
820 the correct answer for non-constant rotate amounts when performing
821 an inline 32-bit rotation.
823 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
824 there is only one thread in the session
825 ** bug fix: memory leak for streams created in one thread and
826 written to in another
827 ** bug fix: lockup when compiled with gcc4
828 ** bug fix: race that allows the gc to be triggered when gc is
830 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
831 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
833 ** bug fix: finished threads are gc'ed properly
834 * fixed some bugs revealed by Paul Dietz' test suite:
835 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
836 has arguments with a different set of applicable methods from
837 the orignal arguments.
838 ** The type error thrown by MAP now has a correct expected-type
840 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
841 name a compiled function.
842 ** Three MISC tests where a large bignum was improperly coerced to
843 a float that couldn't represent that bignum during type
844 derivation were fixed.
845 ** SUBTYPEP can now handle types involving the negation of a
846 list-form FUNCTION type.
847 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
848 as (COMPLEX (AND RATIO (NOT FIXNUM))).
849 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
851 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
852 * New feature: Experimental support for bivalent streams: streams
853 opened with :element-type :default now allow character and binary
854 (unsigned-byte 8) I/O
855 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
856 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
858 * bug fix: no more highly sporadic "couldn't check whether ~S is
859 readable" when reading a stream and an interrupt hits in the middle
860 of a select system call
861 * compiler better recognizes complex arrays (reported by Eduardo
863 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
864 for variables on SBCL built with linkage-tables. (reported by Luis
866 * various error reporting improvements.
867 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
868 (Thanks to Thiemo Seufer)
869 * bug fix: nested reader invokations work correctly
870 * bug fix: it is possible to have simultaneous references to foreign
871 code and foreign data with the same name.
873 ** added x86-64 support
874 ** incompatible change: the threading api now works with thread
875 objects instead of thread ids
876 ** bug fix: threads are protected from signals and interruption when
877 starting up or going down
878 ** bug fix: a race where an exiting thread could lose its stack to gc
879 ** bug fix: don't halt on infinite error in threads if possible
880 ** fixed numerous gc deadlocks introduced in the pthread merge
881 ** bug fix: fixed thread safety issues in read and print
882 ** bug fix: debugger doesn't hang on session lock if interrupted at
883 an inappropriate moment
884 ** bug fix: run-program is now thread safe(r)
885 ** bug fix: inner with-recursive-lock no longer releases the mutex
886 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
887 * fixed some bugs revealed by Paul Dietz' test suite:
888 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
889 EXPECTED-TYPE slots filled.
890 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
892 ** MAKE-CONDITION accepts classes as type-designators.
893 ** COMPILE may never return NIL.
894 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
895 range before calling Unix time functions
897 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
898 * numerous signal handling fixes to increase stability
899 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
900 * minor incompatible change: we now correctly canonize default
901 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
902 as per the MOP, rather than the historical (INITARG INITFUNCTION
903 INITFORM). (reported by Bruno Haible)
904 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
905 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
906 for more information.
907 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
908 pathname is a directory pathname.
909 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
910 * dynamic space size on PPC has been increased to 768Mb. (thanks to
912 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
913 :METACLASS argument in addition to a class name. (reported by
914 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
915 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
916 (reported by Bruno Haible; patch from Gerd Moellmann)
917 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
919 * bug fix: more cleanups to the floating point exception handling on
920 x86-64 (thanks to James Knight)
921 * bug fix: the compiler does not try to constant fold calls of
922 COERCE and BIT-* functions when they return freshly constructed
923 objects. (reported by Nikodemus Siivola and Paul Dietz)
924 * optimization: improved the MIPS versions of generic (in the
925 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
926 * optimization: direct conversion of (unsigned-byte 32) to floats on
927 the PowerPC platform.
928 * optimization: structure instances with raw slots now use less
929 memory, and probably show better memory locality. (thanks to
931 * optimization: DYNAMIC-EXTENT declarations for lists and closures
932 are treated as requests for stack allocation on the x86-64,
933 Alpha-32, PPC and SPARC platforms.
934 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
935 the wrong file descriptor; implementation of SOCKET-OPEN-P.
936 (thanks to Tony Martinez)
938 ** gcing a dead thread can no longer lead to lockups
939 ** threads block signals until they are set up properly
940 ** errno is no longer shared by threads
941 ** interrupt-thread restores the eflags register on x86
942 ** fixed some lockups due to gc/thread interaction
943 * fixed some bugs revealed by Paul Dietz' test suite:
944 ** invalid dotted lists no longer raise a read error when
946 ** signal an error if a symbol that names a declaration is used
947 as the name of a type, or vice versa
948 ** allow using the (declare (typespec var*)) abbreviation for
949 (declare (type typespec var*)) with all type specifiers
950 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
951 ** always return NIL from PPRINT-POP when OBJECT is NIL
952 ** don't signal errors when pretty-printing malformed LABELS,
953 FLET or MACROLET forms
954 ** declarations in a DOLIST body are also in scope for the
956 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
958 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
961 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
962 * fixed cross-compiler leakages that prevented building a 32-bit
963 target with a 64-bit host compiler.
964 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
965 opened with :IF-EXISTS :OVERWRITE.
966 * fixed bug 281: error for an invalid qualifier in a short-form method
967 combination method is not signalled until the faulty method is called.
968 * bug fix: iteration variable type inferrer failed to deal with open
969 intervals. (reported by Alan Shields)
970 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
971 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
972 or more runtime options were provided to the sbcl binary.
973 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
974 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
975 * optimization: there's now a fast-path for fixnum arguments in the
976 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
977 * optimization: the code generated on x86-64 is more compact thanks
978 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
979 * bug fix: floating-point exception handling now partly works on
980 x86-64. (Thanks to James Knight)
981 * improvement to the MIPS backend from Thiemo Seufer: C-style
982 64-bit long long arguments and return values to alien functions
984 * fixed some bugs revealed by Paul Dietz' test suite:
985 ** the type-error signalled from WARN has a filled-in DATUM slot.
986 ** the type-error required when a stream is not associated with
987 a file has the stream as its datum.
988 ** type-errors on single-floats on x86-64 no longer have
989 :INVALID-OBJECT as the datum
990 ** the type-errors signalled for invalid function names now have
991 a correct expected type
992 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
993 for structures defined with a :TYPE. Documentation strings for
994 typed structures are no longer immediately discarded
995 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
996 broadcast streams and synonym streams. FILE-LENGTH now also works
997 on broadcast streams.
999 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
1000 * incompatible change: the --noprogrammer option, deprecated since
1001 version 0.7.5, has been removed. Please use the equivalent
1002 --disable-debugger option instead.
1003 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
1005 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
1006 inoperational for a while now) has been completely removed.
1007 * null lexical environments are now printed as #<NULL-LEXENV>,
1008 significantly reducing the amount of clutter in typical backtraces.
1009 * documentation on weak pointers, finalization, and after GC hooks
1010 has been added to the manual.
1011 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
1012 as other specialized array types, is much faster. SUBSEQ and
1013 COPY-SEQ on such arrays have also been sped up.
1014 * optimization: EQL is now more efficient when at least other argument
1015 is known to be of type (OR FIXNUM (NOT NUMBER)).
1016 * fixed bug: the runtime is now less vulnerable to changes in the
1017 size of the SBCL object on OS X, and virtual memory is reserved for
1018 all spaces that need to be at a fixed address.
1019 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
1020 * fixed bug: finalizers and after GC hooks that cause consing are now
1022 * fixed bug: compiler error messages and summaries are now printed to
1023 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1024 * fixed inference of the upper bound of an iteration variable.
1025 (reported by Rajat Datta).
1026 * fixed bug 211e: calling local functions with duplicated constant
1027 keyword argument no longer causes a bogus style warning about an
1029 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1030 ftype declaration to be disregarded. (reported by Dave Roberts)
1031 * fixed bug 373: caused by erronous compilation of references to alien
1032 variables in the runtime on ppc/darwin.
1033 * fixed bug 376: CONJUGATE type deriver.
1034 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1035 (reported by Baughn on #lisp)
1036 * a cross-compiler bug on non-x86 platforms has been identified and
1037 fixed. (thanks to Bruno Haible)
1038 * improvements to the MIPS runtime code for increased stability.
1039 (thanks to Thiemo Seufer)
1040 * increased the maximimum compact environment size to allow
1041 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1042 * partial workaround for bug 135: don't name the function we're
1043 calling for hairy cases of EVAL, so as not to accumulate one environment
1044 entry per call to EVAL. (reported by Kevin Reid)
1045 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1046 * optimization: type testing for non-vector arrays should be faster.
1047 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1049 * bug fix: setting 31st element of a bit vector to zero did not work
1051 * bug fix: redefining a class definition which failed due to a
1052 previous accessor / function clash now works (but see BUGS entry
1053 #380 for more problems in this area). (thanks to Zach Beane)
1054 * the long form of DEFINE-METHOD-COMBINATION disables method group
1055 checking when given a single method group with pattern *. (thanks
1057 * on x86 compiler supports stack allocation of results of simple
1058 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1059 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1060 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1061 * fixed some bugs related to Unicode integration:
1062 ** the restarts for recovering from input and output encoding
1063 errors only appear when there is in fact such an error to
1065 * fixed some bugs revealed by Paul Dietz' test suite:
1066 ** MISC.549 and similar: late transformation of unsafe type
1067 assertions into derived types caused unexpected code
1069 ** SCALE-FLOAT type deriver is less wrong.
1070 ** type derivers for EXP, LOG and similar functions compute result
1071 types for complex arguments better.
1072 ** (MISC.563) CONJUGATE type deriver works for very restricted
1074 ** out-of-line type testers for character strings are available.
1075 ** EQUAL compiler transform understands specialness of objects
1077 ** accessing double-floats stored on the stack now works on x86-64.
1078 ** debugger internals could sometimes create invalid lispobjs,
1079 resulting in GC crashes.
1080 ** MISC.548: type check weakening can convert required type into
1082 ** initialization forms for bindings are not in scope of free special
1085 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1086 * incompatible change: thread support for non-NPTL systems has
1087 been removed - locking is buggy and unreliable. A threaded
1088 SBCL build will now warn at startup and refuse to create
1089 new threads, unless futex support is detected in the kernel
1090 * incompatible change: the top level REPL now has only an ABORT
1091 restart associated with it, not TOPLEVEL and ABORT as it used to.
1092 TOP and TOPLEVEL are now available as debugger commands for
1093 returning to the top level.
1094 * incompatible change: forms evaluated in the REPL now use the
1095 global optimization policy.
1096 * incompatible change: user- and system-initialization files are
1097 no longer processed with LOAD, but by READ and EVAL; hence the
1098 global optimization policy, startup package, readtable, etc,
1100 * The .fasl file format number has been incremented because of
1101 various incompatible changes.
1102 * internal entry point details and argument counts no longer appear
1103 in backtraces unless explicitly requested by setting
1104 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1105 * built-in and standard functions no longer have names like "top
1106 level local call to FOO".
1107 * fixed bug 32: functions defined in non-null lexical environments
1108 now have more legible printed representation
1109 * fixed bug 33: functions defined in non-null lexical environemnts
1110 are now more amenable to inspection by INSPECT.
1111 * workaround for bug 354: XEPs no longer appear in backtraces unless
1112 explicitly requested.
1113 * fixed bug: receiving the signal which results from attempting to
1114 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1115 SIGBUS on other *BSDs) on architectures where the C stack is also
1116 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1117 to debugger confusion or wild instability. (reported by Cheuksan
1119 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1120 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1121 (reported by Lutz Euler)
1122 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1123 compiled code. (reported by Paul Dietz)
1124 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1125 specialized parameter to a method to something that is not TYPEP
1126 the specializer is now possible.
1127 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1128 face of package deletion.
1129 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1130 pathname. (reported by David Tolpin on c.l.l)
1131 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1132 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1133 STRING-STREAM and FILE-STREAM, respectively.
1134 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1135 than previously; in addition, its test suite is now run on build.
1136 (thanks to Luke Gorrie)
1137 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1138 * optimization: added a immediate representation for single-floats
1140 * fixed some bugs revealed by Paul Dietz' test suite:
1141 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1142 ** attempting to create a package with a colliding nickname causes
1143 correctable errors to be signalled.
1144 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1145 ** several x86-64 backend bugs related to sign-extension of immediate
1148 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1149 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1150 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1152 * fixed loading of multiply forward-referenced layouts.
1153 (thanks to Cheuksan Wang)
1154 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1155 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1156 (see COMPILE-FILE documentation for details of :PRINT option)
1157 * fixed bugs 19 and 317: fixed-format floating point printing is
1158 more accurate. This also fixes a bug reported by Adam Warner
1159 related to the ~@F format directive.
1160 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1162 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1163 dispatch table if the from-char is a dispatch macro character.
1164 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1165 by the wordsize no longer ignore the last word. (reported by Lutz
1167 * fixed bug in type checking of dynamic-extent variables. (reported
1169 * optimization: sequence traversal functions use their freedom to
1170 coerce function designators to functions.
1171 * optimization: code with many calls to CLOS methods specialized on
1172 CLOS classes has had redundant type checks removed.
1173 * optimization: added declarations to speed up operations that access
1174 the internal character database (for example STRING-UPCASE)
1175 * optimization: comparison operations between floats and sufficiently small
1176 fixnums no longer create extra rationals
1177 * fixed some bugs related to Unicode integration:
1178 ** portions of multibyte characters at the end of buffers for
1179 character-based file input are correctly transferred to the
1180 start of the buffer at the next read.
1181 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1182 passing it through to OPEN.
1183 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1184 argument given to internal calls to OPEN.
1185 ** the built-in comment readers (introduced by character sequences
1186 ";" and "#|") are more forgiving to encoding errors; they will
1187 STYLE-WARN and then attempt to resync the stream at a character
1188 boundary. (thanks to Teemu Kalvas)
1189 * fixed some bugs in the x86-64 port:
1190 ** Negative short int return values from c-calls are sign-extended
1192 ** The stack is aligned to 16-bytes for c-calls, as required by
1193 the ABI. (thanks to Cheuksan Wang)
1194 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1195 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1196 for enums. (thanks to Vincent Arkesteijn)
1197 ** Multiple small optimizations and bugfixes for floating point
1199 * fixed some bugs revealed by Paul Dietz' test suite:
1200 ** Space, Tab, Linefeed, Return and Page have the invalid
1201 secondary constituent character trait.
1202 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1204 ** WITH-INPUT-FROM-STRING should only update the index place on
1206 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1207 works more reliably.
1208 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1209 with one-dimensional arrays with fill pointers.
1210 ** TYPE-OF failed on a complex with an integer realpart and a
1212 ** compiler failure during type inference for the code of form
1213 (IF (EQL X (THE ...)) ...) (MISC.535).
1215 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1216 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1217 platform. The port supports 61-bit fixnums, large memory spaces
1218 and reloading shared object files.
1219 * enhancement: saving cores with foreign code loaded is now
1220 supported on ppc/Darwin in addition to the previously supported
1222 * enhancement: the statistical profiler now walks deeper into the
1223 call stack for more meaningful call-graphs and accrued time
1224 reports (x86/x86-64 only). It also now reports time spent in
1226 * enhancement: it is now possible to trace most individual methods
1227 of a generic function in addition to tracing the generic function
1229 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1231 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1232 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1233 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1234 produces an error. (thanks to Vincent Arkesteijn)
1235 * bug fix: NAMESTRING on pathnames with :WILD components in their
1236 directories works correctly. (thanks to Artem V. Andreev)
1237 * fixed bug 125: compiler preserves identity of closures. (reported
1239 * bug fixed: functions with &REST arguments sometimes failed with
1240 "Undefined function" when compiled with (DEBUG 3). (reported by
1242 * bug fix: overflow during compiling of setting element of a bit
1243 vector with constant index and value. (reported by Timmy Douglas)
1244 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1245 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1246 * on x86 compiler partially supports stack allocation of dynamic-extent
1248 * GO and RETURN-FROM do not check the extent of their exit points
1249 when compiled with SAFETY 0.
1250 * fixed some bugs related to Unicode integration:
1251 ** encoding and decoding errors are now much more robustly
1252 handled; it should now be possible to recover even from invalid
1253 input or output to the terminal. (thanks to Teemu Kalvas)
1254 ** provided a first cut at implementing STRING-TO-OCTETS and
1255 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1256 ** altered the SB-MD5 contributed module slightly, changing the
1257 interface just enough for it to be supportable for builds where
1258 lisp characters are not eight bits.
