1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-1.0.2 relative to sbcl-1.0.1:
3 * improvement: experimental support for mach exception handling on
4 x86/macos. requires building with :MACH-EXCEPTION-HANDLER feature
6 * improvement: support for GBK external format.
7 (thanks to Chun Tian (binghe))
8 * improvement: the debugger now displays variables that have been closed
9 over, in code compiled with (DEBUG 2) or higher
10 * improvement: support for executable cores on NetBSD (thanks to
12 * new feature: added a RESTART-FRAME debugger command
13 * new feature: new generic function SB-GRAY:STREAM-FILE-POSITION can
14 be used to provide an implementation for FILE-POSITION on Gray streams
15 (thanks to Eric Marsden)
16 * new feature: add syslog, openlog and closelog support to SB-POSIX
17 (thanks to Richard Kreuter)
18 * optimization: the function call overhead in code compiled with
19 a high DEBUG optimization setting is significantly
20 * bug fix: an error is signaled for attempts to use READ-SEQUENCE
21 for a (SIGNED-BYTE 8) stream and (UNSIGNED-BYTE 8) vector, or vice versa.
22 (thanks to Tony Martinez)
23 * bug fix: the initforms for DEFMETHOD &AUX parameters are only
24 evaluated once (reported by Kevin Reid)
25 * bug fix: the :SHOW-PROGRESS keyword parameter to SB-SPROF:WITH-PROFILING
26 works again (thanks to Kilian Sprotte)
27 * bug fix: an error is signaled for tagbodies with duplicate tags
28 (thanks to Stephen Wilson)
29 * bug fix: NIL can be used as a tagbody tag (thanks to Stephen Wilson)
30 * bug fix: Win32 port can now handle foreign code unwinding Lisp
31 stack frames from alien callbacks.
32 * bug fix: ATANH returned incorrect results on win32 (thanks to Pierre Mai)
33 * bug fix: SBCL works on Linux/ppc systems with a kernel configured to use
34 65k pages (thanks to David Woodhouse)
35 * bug fix: fix SB-POSIX dirent and socket on NetBSD (thanks to Richard Kreuter)
37 changes in sbcl-1.0.1 relative to sbcl-1.0:
38 * new platform: FreeBSD/x86-64, including support for threading.
39 * new feature: the compiler stores cross-referencing information
40 abount function calls (who-calls), macroexpansion (who-macroexpands)
41 and special variables (who-binds, who-sets, who-references) for code
42 compiled with (< SPACE 3). This information is available through the
43 sb-introspect contrib.
44 * new feature: users may subclass SEQUENCE, and have instances of
45 these classes interoperate with standard Common Lisp functions if
46 a number of methods are defined. (This feature is experimental
47 and the interface subject to change based on feedback from SBCL
48 users and the general community)
49 * improvement: sb-sprof traces call stacks to an arbitrary depth on
50 x86/x86-64, rather than the previous fixed depth of 8
51 * improvement: another pthread back-end of mutex "pthread-futex".
52 * bug fix: non-ascii command-line arguments are processed correctly
53 (thanks to Yaroslav Kavenchuk)
54 * bug fix: non-required arguments were not passed correctly when a method
55 defined using DEFMETHOD was called from a mop-generated method using
56 CALL-NEXT-METHOD (reported by Pascal Costanza)
57 * bug fix: recursion is now permitted in accessors through
58 SLOT-UNBOUND. (reported by Pascal Costanza)
59 * bug fix: an error was signaled at startup if the HOME environment
60 variable was defined, but had an empty value (reported by Peter Van Eynde)
61 * bug fix: non ordinary lambda-list keyword in ordinary lambda lists
62 signal a PROGRAM-ERROR, not a BUG.
63 * bug fix: SB-POSIX:READDIR works when built with large file support.
64 * bug fix: ENOUGH-NAMESTRING works relative to #p"/" as well. (thanks
66 * enhancement: DESTRUCTURING-BIND lambda-list uses &BODY instead of &REST
67 for better automatic indentation support. (thanks to Matt Pillsbury)
68 * optimization: loading generic functions no longer takes O(n^2) time,
69 proportional to the amount of methods in the generic function
70 (reported by Todd Sabin and Jeremy Brown)
71 * optimization: the FIND and POSITION family of sequence functions
72 are significantly faster on arrays whose element types have been
74 * improvements to the Windows port:
75 ** Intermittent heap corruption problems have been fixed. (thanks
76 to Alastair Bridgewater)
77 ** TRACE :ENCAPSULATE NIL (and function end breakpoints)
79 ** Lisp is able to unwind foreign exception frames from alien
80 callbacks. (thanks to Alastair Bridgewater)
82 changes in sbcl-1.0 relative to sbcl-0.9.18:
83 * improvement: experimental support for threading on FreeBSD/x86.
84 (thanks to NIIMI Satoshi)
85 * improvement: runtime option --dynamic-space-size can be used
86 to set the size of the dynamic space reserved on startup.
87 * improvement: floating point modes in effect are now saved in
88 core, and restored on startup.
89 * improvement: GET-INTERNAL-REAL-TIME now reports the time since
90 startup, not time since first call to GET-INTERNAL-REAL-TIME.
91 * improvement: SAVE-LISP-AND-DIE explicitly checks that multiple
92 threads are not running after *SAVE-HOOKS* have run.
93 * improvement: writes to CLOS instance slots are type-checked in code
94 compiled with (SAFETY 3)
95 * improvement: floating-point exception handling on FreeBSD (thanks to
97 * improvement: SB-POSIX supports time(2), utime(2) and utimes(2)
98 (thanks to Zach Beane)
99 * improvement: support for files larger than 2GB for CL streams and SB-POSIX
101 * improvement: added support for the Shift-JIS external format.
102 (contributed by NIIMI Satoshi)
103 * improvement: callbacks are supported on Linux/PPC. (thanks to
105 * bug fix: compiler bug triggered by a (non-standard) VALUES
106 declaration in a LET* was fixed. (reported by Kaersten Poeck)
107 * bug fix: file compiler no longer confuses validated and already
108 dumped structurres. (reported by Kaersten Poeck)
109 * bug fix: ADJUST-ARRAY :FILL-POINTER T on an array without a
110 fill-pointer signals a type-error as required. (thanks to
112 * bug fix: disassemly of funcallable instances works.
113 * bug fix: single stepping on PPC.
114 * bug fix: fix thread-safety problems in the type system (generally
115 manifesting as nonsensical errors like "STRING is a bad type specifier
116 for sequences" or "The value 1 is not of type FIXNUM")
117 * bug fix: version components are handled correctly in TRANSLATE-PATHNAME
118 (reported by Josip Gracin)
119 * bug fix: an error is signaled for attempts to displace arrays with
120 incompatible element types (thanks to Mario Mommer)
121 * bug fix: more correct handling of wide characters in debug info
122 (bug reported by Attila Lendvai and fixed by Juho Snellman)
123 * optimization: method calls with &OPTIONAL or &KEY arguments are faster
124 and don't cause extra consing
125 * optimization: MAP and MAP-INTO are significantly faster on vectors
126 whose elements types have been declared.
127 * Improvements to SB-SPROF:
128 ** Support for allocation profiling
129 ** Reduced profiling overhead, especially for long profiling runs
130 * Improvements to the Windows port:
131 ** floating point exceptions are now reported correctly.
132 ** stack exhaustion detection works partially.
133 ** more accurate GET-INTERNAL-REAL-TIME.
134 ** better RUN-PROGRAM behaviour in case of errors in the child
136 ** PROBE-FILE now simplifies pathnames correctly.
137 ** DIRECTORY now works correctly with :WILD-INFERIORS.
138 ** DECODE-UNIVERSAL-TIME works on times before 00:00:00 January
139 1st 1970 (although time timezone data will be unreliable).
140 ** Hyperbolic functions TANH, ASINH, ACOSH, and ATANH work.
142 changes in sbcl-0.9.18 (1.0.beta?) relative to sbcl-0.9.17:
143 * enhancement: SB-POSIX now supports cfsetispeed(3), cfsetospeed(3),
144 cfgetispeed(3), cfgetospeed(3), and related constants. (thanks to
146 * bug fix: two potential GC deadlocks affecting threaded builds.
147 * bug fix: (TYPEP #\A '(NOT (MEMBER #\" #\{ #\:))) now correctly
148 returns T (reported by Anton Kazennikov)
149 * bug fix: the STORE-VALUE restart of CHECK-TYPE works correctly
150 with non-variable places
151 * bug fix: remove a race condition in the setting of
152 funcallable-instance functions, this should make threaded CLOS
153 code more stable against memory faults.
154 * bug fix: corruption of specials when unbinding is interrupted by an
155 asynchronous unwind (reported by Hannu Koivisto)
156 * improvement: the debugger will now also display local variables that
157 are only used once, for code compiled with a DEBUG optimization quality
160 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
161 * feature: weak hash tables, see MAKE-HASH-TABLE documentation
162 * incompatible change: External-format support for FFI calls. The
163 SB-ALIEN:C-STRING no longer implies an ASCII
164 external-format. Instead the string is subject to external-format
165 conversion. Additionally return values of type SB-ALIEN:C-STRING
166 are by default of type (SIMPLE-ARRAY CHARACTER), not
167 SIMPLE-BASE-STRING. For an alien type that behaves like the old
168 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
169 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
170 * incompatible change: SB-EXT package no longer contains the
171 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
172 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
174 * incompatible change: the single-stepper is no longer available
175 on Alpha, Mips and Sparc platforms.
176 * minor incompatible change: the direct superclasses of
177 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
178 not (STANDARD-OBJECT FUNCTION). This makes the
179 class-precedence-lists of GENERIC-FUNCTION and
180 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
182 * minor incompatible change: the default stream external format on
183 non-unicode SBCL is no longer detected from the locale, but is always
185 * new feature: Add a version of evaluator that uses an interpreter instead
186 of the compiler. EVAL still uses the compiler by default, to switch it
187 to use the interpreter, set the value of the variable
188 SB-EXT:*EVALUATOR-MODE* to :INTERPRET.
189 * minor incompatible change: the single-stepper REPL has been merged
190 with the normal debugger (see the "Stepping" heading of the debugger help
191 for more details). The debugger command STEP will no longer switch
192 to the single-stepper REPL.
193 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
194 for a type now works.
195 * bug fix: loading of default sysinit file works. (thanks to Leonid
197 * bug fix: better detection of circularities in the file-compiler.
198 (reported by Marco Monteiro)
199 * bug fix: the CL pathname functions now work with files that have
200 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
201 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
202 systems with Unix98 pty semantics.
203 * bug fix: ASDF-INSTALL will now work with bsd tar.
204 * bug fix: ASDF-INSTALL uses GNU tar on Solaris (thanks to Josip
206 * bug fix: timers expiring in dead threads no longer cause a
207 type-error (reported by Paul "Nonny Mouse").
208 * bug fix: thanks to more lightweight single-stepper instrumentation,
209 code compiled with (DEBUG 3) will compile and execute significantly faster,
210 and will have more accurate type-inferencing than before
211 * bug fix: SLOT-VALUE optimizations are no longer done on method parameters
212 whose bindings are modified
213 * improvements to the win32 port (thanks to Yaroslav Kavenchuk):
214 ** bug fix: arguments to RUN-PROGRAM are escaped correctly
215 ** replace dummy implementations of CL:MACHINE-INSTANCE and
216 CL:SOFTWARE-VERSION with proper version
218 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
219 * feature: implemented the READER-METHOD-CLASS and
220 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
221 as specified by AMOP.
222 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
224 * optimization: faster LOGCOUNT implementation on x86 and x86-64
225 (thanks to Lutz Euler)
226 * optimization: hashing of general arrays and vectors has been
227 improved. (reported by Any Fingerhut)
228 * enhancement: SB-INTROSPECT is now able to find definitions of
229 profiled functions. (thanks to Troels Henriksen)
230 * enhancement: compiler-macro expansion applies now to FUNCALL forms
231 as well, allowing compiler-macros for SETF-functions to expand.
232 * enhancement: step-instrumentation no longer wraps known
233 single-value functions with multiple-value context, allowing
234 better type inference.
235 * fixed bug #337: use of MAKE-METHOD in method combination now works
236 even in the presence of user-defined method classes. (reported by
237 Bruno Haible and Pascal Costanza)
238 * fixed bug #339(c): if there are applicable methods not part of any
239 long-form method-combination group, call INVALID-METHOD-ERROR.
240 (reported by Bruno Haible)
241 * fixed bug #361: the :FUNCTION initarg in the protocol for
242 initialization of methods can now be used to override
243 internally-produced optimized functions. (reported by Bruno
245 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
246 system-provided lambda expression no longer cause warnings about
247 unbound #:|pv-table| symbols.
248 * bug fix: improved the handling of type declarations and the
249 detection of violations for keyword arguments with non-constant
251 * bug fix: potentially erronous calls to PATHNAME and
252 MERGE-PATHNAMES were being flushed in some cases.
253 (reported by Richard Kreuter)
254 * bug fix: compiled calls to TYPEP were mishandling obsolete
255 instances. (reported by James Bielman and Attila Lendvai)
256 * bug fix: format strings with ~> without matching ~< no longer
257 trigger an AVER, but signal an understandable error instead.
258 (reported by Antonio Martinez)
259 * bug fix: specifying an output-file without a pathname-type for
260 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
261 of input-file instead of "fasl". (reported by Robert Dodier)
262 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
263 (reported by James Y Knight).
264 * bug fix: compiler-macros-function did not consider the environment
265 argument for shadowing by local functions.
266 * bug fix: compiler-macros expansion was inhibited by local INLINE
268 * bug fix: inline expansions of known functions were subject to
269 step-instrumentation in high debug policies, leading to problems
271 * bug fix: compiler failed to differentiate between different CONS
273 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
274 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
275 * bug fix: #368: incorrect use of expressed vs. upgraded array
277 * bug fix: #306a: more precise unions of array types.
278 * thread-safety improvements:
279 ** CONDITION-WAIT could return early on Linux, if the thread was
280 interrupted and subsequently continued with SIGCONT.
281 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
283 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
284 * added support for the ucs-2 external format. (contributed by Ivan
286 * minor incompatible change: pretty printing of objects of type
287 (cons symbol) is, in the default pprint-dispatch-table, now
288 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
290 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
291 called later in a class's lifetime, possibly as late as when the
292 first instance of the class is created. Previously,
293 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
294 class became finalizeable.