1259 * fixed some bugs revealed by Paul Dietz' test suite:
1260 ** the FORMATTER-generated functions for ~V[ conditionals require
1261 the correct number of arguments.
1262 ** READ-FROM-STRING returns the mandated second value when applied
1263 to displaced strings.
1264 ** the #\Rubout and #\Backspace characters are treated as invalid
1265 constituent characters by the tokenizer.
1267 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1268 * new feature: reloading changed shared object files with
1269 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1270 * new feature: references to foreign variables and functions
1271 can now be compiled and loaded before the corresponding shared
1272 object file is loaded, as long as the foreign definitions are
1273 available at runtime.
1274 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1275 supported. (thanks to Dan Debertin)
1276 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1277 just keywords. (thanks to Vincent Arkesteijn)
1278 * fixed bug #331: structure-class instances corresponding to
1279 DEFSTRUCT forms are now created eagerly.
1280 * fixed bug #345: backtraces from calls to undefined functions work
1281 on x86 as well. Related bug #61 is now also partially fixed on x86
1282 (backtraces from throws to unknown catch tags.)
1283 * bug fix: lambda-list parsing is now stricter vrt. order and number
1284 of lambda-list keywords.
1285 * bug fix: as specified by AMOP, an error is signalled if a
1286 class-option appears multiple times in a DEFCLASS form. (reported
1288 * bug fix: RANDOM can be compiled when the compiler derives the type
1289 of its numeric argument as a disjoint set of small integers.
1290 (reported by Paul Dietz)
1291 * bug fix: starting a core saved with shared objects loaded when
1292 those objects are not available no longer causes threaded SBCL to
1293 hang. (reported by Sean Ross)
1294 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1296 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1297 argument, which must always be NIL. (reported by Kalle Niemitalo)
1298 * bug fix: printing 1.0d+23 no longer results in an error.
1299 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1300 * bug fix: structure slot setters preserve evaluation order. (thanks
1302 * fixed some bugs related to Unicode integration:
1303 ** RUN-PROGRAM can allow its child to take input from a Lisp
1304 stream. (reported by Stefan Scholl)
1305 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1306 locales. (reported by Ken Causey)
1307 * fixed some bugs revealed by Paul Dietz' test suite:
1308 ** INCF, DECF and REMF evaluate their place form as specified in
1310 ** FORMATTER expands ~{ ~} iteration directives with V or #
1311 parameters correctly.
1312 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1313 consequent uses no arguments correctly.
1314 ** the system has a partial understanding of the (COMPLEX RATIO)
1316 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1319 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1320 * new feature: a build-time option (controlled by the :SB-UNICODE
1321 keyword feature, enabled by default) for building the system with
1322 support for the entire 21-bit character space defined by the
1324 * new feature: the system now has rudimentary external-format
1325 support; the primary user-visible change at this time is that
1326 characters with the high bit set (such as lower-case-e-acute) will
1327 print correctly to a terminal in a UTF-8 environment.
1328 * minor incompatible change: BASE-CHAR no longer names a class;
1329 however, CHARACTER continues to do so, as required by ANSI.
1330 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1331 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1332 has been moved to the SB-EXT package (temporarily re-exported from
1334 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1335 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1337 * fixed bug #351: better error-handlind and reporting for malformed
1339 * fixed bug #350: bignum-printing is now more memory-efficient,
1340 allowing printing of very large bignums, eg. (expt 2 10000000).
1341 (reported by Bruno Haible)
1342 * fixed bug #302: better primitive-type selection for intersection
1344 * fixed bug #308: non-graphic characters now all have names, as
1345 required. (reported by Bruno Haible)
1346 * bug fix: redefining a class with different superclasses now correctly
1347 removes it from the direct-subclasses of its previous superclasses.
1348 (reported by David Morse)
1349 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1350 new value now works. (reported by Bruno Haible)
1351 * bug fix: correct canonicalization of multiple non-standard slot
1352 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1353 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1354 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1355 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1356 now exists, an signals an error.
1357 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1358 during macroexpansion for non-iteration variables as well. (reported
1359 by Bruno Haible for CMUCL)
1360 * bug fix: Cyclic structures and unprintable objects in compiler
1361 messages no longer cause errors. (reported by Bruno Haible)
1362 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1363 and READ-FROM-STRING all now return a primary value of NIL if
1364 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1365 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1366 (reported by Bruno Haible for CMUCL)
1367 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1368 do not designate a condition. (reported by Bruno Haible for
1370 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1371 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1372 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1373 permitting resolution in favour of any of the conflicting symbols.
1374 (reported by Bruno Haible for CMUCL)
1375 * bug fix: EQUAL compiler optimizations is less aggressive on
1376 strings which can potentially compare true despite having distinct
1377 specialized array element types.
1378 * bug fix: unit enumerations can be defined without dividing by
1379 zero. (thanks to Vincent Arkesteijn)
1380 * FORMAT compile-time argument count checking has been enhanced.
1381 (report from Bruno Haible for CMUCL)
1382 * a partial workaround for the bug 262: the compiler does not try to
1383 inline-expand a local function doing RETURN-FROM from a deleted
1384 BLOCK. (thanks to Peter Denno for the bug report and to David
1385 Wragg for the simple test case)
1386 * fixed some bugs revealed by Paul Dietz' test suite:
1387 ** RENAME-PACKAGE allows all package designators as new package
1389 ** constraint propagation and lambda variable substitution are
1390 more cautious in dealing with partially deleted code.
1391 ** compiler sometimes lost reoptimization passes.
1392 ** CERROR, given a condition as condition designator, passes its
1393 remaining arguments to the continue format control without
1395 ** Case-altering FORMAT directives work correctly on non-ASCII
1397 ** The REAL type specifier handles bounds outside the floating
1398 point ranges without signalling FLOATING-POINT-OVERFLOW.
1399 ** Functions with IR1-transformations can create intercomponent
1400 references to global functions.
1401 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1403 ** FORMAT ~R treats a nil value for its first parameter correctly.
1405 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1406 * enhancement: saving cores with foreign code loaded is now
1407 supported on x86/NetBSD and sparc/Linux in addition to the previously
1408 supported platforms.
1409 * bug fix: on some platforms repeated installations caused multiple
1410 copies of HTML documentation to be installed -- should not happen
1411 any more. (reported by Stefan Scholl)
1412 * bug fix: parsing self-recursive alien record types multiple times
1413 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1414 original patch by Helmut Eller for CMUCL)
1415 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1416 (thanks to Richard Kreuter)
1417 * bug fix: defining classes whose accessors are methods on existing
1418 generic functions in other (locked) packages no longer signals
1419 bogus package lock violations. (reported by François-René Rideau)
1420 * bug fix: special variables as DEFMETHOD parameters no longer have
1421 associated bogus type declarations. (reported by David Wragg and
1423 * bug fix: read-write consistency on streams of element-type
1424 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1425 * bug fix: redefiniton of the only method of a generic function with
1426 no DEFGENERIC no longer emits a full WARNING. In addition,
1427 redefinition of generic functions with no DEFGENERIC to an
1428 incompatible lambda list now signals an error. (thanks to Zach
1430 * bug fix: DEFGENERIC now works even when there's a function of the
1431 same name in an enclosing lexical environment. (thanks to Zach
1433 * fixed compiler failure, caused by instrumenting code during
1434 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1435 * optimization: added loop analysis and improved register allocation
1436 * fixed some bugs revealed by Paul Dietz' test suite:
1437 ** POSITION on displaced vectors with non-zero displacement
1438 returns the right answer.
1439 ** (SIMPLE-STRING) is a valid type specifier for sequence
1441 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1443 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1444 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1446 ** The justification version of the FORMAT ~< directive treats
1447 non-zero minpad parameter correctly.
1449 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1450 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1451 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1452 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1453 the supported interface.
1454 * new feature: Single-stepping of code compiled with DEBUG 2 or
1455 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1456 * new feature: saving cores with foreign code loaded is now
1457 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1458 Timothy Moore's work for CMUCL)
1459 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1460 parameters to * instead of NIL if no initform is supplied.
1461 (reported by Johan Bockgaard)
1462 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1463 correctly when FUNCALL appears as the car of the form. Note:
1464 despite this FUNCALL forms are not currently subject to
1465 compiler-macro expansion. (port of Raymond Toy's fix for the
1466 same from CMUCL, reported by Johan Bockgaard)
1467 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1468 (thanks for Teemu Kalvas)
1469 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1470 case instead of terminating the iteration (thanks for Julian
1471 Squires, Sean Champ and Raymond Toy)
1472 * bug fix: incorrect expansion of defgeneric that caused a style
1473 warning. (thanks for Zach Beane)
1474 * bug fix: slot accessor effective method computation works properly
1475 for classes with multiple non-standard applicable methods on
1476 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1477 * on x86 compiler supports stack allocation of results of LIST and
1478 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1479 CMUCL implementation by Gerd Moellmann)
1480 * fixed some bugs revealed by Paul Dietz' test suite:
1481 ** FORMAT strings with both the ~<~:;~> form of the justification
1482 directive and pretty-printing directives cause an error.
1484 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1485 * incompatible change: the internal functions
1486 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1487 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1488 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1489 instead of the old functions.
1490 * new feature: on platforms where "dladdr" is available foreign
1491 function names now appear in backtraces. (based on Helmut Eller's
1493 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1494 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1496 * documentation: ASDF manual is now distributed with SBCL.
1497 * bug fix: SBCL can now load its contributed modules with REQUIRE
1498 even if the system-provided entries have been removed from
1499 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1501 * bug fix: backtraces involving undefined functions or assembly
1502 routines are more informative. (thanks to Brian Downing)
1503 * bug fix: mutually referent alien structures now work correctly.
1504 (reported by Rick Taube)
1505 * bug fix: structures defined by WITH-ALIEN can be referred to
1506 within other definitions in the same WITH-ALIEN.
1507 * bug fix: division operators (MOD, TRUNCATE and the like) with
1508 constant zero divisors and integer dividends no longer generate
1510 * bug fix: provide default methods for INPUT-STREAM-P and
1511 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1512 * bug fix: improve the detection and resolution of MOP metacycles
1513 (where effective method computation depends on the generic function
1514 having its effective method computed). (reported by Bruno Haible)
1515 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1516 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1517 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1518 * bug fix: pathnames with relative directory components can be
1519 represented relative to default pathnames.
1520 * optimization: in taking the GCD of bignums, reduce the two bignums
1521 to approximately the same size (using Euclid's algorithm) before
1522 applying the more sophisticated binary GCD. (thanks to Juho
1524 * optimization: COUNT on bitvectors now operates word-at-a-time.
1525 * optimization: ASH with a positive, but not necessarily constant,
1526 (leftwards) shift, when in a modular context, is compiled to a
1528 * fixed some bugs revealed by Paul Dietz' test suite:
1529 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1530 if the corresponding argument is NIL.
1531 ** FORMAT directives accepting a minpad parameter treat negative
1533 ** literal commas as character parameters to FORMAT directives are
1535 ** literal spaces directly after ~<Newline> directives within a
1536 format-logical-block (~:< ~@:>) do not induce :FILL-style
1537 conditional newlines.
1538 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1539 as specified: it no longer includes conditional newlines).
1540 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1542 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1543 :PER-LINE-PREFIX argument does not evaluate to a string.
1545 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1546 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1547 the manual for details; for now, package locks can be disabled by
1548 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1549 if no major problems are found then it is likely that they will be
1550 compiled in unconditionally.
1551 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1552 unsupported operators on all platforms. To load a shared library
1553 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1554 object file, link it into a shared library outside of SBCL and
1555 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1557 * bug workaround: ROOM T and the GC were not working together
1558 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1559 reporting facility aren't preserved. That reporting has been
1560 disabled, so now until and unless someone figures out how to make it
1561 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1562 * minor incompatible change: as threatened around sbcl-0.8.0, the
1563 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1564 The symbols are also exported from SB-PCL for backwards
1565 compatibility, but more so than before SB-PCL should be treated as
1566 an implementation-internal package.
1567 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1569 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1570 It is known as of this release not to work on the Alpha, however.
1571 * fixed bug #167: errors signalled due to illegal syntax in method
1572 bodies are now more legible.
1573 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1574 designators and can hence be used with TYPEP.
1575 * fixed bug #333: CHECK-TYPE now ensures that the type error
1576 signalled, if any, has the right object to be accessed by
1577 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1578 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1579 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1581 * fixed bug #334: programmatic addition of slots using specialized
1582 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1583 and :ALLOCATION :CLASS slots.
1584 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1585 float exponents. (rereported by Peter Seibel)
1586 * fixed a bug: #\Space (and other whitespace characters) are no
1587 longer considered to be macro characters in standard syntax by
1588 GET-MACRO-CHARACTER.
1589 * fixed bug: initialization of condition class metaobjects no longer
1590 causes an instance of the condition to be created. (reported by
1592 * fixed bug: it is now possible to have slots such that
1593 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1594 description is neither :INSTANCE nor :CLASS.
1595 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1596 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1597 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1598 non-standard SLOT-DEFINITION-ALLOCATION values.
1599 * fixed bugs: various race conditions handling exiting threads.
1600 CL-PPCRE's thread torture test now passes.
1601 * fixed arguably-a-bug: GC time proportional to number of threads in
1602 system even when most of them are idle
1603 * optimization: improved performance of BIT and SBIT on bit-vectors.
1604 * .fasl file incompatibility: The fasl file version number has
1605 been incremented because of changes associated with package locks.
1607 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1608 * minor incompatible change: the system no longer provides
1609 optimization hints (conditions which are TYPEP
1610 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1611 mode; these hints are still emitted when higher SPEED optimization
1613 * new contrib module: a sampling profiler (profiling by statistical
1614 sampling, rather than by instrumenting functions) is available as
1615 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1616 * the behaviour of the standard function ED is now customizeable by
1617 third parties through a hook variable: see ED's documentation
1618 string for information on the protocol.
1619 * the compiler no longer emits efficiency notes for (FUNCALL X)
1620 when the type of X is uncertain under default optimization
1622 * fixed bug 276: mutating a binding of a specialized parameter to a
1623 method to something that is not TYPEP the specializer is now
1625 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1626 exhibit the required behaviour on the x86 platform. (thanks to
1627 Peter van Eynde, Eric Marsden and Bruno Haible)
1628 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1629 even for difficult arguments. (reported by Peter Graves)
1630 * fixed bug 141a: the backquote printer now descends quoted
1632 * fixed another bug in backquote printing: no more destructive
1633 modification of the form's list structure. (reported by Brian
1635 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1636 move between its address being taken and the call to
1637 interrupt_thread, fixing a crashing race condition.
1638 * the SB-POSIX contrib implementation has been adjusted so that it
1639 no longer exhibits ridiculously poor performance when constructing
1640 instances corresponding to C structs.
1642 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1643 * minor incompatible change: the sb-grovel contrib now treats C
1644 structures as alien (in the sense of SB-ALIEN) objects rather than
1645 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1646 has implications for memory management of client code
1647 (sb-grovel-returned objects must now be manually managed) and for
1648 type safety (alien objects now have full types).
1649 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1650 used to control emission of compiler diagnostics, rather than the
1651 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1652 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1653 quality should be considered deprecated.
1654 * install.sh now installs the user manual as well
1655 * (not quite a new documentable feature, but worth considering in
1656 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1657 beginnings of a semantically meaningful condition hierarchy is
1658 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1660 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1661 designator as the defaults argument.
1662 * fixed bug: Displaced arrays whose displaced-to array has become
1663 too small now cause ARRAY-DIMENSION to signal an error, providing
1664 for safer bounds-checking. (reported by Bruno Haible)
1665 * fixed bug: DEFCLASS slot definitions with identical :READER and
1666 :WRITER names now signal a reasonable error. (reported by Thomas
1668 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1670 * fixed bug: Invalid slot specification errors now print correctly.
1671 (thanks to Zach Beane)
1672 * fixed bug 320: Shared to local slot value transfers in class
1673 redefinitions now happen corrently from superclasses as
1674 well. (reported by Bruno Haible)
1675 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1677 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1678 improper lists correctly. (reported by Bruno Haible)
1679 * fixed bug 313: source-transform for <fun-name> was erroneously
1680 applied to a call of a value of a variable with name <fun-name>.
1681 (reported by Antonio Menezes Leitao)
1682 * fixed bug 307: The obsolete instance protocol ensures that
1683 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1684 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1685 remove dead unknown-values globs from the middle of the stack.
1686 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1687 easier to resume long computations after using *BREAK-ON-SIGNALS*
1688 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1689 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1690 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1691 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1692 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1693 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1695 * fixed a (fairly theoretical) bug in string printing: if
1696 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1697 does not have array-element-type equal to the most general string
1699 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1700 function name. (thanks to Nikodemus Siivola)
1701 * fixed bug: compile-time format string checker failed on
1702 non-closed ~{. (reported by Thomas F Burdick)
1703 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1704 don't warn on legal loop code involving more than one aggregate
1705 boolean. (thanks to Nikodemus Siivola)
1706 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1707 in keyword position to MAKE-PACKAGE was not being reported as
1709 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1710 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1712 * fixed bugs 280 and 312: the checking for multiple definitions in a
1713 file is less likely to become confused by uses of inline
1715 * fixed bug: the #S reader macro performs the keyword coercion
1716 specified for slot names. (reported by Kalle Niemitalo)
1717 * fixed bug: lambda lists may contain symbols whose names start with
1718 & but are not lambda-list-keywords; their occurrence triggers a
1720 * fixed bug 321: define-method-combination argument lambda lists do
1721 not cause invalid code to be generated when &optional or &aux
1722 variables have default values. (reported by Bruno Haible)
1723 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1724 have CLOS classes; this appears also to have fixed reported
1725 instability in stack exhaustion detection.
1726 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1727 MUFFLE-WARNING when no associated restart is present is now
1729 * optimization: rearranged the expansion of various defining macros
1730 so that each expands into only one top-level form in a
1731 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1733 * optimization: used a previously little-used slot in symbols to
1734 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1736 * fixed some bugs revealed by Paul Dietz' test suite:
1737 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1739 ** Fixed an optimization bug related to inheritance of initforms
1740 from local to shared slots.
1741 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1742 ** CAST optimizer forgot to flush argument derived type.
1743 ** print/read consistency on floats is now orders of magnitude
1744 more likely. (thanks also to Bruno Haible for a similar report
1746 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1747 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1748 for objects of type REAL. Make it so.
1749 ** SXHASH is specified (infelicitously) to respect similarity,
1750 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1751 it so. (thanks to Markus Ziegler)
1752 ** on the Alpha, the compiler succeeds in compiling functions
1753 returning a known number of arguments greater than 63.
1754 ** fixed handling of invalid NIL arguments in keyword position
1756 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1758 ** provided a readably-printable representation for RANDOM-STATE
1760 ** ensured that pathnames, where they have a namestring, always
1761 print using #P"..." syntax.
1763 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1764 * Support for the forthcoming 2.0 version of the NetBSD kernel
1765 running on x86 hardware has been added. (thanks to Perry
1766 E. Metzger most immediately, and others for their past work)
1767 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1768 current release version; SBCL's OpenBSD support had been broken
1769 since about the time of OpenBSD's switch to ELF binary format.)
1770 * [placeholder for DX summary]
1771 ** user code with &REST lists declared dynamic-extent, under high
1772 speed or space and low safety and debug optimization policy.
1773 * The manual has been converted to Texinfo format and the debugger
1774 chapter from the cmucl manual has been added.
1775 * A facility has been added to extract documentation strings from
1776 sbcl and store them as Texinfo-formatted snippets for inclusion in
1777 the manual (via Texinfo's @include directive)
1778 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1779 the test case to Dave Roberts)
1780 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1781 pointers. (reported by Sean Ross)
1782 * bug fix: PROFILE output is printed nicely even for large numerical
1783 values. (thanks to Zach Beane)
1784 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1785 greater than 32 handle EOF correctly.
1786 * bug fix: on X86 an immediate argument of the IMUL instruction is
1787 correctly printed in disassembly. (thanks to Lutz Euler)
1788 * bug fix: class slots in redefined classes preserve their old
1789 values. (thanks to Bruno Haible and Nikodemus Siivola)
1790 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1791 succeeds. (reported by Marco Baringer)
1792 * fixed some bugs revealed by Paul Dietz' test suite:
1793 ** READ-SEQUENCE now works on ECHO-STREAMs.
1794 ** RATIONALIZE works more according to its specification. (thanks
1797 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1798 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1799 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1800 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1801 should matter to you only if you rebind the printer control
1802 variables and then find you want different bindings in the
1803 debugger than in the ordinary execution of your program.)
1804 * The runtime build system has been tweaked to support building
1805 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1806 assembler and linker. (thanks to Nikodemus Siivola)
1807 * Unbound, undefined, undeclared variables now trigger full
1808 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1809 more often programmer error than deliberate exploitation of undefined
1811 * optimization: the hash algorithm for strings has changed to one
1812 that is less vulnerable to spurious collisions. (thanks to Juho
1814 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1815 needless bounds checking. (thanks to Juho Snellman)
1816 * optimization: implemented multiplication as a modular
1817 (UNSIGNED-BYTE 32) operation on the PPC backend.
1818 * fixed some bugs revealed by Paul Dietz' test suite:
1819 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1821 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1822 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1823 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1825 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1827 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1829 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1830 their output stream on EOF from read.
1831 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1832 have been read to end-of-file.
1833 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1835 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1836 description of determination of which consecutive characters
1838 ** Printing the "Space" character with escaping on now yields "#\\ ",
1839 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1840 ** Reading floating-point numbers with *READ-BASE* set to a number
1841 less than 10 works correctly.
1842 ** Reading floating-point numbers with *READ-BASE* set to a number
1843 more than 10 works correctly.
1844 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1845 the readtable currently in effect.
1847 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1848 * minor incompatible change: parsing of namestrings on a physical
1849 (Unix) host has changed; numbers after the final #\. in a
1850 namestring are no longer interpreted as a version field. This is
1851 intented to be largely invisible to the user, except that the
1852 meaning of the namestring "*.*.*" has changed: it now refers to a
1853 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1854 should usually be replaced by
1855 (make-pathname :name :wild :type :wild :version :wild)
1856 with the added benefit that this is more likely to be portable.
1857 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1858 option to OPEN now signals an error if the file being opened
1859 exists; this may have an impact on existing code.
1860 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1861 signals received, on the PowerPC platforms (both Linux and
1862 Darwin). (thanks to Pierre Mai for pointing out the location of
1864 * several fixes on OS X: The system now builds and runs cleanly on
1865 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1866 (thanks to Brian Mastenbrook)
1867 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1868 second-resolution: integer multiples of 1/3600 between -24 and 24.
1869 (thanks to Vincent Arkesteijn)
1870 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1871 type when called with 1 argument; PEEK-CHAR checked type of
1872 PEEK-TYPE only after having read first character from a
1873 stream. (reported by Peter Graves)
1874 * bug fix: the garbage collector now has much better locality
1875 behaviour, and in particular no longer treats all memory as being
1876 exhausted when many small objects point to each other in a deeply
1878 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1879 recognized as being TYPEP their class.
1880 * bug fix: the PUSHNEW documentation string has been corrected.
1881 (thanks to Vincent Arkesteijn)
1882 * bug fix: defaulting of the value for the last of an atypically
1883 large number of multiple values being bound was not being
1884 performed correctly on the Alpha or PPC platforms
1885 * optimization: implemented multiplication as a modular
1886 (UNSIGNED-BYTE 32) operation on the x86 backend.
1887 * optimization: SEARCH on simple-base-strings can now be open-coded.
1888 (see also contrib/compiler-extras.lisp for inspiration for
1889 teaching the compiler about the Boyer-Moore algorithm).
1890 * value, returned by MAX (and MIN) called with several EQUALP, but
1891 not EQL, arguments now does not depend on compiler settings.
1892 * fixed some bugs revealed by Paul Dietz' test suite:
1893 ** in stack analysis liveness information is propagated from
1894 non-local entry points.
1895 ** pathwise CAST removing failed when the CAST node did not start
1897 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1898 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1900 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1901 host is already defined.
1902 ** RENAME-FILE works on streams instead of signalling an internal
1904 ** PEEK-CHAR uses the current readtable when determining whether
1905 or not a character is whitespace.
1906 ** MERGE-PATHNAMES handles the case when the pathname does not
1907 specify a name while the default-pathname specifies a version.
1908 ** Pathnames now stand a better chance of respecting print/read
1910 ** Attempting to use standardized file system operators with a
1911 pathname with invalid :DIRECTORY components signals a
1913 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1915 ** DIRECTORY on logical pathnames is more correct.
1916 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1917 signal a TYPE-ERROR if their argument is not a stream.
1918 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1919 designator argument does not designate a stream.
1920 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1921 examining the synonym.
1922 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1924 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1925 element-type larger than ([UN]SIGNED-BYTE 32).
1927 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1928 * When built with the :SB-FUTEX feature, threaded builds now take
1929 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1930 for faster/more reliable mutex and condition variable support.
1931 * Incompatible change (but one you probably shouldn't have been using
1932 anyway): the interface and code for arbitrating between multiple
1933 threads in the same user session has been redesigned.
1934 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1935 error when called without an explicit environment argument.
1936 (thanks to Nikodemus Siivola)
1937 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1938 confused about their position. (thanks to Adam Warner and Gerd
1940 * bug fix: Pretty printing backquoted forms with unquotations in the
1941 argument list position of various code constructs such as LAMBDA
1942 now works correctly. (reported by Paul Dietz)
1943 * bug fix: Pretty printing unquotations no longer loses all
1944 stream position information.
1945 * optimization: performance of string output streams is now less
1946 poor for multiple small sequence writes.
1947 * optimization: performance of CSUBTYPEP in the presence of complex
1948 expressions involving CONS and NOT many times has been improved.
1949 (reported by Paul Dietz)
1950 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1952 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1953 streams for the REPL, for the debugger, and so on. (thanks to
1955 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1956 * fixed some bugs revealed by Paul Dietz' test suite:
1957 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1958 is ignored if it is too small, rather than propagating through
1959 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1960 ** extremely complex negations of CONS types were not being
1961 sufficiently canonicalized, leading to inconsistencies in
1963 ** VALUES tranformer lost derived type.
1965 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1966 * fixed a bootstrapping bug: the build process no longer assumes
1967 that the various BOOLE-related constants have the same value in
1968 host and target lisps. (noted by Paul Dietz' test suite on an
1969 SBCL binary built from CLISP)
1970 * The system can now be dynamically linked on the MIPS platform,
1971 which enables dynamic loading of foreign code from Lisp. (thanks
1972 to Ralf Baechle for discussions on the MIPS ABI)
1973 * The system now records debugging information for its own source
1974 files in a filesystem-position-independent manner, relative to
1975 the "SYS" logical host.
1976 * fixed a compiler bug: MV-LET convertion did not check references
1977 to the "max args" entry point. (reported by Brian Downing)
1978 * tweaked disassembly notes to be less confident about proclaiming
1979 some instruction as an LRA. (thanks to Brian Downing)
1980 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1981 now each have their own history, command character, and other
1982 characteristics. (thanks to David Lichteblau)
1983 * fixed some bugs revealed by Paul Dietz' test suite:
1984 ** compiler failure in compiling LOGAND expressions including a
1986 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1987 shift greater than 32.
1988 ** FLUSH-DEST did not mark blocks for type check regeneration.
1989 ** HANDLER-CASE failed to accept declarations in handler clauses
1990 in some circumstances.
1992 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1993 * New code in contrib/sb-introspect (still probably not entirely
1994 stable yet) provides some support for smart Lisp development
1995 environments like SLIME.
1996 * The conditions signalled for errors occurring when loading .fasl
1997 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1998 in a way which should help ASDF recover gracefully.
1999 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
2000 stuff has been cleaned up. If you code contrib/ stuff, this might
2001 affect you, and you can look at contrib/README, contrib/STANDARDS,
2002 and/or the 0.8.4.27 diff to check.
2003 * In full calls the compiler now does not generate checks for declared
2004 argument types for all arguments.
2005 * various threading fixes
2006 ** and some experimental patches which didn't make it into
2007 the main tree for this release, but which are shipped in
2008 contrib/experimental-thread.patch as a possible fix for some
2009 failures (deadlock, spinning...) in GC-intensive multithreaded
2011 * fixed PPC build problem (source code incompatibility of different
2012 library versions): added offsetof() hackery which attempts to divine
2013 where glibc maintainers put uc_mcontext today
2014 * fixed bug 282: compiler does not trust type assertions while passing
2015 arguments to a full call.
2016 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
2017 &OPTIONAL VALUES type parameter.
2018 * fix bug 214: algorithm for noting rejected templates is now more
2019 similar to that of template seletion. (also reported by rydis on
2021 * fixed bug 141b: printing backquoted information readably and prettily
2022 inserts a space where necessary.
2023 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2024 (CHARACTER 10) are now reported as errors, rather than propagated
2025 as unknown types. (reported by piso on #lisp)
2026 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2027 with values NIL and :ERROR. (thanks to Milan Zamazal)
2028 * fixed bug 191c: CLOS now does proper keyword argument checking as
2029 described in CLHS 7.6.5 and 7.6.5.1.
2030 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2031 longer raise an error; further, using a list as a for-as-arithmetic
2032 counter now raises a meaningful error.
2033 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2034 now understood by sequence creation functions such as MAKE-SEQUENCE
2036 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2037 a STRING-INPUT-STREAM.
2038 * compiler enhancement: SIGNUM is now better able to derive the type
2040 * type declarations inside WITH-SLOTS are checked. (reported by
2042 * fixed some bugs revealed by Paul Dietz' test suite:
2043 ** incorrect optimization of TRUNCATE for a positive first
2044 argument and negative second.
2045 ** compiler failure in let-convertion during flushing dead code.
2046 ** compiler failure while deriving type of TRUNCATE on an
2047 interval, containing 0.
2048 ** ASH of a negative bignum by a negative bignum count now returns
2050 ** intersection of CONS types now canonicalizes properly, fixing
2051 inconsistencies in SUBTYPEP.
2053 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2054 * incompatible change: The --disable-debugger command line
2055 option now clobbers the debugger at a more fundamental
2056 level, by redefining #'INVOKE-DEBUGGER instead of by
2057 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2058 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2059 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2060 enter the debugger REPL and then suffer recursive errors
2061 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2062 while under the new system, BREAK in --disable-debugger mode
2063 terminates the system just as an unhandled error would.
2064 * fixed compiler performance when processing loops with a step >1;
2065 * bug fix: DOCUMENTATION now retrieves generic function
2066 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2067 support has been systematized, and now supports the methods
2068 specified by ANSI, along with a default method and a method for
2069 slot documentation. (reported by Nathan Froyd)
2070 * bug fix: effective methods associated with a generic function are
2071 no longer cached over a change of that generic function's method
2072 combination. (reported by Andreas Fuchs)
2073 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2074 names a non-existent directory. (thanks to Andreas Fuchs)
2075 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2076 the PPC platform, lead to stack corruption; this has been fixed.
2077 (reported by Rainer Joswig)
2078 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2079 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2080 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2081 arithmetic in general, is now much more reliable on the PPC
2083 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2084 platform now returns the right answer.
2085 * optimization: restored some effective method precomputation in
2086 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2087 precomputation is now tunable.
2088 * optimization: compiler-internal data structure use has been
2089 reviewed, and changes have been made that should improve the
2090 performance of the compiler by about 20%.
2091 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2092 simple-base-strings and simple-bit-vectors is improved.
2093 * optimization: the optimization of 32-bit logical and arithmetic
2094 functions introduced in version 0.8.3 on the x86 has been
2095 implemented on the mips, ppc and sparc platforms; an
2096 implementation of the same facility, but for 64-bit arithmetic,
2097 has been added for the alpha.
2098 * microoptimization: the compiler is better able to make use of the
2099 x86 LEA instruction for multiplication by constants.
2100 * bug fix: in some situations compiler did not report usage of
2101 generic arithmetic in (SPEED 3) policy.
2102 * bug 145b fix: compiler used wrong type specifier while converting
2103 MEMBER-types to numeric.
2104 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2106 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2107 index argument now works properly on the Alpha platform.
2108 * bug fix: floating point exception treatment on the Alpha platform
2110 * bug fix: FILE-POSITION works much better on string input and
2111 output streams. (thanks to Nikodemus Siivola)
2112 * bug fix: many threading/garbage collection symptoms sorted.