295 * fixed bug: FILE-POSITION sometimes returned inconsistent results
296 for multibyte external-format streams. (thanks to "vbzoli")
297 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
298 with :ALLOCATION :CLASS inherited from superclasses of the
300 * fixed bug: anonymous classes can now be created using the :NAME
301 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
302 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
303 * fixed bug: core-files saved with :EXECUTABLE T can again be
304 executed when SBCL_HOME isn't set. (reported by James Knight)
305 * fixed bug: toplevel LOCALLY forms with declarations could
306 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
307 * fixed bug: printing from several different threads using different
308 values of *print-case* could cause invalid output, due to
309 some internal special variables of the printer not being bound
310 thread-locally (reported by Max Mikhanosha)
311 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
312 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
313 (reported by Pascal Costanza)
314 * fixed bug: FINALIZE-INHERITANCE is called from
315 REINITIALIZE-INSTANCE on classes when the class has previously
316 been finalized, as required by AMOP.
317 * minor code generation optimizations:
318 ** better register allocation in CLOS dispatching functions
319 ** overflow detection when coercing signed bytes to fixnums on x86-64
320 ** is now implemented with one IMUL instruction instead of three shifts
321 ** more efficient bit-vector access on x86 and x86-64
322 ** more efficient access to raw structure slots on x86 and x86-64
323 * fixed some bugs revealed by Paul Dietz' test suite:
324 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
327 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
328 * feature: thread support on Solaris/x86, and experimental thread support
330 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
332 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
333 default method for SLOT-UNBOUND.
334 * minor incompatible change: prevent the user from specializing the
335 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
336 somewhat counter to the intent of the protocol, I (CSR) think, and
337 additionally it just doesn't work in SBCL as currently
338 implemented, thanks to optimizations (that are always valid for
339 the other three SLOT-VALUEish functions, but not for the setter).
340 * minor incompatible change: the :SB-LDB feature is now enabled by
341 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
342 the low-level debugger.
343 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
344 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
345 down to an SB-SYS:FD-STREAM, allowing direct communication in
346 more cases. Temporary files and pipes are still used as a fallback
348 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
349 on the non-reentrant C functions localtime(3) and gmtime(3).
350 * bug fix: global optimization policy was not visible in LOCALLY and
352 * bug fix: class objects can be used as specializers in methods.
353 (reported by Pascal Costanza)
354 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
355 with a type error. (reported by blitz_ on #lisp)
356 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
357 failed with a type error. (reported by Pascal Bourguignon)
358 * bug fix: merging pathnames against defaults with :DIRECTORY
359 starting with '(:RELATIVE :BACK) should preserve the :BACK.
360 (reported by James Y Knight)
361 * bug fix: saving large (>2GB) cores on x86-64 now works
362 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
363 constant index and a (simple-array (signed-byte 32)) array
364 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
365 error (patch by Robert J. Macomber)
366 * feature: TIME now displays the amount of run-time spent in GC
367 * bug fix: The debugger now does a better job of respecting
368 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
369 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
370 workaround for bug 403.)
371 * bug fix: floating point exception handling now works on FreeBSD
372 (thanks to NIIMI Satoshi)
373 * fixed some bugs revealed by Paul Dietz' test suite:
374 ** MISC.641: LET-conversion were not supposed to work in late
376 * improvements to the Win32/x86 port:
377 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
378 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
380 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
381 * new feature: source path information is generated for macro-expansion
382 errors for use in IDE's like Slime (thanks to Helmut Eller)
383 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
384 * bug fix: compiler does not loop forever on an invalid type in
386 * improvement: compilation of most CLOS applications is significantly
388 * optimization: added a limited bytecode compiler for simple toplevel
389 forms, speeding up compilation and FASL loading
390 * bug fix: the statistical profiler now properly distinguishes anonymous
393 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
394 * minor incompatible change: in sbcl-0.9.11 (but not earlier
395 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
396 into subclasses of the finalized class. Now user calls to
397 FINALIZE-INHERITANCE finalize just the one class, and calls by the
398 system return before any subclasses are finalized.
399 * minor incompatible change: The reader no longer ignores errors
400 regarding non-existent packages in #+ and #- feature tests.
401 * new feature: command line options --no-sysinit, --no-userinit to
402 inhibit loading the corresponding init files
403 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
404 for sending data through UDP sockets (thanks to François-René Rideau)
405 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
406 error is signalled instead (thanks to François-René Rideau)
407 * minor incompatible change: Error signalling behaviour of lexical
408 operations violating package locks has changed slightly. Refer to
409 documentation on package locks for details.
410 * enhancement: EVAL can process IF-expressions without resorting to the
412 * bug fix: Some operations involving SIN, COS, or TAN weren't being
413 constant-folded properly due to missing out-of-line functions.
414 (reported by Mika Pihlajamäki)
415 * bug fix: LISTEN sometimes returned T even in cases where no data was
416 immediately available from the stream
417 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
418 were reversed. (reported by Levente Mészáros)
419 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
420 when given a vector displaced to an adjustable array. (reported
422 * bug fix: derivation of float boundaries from numbers outside the
423 appropriate float range (reported by John Wiseman)
424 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
426 * bug fix: precomputing cache entries for generic functions with
427 some subclasses of specializers as yet invalid does not attempt to
428 fill a cache line with a negative offset. (reported by Levente
430 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
431 allow condition class objects as arguments to DOCUMENTATION and
432 (SETF DOCUMENTATION); only find and set documentation for
433 structure names for the STRUCTURE doc-type. (suggested by Gary
435 * improvements to the Win32/x86 port:
436 ** (user-homedir-pathname) and default initialization file
437 locations now know about the user's "Documents and Settings"
438 directory (thanks to Yaroslav Kavenchuk)
439 ** run-program is implemented (thanks to Mike Thomas)
440 ** sockets support (thanks to Timothy Ritchey)
441 ** better backtrace support (thanks to Alastair Bridgewater)
442 ** sb-grovel supported
443 ** asdf-install and sb-posix work somewhat
444 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
445 * improvements to the Solaris/x86 port:
446 ** works on Solaris 11/Solaris Express
447 ** floating-point exception handling support
448 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
449 * fixed some bugs revealed by Paul Dietz' test suite:
450 ** REMOVE-METHOD returns its generic function argument even when
451 the method is not one of the generic functions' methods.
452 ** objects declared as MEMBER types can be admissible arguments to
454 ** printing characters should simply be printed by the FORMAT ~:C
456 ** compiler failure when compiling functions with hairy constant
457 defaults for optional parameters.
458 ** compiler produces wrong code when MAYBE-INLINE-expanding a
459 function, which is already optimized.
461 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
462 * new platform: experimental support for SBCL x86/Darwin, including
463 MacOS X 10.4.5 on Intel.
464 * new feature: Unicode character names are now known to the system
465 (through CHAR-NAME and NAME-CHAR).
466 * new feature: the filesystem location of SBCL's core file is
467 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
468 * minor incompatible change: the contrib modules SB-POSIX and
469 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
470 this change is to make it easier to distribute
471 location-independent binaries.
472 * bug fix: as implied by AMOP, standardized classes no longer have
473 slots named by external symbols of public packages. (reported by
475 * optimization: calling functions via constant symbols -- as in
476 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
477 via the function object as in (FUNCALL #'FOO).
478 * optimization: on x86 and x86-64, the calling convention for the
479 case of dealing with an unknown number of values has been altered
480 to be friendlier to the prediction heuristics implemented,
481 particularly with reference to CALL and RET pairing. (thanks to
482 Alastair Bridgewater)
483 * optimization: on x86, the code for access to thread-local symbol
484 values and binding/unbinding of thread-local symbols is smaller.
485 (thanks to Alastair Bridgewater)
486 * enhancement: CONSTANTP is now able to determine constantness of
487 more complex forms, including calls to constant-foldable standardized
488 functions and some special forms beyond QUOTE.
489 * fixed bug: occasional GC crashes on Solaris/x86
490 * optimization: x86-64 supports stack allocation of results of simple
491 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
492 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
493 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
494 (thanks to James Knight)
495 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
496 by the standard (thanks to Stephen Compall)
498 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
499 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
500 be used for bundling the runtime and the core file into one
501 executable binary. This feature is not currently supported on all SBCL
502 platforms. (thanks to James Bielman and NIIMI Satoshi)
503 * new feature: a generational or ephemeral garbage collector is now
504 the default on the PowerPC platform (both Linux and Darwin). The
505 old Cheney (stop and copy) collector is a build-time option.
506 (thanks to Cyrus Harmon, after Raymond Toy)
507 * minor incompatible change: the method by which SBCL finds its
508 contributed modules has changed; it no longer relies on symbolic
509 links from an $SBCL_HOME/systems directory, but searches directly
510 in the subdirectories of $SBCL_HOME.
511 * enhancement: the dynamic heap size on the Linux/PPC platform is
512 markedly larger, even using the older Cheney garbage collector.
513 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
514 certain complicated string types. (reported by Gary King)
515 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
516 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
517 * fixed bug: DOCUMENTATION on structures no longer signals an
518 error if no structure type of the right name exists. (reported by
520 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
521 applicable non-standard primary or :AROUND method on
522 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
523 no longer cause unbound variable errors. (reported by Kevin Reid)
524 * fixed bug: metacircle resolution in cases where methods have slots
525 added before the slots from STANDARD-METHOD. (reported by Jean
527 * fixed bug: the Power PC assembly code for calling into lisp
528 sometimes computed its return address wrongly (depending on the
529 15th bit of the address, fixed at link time). This used to
530 manifest itself in a segmentation violation while building PCL.
531 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
532 many others over the years)
533 * fixed bug: ROOM no longer reports silly numbers for stack usage if
534 the stack pointer (treated as a signed integer) is negative.
535 (thanks to Peter van Eynde)
537 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
538 * new platform: experimental support for the Windows operating
539 system has been added. (thanks to Alastair Bridgewater)
540 * fixed several bugs in and robustified the PPC FFI (including
541 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
542 * bug fix: allow non-simple string symbol names (reported by Paul
544 * bug fix: interrupt handling on NetBSD (thanks to Richard M
546 * bug fix: saving a core corrupted callbacks on x86/x86-64
547 * bug fix: closed a loophole in metacircularity detection and
548 grounding in the PCL implementation of CLOS.
549 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
550 (reported by tomppa on #lisp)
551 * bug fix: clear the direction flag on Lisp -> C transitions, as
552 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
553 (reported by Andrej Grozin and Hendrik Maryns)
554 * optimization: major improvements to GC efficiency on GENCGC platforms
555 * optimization: faster implementation of EQUAL
556 * optimization: emit more efficient opcodes for some common
557 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
558 * optimization: several other minor code-generation improvements on x86-64
559 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
561 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
562 * minor incompatible change: (SETF CLASS-NAME) and (SETF
563 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
564 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
565 by ANSI as a generic function, and for consistency (SETF
566 GENERIC-FUNCTION-NAME) is treated likewise.
567 * fixed bug #233.b: make constraint propagation notice when a variable
568 value is changed after it is referenced but before it is used
569 * fixed bug #296: no more arbitrary behaviour from filenames with
570 odd characters as --load arguments. (reported by Adam Warner)
571 * fixed bug #390: :CHARACTER-SET pathname components now work as
572 expected. (reported by Tim Daly Jr)
573 * fixed bug #391: complicated :TYPE intersections in slot
574 definitions no longer cause an error in PCL internals.
575 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
576 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
577 returning the number of octets which would be written to the
578 file-stream. (thanks to Robert J. Macomber)
579 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
580 arguments to the internal stream functions. (thanks to David
582 * bug fix: stack exhaustion is now handled on the x86/SunOS
583 platform. (thanks to Daisuke Homma)
584 * bug fix: more accurate ROOM results on GENCGC platforms
585 * optimization: improved type inference for arithmetic-for
586 index variables in LOOP
587 * optimization: faster floating-point SQRT on x86-64
588 * fixed some bugs revealed by Paul Dietz' test suite:
589 ** DOCUMENTATION returns NIL instead of "" for method combinations
590 that don't have a docstring
592 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
593 * minor incompatible change: (SETF CLASS-NAME) and (SETF
594 GENERIC-FUNCTION-NAME) are no longer generic functions, and
595 therefore are not customizeable by user code (as seems to be at
596 least permitted and maybe required by AMOP). As a consolation,
597 however, the SBCL implementation of these functions now calls
598 REINITIALIZE-INSTANCE as specified by AMOP.
599 * bug fix: slot-definition documentation is propagated as per ANSI
600 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
601 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
602 argument, as do eslotd initialization methods. (from Pascal
603 Costanza's "Closer" project)
604 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
605 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
607 * bug fix: it is now possible to have more than one subclass of
608 STANDARD-GENERIC-FUNCTION without causing stack overflow.
609 (reported by Bruno Haible, Pascal Costanza and others)
610 * bug fix: the dependent update protocol now works for generic
611 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
613 * bug fix: condition-class instances corresponding to
614 DEFINE-CONDITION forms are now created eagerly. (reported by
615 Kalle Olavi Niemitalo on comp.lang.lisp)
616 * bug fix: floating point printing is more accurate in some
617 circumstances. (thanks to Simon Alexander)
618 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
619 merged with *DEFAULT-PATHNAME-DEFAULTS*.
620 * bug fix: callbacks on OS X now preserve stack-alignment.
621 (thanks to Cyrus Harmon)
622 * enhancement: the x86-64 disassembler is much better at
623 disassembling SSE instructions. (thanks to Lutz Euler)
624 * enhancement: CHECK-TYPE on a variable now allows type-inference
625 (thanks to Vincent Arkesteijn)
626 * optimization: improved performance of EUC-JP external format.
627 (thanks to NIIMI Satoshi)
628 * optimization: performance improvements to IO on file streams of
629 :ELEMENT-TYPE CHARACTER
630 * optimization: much faster memory allocation on x86-64
631 * optimization: faster garbage collections (latency and throughput)
633 * optimization: faster bignum subtraction and division on x86 and x86-64
634 (thanks to Lutz Euler)
635 * optimization: more accurate type-inference for loops that use a
636 floating point index variable or a negative step.
638 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
639 * bug fix: add a workaround to SBCL looping infinitely at startup on
640 Linux kernels with apparently buggy implementations of personality().
641 (thanks to Svein Ove Aas)
642 * bug fix: Unicode symbols are correctly printed in LDB backtraces
643 (thanks to David Lichteblau)
644 * bug fix: local bindings shadow global symbol macros for
645 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
646 * optimization: non-open coded uses of numeric comparison operators
647 (e.g. >) no longer cons when called with more than one parameter
648 on platforms supporting dynamic-extent allocation.