2113 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2114 might be pseudo-atomic.
2115 * internal change: Stopping for GC is now done with signals not
2116 ptrace. GC is now done in whichever thread wanted it, instead of
2118 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2120 * bug fix: result form in DO is not contained in the implicit
2122 * incompatible change: ICR structure is changed; the value part of
2123 CONTINUATION is now called LVAR; corresponding functions are
2124 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2125 * added type deriver for ISQRT (thanks to Robert E. Brown).
2126 * bug fix: better support for loading from the command line when an
2127 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2129 * fixed some bugs revealed by Paul Dietz' test suite:
2130 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2131 ** ROUND and FROUND now give the right answer when given very
2132 small float arguments.
2133 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2135 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2136 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2137 ** LCM should return a non-negative integer.
2138 ** PARSE-INTEGER returned the index of a terminator instead of the
2139 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2140 ** PARSE-INTEGER returned an incorrect index being applied to a
2142 ** LCM with two arguments of 0 returns 0 rather than signalling
2144 ** unsigned addition of a 32-bit constant with the high bit set no
2145 longer causes an internal compiler error on the x86.
2146 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2147 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2148 with negative last argument.
2149 ** byte specifiers with zero size and position no longer cause
2150 an error during type derivation.
2151 ** bignum multiplication on the Alpha platform now returns the
2153 * porting: The system now builds on SuSE AMD64, although it still
2154 generates a 32-bit binary.
2155 * .fasl file incompatibility: The fasl file version number has
2156 been incremented (because of the changes to internal compiler
2157 data structures referred to above).
2159 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2160 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2161 more accurately, on the Darwin kernel running on PowerPC hardware.
2162 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2163 * Compiler code deletion notes now signal a condition of type
2164 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2165 an associated MUFFLE-WARNING restart.
2166 * The compiler now performs limited argument count validation of
2167 constant format strings in FORMAT, and where appropriate in ERROR,
2168 CERROR and WARN. (thanks to Gerd Moellmann)
2169 * New ASDF-INSTALL contrib can be used for automatic download and
2170 installation of third-party Lisp code from CCLAN or other sites
2172 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2173 which forces another thread to execute a function supplied by the
2175 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2176 now accept and act upon their :ELEMENT-TYPE keyword argument.
2177 (reported by Martin Atzmueller, Edi Weitz)
2178 * bug fix: FILE-POSITION now accepts position designators up to
2179 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2180 is the greater. (thanks to Patrik Nordebo)
2181 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2182 Kalle Olavi Niemitalo)
2183 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2184 printable. (reported by Eric Marsden)
2185 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2186 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2187 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2188 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2189 (thanks to Stig Sandoe)
2190 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2191 UNREAD-CHAR now correctly includes the unread character in the
2192 target sequence. (thanks to Gerd Moellmann)
2193 * bug fix in threaded builds: the system can now be suspended and
2194 resumed by shell job control with minimal disruption.
2195 * bug fixes in times and timezones >2038AD
2196 * better handling of "where is GNU make?" problem in build scripts
2197 (thanks to Nikodemus Siivola)
2198 * new optimization: inside a named function any reference to a
2199 function with the same name is considered to be a self-reference;
2200 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2201 optimization quality.
2202 * new optimization on x86: logical functions and + now have
2203 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2204 used when the result is truncated to 32 bits.
2205 * VALUES declaration is partially enabled.
2206 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2207 * fixed some bugs revealed by Paul Dietz' test suite:
2208 ** The system now obeys the constraint imposed by
2209 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2210 types form a lattice under type intersection.
2211 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2212 ** ASSOC now ignores NIL elements in an alist.
2213 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2214 and (1+ MOST-POSITIVE-FIXNUM) answers.
2215 ** The addition of a method with invalid qualifiers to a generic
2216 function does not cause an error to be signalled immediately;
2217 a warning is signalled, and the error is generated only on
2218 calling the generic function.
2219 * changed .fasl file version number, in part to add type codes for
2220 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2221 obscure ANSI requirements
2223 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2224 * fixed bug 148: failure to inline-expand a local function left
2225 garbage, confusing the compiler.
2226 * fixed bugs 3cd: structure slot readers perform type check if the
2227 slot can have an invalid value (i.e. it is either not initialized
2228 or can be written with a less specific slot writer).
2229 * bug fix: the compiler now traps array references to elements off
2230 the end of an array; previously, the bounds checking in some
2231 circumstances could go off-by-one.
2232 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2234 * type declarations for array element types now obey the description
2235 on the CLHS page "Declaration TYPE", as per discussions on
2236 sbcl-help around 2003-05-08. This means that a declaration
2237 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2238 declaration, all references to BAR will be asserted or assumed
2239 (with THE, so dependent on compiler policy) to involve objects of
2240 type FOO. Note that no such declaration is implied in
2241 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2242 * declared types of functions from the "Conditions"
2243 chapter. (reported by Paul Dietz)
2244 * bug fix: CERROR accepts a function as its first argument.
2245 * bug fix: NTH an NTHCDR accept a bignum as index
2246 arguments. (reported by Adam Warner)
2247 * optimization: character compare routines now optimize comparing
2248 against a constant character. (reported by Gilbert Baumann)
2249 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2250 argument now works properly on the MIPS platform.
2251 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2252 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2253 declared to have a functional type, does not cause an error
2255 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2256 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2257 * fixed bug: if last continuation of a deleted block has a
2258 destination, this destination should be deleted too. (reported by
2260 * fixed a bug in the bootstrap process: the host compiler's values
2261 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2262 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2263 test case from Patrik Nordebo)
2264 * improved the ability of the disassembler on the PPC platform to
2265 provide helpful disassembly notes.
2266 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2267 the class in more cases than previously.
2268 * bug fix: FILE-POSITION now understands :START and :END for
2269 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2270 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2271 8) data. (thanks to David Lichteblau)
2272 * bug fix: it is possible to add a method to a generic function
2273 without lambda list.
2274 * bug fix: reader failed to signal END-OF-FILE inside an
2275 object representation. (reported by Nikodemus Siivola)
2276 * fixed some bugs revealed by Paul Dietz' test suite:
2277 ** LAST and [N]BUTLAST should accept a bignum.
2278 ** condition slot accessors are methods.
2279 ** (VECTOR NIL) is a subtype of STRING.
2281 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2282 * minor incompatible change: some nonsensical specialized lambda
2283 lists (used in DEFMETHOD) which were previously ignored now signal
2285 * minor incompatible change: the system is now aware of the types of
2286 variables in the COMMON-LISP package, and will signal errors for
2287 most violations of these type constraints (where previously they
2288 were silently accepted).
2289 * minor incompatible change: COMPILE-FILE now uses the freedom
2290 afforded (ANSI 3.2.2.3) to use derived function types for
2291 functions defined in the same file. This also permits the system
2292 to warn on static type mismatches and function
2293 redefinition. (Currently it does not work with high DEBUG level.)
2294 * minor incompatible change: VALUES declaration is disabled.
2295 * When issuing notes, the compiler now signals a condition of type
2296 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2297 restart for use in user handlers. It is expected that the
2298 COMPILER-NOTE condition will eventually become a condition
2299 supertype to a hierarchy of note types, which will then be
2300 handleable in a similar fashion. However, at the moment, no such
2301 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2302 but it's an implementation detail, not a classification for the
2304 * Changes in type checking closed the following bugs:
2305 ** type checking of unused values (192b, 194d, 203);
2306 ** template selection based on unsafe type assertions (192c, 236);
2307 ** type checking in branches (194bc).
2308 * A short form of VALUES type specifier has ANSI meaning (it has
2309 increased the number of situations when SBCL cannot perform type
2311 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2312 keywords or constants is permissible.
2313 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2314 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2315 classes. (thanks to Antonio Martinez)
2316 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2317 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2318 argument to a pretty-print stream. Instead, it leaves any such
2319 operations to DESCRIBE-OBJECT methods.
2320 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2322 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2323 lambda lists. (thanks to David Lichteblau)
2324 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2325 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2326 (thanks to Gerd Moellmann)
2327 * fixed bug 246: increased compilation speed of long
2328 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2330 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2332 * DEFCONSTANT now throws a condition of type
2333 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2334 constant to a non-EQL value; CONTINUE and ABORT restarts
2335 respectively change and preserve the value.
2336 * fixed bug 63: The code walker, part of the implementation of CLOS,
2337 is now better at handling symbol macros.
2338 * bug fix: There is no longer an internal implementation type named
2339 CL:LENGTH. (reported by Raymond Toy)
2340 * bug fix: In macro-like defining macros/special operators the
2341 implicit block does not enclose the lambda list.
2342 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2343 enclosed in parentheses) are not suitable as type specifiers, and
2344 their use properly signals an error now.
2345 * bug fix: An argument count mismatch for a type specifier in code
2346 being compiled no longer causes an unhandled error at compile
2347 time, but signals a compile-time warning.
2348 * fixed simple vector readable printing
2349 * bug fix: DESCRIBE takes more care over whether the class
2350 precedence list slot of a class is bound before accessing it.
2351 (reported by Markus Krummenacker)
2352 * bug fix: FORMATTER can successfully compile pretty-printer format
2353 strings which use variants of the ~* directive inside.
2354 * bug fix: SEARCH now applies its TEST predicate to the elements of
2355 the arguments in the correct order. (thanks to Wolfhard Buss)
2356 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2357 in type inference. (thanks to Robert E. Brown)
2358 * bug fix: Reading in symbols with an explicit package name of ""
2359 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2360 * fixed some bugs revealed by Paul Dietz' test suite:
2361 ** NIL is now allowed as a structure slot name.
2362 ** Arbitrary numbers, not just REALs, are allowed in certain
2363 circumstances in LOOP for-as-arithmetic clauses.
2364 ** Multiple class redefinitions before slot access no longer
2365 causes a type error.
2366 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2367 association between the name and a class.
2368 ** Generic functions with non-standard method-combination and over
2369 five methods all of which return constants no longer return NIL
2370 after the first few invocations. (thanks to Gerd Moellmann)
2371 ** CALL-NEXT-METHOD with no arguments now passes the original
2372 values of the arguments, even in the presence of assignment.
2373 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2375 ** DISASSEMBLE works with closures and funcallable instances.
2376 ** ADD-METHOD now returns the generic function, not the new method.
2377 ** FIND-METHOD signals an error if the lengths of the specializers
2378 is incompatible with the generic function, even if the ERRORP
2380 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2381 which its argument is a member.
2382 ** DEFCLASS only redefines the class named by its class-name
2383 argument if that name is the proper name of the class;
2384 otherwise, it creates a new class.
2385 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2386 of the UNBOUND-SLOT condition to the name of the slot.
2387 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2388 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2389 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2390 SLOT-MAKUNBOUND in the specified fashion.
2392 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2393 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2394 cross-compilation host. As a consequence, we can now bootstrap our
2395 way up to SBCL starting with a bare gcc toolchain and human-readable
2396 source code (first the source to CLISP, then the source to SBCL).
2397 * A contributed module containing a partial implementation of the
2398 simple-streams interface has been included. (thanks to Rudi
2400 * A contributed module implementing the RFC1321 Message Digest
2401 Algorithm, known as MD5, has been included.
2402 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2403 no longer has any effect, as the code controlled by this feature
2404 has been deleted. (As far as we know, no-one has ever built using
2405 this feature, and its semantics were confused in any case).
2406 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2407 work on conditions (as required by the ANSI specification),
2408 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2409 expected behaviour on conditions. Users should note, however,
2410 that such behaviour is not required by the ANSI specification,
2411 and so use of this behaviour may render their code unportable.
2412 * fixed some bugs revealed by Paul Dietz' test suite:
2413 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2415 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2416 ** SXHASH on condition objects no longer returns NIL.
2417 ** :ALLOCATION :CLASS slots are better treated; their values are
2418 updated on class redefinition, and initforms inherited from
2419 superclasses are applied.
2420 ** REMOVE-METHOD returns its generic function argument even when
2421 no method was removed.
2422 ** SHARED-INITIALIZE now initializes the values of the requested
2423 slots, including those with :ALLOCATION :CLASS.
2424 ** ALLOCATE-INSTANCE now works on structure classes defined via
2425 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2427 ** SLOT-EXISTS-P now works on conditions, as well as structures
2429 ** MAKE-LOAD-FORM now has the required methods on
2430 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2431 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2432 keyword, and now implements the SLOT-NAMES argument.
2433 ** methods with &OPTIONAL arguments no longer allow too many
2434 arguments to be passed in the call without error.
2435 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2436 option is consistent with the required arguments of the generic
2437 function lambda list.
2438 * bug fix: REQUIRE accepts a string designator. (Thanks to
2440 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2441 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2442 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2443 * bug fix: The compiler now checks for duplicated variables in macro
2445 * bug fix: SETQ on globals returns the correct value.
2446 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2447 not a COMPILER-ERROR (followed by some other strange error on
2448 choosing the CONTINUE restart).
2449 * bug fix: make.sh and friends are now more consistent in the way that
2450 they look for GNU "make".
2452 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2453 * experimental native threads support (on x86 Linux >=2.4 only).
2454 This is not compiled in by default: you need to add :SB-THREAD to
2455 the target features. See the "Beyond ANSI" chapter of the manual
2457 * fix for longstanding nonANSIism: The old distinction between
2458 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2459 The return value from CL:FIND-CLASS is now a CLOS class, and
2460 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2461 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2462 * An interface to the MetaObject Protocol, as described in Kiczales,
2463 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2464 MIT Press, 1991, is available from the SB-MOP package.
2465 * incompatible change: the SB-PCL package should now be considered
2466 a private implementation detail, and no longer a semi-private MOP
2468 * minor incompatible change: due to rearrangement for threads, the
2469 control stack and binding stack are now allocated at arbitrary
2470 addresses instead of being hardcoded per-port. Users affected by
2471 this probably have to be doing advanced things with shared
2472 libraries, and will know who they are.
2473 * minor incompatible change: Previously, all --eval forms used were
2474 processed with READ before any of them were processed with EVAL.
2475 Now each --eval form is processed with both READ and EVAL before
2476 the next --eval form is processed. (Thus package operations like
2477 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2478 work as the user might reasonably expect.)
2479 * minor incompatible change: *STANDARD-INPUT* is now only an
2480 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2482 * minor incompatible change: Y-OR-N-P is now character-oriented,
2483 not line oriented. Also, YES-OR-NO-P now works without errors.
2484 (thanks to Antonio Martinez)
2485 * sb-aclrepl module improvements: an integrated inspector, added
2486 repl features, and a bug fix to :trace command.
2487 * Known functions, which cannot be open coded by the backend, are
2488 considered to be able to check types of their arguments. (fixing
2489 a bug report by Nathan J. Froyd)
2490 * fixed a bug in computing method discriminating functions: It is
2491 now possible to define methods specialized on classes which have
2492 forward-referenced superclasses. (thanks to Gerd Moellmann)
2493 * fixed evaluation order in optional entries (reported by Gilbert
2495 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2496 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2497 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2498 required-by-AMOP NAME argument, as well as CLASS and
2499 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2500 * fixed bug 20: DEFMETHOD can define methods using names that are
2501 not the proper names of classes to designate class specializers.
2502 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2503 Unix file descriptors, instead of blowing up. (thanks to Antonio
2505 * Garbage collection refactoring: user-visible change is that a
2506 call to the GC function during WITHOUT-GCING will not do garbage
2507 collection until the end of the WITHOUT-GCING. If you were doing
2508 this you were probably losing anyway.
2509 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2510 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2511 TYPEP the latter but not the former.
2512 * The compiler issues a full WARNING for calls to undefined functions
2513 with names from the CL package.
2514 * MAP-INTO for a vector destination is open coded. (reported by
2515 Brian Downing on c.l.l)
2516 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2517 documentation string.
2518 * fixed some bugs revealed by Paul Dietz' test suite:
2519 ** COPY-ALIST now signals an error if its argument is a dotted
2521 ** Condition slots are now accessed more correctly in the presence
2522 of multiple initargs for a given slot.
2523 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2524 exclude restarts of the same name associated with a different
2526 ** DEFCLASS of forward-referenced classes with another
2527 forward-referenced class in the superclasses list no longer
2529 ** Condition slots are now initialized once each, not multiple
2530 times. (thanks to Gerd Moellmann)
2531 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2533 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2534 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2535 arguments contain duplicated elements.