649 * enhancement: saving cores with foreign code loaded is now supported
650 on MIPS/Linux in addition to the previously supported platforms.
651 * bug fix: FIND-RESTART now tests for activity, not applicability when given
652 a restart object as identifier. (reported by Helmut Eller for CMUCL)
653 * bug fix: division by zero in sb-sprof when no samples were collected
654 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
655 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
656 (thanks to Alexander Kjeldaas)
657 * new feature: ignoring the return values of destructive functions
658 like NREVERSE signals a compile-time style-warning.
659 (thanks to Kevin Reid)
660 * new feature: the alignment of alien structure fields can be
661 explicitly specified. (thanks to Cyrus Harmon)
662 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
663 and dump core on SIGQUIT
665 ** incompatible change: threads do not inherit values of specials
666 from their parents (see manual)
667 ** bug fix: threads stacks belonging to dead threads are freed by the
668 next exiting thread, no need to gc to collect thread stacks anymore
669 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
670 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
671 ** bug fix: dynamic variable and thread start related gc lossage
672 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
674 * fixed some bugs revealed by Paul Dietz' test suite:
675 ** SUBTYPEP is slightly more accurate on heinously complicated
676 CONS types where some of the members have uncertain (in the
677 NIL, NIL sense) type relationships to each other.
679 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
680 no longer purified unless :PURIFY T is explicitly specified.
681 ** Non-purified cores are significantly smaller than before
683 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
684 * new feature: timers based on Zach Beane's excellent timer package
685 * added support for the following external formats: koi8-u,
686 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
687 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
688 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
689 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
690 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
691 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
692 * incompatible change: a threaded SBCL will no longer revert to
693 non-threaded mode on non-NPTL systems, but refuse to start entirely.
694 * bug fix: interrupts are disabled until startup is complete; no
695 more sigsegvs when receiving a signal to soon
696 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
698 * bug fix: add a workaround for the memory randomization features in
699 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
700 workaround will only be in effect on systems with the proc filesystem
702 * bug fix: printing objects of type HASH-TABLE signals a
703 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
705 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
706 have been running for over 50 days. (reported by Gilbert Baumann)
707 * bug fix: the logic for getting names of functions gets less
708 confused when confronded with alternate-metaclass
709 funcallable-instances. (reported by Cyrus Harmon)
710 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
711 consistent, even on internal alternate-metaclass objects.
712 * bug fix: flush closure information collected by physical
713 environment analysis prepass before the main pass. (bug reported
715 * bug fix: compiler pack phase does not modify a hashtable, which is
716 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
717 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
718 may be expanded. (reported by Alexander <malishev> on c.l.l)
719 * bug fix: no more style-warnings when compiling code using LAST
720 under high-SPEED low-SPACE settings. (reported by David Wallin
722 * fixed some incompatibilities between SBCL's MOP and the MOP
724 ** the METAOBJECT class is now implemented;
725 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
726 STANDARD-OBJECT, as required;
727 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
728 now compatible (as required by VALIDATE-SUPERCLASS); there
729 remains a consistency requirement over the presence or absence
730 of the FUNCTION class in the superclass list of finalized
731 classes; see the manual for more details;
732 ** the :AROUND method for COMPUTE-SLOTS on
733 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
734 requested slot ordering.
736 ** bug fix: parent thread now can be gc'ed even with a live
738 ** bug fix: binding a special with PROGV to no value is not
739 the same as that symbol not having been bound (thanks to
741 ** bug fix: binding specials is thread safe (thanks to
743 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
744 and SB-SPROF do not die with 'no handler for signal XX in
745 interrupt_handle_now(..)' anymore
746 ** bug fix: WITH-TIMEOUT works with multiple running threads
747 * fixed some bugs revealed by Paul Dietz' test suite:
748 ** ENSURE-GENERIC-FUNCTION should take a method class object for
749 the :method-class keyword argument.
751 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
752 * new port: the Solaris operating system on x86 processors is now
753 mostly supported, though some rough edges in the environment
754 remain. (thanks to Daisuke Homma)
755 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
756 space for its heap. (thanks to Thiemo Seufer)
757 * optimization: SBCL on MIPS platforms now supports dynamic-extent
758 closures. (thanks to Thiemo Seufer)
759 * minor incompatible change: eof selects abort in the debugger.
760 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
761 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
763 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
764 deprecated, and will go away in a future revision of SBCL.
765 * minor incompatible change: GC-ON and GC-OFF are no longer
766 implemented with a counter, it does not matter how many times gc
767 is switched on or off
768 * bug fix: discriminating functions for generic function classes
769 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
770 make invalid assumptions about method precedence order. (reported
772 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
773 platforms. (thanks to Thiemo Seufer)
774 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
775 * bug fix: correct pathname printing with printer escaping is on.
776 (thanks to Kevin Reid)
777 * bug fix: complex VOP definitions in "user-space" no longer trigger
778 package locks. (reported by Zach Beane)
779 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
780 for funcallable-instance functions, and is no different from
782 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
783 characters in the current readtable. (reported by Nicholas Neuss)
784 * bug fix: SUBTYPEP on various CONS types returns more a more
785 accurate acknowledgment of its certainty.
786 * optimizations: REMOVE-DUPLICATES now runs in linear time on
787 lists in some cases. This partially fixes bug 384.
788 * flush all standard streams before prompting in the REPL and the
790 * bug fix: signal handling and triggering gc do not conflict
791 directly anymore, in particular a high frequency sb-sprof does
792 not prevent gc from running
793 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
794 approximation for timezone and DST information between the
795 universal time epoch and the smallest negative 32-bit time_t.
796 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
797 year 1899 so long as the encoded time is non-negative.
798 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
799 the correct answer for non-constant rotate amounts when performing
800 an inline 32-bit rotation.
802 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
803 there is only one thread in the session
804 ** bug fix: memory leak for streams created in one thread and
805 written to in another
806 ** bug fix: lockup when compiled with gcc4
807 ** bug fix: race that allows the gc to be triggered when gc is
809 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
810 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
812 ** bug fix: finished threads are gc'ed properly
813 * fixed some bugs revealed by Paul Dietz' test suite:
814 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
815 has arguments with a different set of applicable methods from
816 the orignal arguments.
817 ** The type error thrown by MAP now has a correct expected-type
819 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
820 name a compiled function.
821 ** Three MISC tests where a large bignum was improperly coerced to
822 a float that couldn't represent that bignum during type
823 derivation were fixed.
824 ** SUBTYPEP can now handle types involving the negation of a
825 list-form FUNCTION type.
826 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
827 as (COMPLEX (AND RATIO (NOT FIXNUM))).
828 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
830 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
831 * New feature: Experimental support for bivalent streams: streams
832 opened with :element-type :default now allow character and binary
833 (unsigned-byte 8) I/O
834 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
835 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
837 * bug fix: no more highly sporadic "couldn't check whether ~S is
838 readable" when reading a stream and an interrupt hits in the middle
839 of a select system call
840 * compiler better recognizes complex arrays (reported by Eduardo
842 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
843 for variables on SBCL built with linkage-tables. (reported by Luis
845 * various error reporting improvements.
846 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
847 (Thanks to Thiemo Seufer)
848 * bug fix: nested reader invokations work correctly
849 * bug fix: it is possible to have simultaneous references to foreign
850 code and foreign data with the same name.
852 ** added x86-64 support
853 ** incompatible change: the threading api now works with thread
854 objects instead of thread ids
855 ** bug fix: threads are protected from signals and interruption when
856 starting up or going down
857 ** bug fix: a race where an exiting thread could lose its stack to gc
858 ** bug fix: don't halt on infinite error in threads if possible
859 ** fixed numerous gc deadlocks introduced in the pthread merge
860 ** bug fix: fixed thread safety issues in read and print
861 ** bug fix: debugger doesn't hang on session lock if interrupted at
862 an inappropriate moment
863 ** bug fix: run-program is now thread safe(r)
864 ** bug fix: inner with-recursive-lock no longer releases the mutex
865 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
866 * fixed some bugs revealed by Paul Dietz' test suite:
867 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
868 EXPECTED-TYPE slots filled.
869 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
871 ** MAKE-CONDITION accepts classes as type-designators.
872 ** COMPILE may never return NIL.
873 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
874 range before calling Unix time functions
876 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
877 * numerous signal handling fixes to increase stability
878 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
879 * minor incompatible change: we now correctly canonize default
880 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
881 as per the MOP, rather than the historical (INITARG INITFUNCTION
882 INITFORM). (reported by Bruno Haible)
883 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
884 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
885 for more information.
886 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
887 pathname is a directory pathname.
888 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
889 * dynamic space size on PPC has been increased to 768Mb. (thanks to
891 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
892 :METACLASS argument in addition to a class name. (reported by
893 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
894 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
895 (reported by Bruno Haible; patch from Gerd Moellmann)
896 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
898 * bug fix: more cleanups to the floating point exception handling on
899 x86-64 (thanks to James Knight)
900 * bug fix: the compiler does not try to constant fold calls of
901 COERCE and BIT-* functions when they return freshly constructed
902 objects. (reported by Nikodemus Siivola and Paul Dietz)
903 * optimization: improved the MIPS versions of generic (in the
904 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
905 * optimization: direct conversion of (unsigned-byte 32) to floats on
906 the PowerPC platform.
907 * optimization: structure instances with raw slots now use less
908 memory, and probably show better memory locality. (thanks to
910 * optimization: DYNAMIC-EXTENT declarations for lists and closures
911 are treated as requests for stack allocation on the x86-64,
912 Alpha-32, PPC and SPARC platforms.
913 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
914 the wrong file descriptor; implementation of SOCKET-OPEN-P.
915 (thanks to Tony Martinez)
917 ** gcing a dead thread can no longer lead to lockups
918 ** threads block signals until they are set up properly
919 ** errno is no longer shared by threads
920 ** interrupt-thread restores the eflags register on x86
921 ** fixed some lockups due to gc/thread interaction
922 * fixed some bugs revealed by Paul Dietz' test suite:
923 ** invalid dotted lists no longer raise a read error when
925 ** signal an error if a symbol that names a declaration is used
926 as the name of a type, or vice versa
927 ** allow using the (declare (typespec var*)) abbreviation for
928 (declare (type typespec var*)) with all type specifiers
929 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
930 ** always return NIL from PPRINT-POP when OBJECT is NIL
931 ** don't signal errors when pretty-printing malformed LABELS,
932 FLET or MACROLET forms
933 ** declarations in a DOLIST body are also in scope for the
935 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
937 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
940 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
941 * fixed cross-compiler leakages that prevented building a 32-bit
942 target with a 64-bit host compiler.
943 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
944 opened with :IF-EXISTS :OVERWRITE.
945 * fixed bug 281: error for an invalid qualifier in a short-form method
946 combination method is not signalled until the faulty method is called.
947 * bug fix: iteration variable type inferrer failed to deal with open
948 intervals. (reported by Alan Shields)
949 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
950 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
951 or more runtime options were provided to the sbcl binary.
952 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
953 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
954 * optimization: there's now a fast-path for fixnum arguments in the
955 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
956 * optimization: the code generated on x86-64 is more compact thanks
957 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
958 * bug fix: floating-point exception handling now partly works on
959 x86-64. (Thanks to James Knight)
960 * improvement to the MIPS backend from Thiemo Seufer: C-style
961 64-bit long long arguments and return values to alien functions
963 * fixed some bugs revealed by Paul Dietz' test suite:
964 ** the type-error signalled from WARN has a filled-in DATUM slot.
965 ** the type-error required when a stream is not associated with
966 a file has the stream as its datum.
967 ** type-errors on single-floats on x86-64 no longer have
968 :INVALID-OBJECT as the datum
969 ** the type-errors signalled for invalid function names now have
970 a correct expected type
971 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
972 for structures defined with a :TYPE. Documentation strings for
973 typed structures are no longer immediately discarded
974 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
975 broadcast streams and synonym streams. FILE-LENGTH now also works
976 on broadcast streams.
978 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
979 * incompatible change: the --noprogrammer option, deprecated since
980 version 0.7.5, has been removed. Please use the equivalent
981 --disable-debugger option instead.
982 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
984 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
985 inoperational for a while now) has been completely removed.
986 * null lexical environments are now printed as #<NULL-LEXENV>,
987 significantly reducing the amount of clutter in typical backtraces.
988 * documentation on weak pointers, finalization, and after GC hooks
989 has been added to the manual.
990 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
991 as other specialized array types, is much faster. SUBSEQ and
992 COPY-SEQ on such arrays have also been sped up.
993 * optimization: EQL is now more efficient when at least other argument
994 is known to be of type (OR FIXNUM (NOT NUMBER)).
995 * fixed bug: the runtime is now less vulnerable to changes in the
996 size of the SBCL object on OS X, and virtual memory is reserved for
997 all spaces that need to be at a fixed address.
998 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
999 * fixed bug: finalizers and after GC hooks that cause consing are now
1001 * fixed bug: compiler error messages and summaries are now printed to
1002 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
1003 * fixed inference of the upper bound of an iteration variable.
1004 (reported by Rajat Datta).
1005 * fixed bug 211e: calling local functions with duplicated constant
1006 keyword argument no longer causes a bogus style warning about an
1008 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
1009 ftype declaration to be disregarded. (reported by Dave Roberts)
1010 * fixed bug 373: caused by erronous compilation of references to alien
1011 variables in the runtime on ppc/darwin.
1012 * fixed bug 376: CONJUGATE type deriver.
1013 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
1014 (reported by Baughn on #lisp)
1015 * a cross-compiler bug on non-x86 platforms has been identified and
1016 fixed. (thanks to Bruno Haible)
1017 * improvements to the MIPS runtime code for increased stability.
1018 (thanks to Thiemo Seufer)
1019 * increased the maximimum compact environment size to allow
1020 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
1021 * partial workaround for bug 135: don't name the function we're
1022 calling for hairy cases of EVAL, so as not to accumulate one environment
1023 entry per call to EVAL. (reported by Kevin Reid)
1024 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
1025 * optimization: type testing for non-vector arrays should be faster.