2536 ** RESTART-CASE understands local macros.
2537 ** RESTART-CASE associates exactly its own restarts with a condition.
2538 ** ENDP in safe mode checks its argument to be of type LIST.
2539 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2540 in question is unbound.
2541 ** Optimized MAKE-INSTANCE functions no longer cause internal
2542 assertion failures in the presence of duplicate initargs.
2543 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2544 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2545 propagated even in the ctor optimized implementation of
2547 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2549 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2550 * a better implementation of SXHASH on (simple) bit vectors,
2551 measured both in execution speed and in distribution of results
2552 over the positive fixnums, has been installed. Likewise, a better
2553 implementation of EQUAL for simple bit vectors is now available.
2554 * fixed CEILING optimization for a divisor of form 2^k.
2555 * fixed bug 240 (emitting extra style warnings "using the lexical
2556 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2557 by Antonio Martinez)
2558 * fixed SXHASH, giving different results for NIL depending on type
2559 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2560 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2561 variable. (found by Rolf Wester)
2562 * SBCL does not ignore type declarations for special
2563 variables. (reported by rif on c.l.l 2003-03-05)
2564 * some bug fixes in contrib/sb-aclrepl/
2565 * fixed some bugs revealed by Paul Dietz' test suite:
2566 ** a bug in the CONS type specifier, whereby the CAR and CDR
2567 types got intertwined, has been fixed;
2568 ** the type system is now able to reason about the interaction
2569 between INTEGER and RATIO types more completely;
2570 ** APPEND, [N]REVERSE and NRECONC check that those their
2571 arguments, which must be proper lists, are really so;
2572 ** An array specialized to be unable to hold elements has been
2573 implemented, as required -- yes, really -- by ANSI;
2574 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2575 on malformed property lists;
2577 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2578 * incompatible packaging change: in line with Unix convention,
2579 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2580 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2581 * REQUIRE and PROVIDE are now optionally capable of doing something
2582 useful. See the documentation string for REQUIRE.
2583 * infrastructure for a managed SBCL contrib system: contributed
2584 modules in this release include:
2585 ** the ASDF system definition facility;
2586 ** an interface to the BSD Sockets API;
2587 ** an ACL-like convenience interface to the repl;
2588 (thanks to Kevin Rosenberg)
2589 ** an implementation of ROTATE-BYTE, with efficient implementation
2591 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2592 gives a closer approximation to the right answer than previously.
2593 (thanks to Raymond Toy)
2594 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2595 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2596 optional environment arguments, as required by ANSI.
2597 * fixed bugs in other functions taking environment objects, allowing
2598 calls with an explicit NIL environment argument to be compiled
2600 * fixed bug 228: primary return values from
2601 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2602 COMPILE or FUNCTION.
2603 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2604 :TYPE will no longer signal errors on innocuous objects.
2605 * fixed bug 231b: SETQ is better at respecting type declarations in
2606 the lexical environment.
2607 * fixed a bug in DEFCLASS: classes named by symbols with no or
2608 unprintable packages can now be defined.
2609 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2610 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2611 * fixed compiler failure related to checking types of functions.
2612 (reported by Robert E. Brown)
2613 * the compiler is now much more consistent in its error-checking
2614 treatment of bounding index arguments to sequence functions: in
2615 (SAFETY 3) code, errors will be signalled in almost all cases if
2616 invalid sequence bounding indices are passed to functions defined
2617 by ANSI to operate on sequences.
2618 * fixed a bug in the build procedure: documentation of SBCL-specific
2619 packages is now preserved and available in the final Lisp image.
2620 * lifted FDEFINITION lookup out of loops in the implementation of
2621 many list operations. (thanks to Robert E. Brown)
2622 * fixed a bug in the reader: the #n# reader macro now works for
2623 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2624 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2625 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2626 E. Brown and Tony Martinez respectively)
2627 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2628 SLOT-BOUNDP: in particular, it is now able to optimize them much
2629 better, and is now not vulnerable to having packages renamed.
2630 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2631 always called when a slot is not present in an instance. (thanks
2633 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2634 longer signal an error on unreading a character following EOF on
2635 the previous constituent stream. (thanks to Tony Martinez)
2636 * fixed some bugs revealed by Paul Dietz' test suite:
2637 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2638 not just nonnegative fixnums;
2639 ** the logical bit-array operators such as BIT-AND now accept an
2640 explicit NIL for their "opt-arg" argument (to indicate a
2641 freshly-consed result bit-array);
2642 ** ELT now signals an error on an invalid sequence index in safe
2644 ** the type system is now cleverer about negations of numeric
2645 types, and consequently understands the BIGNUM and RATIO types
2647 ** the type system is now cleverer about the interaction between
2648 INTEGER and RATIO types: while bugs still remain, many more
2649 cases are accurately computed;
2650 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2651 if it is in the last clause;
2652 ** CONSTANTLY now correctly returns a side-effect-free function in
2654 ** DECLARE is no longer treated as a special-operator; in
2655 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2656 * incremented fasl file version number due to the change in the
2657 DEFSTRUCT-SLOT-DESCRIPTION structure.
2659 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2660 * minor incompatible change: code processed by the "interpreter" or
2661 EVAL now has a compilation optimization policy of (DEBUG 2)
2662 (changed from (DEBUG 1)) to improve debuggability of interactive
2663 development, and to allow the use of the debug RETURN command in
2665 * an experimental implementation of the RETURN command for the
2666 debugger has been included. (thanks to Frederik Kuivinen)
2667 * fixed bug 62: constraints were not propagated into a loop.
2668 * fixed bug in embedded calls of SORT (reported and investigated by
2670 * fixed some bugs revealed by Paul Dietz' test suite:
2671 ** printing and reading of arrays with some dimensions having
2672 length 0 (thanks to Gerd Moellmann);
2673 ** BOA constructor with &AUX argument without a default value does
2674 not cause a type error;
2675 ** CONSTANTP now returns true for all self-evaluating objects.
2677 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2678 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2679 accessors that are related by inheritance, as specified in the
2680 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2681 Valtteri Vuorikoski)
2682 * The compiler is now able to inline functions that were defined in
2683 a complex lexical environment (e.g. inside a MACROLET).
2684 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2686 * Support for the upcoming FreeBSD-5.0 release has been included.
2687 (thanks to Dag-Erling Smorgrav)
2688 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2689 effect when it is not in a toplevel context.
2690 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2691 stand a better chance of being correct. (thanks to Gerd
2693 * fixed bug in COERCE, which now signals an error on coercing a
2694 rational to a bounded real type which excludes the expected
2696 * The compiler is now able to derive types more accurately from the
2697 COERCE and COMPILE functions.
2698 * fixed bug 223: functional binding is considered to be constant
2699 only for symbols in the CL package.
2700 * fixed bug 231: SETQ did not check the type of a variable being set
2701 (reported by Robert E. Brown)
2702 * A new optimization for MAKE-INSTANCE has been included, fixing
2703 various bugs (including relating to :ALLOCATION :CLASS slots and
2704 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2706 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2707 ** As required by ANSI, LOOP now disallows anonymous collection
2708 clauses such as COLLECT I in conjunction with aggregate boolean
2709 clauses such as THEREIS (= I 1);
2710 ** LOOP now signals an error when any variable is reused in the
2711 same loop (including the potentially useful construct analogous
2712 to WITH A = 1 WITH A = (1+ A);
2713 ** IT is only a special loop symbol within the first clause of a
2714 conditional loop clause;
2715 ** LOOP with a typed iteration variable over a hashtable now
2716 signals a type error iff it should.
2717 * fixed some other bugs revealed by Paul Dietz' test suite:
2718 ** FILE-STREAM now names the class previously known as FD-STREAM;
2719 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2720 argument) no longer signals an error;
2721 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2722 of the slot symbol, rather than using the current package
2723 ((:CONC-NAME "") continues to intern the slot's name in the
2725 * incremented fasl file version number, because of the incompatible
2726 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2727 of the new implementation of DEFINE-COMPILER-MACRO.
2729 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2730 * Support for building SBCL for MIPS platforms running in
2731 little-endian mode has now been checked in, and basic
2732 functionality on said platforms verified.
2733 * minor incompatible change: PCL now records the pathname of a file
2734 in which methods and the like are defined, rather than its
2736 * minor incompatible change: TRUENAME now considers the truename of
2737 a file naming a directory to be the pathname with :DIRECTORY
2738 component indicating that directory.
2739 * minor incompatible change: a NAMED clause in the extended form of
2740 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2741 reason for the previous behaviour is unclear.
2742 * more systematization and improvement of CLOS and MOP conformance
2743 in PCL (thanks to Gerd Moellman and Pierre Mai):
2744 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2746 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2747 primary methods with no specializers;
2748 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2750 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2751 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2752 FORWARD-REFERENCED-CLASSes; error reporting on
2753 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2755 ** SXHASH on CLOS instances now uses a slot internal to the
2756 instance to return different numbers on distinct instances,
2757 while preserving the same return value through invocations of
2759 ** DEFMETHOD signals errors when methods with longer incongruent
2760 lambda lists are added to generic functions;
2761 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2762 CLASS, as specified in AMOP;
2763 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2764 on the order returned by the primary method for classes of
2765 class STANDARD-CLASS;
2766 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2767 * fixed some bugs shown by Paul Dietz' test suite:
2768 ** DOLIST puts its body in TAGBODY;
2769 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2771 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2772 value producing form;
2773 ** if more variables are given to PROGV than values, extra
2774 variables are bound and made to have no value;
2775 ** NSUBSTITUTE on list arguments gets the right answer with
2777 ** ELT signals an error of type TYPE-ERROR when the index argument
2778 is not a valid sequence index;
2779 ** LOOP signals (at macroexpansion time) an error of type
2780 PROGRAM-ERROR when duplicate variable names are found;
2781 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2782 ** FUNCALL of special-operators now cause an error of type
2784 ** PSETQ now works as required in the presence of side-effecting
2785 symbol-macro places;
2786 ** NCONC accepts any object as its last argument;
2787 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2789 ** loop-for-as-package does not require a package to be explicitely
2791 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2793 * fixed bug 166: compiler preserves "there is a way to go"
2794 invariant when deleting code.
2795 * fixed bug 172: macro lambda lists with required arguments after
2796 &REST arguments now cause an error to be signalled. (thanks to
2798 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2799 bug. (thanks to Matthew Danish)
2800 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2802 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2803 arguments are lexically rebound. (thanks to Gerd Moellmann and
2805 * fixed bug 194: error messages are now more informative when there
2806 is no primary method applicable in a call to a generic
2807 function. (thanks to Gerd Moellmann)
2808 * fixed bug in command line argument checking (thanks to Julian
2810 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2811 (thanks to Matthew Danish)
2812 * incremented fasl file version number, because of the
2813 SXHASH-related changes in the layout of CLOS data structures
2815 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2816 * minor incompatible change: The runtime (the Unix executable named
2817 "sbcl") is now much pickier about the .core files it will load.
2818 Essentially it now requires .core files to descend from the same
2819 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2820 as the runtime does. (The intent is to prevent the crashes which
2821 can occur, and which can even be reported as mysterious failures,
2822 when people patch the sources or change the build parameters
2823 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2824 sbcl and .core files.)
2825 * fixed bug: VALUES-LIST is no longer optimized away.
2826 * fixed bug 142: The FFI conversion of C string values to Lisp
2827 string values no longer conses excessively. (thanks to Nathan
2828 Froyd porting Raymond Toy's fix to CMU CL)
2829 * began to systematize and improve MOP conformance in PCL (thanks to
2830 Nathan Froyd, Gerd Moellman and Pierre Mai):
2831 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2833 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2834 ** FINALIZE-INHERITANCE is now called on class finalization;
2835 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2836 argument precedence order.
2837 * fixed bug 202: The compiler no longer fails on functions whose
2838 derived types contradict their declared type.
2839 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2840 so it can be non-toplevel.
2841 * The fasl file version number has changed (because of the new
2842 implementation of DEFMACRO).
2843 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2844 safe code, that any length requirement by their type-specifier
2845 argument is valid. The exceptions to this are described in bug
2847 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2848 functions now check the types of their inputs as required by ANSI.
2849 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2850 introduced symbol is DECLAREd to be SPECIAL.
2851 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2852 * fixed bug: :COUNT argument to sequence functions may be negative
2853 * fixed bug: body of DO-SYMBOLS may contain declarations
2854 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2855 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2856 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2857 symbol macro only once
2858 * fixed printing of call frame when argument list is unavailable
2859 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2860 * compiler no longer signals WARNING on unknown keyword
2863 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2864 * A beta-quality port to the mips architecture running Linux,
2865 based on the old CMUCL backend, has been made. It has been tested
2866 on a big-endian kernel, and works sufficiently well to be able to
2867 rebuild itself; it has not been tested in little-endian mode.
2868 * fixed an inconsistency between gencgc.c and purify.c which made
2869 dumping/loading .core files unreliable
2870 * fixed bug 120a: The compiler now deals correctly with IFs where
2871 the consequent is the same as the alternative, instead of
2872 misderiving the return type. (thanks to Alexey Dejneka)
2873 * fixed bug 113: Logical pathnames are now dumpable (the logical
2874 host is resolved at load-time, throwing an error if it is not
2876 * fixed bug 174: FORMAT's error message is slightly clearer when a
2877 non-printing character is used in a format directive.
2878 * fixed several bugs in compiler checking of type declarations, i.e.
2879 violations of the Python "declarations are assertions" principle
2880 (thanks to Alexey Dejneka)
2881 * fixed several bugs in PCL's error checking (thanks to Gerd
2883 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2885 * fixed bug in compilation of functions as first class values
2886 (thanks to Antonio Martinez-Shotton)
2887 * The compiler's handling TYPE-ERRORs which it can prove will
2888 inevitably happen at runtime has been cleaned up and corrected
2889 in several ways. (thanks to Alexey Dejneka)
2890 * improved argument type checking for various basic arithmetic
2891 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2892 lavished on them in the past that they can be compiled in many
2893 ways in different special cases
2894 * fixed bug 181: compiler checks validity of user supplied type
2896 * cleaned up code flushing in optimization: Function calls which
2897 should signal errors for safety purposes (e.g. which ANSI says
2898 should signal errors when their arguments are of incorrect type)
2899 are no longer optimized away.
2900 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2901 * incremented fasl file version number, because changes in the
2902 implementation of sequence functions like COERCE caused
2903 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2904 in old inline expansions) to become undefined. (Actually these
2905 changes were later undone, so we might very well be binary
2906 compatible with 0.7.7 after all, but leaving the version number
2907 incremented seemed like the simplest and most conservative
2910 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2911 * An alpha-quality port to the parisc architecture running Linux,
2912 based on the old CMUCL backend, has been made. This, even more so
2913 than the other backends, should be considered still a work in
2914 progress; known problems include that the Linux kernel in 64-bit
2915 mode does not propagate the correct sigcontext structure to
2916 userspace, and consequently SBCL on a parisc64 kernel will not
2918 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2919 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2920 suppressed.) Also now that the compiler is looking at declarations
2921 in the environment, it checks optimization declarations as well,
2922 and suppresses inlining when (> DEBUG SPEED).
2923 * More fixes have been made to treatment of floating point exception
2924 treatment and other Unix signals. In particular, floating point
2925 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2926 * The detection and handling of control stack exhaustion (infinite
2927 or very deeply nested recursion) has changed. Stack exhaustion
2928 detection is now done by write-protecting pages at the OS level
2929 and applies at all optimization settings; when found, a
2930 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2931 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2932 be caught using IGNORE-ERRORS.
2933 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2934 correctly when transforming recursive LABELS functions to LETs.
2935 (thanks to Alexey Dejneka)
2936 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2937 that are names of constants or global variables.
2938 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2939 alien routines with docstrings.
2940 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2941 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2943 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2944 (thanks to Eric Marsden)
2945 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2946 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2947 * Bug fix: The SPARC backend can now compile functions involving
2948 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2949 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2950 to a functions expecting a PATHNAME-DESIGNATOR.
2951 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2952 lambda lists. (thanks to Alexey Dejneka)
2953 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2954 functions so that EXPORTing the name of the class doesn't cause
2955 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2956 to Antonio Martinez for reporting this)
2957 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2958 constant-folded without failing an assertion. (thanks to Einar
2959 Floystad Dorum for reporting this)
2960 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2961 types are now dealt with more correctly by the compiler.
2962 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2963 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2964 behaviour analogous to RENAME-FILE. This puts its behaviour more
2965 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2967 * The fasl file version number has changed again. (because of the
2968 bug fix involving the names of PCL MAKE-INSTANCE functions)
2970 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2971 * bug fix: Floating point exceptions are treated much more
2972 consistently on the x86/Linux and PPC/Linux platforms.
2973 * Array initialization with :INITIAL-ELEMENT is now much faster for
2974 cases when the compiler cannot open code the array creation, but
2975 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2976 array accesses have also seen a speed increase.
2977 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2978 specified. (This was at the root of some bad interactions between
2979 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2981 * bug fix: Internal error arguments for undefined functions are now
2982 computed correctly on the PPC/Linux platform.
2983 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2984 Raymond Toy's patch for CMU CL)
2985 * Support for the Solaris 9 operating environment has been included
2986 (thanks to Daniel Merritt)
2987 * A very ugly but hopefully complete draft of the missing FFI chapter
2988 of the manual has been created by reformatting the corresponding
2989 CMU CL manual chapter into (currently very ugly and incoherent)
2990 DocBook and bringing it up to date for SBCL behavior. Thus, the
2991 manual is now essentially complete, at least by my extreme
2992 once-and-only-once standards, whereby it's acceptable to refer to
2993 the doc strings of SB-EXT functions as the primary documentation.
2994 * The fasl file version number has changed again, due to cleanup of
2995 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2996 is no longer a static symbol.)
2998 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2999 * SBCL now builds with OpenMCL (version 0.12) as the
3000 cross-compilation host; also, more progress has been made toward
3001 bootstrapping under CLISP.
3002 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
3004 * bug 158 fixed: The compiler can now deal with integer loop
3005 increments different from 1; fixing this turned out also to fix
3007 * bug 169 fixed: no more bogus warnings about using lexical bindings
3008 despite the presence of perfectly good SPECIAL declarations (thanks
3009 to David Lichteblau)
3010 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
3011 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
3012 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
3014 * bug 180 fixed: Method combination specifications no longer ignore
3015 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
3016 * bug fix: Structure type predicate functions now check their argument
3017 count as they should.
3018 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
3019 correctly. (thanks to Pierre Mai)
3020 * minor incompatible change: The --noprogrammer option is deprecated
3021 in favor of the new --disable-debugger option, which is very similar.
3022 (The major difference is that it takes effect at a slightly different
3023 time at startup, causing handling of errors in --sysinit and
3024 --userinit files will be affected differently.) The
3025 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3026 been added to allow this functionality to be controlled from ordinary
3027 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3028 maintainers, who might want to run non-interactive scripts to
3029 build SBCL cores which will later be used interactively.)
3030 * minor incompatible change: The LOAD function no longer, when given
3031 a wild pathname to load, loads all files matching that pathname.
3032 Instead, an error of type FILE-ERROR is signalled.
3034 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3035 * bug 147 fixed: The compiler preserves its block link/count
3036 invariants more correctly now so that it doesn't crash. (thanks
3038 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3040 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3041 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3042 no-value-for-this-character as NIL (as specified by ANSI).
3043 * HOST-NAMESTRING on physical pathnames now returns a string that is
3044 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3045 (thanks to Christophe Rhodes)
3046 * The Alpha port handles icache flushing more correctly. (thanks to
3048 * More progress has been made toward bootstrapping under CLISP. (thanks
3049 to Christophe Rhodes)
3050 * The fasl file format has changed again, because dynamic loading
3051 on OpenBSD (which has non-ELF object files) motivated some cleanups
3052 in the way that foreign symbols are transformed and passed around.
3053 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3054 is no longer treated as whitespace by the reader, but instead as
3055 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3056 |A
\7fB|, instead of A as it used to.
3058 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3059 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3060 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3061 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3062 this as of 0.7.1.45, but was left out of the previous news section
3063 (thanks to Dan Barlow)
3064 * SBCL now runs on the Solaris operating system on SPARC architectures
3065 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3066 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3067 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3068 * SPARC backend cleanups, allowing builds of cores optimized for V8
3069 and V9 SPARCS, and also emission of code targeted to a particular
3070 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3072 * SBCL is closer to bootstrapping under CLISP, thanks to various
3073 fixes by Christophe Rhodes.
3074 * The fasl file format has changed again, to allow the compiler's
3075 INFO database to support symbol macros.
3076 * The user manual (in doc/) is formatted into HTML more nicely.
3077 (thanks to coreythomas)
3078 * The system is smarter about SUBTYPEP relationships, especially
3079 those involving NOT types (including types such as ATOM which are
3080 represented internally using NOT types). Thus SUBTYPEP is less
3081 likely to return (VALUES NIL NIL) in general, and in particular
3082 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3084 * The fasl file format has changed again, because the internal
3085 representation of types now includes a new slot to support the new
3086 SUBTYPEP-of-NOT-types logic.
3087 * (not a change in the main branch of SBCL, but a related prototype
3088 which can hopefully be merged into the main branch of SBCL in the
3089 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3090 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3091 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3092 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3093 dispatch macro character combinations. (thanks to Alexey Dejneka)
3094 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3095 that superficially look like logical namestrings correctly.
3096 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3098 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3099 default supplied-p) for &optional and &key arguments. (thanks to
3102 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3103 * incompatible change: The compiler is now less aggressive about
3104 tail call optimization, doing it only when (> SPACE DEBUG) or
3105 (> SPEED DEBUG). (This is an incompatible change because there are
3106 programs which relied on the old CMU-CL-style behavior to optimize
3107 away their unbounded recursion which will now die of stack overflow.)
3108 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3109 for non-GENCGC systems has been increased to 20M (since that
3110 seems much closer to the likely performance optimum for modern
3111 systems than the old 4M value was)
3112 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3113 and *DEBUG-PRINT-LEVEL*
3114 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3115 of CMU CL's support for SPARC, and various endianness and other
3116 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3117 * new syntactic sugar for the Unix command line: --load foo.bar is now
3118 an alternate notation for --eval '(load "foo.bar")'.
3120 ** The system now detects stack overflow and handles it gracefully,
3121 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3122 optimization settings. (This is a good thing in general, and
3123 its introduction in this version should be particularly timely
3124 for anyone whose code fails because of suppression of tail
3126 ** The system now hunts for the C variable "environ" in a more
3127 devious way, to avoid segfaults when the C library version differs
3128 between compile time and run time. (thanks to Christophe Rhodes)
3129 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3130 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3131 ** The compiler no longer issues bogus style warnings for undefined
3132 classes in the same source file as the DEFCLASSes which defined
3133 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3135 ** fixes in CONDITION class precedence list for undefined function
3136 errors (thanks to Alexei Dejneka)
3137 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3138 correctly. (thanks to Dan Barlow)
3139 ** portability fixes aiming at bootstrapping under CLISP (thanks
3140 to Dave McDonald and Christophe Rhodes)
3141 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3142 ** fixes in type translation and and type inference (thanks to
3144 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3145 ** various fixes in the new ports (thanks to Dan Barlow)
3146 * several changes related to debugging:
3147 ** suppression of tail recursion, as noted above
3148 ** stack overflow detection, as noted above
3149 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3150 is now the default. (For some time encapsulation has been more
3151 reliable than the breakpoint-based :ENCAPSULATE NIL
3152 implementation, at least on X86 systems; and I just noticed that
3153 encapsulation also seems closer to the spirit of the ANSI
3156 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3158 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3159 up properly again. (There was a packaging bug in 0.7.0 which
3160 left their definitions in SB-SYS::LOAD-FOREIGN and
3161 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3162 vital for most things which interface to C-level interfaces,
3163 like extensions working with sockets or databases or
3164 Perl-compatible regexes or whatever, and the need to fix
3165 this bug is the main reason that 0.7.1 was released so
3167 ** DEFGENERIC is now choosier about the methods it redefines, so that
3168 reLOADing a previously-LOADed file containing DEFGENERICs does
3169 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3170 cycle now works as it should. (thanks to Alexey Dejneka)
3171 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3172 fixed by Christophe Rhodes. (He actually submitted this patch
3173 months ago, and I delayed until after 0.7.0.)
3174 ** Bug 111 (internal compiler confusion about runtime checks on
3175 FUNCTION types) was fixed by Alexey Dejneka.
3176 * Some internal cleanups (getting rid of variables which aren't
3177 needed now that the byte interpreter is gone) caused the fasl
3178 file format number to change again.
3180 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3181 * major incompatible change: The default fasl file extension, i.e. the
3182 default extension for files produced by COMPILE-FILE, has changed
3183 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3185 ** There are many changes in the implementation of the compiler.
3186 SBCL is now essentially a compiler-only implementation of ANSI
3187 Common Lisp. EVAL still "interprets" a few special cases, but
3188 almost all the interesting cases are handled by creating
3189 a LAMBDA expression, calling COMPILE on it, then calling
3190 FUNCALL on the result.
3191 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3192 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3193 Since the code is newer, there might still be some new bugs
3194 (though not as many as before Martin Atzmueller's fixes:-). But
3195 the new code is substantially simpler and clearer, and hopefully
3196 any remaining bugs will be simpler, less fundamental, and more
3197 fixable then the bugs in the old code.
3198 ** The revised compiler is still a little unsteady on its feet.
3200 *** The debugging information it produces (particularly the names
3201 of FUNCTION objects) is sometimes much less useful than what
3202 the old compiler produced.
3203 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3204 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3205 within a MACROLET) has been temporarily weakened.
3206 ** There are new compiler optimizations for various functions:
3207 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3208 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3209 *** the math functions TRUNCATE, FLOOR, and CEILING
3210 *** the function-of-all-trades COERCE
3211 Mostly these should be transparent, but there's one
3212 potentially-annoying problem (bug 117): when the compiler
3213 inline-expands a function and does type analysis on the result,
3214 it can create control paths which have type mismatches, and
3215 when it can't prove that those control paths aren't taken,
3216 it will issue WARNINGs about the type mismatches. This is
3217 a particular problem in practice for the new sequence functions.
3218 It's not clear how this should be fixed, and for now, a
3219 workaround is given in the entry for 117 in the BUGS file.
3220 ** (Because of the interaction between the two previous items --
3221 occasional inlining problems and new inline expansions -- some
3222 of the new sequence function optimizations won't really kick in
3223 completely until debugging information, and then inlining, are
3224 straightened out in some future version.)
3225 * minor incompatible changes:
3226 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3227 DIRECTORY no longer implicitly promotes NIL slots of its
3228 pathname argument to :WILD. In particular, when you ask for the
3229 contents of a directory (which you used to be able to do without
3230 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3231 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3232 ** changes in behavior that ANSI explicitly defines to be
3233 implementation dependent:
3234 *** The new compiler-only implementation still conforms with ANSI,
3235 but acts a little different than before. Besides the obvious
3236 changes in performance tradeoffs (that the cost per form passed
3237 to EVAL has gone up, and the cost per form executed by EVAL
3238 has gone down), the behavior of the system changes a little
3239 because there are no longer any interpreted function objects.
3240 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3241 e.g. doing COMPILE on the output of interactive DEFUN is
3243 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3245 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3246 has changed from (:CL) to NIL.
3247 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3248 e.g. #\Newline and #\Tab, don't have names specified in the
3249 ANSI Common Lisp standard, is now based on their ASCII symbolic
3250 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3251 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3252 are no longer used for output.
3253 ** changes in internal implementation constants:
3254 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3255 4 million. (If your application spends a lot of time GCing and
3256 you have a lot of RAM, you might want to experiment with
3257 increasing it even more.)
3258 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3259 However, almost all old code should still continue to work without
3260 immediate update, as SB-C-CALL is now a (deprecated) nickname
3262 ** Old operator names in the style DEF-FOO are now deprecated in
3263 favor of new corresponding names DEFINE-FOO, for consistency with
3264 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3265 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3266 internal symbols, but a few supported extensions like
3267 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3268 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3269 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3270 etc. as you get deeper into recursive calls to the debugger
3271 command loop, instead of the old "5]", "5]]", "5]]]"
3272 sequence. (I was motivated to do this when squabbles between
3273 ILISP and SBCL left me very deeply nested in the debugger. In the
3274 short term, this change will probably provoke more ILISP/SBCL
3275 squabbles, but hopefully it will be an improvement in the long run.)
3276 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3277 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3278 function, since declaiming return types involving aliens is
3279 (1) annoyingly messy to do by hand and (2) vital to efficient
3280 compilation of code which calls such functions.
3281 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3282 longer reexported by the SB-EXT package. They're solely useful
3283 for alien code, so it seems more logical that you should get
3284 them from the SB-ALIEN package, not in SB-EXT.
3285 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3286 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3287 features. Instead, the code that they used to control is always
3288 built into the system.
3289 * many other bug fixes
3290 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3291 advantage of the new EVAL-WHEN stuff and to clean them up in
3292 general, and they are now more ANSI-compliant in a number of
3293 ways. Martin Atzmueller is responsible for a lot of this.
3294 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3296 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3297 *** correct ERROR type for various file operations
3298 *** some fixes for Lisp streams
3299 *** DEFMETHOD syntax checking
3300 *** changing old weird representation of debug information as
3301 strings (which, among their other deficiencies, don't transform
3302 correctly when you rename packages, and don't change their
3303 print representation when you change things like *PACKAGE*
3304 and *PRINT-LENGTH*) to symbols and lists of symbols
3305 He also made several improvements and fixed several bugs in DESCRIBE.
3306 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3308 *** misbehavior of WRITE-STRING/WRITE-LINE
3309 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3310 and several other LOOP problems as well
3311 *** DIRECTORY when similar filenames are present
3312 *** DEFGENERIC with :METHOD options
3313 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3314 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3315 *** argument ordering in FIND with :TEST option
3316 *** mishandled package designator argument in APROPOS-LIST
3317 *** various problems in the backquote readmacro
3318 *** a bug in APROPOS
3319 *** probably some others that I'm not describing very well here,
3320 since the CVS log documents them by reference to sbcl-devel
3321 messages, and the SourceForge archives aren't working well.:-(
3322 ** Dan Barlow improved the Alpha port (and is making progress on the
3323 PPC port, for those of you who think different).
3324 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3325 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3326 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3327 continues to work on the SPARC port (for those of you in a position
3328 to look down upon our little PC-compatible boxes from a great height).
3329 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3330 and uses that copy, rather than the current dynamic value, when
3331 it's trying to decide whether to truncate output. Thus e.g.
3332 (let ((*print-lines* 50))
3333 (pprint-logical-block (stream nil)
3335 (let ((*print-lines* 8))
3336 (print (aref possiblybigthings i) stream)))))
3337 should now truncate the logical block only at 50 lines, instead of
3338 often truncating it at 8 lines, as it did before.
3339 * The doc/cmucl/ directory, containing old CMU CL documentation
3340 from the time of the fork, is no longer part of the base system.
3341 SourceForge has shut down its anonymous FTP service, and with it
3342 my original plan for distributing the old CMU CL documentation
3343 there. For now, if you need these files you can download an old
3344 SBCL source release and extract them from it.
3345 * The fasl file version number changed again, for dozens of reasons,
3346 some of which are apparent above.
3348 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3349 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3350 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3351 patch, so that DISASSEMBLE output is much nicer.
3352 * The code in the SB-PROFILE package now seems reasonably stable.
3353 I still haven't decided what the final interface should look like
3354 (I'd like PROFILE to interact cleanly with TRACE, since both
3355 facilities use function encapsulation) but if you have a need
3356 for profiling now, you can probably use it successfully with
3357 the current CMU-CL-style interface.
3358 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3359 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3360 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3361 link now returns the dangling link itself, and for similar
3362 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3363 link itself. (In these cases the old code signalled an error and
3364 looped endlessly, respectively.) Thus, DIRECTORY now works even
3365 in the presence of dangling and cyclic symbolic links.
3366 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3367 is now a supported extension again, since the consensus on
3368 sbcl-devel was that it can be useful for ordinary development
3369 work, not just for debugging SBCL itself.
3370 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3371 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3372 that currently-defined functions might be redefined later with
3373 different return types.
3374 * Hash tables can be printed readably, as inspired by CMU CL code
3375 of Eric Marsden and SBCL code of Martin Atzmueller.
3376 * better error handling in CLOS method combination, thanks to
3377 Martin Atzmueller porting Pierre Mai's CMU CL patches
3378 * more overflow fixes for >16Mbyte I/O buffers
3379 * A bug in READ has been fixed, so that now a single Ctrl-D
3380 character suffices to cause end-of-file on character streams.