1026 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
1028 * bug fix: setting 31st element of a bit vector to zero did not work
1030 * bug fix: redefining a class definition which failed due to a
1031 previous accessor / function clash now works (but see BUGS entry
1032 #380 for more problems in this area). (thanks to Zach Beane)
1033 * the long form of DEFINE-METHOD-COMBINATION disables method group
1034 checking when given a single method group with pattern *. (thanks
1036 * on x86 compiler supports stack allocation of results of simple
1037 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
1038 * support for building and running on Mac OS X 10.4 "Tiger" has been added
1039 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
1040 * fixed some bugs related to Unicode integration:
1041 ** the restarts for recovering from input and output encoding
1042 errors only appear when there is in fact such an error to
1044 * fixed some bugs revealed by Paul Dietz' test suite:
1045 ** MISC.549 and similar: late transformation of unsafe type
1046 assertions into derived types caused unexpected code
1048 ** SCALE-FLOAT type deriver is less wrong.
1049 ** type derivers for EXP, LOG and similar functions compute result
1050 types for complex arguments better.
1051 ** (MISC.563) CONJUGATE type deriver works for very restricted
1053 ** out-of-line type testers for character strings are available.
1054 ** EQUAL compiler transform understands specialness of objects
1056 ** accessing double-floats stored on the stack now works on x86-64.
1057 ** debugger internals could sometimes create invalid lispobjs,
1058 resulting in GC crashes.
1059 ** MISC.548: type check weakening can convert required type into
1061 ** initialization forms for bindings are not in scope of free special
1064 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
1065 * incompatible change: thread support for non-NPTL systems has
1066 been removed - locking is buggy and unreliable. A threaded
1067 SBCL build will now warn at startup and refuse to create
1068 new threads, unless futex support is detected in the kernel
1069 * incompatible change: the top level REPL now has only an ABORT
1070 restart associated with it, not TOPLEVEL and ABORT as it used to.
1071 TOP and TOPLEVEL are now available as debugger commands for
1072 returning to the top level.
1073 * incompatible change: forms evaluated in the REPL now use the
1074 global optimization policy.
1075 * incompatible change: user- and system-initialization files are
1076 no longer processed with LOAD, but by READ and EVAL; hence the
1077 global optimization policy, startup package, readtable, etc,
1079 * The .fasl file format number has been incremented because of
1080 various incompatible changes.
1081 * internal entry point details and argument counts no longer appear
1082 in backtraces unless explicitly requested by setting
1083 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
1084 * built-in and standard functions no longer have names like "top
1085 level local call to FOO".
1086 * fixed bug 32: functions defined in non-null lexical environments
1087 now have more legible printed representation
1088 * fixed bug 33: functions defined in non-null lexical environemnts
1089 are now more amenable to inspection by INSPECT.
1090 * workaround for bug 354: XEPs no longer appear in backtraces unless
1091 explicitly requested.
1092 * fixed bug: receiving the signal which results from attempting to
1093 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
1094 SIGBUS on other *BSDs) on architectures where the C stack is also
1095 the Lisp stack (x86 and x86-64) from foreign code no longer leads
1096 to debugger confusion or wild instability. (reported by Cheuksan
1098 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
1099 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
1100 (reported by Lutz Euler)
1101 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
1102 compiled code. (reported by Paul Dietz)
1103 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
1104 specialized parameter to a method to something that is not TYPEP
1105 the specializer is now possible.
1106 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
1107 face of package deletion.
1108 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
1109 pathname. (reported by David Tolpin on c.l.l)
1110 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
1111 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
1112 STRING-STREAM and FILE-STREAM, respectively.
1113 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
1114 than previously; in addition, its test suite is now run on build.
1115 (thanks to Luke Gorrie)
1116 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
1117 * optimization: added a immediate representation for single-floats
1119 * fixed some bugs revealed by Paul Dietz' test suite:
1120 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
1121 ** attempting to create a package with a colliding nickname causes
1122 correctable errors to be signalled.
1123 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
1124 ** several x86-64 backend bugs related to sign-extension of immediate
1127 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
1128 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
1129 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
1131 * fixed loading of multiply forward-referenced layouts.
1132 (thanks to Cheuksan Wang)
1133 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
1134 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
1135 (see COMPILE-FILE documentation for details of :PRINT option)
1136 * fixed bugs 19 and 317: fixed-format floating point printing is
1137 more accurate. This also fixes a bug reported by Adam Warner
1138 related to the ~@F format directive.
1139 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
1141 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
1142 dispatch table if the from-char is a dispatch macro character.
1143 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
1144 by the wordsize no longer ignore the last word. (reported by Lutz
1146 * fixed bug in type checking of dynamic-extent variables. (reported
1148 * optimization: sequence traversal functions use their freedom to
1149 coerce function designators to functions.
1150 * optimization: code with many calls to CLOS methods specialized on
1151 CLOS classes has had redundant type checks removed.
1152 * optimization: added declarations to speed up operations that access
1153 the internal character database (for example STRING-UPCASE)
1154 * optimization: comparison operations between floats and sufficiently small
1155 fixnums no longer create extra rationals
1156 * fixed some bugs related to Unicode integration:
1157 ** portions of multibyte characters at the end of buffers for
1158 character-based file input are correctly transferred to the
1159 start of the buffer at the next read.
1160 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
1161 passing it through to OPEN.
1162 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
1163 argument given to internal calls to OPEN.
1164 ** the built-in comment readers (introduced by character sequences
1165 ";" and "#|") are more forgiving to encoding errors; they will
1166 STYLE-WARN and then attempt to resync the stream at a character
1167 boundary. (thanks to Teemu Kalvas)
1168 * fixed some bugs in the x86-64 port:
1169 ** Negative short int return values from c-calls are sign-extended
1171 ** The stack is aligned to 16-bytes for c-calls, as required by
1172 the ABI. (thanks to Cheuksan Wang)
1173 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
1174 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
1175 for enums. (thanks to Vincent Arkesteijn)
1176 ** Multiple small optimizations and bugfixes for floating point
1178 * fixed some bugs revealed by Paul Dietz' test suite:
1179 ** Space, Tab, Linefeed, Return and Page have the invalid
1180 secondary constituent character trait.
1181 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
1183 ** WITH-INPUT-FROM-STRING should only update the index place on
1185 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1186 works more reliably.
1187 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1188 with one-dimensional arrays with fill pointers.
1189 ** TYPE-OF failed on a complex with an integer realpart and a
1191 ** compiler failure during type inference for the code of form
1192 (IF (EQL X (THE ...)) ...) (MISC.535).
1194 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1195 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1196 platform. The port supports 61-bit fixnums, large memory spaces
1197 and reloading shared object files.
1198 * enhancement: saving cores with foreign code loaded is now
1199 supported on ppc/Darwin in addition to the previously supported
1201 * enhancement: the statistical profiler now walks deeper into the
1202 call stack for more meaningful call-graphs and accrued time
1203 reports (x86/x86-64 only). It also now reports time spent in
1205 * enhancement: it is now possible to trace most individual methods
1206 of a generic function in addition to tracing the generic function
1208 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1210 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1211 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1212 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1213 produces an error. (thanks to Vincent Arkesteijn)
1214 * bug fix: NAMESTRING on pathnames with :WILD components in their
1215 directories works correctly. (thanks to Artem V. Andreev)
1216 * fixed bug 125: compiler preserves identity of closures. (reported
1218 * bug fixed: functions with &REST arguments sometimes failed with
1219 "Undefined function" when compiled with (DEBUG 3). (reported by
1221 * bug fix: overflow during compiling of setting element of a bit
1222 vector with constant index and value. (reported by Timmy Douglas)
1223 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1224 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1225 * on x86 compiler partially supports stack allocation of dynamic-extent
1227 * GO and RETURN-FROM do not check the extent of their exit points
1228 when compiled with SAFETY 0.
1229 * fixed some bugs related to Unicode integration:
1230 ** encoding and decoding errors are now much more robustly
1231 handled; it should now be possible to recover even from invalid
1232 input or output to the terminal. (thanks to Teemu Kalvas)
1233 ** provided a first cut at implementing STRING-TO-OCTETS and
1234 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1235 ** altered the SB-MD5 contributed module slightly, changing the
1236 interface just enough for it to be supportable for builds where
1237 lisp characters are not eight bits.
1238 * fixed some bugs revealed by Paul Dietz' test suite:
1239 ** the FORMATTER-generated functions for ~V[ conditionals require
1240 the correct number of arguments.
1241 ** READ-FROM-STRING returns the mandated second value when applied
1242 to displaced strings.
1243 ** the #\Rubout and #\Backspace characters are treated as invalid
1244 constituent characters by the tokenizer.
1246 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1247 * new feature: reloading changed shared object files with
1248 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1249 * new feature: references to foreign variables and functions
1250 can now be compiled and loaded before the corresponding shared
1251 object file is loaded, as long as the foreign definitions are
1252 available at runtime.
1253 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1254 supported. (thanks to Dan Debertin)
1255 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1256 just keywords. (thanks to Vincent Arkesteijn)
1257 * fixed bug #331: structure-class instances corresponding to
1258 DEFSTRUCT forms are now created eagerly.
1259 * fixed bug #345: backtraces from calls to undefined functions work
1260 on x86 as well. Related bug #61 is now also partially fixed on x86
1261 (backtraces from throws to unknown catch tags.)
1262 * bug fix: lambda-list parsing is now stricter vrt. order and number
1263 of lambda-list keywords.
1264 * bug fix: as specified by AMOP, an error is signalled if a
1265 class-option appears multiple times in a DEFCLASS form. (reported
1267 * bug fix: RANDOM can be compiled when the compiler derives the type
1268 of its numeric argument as a disjoint set of small integers.
1269 (reported by Paul Dietz)
1270 * bug fix: starting a core saved with shared objects loaded when
1271 those objects are not available no longer causes threaded SBCL to
1272 hang. (reported by Sean Ross)
1273 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1275 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1276 argument, which must always be NIL. (reported by Kalle Niemitalo)
1277 * bug fix: printing 1.0d+23 no longer results in an error.
1278 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1279 * bug fix: structure slot setters preserve evaluation order. (thanks
1281 * fixed some bugs related to Unicode integration:
1282 ** RUN-PROGRAM can allow its child to take input from a Lisp
1283 stream. (reported by Stefan Scholl)
1284 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1285 locales. (reported by Ken Causey)
1286 * fixed some bugs revealed by Paul Dietz' test suite:
1287 ** INCF, DECF and REMF evaluate their place form as specified in
1289 ** FORMATTER expands ~{ ~} iteration directives with V or #
1290 parameters correctly.
1291 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1292 consequent uses no arguments correctly.
1293 ** the system has a partial understanding of the (COMPLEX RATIO)
1295 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1298 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1299 * new feature: a build-time option (controlled by the :SB-UNICODE
1300 keyword feature, enabled by default) for building the system with
1301 support for the entire 21-bit character space defined by the
1303 * new feature: the system now has rudimentary external-format
1304 support; the primary user-visible change at this time is that
1305 characters with the high bit set (such as lower-case-e-acute) will
1306 print correctly to a terminal in a UTF-8 environment.
1307 * minor incompatible change: BASE-CHAR no longer names a class;
1308 however, CHARACTER continues to do so, as required by ANSI.
1309 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1310 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1311 has been moved to the SB-EXT package (temporarily re-exported from
1313 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1314 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1316 * fixed bug #351: better error-handlind and reporting for malformed
1318 * fixed bug #350: bignum-printing is now more memory-efficient,
1319 allowing printing of very large bignums, eg. (expt 2 10000000).
1320 (reported by Bruno Haible)
1321 * fixed bug #302: better primitive-type selection for intersection
1323 * fixed bug #308: non-graphic characters now all have names, as
1324 required. (reported by Bruno Haible)
1325 * bug fix: redefining a class with different superclasses now correctly
1326 removes it from the direct-subclasses of its previous superclasses.
1327 (reported by David Morse)
1328 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1329 new value now works. (reported by Bruno Haible)
1330 * bug fix: correct canonicalization of multiple non-standard slot
1331 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1332 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1333 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1334 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1335 now exists, an signals an error.
1336 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1337 during macroexpansion for non-iteration variables as well. (reported
1338 by Bruno Haible for CMUCL)
1339 * bug fix: Cyclic structures and unprintable objects in compiler
1340 messages no longer cause errors. (reported by Bruno Haible)
1341 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1342 and READ-FROM-STRING all now return a primary value of NIL if
1343 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1344 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1345 (reported by Bruno Haible for CMUCL)
1346 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1347 do not designate a condition. (reported by Bruno Haible for
1349 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1350 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1351 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1352 permitting resolution in favour of any of the conflicting symbols.
1353 (reported by Bruno Haible for CMUCL)
1354 * bug fix: EQUAL compiler optimizations is less aggressive on
1355 strings which can potentially compare true despite having distinct
1356 specialized array element types.
1357 * bug fix: unit enumerations can be defined without dividing by
1358 zero. (thanks to Vincent Arkesteijn)
1359 * FORMAT compile-time argument count checking has been enhanced.
1360 (report from Bruno Haible for CMUCL)
1361 * a partial workaround for the bug 262: the compiler does not try to
1362 inline-expand a local function doing RETURN-FROM from a deleted
1363 BLOCK. (thanks to Peter Denno for the bug report and to David
1364 Wragg for the simple test case)
1365 * fixed some bugs revealed by Paul Dietz' test suite:
1366 ** RENAME-PACKAGE allows all package designators as new package
1368 ** constraint propagation and lambda variable substitution are
1369 more cautious in dealing with partially deleted code.
1370 ** compiler sometimes lost reoptimization passes.
1371 ** CERROR, given a condition as condition designator, passes its
1372 remaining arguments to the continue format control without
1374 ** Case-altering FORMAT directives work correctly on non-ASCII
1376 ** The REAL type specifier handles bounds outside the floating
1377 point ranges without signalling FLOATING-POINT-OVERFLOW.
1378 ** Functions with IR1-transformations can create intercomponent
1379 references to global functions.
1380 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1382 ** FORMAT ~R treats a nil value for its first parameter correctly.
1384 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1385 * enhancement: saving cores with foreign code loaded is now
1386 supported on x86/NetBSD and sparc/Linux in addition to the previously
1387 supported platforms.