3381 In particular, now you only need one Ctrl-D at the command
3382 line (not two) to exit SBCL.
3383 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3385 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3386 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3387 does the right thing.
3388 * The implementation of some type tests, especially for CONDITION
3389 types, is now tidier and maybe faster, due to CMU CL code
3390 originally by Douglas Crosher, ported by Martin Atzmueller.
3391 * Some math functions have been fixed, and there are new
3392 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3393 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3394 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3395 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3396 likely to be built into sbcl-0.7.0.)
3397 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3398 CL, ported to SBCL by Martin Atzmueller)
3399 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3400 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3401 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3402 * There's a new slam.sh hack to shorten the edit/compile/debug
3403 cycle for low-level changes to SBCL itself, and a new
3404 :SB-AFTER-XC-CORE target feature to control the generation of
3405 the after-xc.core file needed by slam.sh.
3406 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3407 COMPILE-FILE is no longer supported, so that now every function
3408 gets an entry point, so that block compilation looks a little
3409 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3410 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3411 returns the number of bytes consed since the system started,
3412 rather than the number consed since the first time the function
3413 was called. (The new definition parallels ANSI functions like
3414 CL:GET-INTERNAL-RUN-TIME.)
3415 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3416 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3417 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3418 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3419 * Fasl file version numbers are now independent of the target CPU,
3420 since historically most system changes which required version
3421 number changes have affected all CPUs equally. Similarly,
3422 the byte fasl file version is now equal to the ordinary
3425 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3426 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3427 is no longer recognized. I apologize for this, because it was
3428 listed in SB-EXT as a supported extension, but I found that
3429 its existing behavior was poorly specified, as well as incorrectly
3430 specified, and it looked like too much of a mess to straighten it
3431 out. I have enough on my hands trying to get ANSI stuff to work..
3432 * many patches ported from CMU CL by Martin Atzmueller, with
3433 half a dozen bug fixes in pretty-printing and the debugger, and
3434 half a dozen others elsewhere
3435 * fixed bug 13: Floating point infinities are now supported again.
3436 They might still be a little bit flaky, but thanks to bug reports
3437 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3438 as flaky as they were.
3439 * The --noprogrammer command line option is now supported. (Its
3440 behavior is slightly different in detail from what the old man
3441 page claimed it would do, but it's still appropriate under the
3442 same circumstances that the man page talks about.)
3443 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3444 are now supported, and enabled by default. Thus, the compiler can
3445 handle many floating point and complex operations much less
3446 inefficiently. (Thus e.g. you can implement a complex FFT
3448 * The compiler now detects type mismatches between DECLAIM FTYPE
3449 and DEFUN better, and implements CHECK-TYPE more correctly, and
3450 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3451 thanks to patches from Martin Atzmueller.
3452 * various fixes to make the cross-compiler more portable to
3453 ANSI-conforming-but-different cross-compilation hosts (notably
3454 Lispworks for Windows, following bug reports from Arthur Lemmens)
3455 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3456 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3457 fixed thanks to Pierre Mai's CMU CL patches.
3458 * fixes to keep the system from overflowing internal counters when
3459 it tries to use i/o buffers larger than 16M bytes
3460 * fixed bug 45a: Various internal functions required to support
3461 complex special functions have been merged from CMU CL sources.
3462 (When I was first setting up SBCL, I misunderstood a compile-time
3463 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3464 * improved support for type intersection and union, fixing bug 12
3465 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3466 more obscure bugs as well
3467 * some steps toward byte-compiling non-performance-critical
3468 parts of the system, courtesy of patches from Martin Atzmueller
3469 * Christophe Rhodes has made some debian packages of sbcl at
3470 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3471 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3472 stable, but are nonetheless usable. When he's ready, I'd be happy
3473 to add them to the SourceForge "File Releases" section. (And if
3474 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3475 * new fasl file format version number (because of changes in
3476 internal representation of (OR ..) types to accommodate the new
3477 support for (AND ..) types, among other things)
3479 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3480 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3482 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3484 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3485 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3486 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3487 * DESCRIBE now works on structure objects again.
3488 * Most function call argument type mismatches are now handled as
3489 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3490 know whether the function will be redefined before the call is
3491 executed. (The compiler could flag local calls with full WARNINGs,
3492 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3493 it doesn't keep track of enough information to know whether calls
3494 are local in this sense.)
3495 * Compiler output is now more verbose, with messages truncated
3496 later than before. (There should be some supported way for users
3497 to override the default verbosity, but I haven't decided how to
3498 provide it yet, so this behavior is still controlled by the internal
3499 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3500 src/compiler/ir1util.lisp.)
3501 * Fasl file format version numbers have increased again, because
3502 support for the Gray streams extension changes the layout of the
3503 system's STREAM objects.
3504 * The Gray subclassable streams extension now works, thanks to a
3505 patch from Martin Atzmueller.
3506 * The full LOAD-FOREIGN extension (not just the primitive
3507 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3508 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3509 but like most other programs, it defaults to copying the Unix
3510 environment from the original process instead of starting the
3511 new process in an empty environment.
3512 * Extensions which manipulate the Unix environment now support
3513 an :ENVIRONMENT keyword option which doesn't smash case or
3514 do other bad things. The CMU-CL-style :ENV option is retained
3515 for porting convenience.
3516 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3517 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3519 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3521 * A patch from Martin Atzmueller seems to have solved the SIGINT
3522 problem, and as far as we know, signal-handling now works cleanly.
3523 (If you find any new bugs, please report them!)
3524 * The system no longer defaults Lisp source file names to types
3525 ".l", ".cl", or ".lsp", but only to ".lisp".
3526 * The compiler no longer uses special default file extensions for
3527 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3528 seems to expect a single default extension for all compiled code,
3529 and there's no compelling reason to try to stretch the standard
3530 to allow two different extensions.) Instead, byte-compiled files
3531 default to the same extension as native-compiled files.
3532 * Fasl file format version numbers have increased again, because
3533 a rearrangement of internal implementation packages made some
3534 dumped symbols in old fasl files unreadable in new cores.
3535 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3536 many fewer weird special cases.
3537 * Bug #17 (differing COMPILE-FILE behavior between logical and
3538 physical pathnames) has been fixed, and some related misbehavior too,
3539 thanks to a patch from Martin Atzmueller.
3540 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3541 by Tim Moore, ported to SBCL by Martin Atzmueller.
3542 * Martin Atzmueller fixed several filesystem-related problems,
3543 including bug #36, in part by porting CMU CL patches, which were
3544 written in part by Paul Werkowski.
3545 * More compiler warnings in src/runtime/ are gone, thanks to
3546 more patches from Martin Atzmueller.
3547 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3550 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3552 * DESCRIBE now works on CONDITION objects.
3553 * The debugger now handles errors which arise when trying to print
3554 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3556 * The build system now uses an additional file, customize-target-features.lisp,
3557 to allow local modifications to the target *FEATURES* list. (The point of
3558 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3559 debugging features enabled, without having to worry about propagating it
3560 into everyone's system when I do a "cvs update".) When no
3561 customize-target-features.lisp file exists, the target *FEATURES* list
3562 should be constructed the same way as before.
3563 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3564 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3565 warn about behavior which is undefined under the ANSI spec. Note: This
3566 is specified by ANSI, but it's not very popular with programmers.
3567 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3568 macro in the SBCL sources for an example of a workaround which you
3569 might use to make portable ANSI-standard code which does what you want.
3570 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3571 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3572 when it's used as a non-toplevel form. (This is required in order
3573 to implement the DEFCONSTANT-EQX macro.)
3574 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3575 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3576 funny things, due to bugs in the implementation of EVAL-WHEN
3577 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3578 non-toplevel DEFCONSTANTs are uncommon.)
3579 * The core file version number and fasl file version number have been
3580 incremented, because the old noncompliant DEFCONSTANT behavior involved
3581 calling functions which no longer exist, and because I also took the
3582 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3583 * fixed bug 1 (error handling before read-eval-print loop starts), and
3584 redid debugger restarts and related debugger commands somewhat while
3586 ** The QUIT debugger command is gone, since it did something
3587 rather different than the SB-EXT:QUIT command, and since it never
3588 worked properly outside the main toplevel read/eval/print loop.
3589 Invoking the new TOPLEVEL restart provides the same functionality.
3590 ** The GO debugger command is also gone, since you can just invoke
3591 the CONTINUE restart directly instead.
3592 ** The TOP debugger command is also gone, since it's redundant with the
3593 FRAME 0 command, and since it interfered with abbreviations for the
3595 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3597 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3598 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3599 some porting work by Martin Atzmueller)
3600 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3601 with a boatload of patches which he ported to SBCL. Now that those
3603 ** The system tries to make sure that its low-priority messages
3604 are prefixed by semicolons, to help people who like to use
3605 syntax highlighting in their ILISP buffer. (This patch
3606 was originally due to Raymond Toy.)
3607 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3608 patches originally written by Raymond Toy.
3609 ** The compiler understands coercion between single-value and
3610 multiple-VALUES type expressions better, getting rid of some very
3611 weird behavior, thanks to patches originally by Robert MacLachlan
3612 and Douglas Crosher.
3613 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3614 lambda lists, thanks to a patch originally by Pierre Mai.
3615 ** The system no longer bogusly warns about "abbreviated type
3617 ** The compiler gets less confused by inlining and RETURN-FROM,
3618 thanks to some patches originally by Tim Moore.
3619 ** The system no longer hangs when dumping circular lists to fasl
3620 files, thanks to a patch originally from Douglas Crosher.
3621 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3622 undefined function error.
3623 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3624 in structure slot initforms) for now, documented workaround instead:-|
3625 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3626 * fixed bug 5: added stubs for various Gray stream functions called
3627 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3628 installed, at least appropriate type errors are generated
3629 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3630 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3631 consistently in DEFMETHOD forms.
3632 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3633 it doesn't seem to affect SBCL after all
3634 * The C runtime system now builds with better optimization and many
3635 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3637 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3639 * The system is now under CVS at SourceForge (instead of the
3640 CVS repository on my home machine).
3641 * The new signal handling code has been tweaked to treat register
3642 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3643 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3644 downstream, e.g. in debug-int.lisp, has implicit dependencies
3645 on the unsignedness of integer representation of machine words,
3646 and that caused the system to bomb out with infinite regress
3647 when trying to recover from type errors involving signed values,
3648 e.g. (BUTLAST '(1 2 3) -1).)
3649 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3650 (This was one of the bugs Peter Van Eynde reported back in July.)
3651 * The system now uses code inspired by Colin Walters' O(N)
3652 implementation of MAP (from the cmucl-imp@cons.org mailing
3653 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3654 inline the MAP operation, and there is more than one
3655 sequence argument to the MAP call (so that it can't just
3656 do ETYPECASE once and for all based on the type of the
3657 single sequence argument). (The old non-inline implementation
3658 of the general M-argument sequence-of-length-N case required
3659 O(M*N*N) time when any of the sequence arguments were LISTs.)
3660 * The QUIT :UNIX-CODE keyword argument has been renamed to
3661 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3662 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3664 * Raymond Wiker's patches to port dynamic loading from Linux to
3665 FreeBSD have been added.
3666 * The BUGS file is now more nearly up to date, thanks in large part
3667 to Martin Atzmueller's review of it.
3668 * The debugger now flushes standard output streams before it begins
3669 its output ("debugger invoked" and so forth).
3670 * The core version number and fasl file version number have both
3671 been incremented, because of incompatible changes in the layout
3673 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3674 used to not be called for a saved Lisp image.)
3675 * Martin Atzmueller's version of a patch to fix a compiler crash,
3676 as posted on sbcl-devel 13 September 2000, has been installed.
3677 * Instead of installing Martin Atzmueller's patch for the
3678 compiler transform for SUBSEQ, I deleted the compiler transform,
3679 and transforms for some similar consing operations.
3680 * A bug in signal handling which kept TRACE from working on OpenBSD
3682 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3683 away by constant folding
3684 * The system now defines its address space constants in one place
3685 (in the Lisp sources), and propagates them automatically elsewhere
3686 (through GENESIS and the sbcl.h file). Therefore, patching the
3687 address map is less unnecessarily tedious and error-prone. The
3688 Lisp names of address space constants have also been systematized.
3689 * CVS tags like dollar-Header-dollar have been removed from
3690 the sources, because they have never saved me trouble and
3691 they've been source of trouble working with patches and other
3692 diff-related operations.
3693 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3694 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3696 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3698 * The system has been ported to OpenBSD.
3699 * The system now compiles with a simple "sh make.sh" on the systems
3700 that it's supported on. I.e., now you no longer need to tweak
3701 text in the target-features.lisp-expr and symlinks in src/runtime/
3702 by hand, the make.sh takes care of it for you.
3703 * The system is no longer so grossly inefficient when compiling code
3704 involving vectors implemented as general (not simple) vectors (VECTOR T),
3705 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3706 which dares to use the various sequence functions on non-simple
3707 vectors, takes less of a performance hit.
3708 * There is now a primitive type predicate VECTOR-T-P
3709 to test for the (VECTOR T) type, so that e.g.
3710 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3711 can now be compiled with some semblance of efficiency. (The old code
3712 turned the type declaration into a full call to %TYPEP at runtime!)
3713 * AREF on (VECTOR T) is still not fast, since it's still compiled
3714 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3715 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3716 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3717 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3718 without making full calls to SUBTYPEP at runtime.
3719 (Some analogous efficiency issues for non-simple vectors specialized to
3720 element types other than T, or for non-simple multidimensional arrays,
3721 have not been addressed. They could almost certainly be handled the
3722 same way if anyone is motivated to do so.)
3723 * The changes in array handling break binary compatibility, so
3724 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3725 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3726 instead of (VALUES T T).
3727 * By following the instructions that Dan Barlow posted to sbcl-devel
3728 on 2 July 2000, I was able to enable primitive dynamic object
3729 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3730 functionality is not implemented (since it calls ld to resolve
3731 library references automatically, requiring RUN-PROGRAM for its
3732 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3733 try to resolve library references) is now supported.
3734 * The system now flushes the standard output streams when it terminates,
3735 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3736 them at several other probably-convenient times, e.g. in each pass of
3737 the toplevel read-eval-print loop, and after evaluating a form given
3738 as an "--eval" command-line option. (These changes were motivated by a
3739 discussion of stream flushing issues on cmucl-imp in August 2000.)
3740 * The source transform for TYPEP of array types no longer assumes
3741 that an array whose element type is a not-yet-defined type
3742 is implemented as an array of T, but instead punts, so that the
3743 type will be interpreted at runtime.
3744 * There is now some support for cross-compiling in make.sh: each of
3745 the phases of make.sh has its own script. (This should be transparent
3746 to people doing ordinary, non-cross-compile builds.)
3747 * Since my laptop doesn't have hundreds of megabytes of memory like
3748 my desktop machine, I became more motivated to do some items on
3749 my to-do list in order to reduce the size of the system a little:
3750 ** Arrange for various needed-only-at-cold-init things to be
3751 uninterned after cold init. To support this, those things have
3752 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3753 symbols with such names are now uninterned after cold init).
3754 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3755 fasl files for cold load.
3756 ** Remove the old compiler structure pooling code (which used to
3757 be conditional on the target feature :SB-ALLOC) completely.
3758 ** Redo the representation of some data in cold init to be more compact.
3759 (I also looked into supporting byte compiled code at bootstrap time,
3760 which would probably reduce the size of the system a lot, but that
3761 looked too complicated, so I punted for now.)
3762 * The maximum signal nesting depth in the src/runtime/ support code has
3763 been reduced from 4096 to 256. (I don't know any reason for the very
3764 large old value. If the new smaller value turns out to break something,
3765 I'll probably just bump it back up.)
3766 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3768 * Many, many bugs reported by Peter Van Eynde have been added to
3769 the BUGS list; some have even been fixed.
3770 * While enabling dynamic object file loading, I tried to make the
3771 code easier to understand, renaming various functions and variables
3772 with less ambiguous names, and changing some function calling
3773 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3774 * While trying to figure out how to do the OpenBSD port, I tried to
3775 clean up some of the code in src/runtime/. In particular, I dropped
3776 support for non-POSIX signal handling, added various comments,
3777 tweaked the code to reduce the number of compilation warnings, and
3778 renamed some files to increase consistency.