1388 * bug fix: on some platforms repeated installations caused multiple
1389 copies of HTML documentation to be installed -- should not happen
1390 any more. (reported by Stefan Scholl)
1391 * bug fix: parsing self-recursive alien record types multiple times
1392 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1393 original patch by Helmut Eller for CMUCL)
1394 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1395 (thanks to Richard Kreuter)
1396 * bug fix: defining classes whose accessors are methods on existing
1397 generic functions in other (locked) packages no longer signals
1398 bogus package lock violations. (reported by François-René Rideau)
1399 * bug fix: special variables as DEFMETHOD parameters no longer have
1400 associated bogus type declarations. (reported by David Wragg and
1402 * bug fix: read-write consistency on streams of element-type
1403 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1404 * bug fix: redefiniton of the only method of a generic function with
1405 no DEFGENERIC no longer emits a full WARNING. In addition,
1406 redefinition of generic functions with no DEFGENERIC to an
1407 incompatible lambda list now signals an error. (thanks to Zach
1409 * bug fix: DEFGENERIC now works even when there's a function of the
1410 same name in an enclosing lexical environment. (thanks to Zach
1412 * fixed compiler failure, caused by instrumenting code during
1413 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1414 * optimization: added loop analysis and improved register allocation
1415 * fixed some bugs revealed by Paul Dietz' test suite:
1416 ** POSITION on displaced vectors with non-zero displacement
1417 returns the right answer.
1418 ** (SIMPLE-STRING) is a valid type specifier for sequence
1420 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1422 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1423 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1425 ** The justification version of the FORMAT ~< directive treats
1426 non-zero minpad parameter correctly.
1428 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1429 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1430 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1431 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1432 the supported interface.
1433 * new feature: Single-stepping of code compiled with DEBUG 2 or
1434 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1435 * new feature: saving cores with foreign code loaded is now
1436 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1437 Timothy Moore's work for CMUCL)
1438 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1439 parameters to * instead of NIL if no initform is supplied.
1440 (reported by Johan Bockgaard)
1441 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1442 correctly when FUNCALL appears as the car of the form. Note:
1443 despite this FUNCALL forms are not currently subject to
1444 compiler-macro expansion. (port of Raymond Toy's fix for the
1445 same from CMUCL, reported by Johan Bockgaard)
1446 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1447 (thanks for Teemu Kalvas)
1448 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1449 case instead of terminating the iteration (thanks for Julian
1450 Squires, Sean Champ and Raymond Toy)
1451 * bug fix: incorrect expansion of defgeneric that caused a style
1452 warning. (thanks for Zach Beane)
1453 * bug fix: slot accessor effective method computation works properly
1454 for classes with multiple non-standard applicable methods on
1455 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1456 * on x86 compiler supports stack allocation of results of LIST and
1457 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1458 CMUCL implementation by Gerd Moellmann)
1459 * fixed some bugs revealed by Paul Dietz' test suite:
1460 ** FORMAT strings with both the ~<~:;~> form of the justification
1461 directive and pretty-printing directives cause an error.
1463 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1464 * incompatible change: the internal functions
1465 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1466 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1467 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1468 instead of the old functions.
1469 * new feature: on platforms where "dladdr" is available foreign
1470 function names now appear in backtraces. (based on Helmut Eller's
1472 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1473 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1475 * documentation: ASDF manual is now distributed with SBCL.
1476 * bug fix: SBCL can now load its contributed modules with REQUIRE
1477 even if the system-provided entries have been removed from
1478 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1480 * bug fix: backtraces involving undefined functions or assembly
1481 routines are more informative. (thanks to Brian Downing)
1482 * bug fix: mutually referent alien structures now work correctly.
1483 (reported by Rick Taube)
1484 * bug fix: structures defined by WITH-ALIEN can be referred to
1485 within other definitions in the same WITH-ALIEN.
1486 * bug fix: division operators (MOD, TRUNCATE and the like) with
1487 constant zero divisors and integer dividends no longer generate
1489 * bug fix: provide default methods for INPUT-STREAM-P and
1490 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1491 * bug fix: improve the detection and resolution of MOP metacycles
1492 (where effective method computation depends on the generic function
1493 having its effective method computed). (reported by Bruno Haible)
1494 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1495 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1496 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1497 * bug fix: pathnames with relative directory components can be
1498 represented relative to default pathnames.
1499 * optimization: in taking the GCD of bignums, reduce the two bignums
1500 to approximately the same size (using Euclid's algorithm) before
1501 applying the more sophisticated binary GCD. (thanks to Juho
1503 * optimization: COUNT on bitvectors now operates word-at-a-time.
1504 * optimization: ASH with a positive, but not necessarily constant,
1505 (leftwards) shift, when in a modular context, is compiled to a
1507 * fixed some bugs revealed by Paul Dietz' test suite:
1508 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1509 if the corresponding argument is NIL.
1510 ** FORMAT directives accepting a minpad parameter treat negative
1512 ** literal commas as character parameters to FORMAT directives are
1514 ** literal spaces directly after ~<Newline> directives within a
1515 format-logical-block (~:< ~@:>) do not induce :FILL-style
1516 conditional newlines.
1517 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1518 as specified: it no longer includes conditional newlines).
1519 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1521 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1522 :PER-LINE-PREFIX argument does not evaluate to a string.
1524 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1525 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1526 the manual for details; for now, package locks can be disabled by
1527 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1528 if no major problems are found then it is likely that they will be
1529 compiled in unconditionally.
1530 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1531 unsupported operators on all platforms. To load a shared library
1532 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1533 object file, link it into a shared library outside of SBCL and
1534 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1536 * bug workaround: ROOM T and the GC were not working together
1537 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1538 reporting facility aren't preserved. That reporting has been
1539 disabled, so now until and unless someone figures out how to make it
1540 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1541 * minor incompatible change: as threatened around sbcl-0.8.0, the
1542 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1543 The symbols are also exported from SB-PCL for backwards
1544 compatibility, but more so than before SB-PCL should be treated as
1545 an implementation-internal package.
1546 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1548 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1549 It is known as of this release not to work on the Alpha, however.
1550 * fixed bug #167: errors signalled due to illegal syntax in method
1551 bodies are now more legible.
1552 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1553 designators and can hence be used with TYPEP.
1554 * fixed bug #333: CHECK-TYPE now ensures that the type error
1555 signalled, if any, has the right object to be accessed by
1556 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1557 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1558 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1560 * fixed bug #334: programmatic addition of slots using specialized
1561 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1562 and :ALLOCATION :CLASS slots.
1563 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1564 float exponents. (rereported by Peter Seibel)
1565 * fixed a bug: #\Space (and other whitespace characters) are no
1566 longer considered to be macro characters in standard syntax by
1567 GET-MACRO-CHARACTER.
1568 * fixed bug: initialization of condition class metaobjects no longer
1569 causes an instance of the condition to be created. (reported by
1571 * fixed bug: it is now possible to have slots such that
1572 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1573 description is neither :INSTANCE nor :CLASS.
1574 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1575 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1576 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1577 non-standard SLOT-DEFINITION-ALLOCATION values.
1578 * fixed bugs: various race conditions handling exiting threads.
1579 CL-PPCRE's thread torture test now passes.
1580 * fixed arguably-a-bug: GC time proportional to number of threads in
1581 system even when most of them are idle
1582 * optimization: improved performance of BIT and SBIT on bit-vectors.
1583 * .fasl file incompatibility: The fasl file version number has
1584 been incremented because of changes associated with package locks.
1586 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1587 * minor incompatible change: the system no longer provides
1588 optimization hints (conditions which are TYPEP
1589 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1590 mode; these hints are still emitted when higher SPEED optimization
1592 * new contrib module: a sampling profiler (profiling by statistical
1593 sampling, rather than by instrumenting functions) is available as
1594 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1595 * the behaviour of the standard function ED is now customizeable by
1596 third parties through a hook variable: see ED's documentation
1597 string for information on the protocol.
1598 * the compiler no longer emits efficiency notes for (FUNCALL X)
1599 when the type of X is uncertain under default optimization
1601 * fixed bug 276: mutating a binding of a specialized parameter to a
1602 method to something that is not TYPEP the specializer is now
1604 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1605 exhibit the required behaviour on the x86 platform. (thanks to
1606 Peter van Eynde, Eric Marsden and Bruno Haible)
1607 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1608 even for difficult arguments. (reported by Peter Graves)
1609 * fixed bug 141a: the backquote printer now descends quoted
1611 * fixed another bug in backquote printing: no more destructive
1612 modification of the form's list structure. (reported by Brian
1614 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1615 move between its address being taken and the call to
1616 interrupt_thread, fixing a crashing race condition.
1617 * the SB-POSIX contrib implementation has been adjusted so that it
1618 no longer exhibits ridiculously poor performance when constructing
1619 instances corresponding to C structs.
1621 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1622 * minor incompatible change: the sb-grovel contrib now treats C
1623 structures as alien (in the sense of SB-ALIEN) objects rather than
1624 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1625 has implications for memory management of client code
1626 (sb-grovel-returned objects must now be manually managed) and for
1627 type safety (alien objects now have full types).
1628 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1629 used to control emission of compiler diagnostics, rather than the
1630 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1631 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1632 quality should be considered deprecated.
1633 * install.sh now installs the user manual as well
1634 * (not quite a new documentable feature, but worth considering in
1635 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1636 beginnings of a semantically meaningful condition hierarchy is
1637 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1639 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1640 designator as the defaults argument.
1641 * fixed bug: Displaced arrays whose displaced-to array has become
1642 too small now cause ARRAY-DIMENSION to signal an error, providing
1643 for safer bounds-checking. (reported by Bruno Haible)
1644 * fixed bug: DEFCLASS slot definitions with identical :READER and
1645 :WRITER names now signal a reasonable error. (reported by Thomas
1647 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1649 * fixed bug: Invalid slot specification errors now print correctly.
1650 (thanks to Zach Beane)
1651 * fixed bug 320: Shared to local slot value transfers in class
1652 redefinitions now happen corrently from superclasses as
1653 well. (reported by Bruno Haible)
1654 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1656 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1657 improper lists correctly. (reported by Bruno Haible)
1658 * fixed bug 313: source-transform for <fun-name> was erroneously
1659 applied to a call of a value of a variable with name <fun-name>.
1660 (reported by Antonio Menezes Leitao)
1661 * fixed bug 307: The obsolete instance protocol ensures that
1662 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1663 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1664 remove dead unknown-values globs from the middle of the stack.
1665 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1666 easier to resume long computations after using *BREAK-ON-SIGNALS*
1667 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1668 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1669 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1670 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1671 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1672 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1674 * fixed a (fairly theoretical) bug in string printing: if
1675 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1676 does not have array-element-type equal to the most general string
1678 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1679 function name. (thanks to Nikodemus Siivola)
1680 * fixed bug: compile-time format string checker failed on
1681 non-closed ~{. (reported by Thomas F Burdick)
1682 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1683 don't warn on legal loop code involving more than one aggregate
1684 boolean. (thanks to Nikodemus Siivola)
1685 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1686 in keyword position to MAKE-PACKAGE was not being reported as
1688 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1689 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1691 * fixed bugs 280 and 312: the checking for multiple definitions in a
1692 file is less likely to become confused by uses of inline
1694 * fixed bug: the #S reader macro performs the keyword coercion
1695 specified for slot names. (reported by Kalle Niemitalo)
1696 * fixed bug: lambda lists may contain symbols whose names start with
1697 & but are not lambda-list-keywords; their occurrence triggers a
1699 * fixed bug 321: define-method-combination argument lambda lists do
1700 not cause invalid code to be generated when &optional or &aux
1701 variables have default values. (reported by Bruno Haible)
1702 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1703 have CLOS classes; this appears also to have fixed reported
1704 instability in stack exhaustion detection.
1705 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1706 MUFFLE-WARNING when no associated restart is present is now
1708 * optimization: rearranged the expansion of various defining macros
1709 so that each expands into only one top-level form in a
1710 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1712 * optimization: used a previously little-used slot in symbols to
1713 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1715 * fixed some bugs revealed by Paul Dietz' test suite:
1716 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1718 ** Fixed an optimization bug related to inheritance of initforms
1719 from local to shared slots.
1720 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1721 ** CAST optimizer forgot to flush argument derived type.
1722 ** print/read consistency on floats is now orders of magnitude
1723 more likely. (thanks also to Bruno Haible for a similar report
1725 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1726 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1727 for objects of type REAL. Make it so.
1728 ** SXHASH is specified (infelicitously) to respect similarity,
1729 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1730 it so. (thanks to Markus Ziegler)
1731 ** on the Alpha, the compiler succeeds in compiling functions
1732 returning a known number of arguments greater than 63.
1733 ** fixed handling of invalid NIL arguments in keyword position
1735 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1737 ** provided a readably-printable representation for RANDOM-STATE
1739 ** ensured that pathnames, where they have a namestring, always
1740 print using #P"..." syntax.
1742 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1743 * Support for the forthcoming 2.0 version of the NetBSD kernel
1744 running on x86 hardware has been added. (thanks to Perry
1745 E. Metzger most immediately, and others for their past work)
1746 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1747 current release version; SBCL's OpenBSD support had been broken
1748 since about the time of OpenBSD's switch to ELF binary format.)
1749 * [placeholder for DX summary]
1750 ** user code with &REST lists declared dynamic-extent, under high
1751 speed or space and low safety and debug optimization policy.
1752 * The manual has been converted to Texinfo format and the debugger
1753 chapter from the cmucl manual has been added.
1754 * A facility has been added to extract documentation strings from
1755 sbcl and store them as Texinfo-formatted snippets for inclusion in
1756 the manual (via Texinfo's @include directive)
1757 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1758 the test case to Dave Roberts)
1759 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1760 pointers. (reported by Sean Ross)
1761 * bug fix: PROFILE output is printed nicely even for large numerical
1762 values. (thanks to Zach Beane)
1763 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1764 greater than 32 handle EOF correctly.
1765 * bug fix: on X86 an immediate argument of the IMUL instruction is
1766 correctly printed in disassembly. (thanks to Lutz Euler)
1767 * bug fix: class slots in redefined classes preserve their old
1768 values. (thanks to Bruno Haible and Nikodemus Siivola)
1769 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1770 succeeds. (reported by Marco Baringer)
1771 * fixed some bugs revealed by Paul Dietz' test suite:
1772 ** READ-SEQUENCE now works on ECHO-STREAMs.
1773 ** RATIONALIZE works more according to its specification. (thanks
1776 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1777 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1778 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1779 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1780 should matter to you only if you rebind the printer control
1781 variables and then find you want different bindings in the
1782 debugger than in the ordinary execution of your program.)