3779 * To support the new automatic configuration functionality in make.sh,
3780 the source file target-features.lisp-expr has been replaced with the
3781 source file base-target-features.lisp-expr and the machine-generated
3782 file local-target-features.lisp-expr.
3783 * fixed a stupid quoting error in make.sh so that using CMU CL
3784 "lisp -batch" as cross-compilation host works again
3786 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3788 * DESCRIBE no longer tries to call itself recursively to describe
3789 bound/fbound values, so that it no longer fails on symbols which are
3790 bound to themselves (like keywords, T, and NIL).
3791 * DESCRIBE now works on generic functions.
3792 * The printer now prints less-screwed-up representations of closures
3793 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3794 * A private symbol is used instead of the :EMPTY keyword previously
3795 used to mark empty slots in hash tables. Thus
3796 (DEFVAR *HT* (MAKE-HASH-TABLE))
3797 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3798 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3799 now does what ANSI says that it should. (You can still get
3800 similar noncompliant behavior if bang on the hash table
3801 implementation with all the symbols you get back from
3802 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3803 This breaks binary compatibility, since tests for equality to
3804 :EMPTY are wired into things like the macroexpansion of
3805 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3807 * There's now a minimal placeholder implementation for CL:STEP,
3808 as required by ANSI.
3809 * An obscure bug in the interaction of the normal compiler, the byte
3810 compiler, inlining, and structure predicates has been patched
3811 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3812 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3814 * Missing ordinary arguments in a macro call are now detected even
3815 when the macro lambda list contains &KEY or &REST.
3816 * The debugger no longer complains about encountering the top of the
3817 stack when you type "FRAME 0" to explicitly instruct it to go to
3818 the top of the stack. And it now prints the frame you request even
3819 if it's the current frame (instead of saying "You are here.").
3820 * As specified by ANSI, the system now always prints keywords
3821 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3823 * The default initial SIZE of HASH-TABLEs is now smaller.
3824 * Type information from CLOS class dispatch is now propagated
3825 into DEFMETHOD bodies, so that e.g.
3826 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3828 is now basically equivalent to
3829 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3830 (DECLARE (TYPE SINGLE-FLOAT X))
3832 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3833 operation, without having to do run-time type dispatch.
3834 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3835 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3836 * Since I don't seem to be making big file reorganizations very often
3837 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3838 is overflowing my ability to conveniently back them up), I've finally
3839 checked the system into CVS. (The CVS repository is on my home system,
3840 not at SourceForge -- putting it on SourceForge might come later.)
3841 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3842 high-level GC-NOTIFY-FOO functions send their output. (There's
3843 still very little control of where low-level verbose GC functions
3844 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3845 less than it used to -- the GC-NOTIFY-FOO functions are now under
3846 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3847 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3848 in only one place in the source code, and propagates it automatically
3849 everywhere that it's needed. Thus e.g. when I bump the version from
3850 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3851 * The C source files now include boilerplate legalese and documentation
3852 at the head of each file (just as the Lisp source files already did).
3853 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3854 to his page will be replaced with a link to his new CLiki service.
3856 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3858 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3859 * The build process now looks for GNU make under the default name "gmake",
3860 instead of "make" as it used to. If GNU make is not available as "gmake"
3861 on your system, you can change this default behavior by setting the
3862 GNUMAKE environment variable.
3863 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3864 can build without error under CMU CL.
3866 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3868 * There is now a partial SBCL user manual (with some new text and some
3869 text cribbed from the CMU CL manual).
3870 * The beginnings of a profiler have been added (starting with the
3871 CMU CL profiler and simplifying and cleaning up). Eventually the
3872 main interface should be through the TRACE macro, but for now,
3873 it's still accessed through vaguely CMU-CL-style functions and macros
3874 exported from the package SB-PROFILE.
3875 * Some problems left over from porting CMU CL to the new
3876 cross-compilation bootstrap process have been cleaned up:
3877 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3878 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3879 part when getting rid of the compiler *BACKEND* stuff.)
3880 ** The value of *NULL-TYPE* was screwed up, because it was
3881 being initialized before the type system knew the final
3882 definition of the 'NULL type. This screwed up several key
3883 optimizations in the compiler, causing inefficiency in all sorts
3884 of places. (I found it because I wanted to understand why
3885 GET-INTERNAL-RUN-TIME was consing.)
3886 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3887 PROCLAIM FTYPE information. Unfortunately this broke binary
3888 compatibility again, since now the forms output by DEFGENERIC
3889 to refer to functions which didn't exist in 0.6.3.
3890 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3891 can use the new (as of 0.6.3) transform for SOME into MAP into
3893 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3894 values to (INTEGER 0 1000000), so that the time code will no longer
3895 occasionally get blown up by Linux returning 1000000 microseconds
3896 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3897 its output conform to the ANSI spec. (Alas, this makes its output
3898 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3899 * A full call to MAP NIL with a single sequence argument no longer conses.
3900 * fixes to problems pointed out by Martin Atzmueller:
3901 * The manual page no longer talks about multiprocessing as though
3902 it were currently supported.
3903 * The ILISP support patches have been removed from the distribution,
3904 because as of version 5.10.1, ILISP now supports SBCL without us
3905 having to maintain patches.
3906 * added a modified version of Raymond Toy's recent CMU CL patch for
3907 EQUALP comparison of HASH-TABLE
3909 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3911 * The system still can't cross-compile itself with
3912 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3913 entails), but at least it can compile more of itself that way
3914 than it used to be able to, and various buglets which were uncovered
3915 by trying to cross-compile itself that way have now been fixed.
3916 * This release breaks binary compatibility again. This time
3917 at least I've incremented the FASL file format version to 2, so that the
3918 problem can be detected reliably instead of just causing weird errors.
3919 * various new style warnings:
3920 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3921 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3922 ** using the lexical binding of a variable named in the *FOO* style
3923 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3924 as specified by ANSI.
3925 * *RANDOM-STATE* is no longer automatically initialized from
3926 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3927 default behavior of the system is to repeat its behavior every time
3928 it's run. If you'd like to change this behavior, you can always
3929 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3930 old convention there was no comparably easy way to get the system to
3931 repeat its behavior every time it was run.
3932 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3933 FTYPE declarations has been removed, in favor of their ANSI
3934 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3935 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3936 the types of their sequence arguments can be determined at compile time.
3937 This is done through a new open code expansion for MAP which eliminates
3938 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3939 argument types can be determined at compile time.
3940 * The optimizer now transforms COERCE into an identity operation when it
3941 can prove that the coerced object is already of the correct type. (This
3942 can be a win for machine generated code, including the output of other
3943 optimization transforms, such as the MAP transform above.)
3944 * Credit information has been moved from source file headers into CREDITS.
3945 * Source file headers have been made more standard.
3946 * The CASE macro now compiles without complaining even when it has
3949 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3951 * (Note that the way that the PCL macroexpansions were rewritten
3952 to accommodate the change in DEFGENERIC below breaks binary
3953 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3954 not run under sbcl-0.6.2. Once we get out of alpha releases,
3955 i.e. hit release 1.0.0, we'll probably try to maintain binary
3956 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3957 and sbcl-1.4.4. Until then, however, it might be fairly common
3958 for maintenance releases to break binary compatibility.)
3959 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3961 * The reporting of unhandled signals has been changed to print some
3962 explanatory text as well as the report form. (Previously only
3963 the report form was printed.)
3964 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3965 it defines, so that the compiler no longer issues undefined function
3966 warnings for compiled-but-not-yet-loaded generic functions.
3967 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3968 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3969 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3970 section of the ANSI spec are supported.
3971 * The "" nickname for the "KEYWORD" package has been removed.
3972 The reader still handles symbol tokens which begin with a package marker
3973 as keywords, but it doesn't expose its mechanism for doing so in the
3974 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3975 * The system now issues STYLE-WARNINGs for contradictory TYPE
3976 proclamations. (Warnings for contradictory FTYPE proclamations would
3977 be nice too, but those can't be done usefully unless the type system
3978 is made smarter about FUNCTION types.)
3979 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3980 systematized, so that "*target-*.lisp is supposed to exist only on the
3981 target and imply that there's a related file which exists on the
3982 host, and *host-*.lisp is supposed to exist only on the host and imply
3983 that there's a related file which exists on the target. This involves a
3984 lot of renaming. Hopefully the acute confusion caused by the renaming
3985 will be justified by the reduction in chronic confusion..
3986 ** runtime-type.lisp -> early-target-type.lisp
3987 ** target-type.lisp -> late-target-type.lisp
3988 ** early-host-format.lisp -> early-format.lisp
3989 ** late-host-format.lisp -> late-format.lisp
3990 ** host-error.lisp -> misc-error.lisp
3991 ** early-error.lisp -> early-target-error.lisp
3992 ** late-error.lisp -> late-target-error.lisp
3993 ** host-defboot.lisp -> early-defboot.lisp
3994 ** code/misc.lisp -> code/target-misc.lisp
3995 ** code/host-misc.lisp -> code/misc.lisp
3996 ** code/numbers.lisp -> code/target-numbers.lisp
3997 ** code/early-numbers.lisp -> numbers.lisp
3998 ** early-host-type.lisp -> early-type.lisp
3999 ** late-host-type.lisp -> late-type.lisp
4000 ** host-typep.lisp -> typep.lisp
4001 ** load.lisp -> target-load.lisp
4002 ** host-load.lisp -> load.lisp
4003 ** host-disassem.lisp -> disassem.lisp
4004 ** host-insts.lisp -> insts.lisp
4005 ** byte-comp.lisp -> target-byte-comp.lisp
4006 ** host-byte-comp.lisp -> byte-comp.lisp
4007 ** host-signal.lisp -> signal.lisp
4008 ** host-defstruct.lisp -> defstruct.lisp
4009 ** late-target-type.lisp -> deftypes-for-target.lisp
4010 Furthermore, several other previously target-only files foo.lisp (e.g.
4011 hash-table.lisp and random.lisp) have been split into a target-and-host
4012 foo.lisp file and a target-only target-foo.lisp file, with their key type
4013 definitions in the target-and-host part, so that the cross-compiler will
4014 know more about target types.
4015 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
4016 gone away. In their place are various *BACKEND-FOO* variables
4017 corresponding to the slots of the old structure.
4018 * A bug which caused the SB-COLD bootstrap-time package to be propagated
4019 into the target SBCL has been fixed.
4020 * The chill.lisp system for loading cold code into a running SBCL
4022 * Support for the CMU CL "scavenger hook" extension has been removed.
4023 (It was undocumented and unused in the CMU CL sources that SBCL was
4024 derived from, and stale in sbcl-0.6.1.)
4025 * Various errors in the cross-compiler type system were detected
4026 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4027 (enabling various consistency checks). Many of them were fixed,
4028 but some hard problems remain, so the compiler is back to
4029 running without *TYPE-SYSTEM-INITIALIZED* for now.
4030 * As part of the cross-compiler type system cleanup, I implemented
4031 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4032 * I have started adding UNCROSS calls throughout the type system
4033 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4034 on cross-compiler input files will be able to go away, and various
4036 * CONSTANTP now returns true for quoted forms (as explicitly required
4039 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4041 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4042 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4043 (MAKE-LIST -1). In the longer term, it ought to become true
4044 that declarations are assertions even at SAFETY 1. For now, it's not
4045 quite true even at SAFETY 3, but it's at least more nearly true..
4046 (Note that this change seems to increases the size of the system by
4047 O(5%) and to decrease the speed of the compiler by 20% or more.)
4048 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4049 for the problem of printing dozens of lines of distracting information
4050 about low-level system machinery as part of the top stack frame
4051 on entry to the debugger when an undefined function was called.
4052 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4053 is not reset to COMMON-LISP-USER.
4054 * Compilation of stuff related to dyncount.lisp has been made conditional
4055 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4056 smaller. The various dyncount-related symbols have been moved into
4057 a new "SB-DYNCOUNT" package.
4058 * tty-inspect.lisp has been renamed to inspect.lisp.
4059 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4060 feature has gone away. (When we eventually port to other flavors of
4061 libc and/or Unix, we'll try to make the differences between flavors
4062 invisible at the user level.)
4063 * Various other *FEATURES* tags, and/or their associated conditionals,
4064 have been removed if obsolescent, or given better documentation, or
4065 sometimes given more-mnemonic names.
4067 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4069 * tidied up "make.sh" script
4070 * tidied up system directory structure
4071 * better "clean.sh" behavior
4072 * added doc/FOR-CMUCL-DEVELOPERS
4073 * many many small tweaks to output format, e.g. removing possibly-confusing
4074 trailing #\. character in DESCRIBE-INSTANCE
4075 * (EQUALP #\A 'A) no longer signals an error.
4076 * new hashing code, including EQUALP hashing
4077 * tidied up Lisp initialization and toplevel
4078 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4079 * command line argument processing
4080 * added POSIX-GETENV function to deal with Unix-ish environment variables
4081 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4082 terminating SBCL on EOF
4083 * non-verbose GC by default
4084 * There is no more "sbcl" shell script; the sbcl file is now the C
4085 runtime executable (just like CMU CL).
4086 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4088 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4089 symbol and package information as Lisp native symbol and package objects
4090 instead of strings naming symbols and strings naming packages. This way,
4091 whenever packages are renamed (as in warm init), debug information is
4092 transformed along with everything else.
4093 * tweaked the optimization policy declarations which control the building
4094 of SBCL itself. Now, among other things, the system no longer saves
4095 source location debugging information. (This helps two problems at once
4096 by reducing SBCL size and by keeping SBCL from trying to look for its
4097 sources -- which may not exist -- when reporting errors.)
4098 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4099 debugging and testing purposes
4100 * cleaned up printing, making the printer call PRINT-OBJECT for
4101 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4102 methods, giving nearly-ANSI behavior
4103 * converted almost all special variables to use *FOO* naming convention
4104 * deleted PARSE-TIME functionality, since it can be done portably
4105 * moved some files out of cold init into warm init
4106 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4108 * regularized formatting of source files
4109 * added an install.sh script
4110 * fixed ridiculous memory usage of cross-compiler by making
4111 compiler/alloc.lisp not try to do pooling unless it can hook
4112 itself into the GC of the cross-compilation host. Now the system
4113 builds nicely on my old laptop.
4114 * added :SB-ALLOC in target-features.lisp-expr
4115 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4116 was not implemented)
4117 * re-did condition handling and note reporting in the compiler. Notes
4118 are no longer handled by signalling conditions. Style warnings
4119 and warnings are handled more correctly and reported in such a way
4120 that it's easy to find one or the other in your output (so that you
4121 can e.g. figure out which of many problems caused COMPILE-FILE to
4123 * changed the severity of several compiler warnings from full WARNING
4124 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4125 compiler note reporting so that it doesn't use the condition system
4126 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4127 COMPILE-FILE command)
4128 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4129 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4130 no longer supported, since their implementation was deeply intertwingled
4131 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4132 * removed bogus "support" for compiler macros named (SETF FOO), and
4133 removed the compiler macro for SETF INFO (but only after making a fool
4134 of myself on the cmucl-imp mailing list by posting a bogus patch for
4135 DEFINE-COMPILER-MACRO..)
4136 * Compiled files containing forms which have side effects on the Lisp
4137 reader (such as DEFPACKAGE forms) are now handled more correctly.
4138 (Compiler queuing of top level lambdas has been suppressed by setting
4139 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4140 * deleted various currently-unused source files, e.g. gengc.lisp. They
4141 may be added back at some point e.g. when porting to other architectures,
4142 but until they are it's distracting to distribute them and to try to
4144 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4145 there were so many of them they're just distractions, and UNCROSS is
4146 known to be able to handle the current sources
4147 * moved PROFILE functionality into TRACE, so that it will be clear
4148 how the wrapping and unwrapping of functions when you profile them
4149 interacts with the wrapping and unwrapping of functions when you
4150 trace them. (Actually, the functionality isn't there yet, but at least
4151 the interface specification is there. Hopefully, the functionality will
4152 arrive with some future maintenance release.)
4153 * removed host-oops.lisp
4154 * changed signature of QUIT function to allow UNIX-CODE argument
4155 * fixed READ-SEQUENCE bug
4156 * tweaked verbose GC output so that it looks more like the progress
4157 output that ANSI specifies for functions like LOAD
4158 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4159 * added <http://sbcl.sourceforge.com> to the banner information printed by