1783 * The runtime build system has been tweaked to support building
1784 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1785 assembler and linker. (thanks to Nikodemus Siivola)
1786 * Unbound, undefined, undeclared variables now trigger full
1787 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1788 more often programmer error than deliberate exploitation of undefined
1790 * optimization: the hash algorithm for strings has changed to one
1791 that is less vulnerable to spurious collisions. (thanks to Juho
1793 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1794 needless bounds checking. (thanks to Juho Snellman)
1795 * optimization: implemented multiplication as a modular
1796 (UNSIGNED-BYTE 32) operation on the PPC backend.
1797 * fixed some bugs revealed by Paul Dietz' test suite:
1798 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1800 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1801 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1802 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1804 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1806 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1808 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1809 their output stream on EOF from read.
1810 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1811 have been read to end-of-file.
1812 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1814 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1815 description of determination of which consecutive characters
1817 ** Printing the "Space" character with escaping on now yields "#\\ ",
1818 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1819 ** Reading floating-point numbers with *READ-BASE* set to a number
1820 less than 10 works correctly.
1821 ** Reading floating-point numbers with *READ-BASE* set to a number
1822 more than 10 works correctly.
1823 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1824 the readtable currently in effect.
1826 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1827 * minor incompatible change: parsing of namestrings on a physical
1828 (Unix) host has changed; numbers after the final #\. in a
1829 namestring are no longer interpreted as a version field. This is
1830 intented to be largely invisible to the user, except that the
1831 meaning of the namestring "*.*.*" has changed: it now refers to a
1832 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1833 should usually be replaced by
1834 (make-pathname :name :wild :type :wild :version :wild)
1835 with the added benefit that this is more likely to be portable.
1836 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1837 option to OPEN now signals an error if the file being opened
1838 exists; this may have an impact on existing code.
1839 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1840 signals received, on the PowerPC platforms (both Linux and
1841 Darwin). (thanks to Pierre Mai for pointing out the location of
1843 * several fixes on OS X: The system now builds and runs cleanly on
1844 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1845 (thanks to Brian Mastenbrook)
1846 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1847 second-resolution: integer multiples of 1/3600 between -24 and 24.
1848 (thanks to Vincent Arkesteijn)
1849 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1850 type when called with 1 argument; PEEK-CHAR checked type of
1851 PEEK-TYPE only after having read first character from a
1852 stream. (reported by Peter Graves)
1853 * bug fix: the garbage collector now has much better locality
1854 behaviour, and in particular no longer treats all memory as being
1855 exhausted when many small objects point to each other in a deeply
1857 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1858 recognized as being TYPEP their class.
1859 * bug fix: the PUSHNEW documentation string has been corrected.
1860 (thanks to Vincent Arkesteijn)
1861 * bug fix: defaulting of the value for the last of an atypically
1862 large number of multiple values being bound was not being
1863 performed correctly on the Alpha or PPC platforms
1864 * optimization: implemented multiplication as a modular
1865 (UNSIGNED-BYTE 32) operation on the x86 backend.
1866 * optimization: SEARCH on simple-base-strings can now be open-coded.
1867 (see also contrib/compiler-extras.lisp for inspiration for
1868 teaching the compiler about the Boyer-Moore algorithm).
1869 * value, returned by MAX (and MIN) called with several EQUALP, but
1870 not EQL, arguments now does not depend on compiler settings.
1871 * fixed some bugs revealed by Paul Dietz' test suite:
1872 ** in stack analysis liveness information is propagated from
1873 non-local entry points.
1874 ** pathwise CAST removing failed when the CAST node did not start
1876 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1877 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1879 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1880 host is already defined.
1881 ** RENAME-FILE works on streams instead of signalling an internal
1883 ** PEEK-CHAR uses the current readtable when determining whether
1884 or not a character is whitespace.
1885 ** MERGE-PATHNAMES handles the case when the pathname does not
1886 specify a name while the default-pathname specifies a version.
1887 ** Pathnames now stand a better chance of respecting print/read
1889 ** Attempting to use standardized file system operators with a
1890 pathname with invalid :DIRECTORY components signals a
1892 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1894 ** DIRECTORY on logical pathnames is more correct.
1895 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1896 signal a TYPE-ERROR if their argument is not a stream.
1897 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1898 designator argument does not designate a stream.
1899 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1900 examining the synonym.
1901 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1903 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1904 element-type larger than ([UN]SIGNED-BYTE 32).
1906 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1907 * When built with the :SB-FUTEX feature, threaded builds now take
1908 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1909 for faster/more reliable mutex and condition variable support.
1910 * Incompatible change (but one you probably shouldn't have been using
1911 anyway): the interface and code for arbitrating between multiple
1912 threads in the same user session has been redesigned.
1913 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1914 error when called without an explicit environment argument.
1915 (thanks to Nikodemus Siivola)
1916 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1917 confused about their position. (thanks to Adam Warner and Gerd
1919 * bug fix: Pretty printing backquoted forms with unquotations in the
1920 argument list position of various code constructs such as LAMBDA
1921 now works correctly. (reported by Paul Dietz)
1922 * bug fix: Pretty printing unquotations no longer loses all
1923 stream position information.
1924 * optimization: performance of string output streams is now less
1925 poor for multiple small sequence writes.
1926 * optimization: performance of CSUBTYPEP in the presence of complex
1927 expressions involving CONS and NOT many times has been improved.
1928 (reported by Paul Dietz)
1929 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1931 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1932 streams for the REPL, for the debugger, and so on. (thanks to
1934 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1935 * fixed some bugs revealed by Paul Dietz' test suite:
1936 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1937 is ignored if it is too small, rather than propagating through
1938 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1939 ** extremely complex negations of CONS types were not being
1940 sufficiently canonicalized, leading to inconsistencies in
1942 ** VALUES tranformer lost derived type.
1944 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1945 * fixed a bootstrapping bug: the build process no longer assumes
1946 that the various BOOLE-related constants have the same value in
1947 host and target lisps. (noted by Paul Dietz' test suite on an
1948 SBCL binary built from CLISP)
1949 * The system can now be dynamically linked on the MIPS platform,
1950 which enables dynamic loading of foreign code from Lisp. (thanks
1951 to Ralf Baechle for discussions on the MIPS ABI)
1952 * The system now records debugging information for its own source
1953 files in a filesystem-position-independent manner, relative to
1954 the "SYS" logical host.
1955 * fixed a compiler bug: MV-LET convertion did not check references
1956 to the "max args" entry point. (reported by Brian Downing)
1957 * tweaked disassembly notes to be less confident about proclaiming
1958 some instruction as an LRA. (thanks to Brian Downing)
1959 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1960 now each have their own history, command character, and other
1961 characteristics. (thanks to David Lichteblau)
1962 * fixed some bugs revealed by Paul Dietz' test suite:
1963 ** compiler failure in compiling LOGAND expressions including a
1965 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1966 shift greater than 32.
1967 ** FLUSH-DEST did not mark blocks for type check regeneration.
1968 ** HANDLER-CASE failed to accept declarations in handler clauses
1969 in some circumstances.
1971 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1972 * New code in contrib/sb-introspect (still probably not entirely
1973 stable yet) provides some support for smart Lisp development
1974 environments like SLIME.
1975 * The conditions signalled for errors occurring when loading .fasl
1976 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1977 in a way which should help ASDF recover gracefully.
1978 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1979 stuff has been cleaned up. If you code contrib/ stuff, this might
1980 affect you, and you can look at contrib/README, contrib/STANDARDS,
1981 and/or the 0.8.4.27 diff to check.
1982 * In full calls the compiler now does not generate checks for declared
1983 argument types for all arguments.
1984 * various threading fixes
1985 ** and some experimental patches which didn't make it into
1986 the main tree for this release, but which are shipped in
1987 contrib/experimental-thread.patch as a possible fix for some
1988 failures (deadlock, spinning...) in GC-intensive multithreaded
1990 * fixed PPC build problem (source code incompatibility of different
1991 library versions): added offsetof() hackery which attempts to divine
1992 where glibc maintainers put uc_mcontext today
1993 * fixed bug 282: compiler does not trust type assertions while passing
1994 arguments to a full call.
1995 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1996 &OPTIONAL VALUES type parameter.
1997 * fix bug 214: algorithm for noting rejected templates is now more
1998 similar to that of template seletion. (also reported by rydis on
2000 * fixed bug 141b: printing backquoted information readably and prettily
2001 inserts a space where necessary.
2002 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
2003 (CHARACTER 10) are now reported as errors, rather than propagated
2004 as unknown types. (reported by piso on #lisp)
2005 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
2006 with values NIL and :ERROR. (thanks to Milan Zamazal)
2007 * fixed bug 191c: CLOS now does proper keyword argument checking as
2008 described in CLHS 7.6.5 and 7.6.5.1.
2009 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
2010 longer raise an error; further, using a list as a for-as-arithmetic
2011 counter now raises a meaningful error.
2012 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
2013 now understood by sequence creation functions such as MAKE-SEQUENCE
2015 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
2016 a STRING-INPUT-STREAM.
2017 * compiler enhancement: SIGNUM is now better able to derive the type
2019 * type declarations inside WITH-SLOTS are checked. (reported by
2021 * fixed some bugs revealed by Paul Dietz' test suite:
2022 ** incorrect optimization of TRUNCATE for a positive first
2023 argument and negative second.
2024 ** compiler failure in let-convertion during flushing dead code.
2025 ** compiler failure while deriving type of TRUNCATE on an
2026 interval, containing 0.
2027 ** ASH of a negative bignum by a negative bignum count now returns
2029 ** intersection of CONS types now canonicalizes properly, fixing
2030 inconsistencies in SUBTYPEP.
2032 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
2033 * incompatible change: The --disable-debugger command line
2034 option now clobbers the debugger at a more fundamental
2035 level, by redefining #'INVOKE-DEBUGGER instead of by
2036 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
2037 is specified by ANSI to ignore *DEBUGGER-HOOK* and
2038 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
2039 enter the debugger REPL and then suffer recursive errors
2040 because *DEBUG-IO* is also messed up in --disable-debugger mode;
2041 while under the new system, BREAK in --disable-debugger mode
2042 terminates the system just as an unhandled error would.
2043 * fixed compiler performance when processing loops with a step >1;
2044 * bug fix: DOCUMENTATION now retrieves generic function
2045 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
2046 support has been systematized, and now supports the methods
2047 specified by ANSI, along with a default method and a method for
2048 slot documentation. (reported by Nathan Froyd)
2049 * bug fix: effective methods associated with a generic function are
2050 no longer cached over a change of that generic function's method
2051 combination. (reported by Andreas Fuchs)
2052 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
2053 names a non-existent directory. (thanks to Andreas Fuchs)
2054 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
2055 the PPC platform, lead to stack corruption; this has been fixed.
2056 (reported by Rainer Joswig)
2057 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
2058 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
2059 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
2060 arithmetic in general, is now much more reliable on the PPC
2062 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
2063 platform now returns the right answer.
2064 * optimization: restored some effective method precomputation in
2065 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
2066 precomputation is now tunable.
2067 * optimization: compiler-internal data structure use has been
2068 reviewed, and changes have been made that should improve the
2069 performance of the compiler by about 20%.
2070 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
2071 simple-base-strings and simple-bit-vectors is improved.
2072 * optimization: the optimization of 32-bit logical and arithmetic
2073 functions introduced in version 0.8.3 on the x86 has been
2074 implemented on the mips, ppc and sparc platforms; an
2075 implementation of the same facility, but for 64-bit arithmetic,
2076 has been added for the alpha.
2077 * microoptimization: the compiler is better able to make use of the
2078 x86 LEA instruction for multiplication by constants.
2079 * bug fix: in some situations compiler did not report usage of
2080 generic arithmetic in (SPEED 3) policy.
2081 * bug 145b fix: compiler used wrong type specifier while converting
2082 MEMBER-types to numeric.
2083 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
2085 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
2086 index argument now works properly on the Alpha platform.
2087 * bug fix: floating point exception treatment on the Alpha platform
2089 * bug fix: FILE-POSITION works much better on string input and
2090 output streams. (thanks to Nikodemus Siivola)
2091 * bug fix: many threading/garbage collection symptoms sorted.
2092 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
2093 might be pseudo-atomic.
2094 * internal change: Stopping for GC is now done with signals not
2095 ptrace. GC is now done in whichever thread wanted it, instead of
2097 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
2099 * bug fix: result form in DO is not contained in the implicit
2101 * incompatible change: ICR structure is changed; the value part of
2102 CONTINUATION is now called LVAR; corresponding functions are
2103 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
2104 * added type deriver for ISQRT (thanks to Robert E. Brown).
2105 * bug fix: better support for loading from the command line when an
2106 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
2108 * fixed some bugs revealed by Paul Dietz' test suite:
2109 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
2110 ** ROUND and FROUND now give the right answer when given very
2111 small float arguments.
2112 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
2114 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
2115 ** (GCD 0 <negative-integer>) returned <negative-integer>.
2116 ** LCM should return a non-negative integer.
2117 ** PARSE-INTEGER returned the index of a terminator instead of the
2118 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
2119 ** PARSE-INTEGER returned an incorrect index being applied to a
2121 ** LCM with two arguments of 0 returns 0 rather than signalling
2123 ** unsigned addition of a 32-bit constant with the high bit set no
2124 longer causes an internal compiler error on the x86.
2125 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
2126 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
2127 with negative last argument.
2128 ** byte specifiers with zero size and position no longer cause
2129 an error during type derivation.
2130 ** bignum multiplication on the Alpha platform now returns the
2132 * porting: The system now builds on SuSE AMD64, although it still
2133 generates a 32-bit binary.
2134 * .fasl file incompatibility: The fasl file version number has
2135 been incremented (because of the changes to internal compiler
2136 data structures referred to above).
2138 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
2139 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
2140 more accurately, on the Darwin kernel running on PowerPC hardware.
2141 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
2142 * Compiler code deletion notes now signal a condition of type
2143 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
2144 an associated MUFFLE-WARNING restart.
2145 * The compiler now performs limited argument count validation of
2146 constant format strings in FORMAT, and where appropriate in ERROR,
2147 CERROR and WARN. (thanks to Gerd Moellmann)
2148 * New ASDF-INSTALL contrib can be used for automatic download and
2149 installation of third-party Lisp code from CCLAN or other sites
2151 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
2152 which forces another thread to execute a function supplied by the
2154 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
2155 now accept and act upon their :ELEMENT-TYPE keyword argument.
2156 (reported by Martin Atzmueller, Edi Weitz)
2157 * bug fix: FILE-POSITION now accepts position designators up to
2158 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
2159 is the greater. (thanks to Patrik Nordebo)
2160 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
2161 Kalle Olavi Niemitalo)
2162 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
2163 printable. (reported by Eric Marsden)
2164 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
2165 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
2166 * x86 bug fix in control stack exhaustion checking: now shows backtrace
2167 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
2168 (thanks to Stig Sandoe)
2169 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
2170 UNREAD-CHAR now correctly includes the unread character in the
2171 target sequence. (thanks to Gerd Moellmann)
2172 * bug fix in threaded builds: the system can now be suspended and
2173 resumed by shell job control with minimal disruption.
2174 * bug fixes in times and timezones >2038AD
2175 * better handling of "where is GNU make?" problem in build scripts
2176 (thanks to Nikodemus Siivola)
2177 * new optimization: inside a named function any reference to a
2178 function with the same name is considered to be a self-reference;
2179 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
2180 optimization quality.
2181 * new optimization on x86: logical functions and + now have
2182 optimized (UNSIGNED-BYTE 32) versions, which are automatically
2183 used when the result is truncated to 32 bits.
2184 * VALUES declaration is partially enabled.
2185 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2186 * fixed some bugs revealed by Paul Dietz' test suite:
2187 ** The system now obeys the constraint imposed by
2188 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2189 types form a lattice under type intersection.
2190 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2191 ** ASSOC now ignores NIL elements in an alist.
2192 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2193 and (1+ MOST-POSITIVE-FIXNUM) answers.
2194 ** The addition of a method with invalid qualifiers to a generic
2195 function does not cause an error to be signalled immediately;
2196 a warning is signalled, and the error is generated only on
2197 calling the generic function.
2198 * changed .fasl file version number, in part to add type codes for
2199 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2200 obscure ANSI requirements
2202 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2203 * fixed bug 148: failure to inline-expand a local function left
2204 garbage, confusing the compiler.
2205 * fixed bugs 3cd: structure slot readers perform type check if the
2206 slot can have an invalid value (i.e. it is either not initialized
2207 or can be written with a less specific slot writer).
2208 * bug fix: the compiler now traps array references to elements off
2209 the end of an array; previously, the bounds checking in some
2210 circumstances could go off-by-one.
2211 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2213 * type declarations for array element types now obey the description
2214 on the CLHS page "Declaration TYPE", as per discussions on
2215 sbcl-help around 2003-05-08. This means that a declaration
2216 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2217 declaration, all references to BAR will be asserted or assumed
2218 (with THE, so dependent on compiler policy) to involve objects of
2219 type FOO. Note that no such declaration is implied in
2220 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2221 * declared types of functions from the "Conditions"
2222 chapter. (reported by Paul Dietz)
2223 * bug fix: CERROR accepts a function as its first argument.
2224 * bug fix: NTH an NTHCDR accept a bignum as index
2225 arguments. (reported by Adam Warner)
2226 * optimization: character compare routines now optimize comparing
2227 against a constant character. (reported by Gilbert Baumann)
2228 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2229 argument now works properly on the MIPS platform.
2230 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2231 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2232 declared to have a functional type, does not cause an error
2234 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2235 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2236 * fixed bug: if last continuation of a deleted block has a
2237 destination, this destination should be deleted too. (reported by
2239 * fixed a bug in the bootstrap process: the host compiler's values
2240 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2241 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2242 test case from Patrik Nordebo)
2243 * improved the ability of the disassembler on the PPC platform to
2244 provide helpful disassembly notes.
2245 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2246 the class in more cases than previously.
2247 * bug fix: FILE-POSITION now understands :START and :END for
2248 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2249 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2250 8) data. (thanks to David Lichteblau)
2251 * bug fix: it is possible to add a method to a generic function
2252 without lambda list.
2253 * bug fix: reader failed to signal END-OF-FILE inside an
2254 object representation. (reported by Nikodemus Siivola)
2255 * fixed some bugs revealed by Paul Dietz' test suite:
2256 ** LAST and [N]BUTLAST should accept a bignum.
2257 ** condition slot accessors are methods.
2258 ** (VECTOR NIL) is a subtype of STRING.
2260 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2261 * minor incompatible change: some nonsensical specialized lambda
2262 lists (used in DEFMETHOD) which were previously ignored now signal
2264 * minor incompatible change: the system is now aware of the types of
2265 variables in the COMMON-LISP package, and will signal errors for
2266 most violations of these type constraints (where previously they
2267 were silently accepted).
2268 * minor incompatible change: COMPILE-FILE now uses the freedom
2269 afforded (ANSI 3.2.2.3) to use derived function types for
2270 functions defined in the same file. This also permits the system
2271 to warn on static type mismatches and function
2272 redefinition. (Currently it does not work with high DEBUG level.)
2273 * minor incompatible change: VALUES declaration is disabled.
2274 * When issuing notes, the compiler now signals a condition of type
2275 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2276 restart for use in user handlers. It is expected that the
2277 COMPILER-NOTE condition will eventually become a condition
2278 supertype to a hierarchy of note types, which will then be
2279 handleable in a similar fashion. However, at the moment, no such
2280 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2281 but it's an implementation detail, not a classification for the
2283 * Changes in type checking closed the following bugs:
2284 ** type checking of unused values (192b, 194d, 203);
2285 ** template selection based on unsafe type assertions (192c, 236);
2286 ** type checking in branches (194bc).
2287 * A short form of VALUES type specifier has ANSI meaning (it has
2288 increased the number of situations when SBCL cannot perform type
2290 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2291 keywords or constants is permissible.
2292 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2293 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2294 classes. (thanks to Antonio Martinez)
2295 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2296 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2297 argument to a pretty-print stream. Instead, it leaves any such
2298 operations to DESCRIBE-OBJECT methods.
2299 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2301 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2302 lambda lists. (thanks to David Lichteblau)
2303 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2304 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2305 (thanks to Gerd Moellmann)
2306 * fixed bug 246: increased compilation speed of long
2307 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2309 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2311 * DEFCONSTANT now throws a condition of type
2312 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2313 constant to a non-EQL value; CONTINUE and ABORT restarts
2314 respectively change and preserve the value.
2315 * fixed bug 63: The code walker, part of the implementation of CLOS,
2316 is now better at handling symbol macros.
2317 * bug fix: There is no longer an internal implementation type named
2318 CL:LENGTH. (reported by Raymond Toy)
2319 * bug fix: In macro-like defining macros/special operators the
2320 implicit block does not enclose the lambda list.
2321 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2322 enclosed in parentheses) are not suitable as type specifiers, and
2323 their use properly signals an error now.
2324 * bug fix: An argument count mismatch for a type specifier in code
2325 being compiled no longer causes an unhandled error at compile
2326 time, but signals a compile-time warning.
2327 * fixed simple vector readable printing
2328 * bug fix: DESCRIBE takes more care over whether the class
2329 precedence list slot of a class is bound before accessing it.
2330 (reported by Markus Krummenacker)
2331 * bug fix: FORMATTER can successfully compile pretty-printer format
2332 strings which use variants of the ~* directive inside.
2333 * bug fix: SEARCH now applies its TEST predicate to the elements of
2334 the arguments in the correct order. (thanks to Wolfhard Buss)
2335 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2336 in type inference. (thanks to Robert E. Brown)
2337 * bug fix: Reading in symbols with an explicit package name of ""
2338 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2339 * fixed some bugs revealed by Paul Dietz' test suite:
2340 ** NIL is now allowed as a structure slot name.
2341 ** Arbitrary numbers, not just REALs, are allowed in certain
2342 circumstances in LOOP for-as-arithmetic clauses.
2343 ** Multiple class redefinitions before slot access no longer
2344 causes a type error.
2345 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2346 association between the name and a class.
2347 ** Generic functions with non-standard method-combination and over
2348 five methods all of which return constants no longer return NIL
2349 after the first few invocations. (thanks to Gerd Moellmann)
2350 ** CALL-NEXT-METHOD with no arguments now passes the original
2351 values of the arguments, even in the presence of assignment.
2352 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2354 ** DISASSEMBLE works with closures and funcallable instances.
2355 ** ADD-METHOD now returns the generic function, not the new method.
2356 ** FIND-METHOD signals an error if the lengths of the specializers
2357 is incompatible with the generic function, even if the ERRORP
2359 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2360 which its argument is a member.
2361 ** DEFCLASS only redefines the class named by its class-name
2362 argument if that name is the proper name of the class;
2363 otherwise, it creates a new class.
2364 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2365 of the UNBOUND-SLOT condition to the name of the slot.
2366 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2367 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2368 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2369 SLOT-MAKUNBOUND in the specified fashion.
2371 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2372 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2373 cross-compilation host. As a consequence, we can now bootstrap our
2374 way up to SBCL starting with a bare gcc toolchain and human-readable
2375 source code (first the source to CLISP, then the source to SBCL).
2376 * A contributed module containing a partial implementation of the
2377 simple-streams interface has been included. (thanks to Rudi
2379 * A contributed module implementing the RFC1321 Message Digest
2380 Algorithm, known as MD5, has been included.
2381 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2382 no longer has any effect, as the code controlled by this feature
2383 has been deleted. (As far as we know, no-one has ever built using
2384 this feature, and its semantics were confused in any case).
2385 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2386 work on conditions (as required by the ANSI specification),
2387 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2388 expected behaviour on conditions. Users should note, however,
2389 that such behaviour is not required by the ANSI specification,
2390 and so use of this behaviour may render their code unportable.
2391 * fixed some bugs revealed by Paul Dietz' test suite:
2392 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2394 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2395 ** SXHASH on condition objects no longer returns NIL.
2396 ** :ALLOCATION :CLASS slots are better treated; their values are
2397 updated on class redefinition, and initforms inherited from
2398 superclasses are applied.
2399 ** REMOVE-METHOD returns its generic function argument even when
2400 no method was removed.
2401 ** SHARED-INITIALIZE now initializes the values of the requested
2402 slots, including those with :ALLOCATION :CLASS.
2403 ** ALLOCATE-INSTANCE now works on structure classes defined via
2404 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2406 ** SLOT-EXISTS-P now works on conditions, as well as structures
2408 ** MAKE-LOAD-FORM now has the required methods on
2409 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2410 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2411 keyword, and now implements the SLOT-NAMES argument.
2412 ** methods with &OPTIONAL arguments no longer allow too many
2413 arguments to be passed in the call without error.
2414 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2415 option is consistent with the required arguments of the generic
2416 function lambda list.
2417 * bug fix: REQUIRE accepts a string designator. (Thanks to
2419 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2420 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2421 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2422 * bug fix: The compiler now checks for duplicated variables in macro
2424 * bug fix: SETQ on globals returns the correct value.
2425 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2426 not a COMPILER-ERROR (followed by some other strange error on
2427 choosing the CONTINUE restart).
2428 * bug fix: make.sh and friends are now more consistent in the way that
2429 they look for GNU "make".
2431 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2432 * experimental native threads support (on x86 Linux >=2.4 only).
2433 This is not compiled in by default: you need to add :SB-THREAD to
2434 the target features. See the "Beyond ANSI" chapter of the manual
2436 * fix for longstanding nonANSIism: The old distinction between
2437 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2438 The return value from CL:FIND-CLASS is now a CLOS class, and
2439 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2440 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2441 * An interface to the MetaObject Protocol, as described in Kiczales,
2442 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2443 MIT Press, 1991, is available from the SB-MOP package.
2444 * incompatible change: the SB-PCL package should now be considered
2445 a private implementation detail, and no longer a semi-private MOP
2447 * minor incompatible change: due to rearrangement for threads, the
2448 control stack and binding stack are now allocated at arbitrary
2449 addresses instead of being hardcoded per-port. Users affected by
2450 this probably have to be doing advanced things with shared
2451 libraries, and will know who they are.
2452 * minor incompatible change: Previously, all --eval forms used were
2453 processed with READ before any of them were processed with EVAL.
2454 Now each --eval form is processed with both READ and EVAL before
2455 the next --eval form is processed. (Thus package operations like
2456 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2457 work as the user might reasonably expect.)
2458 * minor incompatible change: *STANDARD-INPUT* is now only an
2459 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2461 * minor incompatible change: Y-OR-N-P is now character-oriented,
2462 not line oriented. Also, YES-OR-NO-P now works without errors.
2463 (thanks to Antonio Martinez)
2464 * sb-aclrepl module improvements: an integrated inspector, added
2465 repl features, and a bug fix to :trace command.
2466 * Known functions, which cannot be open coded by the backend, are
2467 considered to be able to check types of their arguments. (fixing
2468 a bug report by Nathan J. Froyd)
2469 * fixed a bug in computing method discriminating functions: It is
2470 now possible to define methods specialized on classes which have
2471 forward-referenced superclasses. (thanks to Gerd Moellmann)
2472 * fixed evaluation order in optional entries (reported by Gilbert
2474 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2475 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2476 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2477 required-by-AMOP NAME argument, as well as CLASS and
2478 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2479 * fixed bug 20: DEFMETHOD can define methods using names that are
2480 not the proper names of classes to designate class specializers.
2481 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2482 Unix file descriptors, instead of blowing up. (thanks to Antonio
2484 * Garbage collection refactoring: user-visible change is that a
2485 call to the GC function during WITHOUT-GCING will not do garbage
2486 collection until the end of the WITHOUT-GCING. If you were doing
2487 this you were probably losing anyway.
2488 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2489 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2490 TYPEP the latter but not the former.
2491 * The compiler issues a full WARNING for calls to undefined functions
2492 with names from the CL package.
2493 * MAP-INTO for a vector destination is open coded. (reported by
2494 Brian Downing on c.l.l)
2495 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2496 documentation string.
2497 * fixed some bugs revealed by Paul Dietz' test suite:
2498 ** COPY-ALIST now signals an error if its argument is a dotted
2500 ** Condition slots are now accessed more correctly in the presence
2501 of multiple initargs for a given slot.
2502 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2503 exclude restarts of the same name associated with a different
2505 ** DEFCLASS of forward-referenced classes with another
2506 forward-referenced class in the superclasses list no longer
2508 ** Condition slots are now initialized once each, not multiple
2509 times. (thanks to Gerd Moellmann)
2510 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2512 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2513 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2514 arguments contain duplicated elements.
2515 ** RESTART-CASE understands local macros.
2516 ** RESTART-CASE associates exactly its own restarts with a condition.
2517 ** ENDP in safe mode checks its argument to be of type LIST.
2518 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2519 in question is unbound.
2520 ** Optimized MAKE-INSTANCE functions no longer cause internal
2521 assertion failures in the presence of duplicate initargs.
2522 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2523 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2524 propagated even in the ctor optimized implementation of
2526 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2528 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2529 * a better implementation of SXHASH on (simple) bit vectors,
2530 measured both in execution speed and in distribution of results
2531 over the positive fixnums, has been installed. Likewise, a better
2532 implementation of EQUAL for simple bit vectors is now available.
2533 * fixed CEILING optimization for a divisor of form 2^k.
2534 * fixed bug 240 (emitting extra style warnings "using the lexical
2535 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2536 by Antonio Martinez)
2537 * fixed SXHASH, giving different results for NIL depending on type
2538 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2539 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2540 variable. (found by Rolf Wester)
2541 * SBCL does not ignore type declarations for special
2542 variables. (reported by rif on c.l.l 2003-03-05)
2543 * some bug fixes in contrib/sb-aclrepl/
2544 * fixed some bugs revealed by Paul Dietz' test suite:
2545 ** a bug in the CONS type specifier, whereby the CAR and CDR
2546 types got intertwined, has been fixed;
2547 ** the type system is now able to reason about the interaction
2548 between INTEGER and RATIO types more completely;
2549 ** APPEND, [N]REVERSE and NRECONC check that those their
2550 arguments, which must be proper lists, are really so;
2551 ** An array specialized to be unable to hold elements has been
2552 implemented, as required -- yes, really -- by ANSI;
2553 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2554 on malformed property lists;
2556 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2557 * incompatible packaging change: in line with Unix convention,
2558 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2559 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2560 * REQUIRE and PROVIDE are now optionally capable of doing something
2561 useful. See the documentation string for REQUIRE.
2562 * infrastructure for a managed SBCL contrib system: contributed
2563 modules in this release include:
2564 ** the ASDF system definition facility;
2565 ** an interface to the BSD Sockets API;
2566 ** an ACL-like convenience interface to the repl;
2567 (thanks to Kevin Rosenberg)
2568 ** an implementation of ROTATE-BYTE, with efficient implementation
2570 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2571 gives a closer approximation to the right answer than previously.
2572 (thanks to Raymond Toy)
2573 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2574 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2575 optional environment arguments, as required by ANSI.
2576 * fixed bugs in other functions taking environment objects, allowing
2577 calls with an explicit NIL environment argument to be compiled
2579 * fixed bug 228: primary return values from
2580 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2581 COMPILE or FUNCTION.
2582 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2583 :TYPE will no longer signal errors on innocuous objects.
2584 * fixed bug 231b: SETQ is better at respecting type declarations in
2585 the lexical environment.
2586 * fixed a bug in DEFCLASS: classes named by symbols with no or
2587 unprintable packages can now be defined.
2588 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2589 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2590 * fixed compiler failure related to checking types of functions.
2591 (reported by Robert E. Brown)
2592 * the compiler is now much more consistent in its error-checking
2593 treatment of bounding index arguments to sequence functions: in
2594 (SAFETY 3) code, errors will be signalled in almost all cases if
2595 invalid sequence bounding indices are passed to functions defined
2596 by ANSI to operate on sequences.
2597 * fixed a bug in the build procedure: documentation of SBCL-specific
2598 packages is now preserved and available in the final Lisp image.
2599 * lifted FDEFINITION lookup out of loops in the implementation of
2600 many list operations. (thanks to Robert E. Brown)
2601 * fixed a bug in the reader: the #n# reader macro now works for
2602 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2603 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2604 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2605 E. Brown and Tony Martinez respectively)
2606 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2607 SLOT-BOUNDP: in particular, it is now able to optimize them much
2608 better, and is now not vulnerable to having packages renamed.
2609 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2610 always called when a slot is not present in an instance. (thanks
2612 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2613 longer signal an error on unreading a character following EOF on
2614 the previous constituent stream. (thanks to Tony Martinez)
2615 * fixed some bugs revealed by Paul Dietz' test suite:
2616 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2617 not just nonnegative fixnums;
2618 ** the logical bit-array operators such as BIT-AND now accept an
2619 explicit NIL for their "opt-arg" argument (to indicate a
2620 freshly-consed result bit-array);
2621 ** ELT now signals an error on an invalid sequence index in safe
2623 ** the type system is now cleverer about negations of numeric
2624 types, and consequently understands the BIGNUM and RATIO types
2626 ** the type system is now cleverer about the interaction between
2627 INTEGER and RATIO types: while bugs still remain, many more
2628 cases are accurately computed;
2629 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2630 if it is in the last clause;
2631 ** CONSTANTLY now correctly returns a side-effect-free function in
2633 ** DECLARE is no longer treated as a special-operator; in
2634 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2635 * incremented fasl file version number due to the change in the
2636 DEFSTRUCT-SLOT-DESCRIPTION structure.
2638 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2639 * minor incompatible change: code processed by the "interpreter" or
2640 EVAL now has a compilation optimization policy of (DEBUG 2)
2641 (changed from (DEBUG 1)) to improve debuggability of interactive
2642 development, and to allow the use of the debug RETURN command in
2644 * an experimental implementation of the RETURN command for the
2645 debugger has been included. (thanks to Frederik Kuivinen)
2646 * fixed bug 62: constraints were not propagated into a loop.
2647 * fixed bug in embedded calls of SORT (reported and investigated by
2649 * fixed some bugs revealed by Paul Dietz' test suite:
2650 ** printing and reading of arrays with some dimensions having
2651 length 0 (thanks to Gerd Moellmann);
2652 ** BOA constructor with &AUX argument without a default value does
2653 not cause a type error;
2654 ** CONSTANTP now returns true for all self-evaluating objects.
2656 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2657 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2658 accessors that are related by inheritance, as specified in the
2659 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2660 Valtteri Vuorikoski)
2661 * The compiler is now able to inline functions that were defined in
2662 a complex lexical environment (e.g. inside a MACROLET).
2663 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2665 * Support for the upcoming FreeBSD-5.0 release has been included.
2666 (thanks to Dag-Erling Smorgrav)
2667 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2668 effect when it is not in a toplevel context.
2669 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2670 stand a better chance of being correct. (thanks to Gerd
2672 * fixed bug in COERCE, which now signals an error on coercing a
2673 rational to a bounded real type which excludes the expected
2675 * The compiler is now able to derive types more accurately from the
2676 COERCE and COMPILE functions.
2677 * fixed bug 223: functional binding is considered to be constant
2678 only for symbols in the CL package.
2679 * fixed bug 231: SETQ did not check the type of a variable being set
2680 (reported by Robert E. Brown)
2681 * A new optimization for MAKE-INSTANCE has been included, fixing
2682 various bugs (including relating to :ALLOCATION :CLASS slots and
2683 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2685 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2686 ** As required by ANSI, LOOP now disallows anonymous collection
2687 clauses such as COLLECT I in conjunction with aggregate boolean
2688 clauses such as THEREIS (= I 1);
2689 ** LOOP now signals an error when any variable is reused in the
2690 same loop (including the potentially useful construct analogous
2691 to WITH A = 1 WITH A = (1+ A);
2692 ** IT is only a special loop symbol within the first clause of a
2693 conditional loop clause;
2694 ** LOOP with a typed iteration variable over a hashtable now
2695 signals a type error iff it should.
2696 * fixed some other bugs revealed by Paul Dietz' test suite:
2697 ** FILE-STREAM now names the class previously known as FD-STREAM;
2698 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2699 argument) no longer signals an error;
2700 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2701 of the slot symbol, rather than using the current package
2702 ((:CONC-NAME "") continues to intern the slot's name in the
2704 * incremented fasl file version number, because of the incompatible
2705 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2706 of the new implementation of DEFINE-COMPILER-MACRO.
2708 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2709 * Support for building SBCL for MIPS platforms running in
2710 little-endian mode has now been checked in, and basic
2711 functionality on said platforms verified.
2712 * minor incompatible change: PCL now records the pathname of a file
2713 in which methods and the like are defined, rather than its
2715 * minor incompatible change: TRUENAME now considers the truename of
2716 a file naming a directory to be the pathname with :DIRECTORY
2717 component indicating that directory.
2718 * minor incompatible change: a NAMED clause in the extended form of
2719 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2720 reason for the previous behaviour is unclear.
2721 * more systematization and improvement of CLOS and MOP conformance
2722 in PCL (thanks to Gerd Moellman and Pierre Mai):
2723 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2725 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2726 primary methods with no specializers;
2727 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2729 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2730 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2731 FORWARD-REFERENCED-CLASSes; error reporting on
2732 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2734 ** SXHASH on CLOS instances now uses a slot internal to the
2735 instance to return different numbers on distinct instances,
2736 while preserving the same return value through invocations of
2738 ** DEFMETHOD signals errors when methods with longer incongruent
2739 lambda lists are added to generic functions;
2740 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2741 CLASS, as specified in AMOP;
2742 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2743 on the order returned by the primary method for classes of
2744 class STANDARD-CLASS;
2745 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2746 * fixed some bugs shown by Paul Dietz' test suite:
2747 ** DOLIST puts its body in TAGBODY;
2748 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2750 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2751 value producing form;
2752 ** if more variables are given to PROGV than values, extra
2753 variables are bound and made to have no value;
2754 ** NSUBSTITUTE on list arguments gets the right answer with
2756 ** ELT signals an error of type TYPE-ERROR when the index argument
2757 is not a valid sequence index;
2758 ** LOOP signals (at macroexpansion time) an error of type
2759 PROGRAM-ERROR when duplicate variable names are found;
2760 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2761 ** FUNCALL of special-operators now cause an error of type
2763 ** PSETQ now works as required in the presence of side-effecting
2764 symbol-macro places;
2765 ** NCONC accepts any object as its last argument;
2766 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2768 ** loop-for-as-package does not require a package to be explicitely
2770 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2772 * fixed bug 166: compiler preserves "there is a way to go"
2773 invariant when deleting code.
2774 * fixed bug 172: macro lambda lists with required arguments after
2775 &REST arguments now cause an error to be signalled. (thanks to
2777 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2778 bug. (thanks to Matthew Danish)
2779 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2781 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2782 arguments are lexically rebound. (thanks to Gerd Moellmann and
2784 * fixed bug 194: error messages are now more informative when there
2785 is no primary method applicable in a call to a generic
2786 function. (thanks to Gerd Moellmann)
2787 * fixed bug in command line argument checking (thanks to Julian
2789 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2790 (thanks to Matthew Danish)
2791 * incremented fasl file version number, because of the
2792 SXHASH-related changes in the layout of CLOS data structures
2794 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2795 * minor incompatible change: The runtime (the Unix executable named
2796 "sbcl") is now much pickier about the .core files it will load.
2797 Essentially it now requires .core files to descend from the same
2798 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2799 as the runtime does. (The intent is to prevent the crashes which
2800 can occur, and which can even be reported as mysterious failures,
2801 when people patch the sources or change the build parameters
2802 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2803 sbcl and .core files.)
2804 * fixed bug: VALUES-LIST is no longer optimized away.
2805 * fixed bug 142: The FFI conversion of C string values to Lisp
2806 string values no longer conses excessively. (thanks to Nathan
2807 Froyd porting Raymond Toy's fix to CMU CL)
2808 * began to systematize and improve MOP conformance in PCL (thanks to
2809 Nathan Froyd, Gerd Moellman and Pierre Mai):
2810 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2812 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2813 ** FINALIZE-INHERITANCE is now called on class finalization;
2814 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2815 argument precedence order.
2816 * fixed bug 202: The compiler no longer fails on functions whose
2817 derived types contradict their declared type.
2818 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2819 so it can be non-toplevel.
2820 * The fasl file version number has changed (because of the new
2821 implementation of DEFMACRO).
2822 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2823 safe code, that any length requirement by their type-specifier
2824 argument is valid. The exceptions to this are described in bug
2826 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2827 functions now check the types of their inputs as required by ANSI.
2828 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2829 introduced symbol is DECLAREd to be SPECIAL.
2830 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2831 * fixed bug: :COUNT argument to sequence functions may be negative
2832 * fixed bug: body of DO-SYMBOLS may contain declarations
2833 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2834 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2835 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2836 symbol macro only once
2837 * fixed printing of call frame when argument list is unavailable
2838 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2839 * compiler no longer signals WARNING on unknown keyword
2842 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2843 * A beta-quality port to the mips architecture running Linux,
2844 based on the old CMUCL backend, has been made. It has been tested
2845 on a big-endian kernel, and works sufficiently well to be able to
2846 rebuild itself; it has not been tested in little-endian mode.
2847 * fixed an inconsistency between gencgc.c and purify.c which made
2848 dumping/loading .core files unreliable
2849 * fixed bug 120a: The compiler now deals correctly with IFs where
2850 the consequent is the same as the alternative, instead of
2851 misderiving the return type. (thanks to Alexey Dejneka)
2852 * fixed bug 113: Logical pathnames are now dumpable (the logical
2853 host is resolved at load-time, throwing an error if it is not
2855 * fixed bug 174: FORMAT's error message is slightly clearer when a
2856 non-printing character is used in a format directive.
2857 * fixed several bugs in compiler checking of type declarations, i.e.
2858 violations of the Python "declarations are assertions" principle
2859 (thanks to Alexey Dejneka)
2860 * fixed several bugs in PCL's error checking (thanks to Gerd
2862 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2864 * fixed bug in compilation of functions as first class values
2865 (thanks to Antonio Martinez-Shotton)
2866 * The compiler's handling TYPE-ERRORs which it can prove will
2867 inevitably happen at runtime has been cleaned up and corrected
2868 in several ways. (thanks to Alexey Dejneka)
2869 * improved argument type checking for various basic arithmetic
2870 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2871 lavished on them in the past that they can be compiled in many
2872 ways in different special cases
2873 * fixed bug 181: compiler checks validity of user supplied type
2875 * cleaned up code flushing in optimization: Function calls which
2876 should signal errors for safety purposes (e.g. which ANSI says
2877 should signal errors when their arguments are of incorrect type)
2878 are no longer optimized away.
2879 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2880 * incremented fasl file version number, because changes in the
2881 implementation of sequence functions like COERCE caused
2882 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2883 in old inline expansions) to become undefined. (Actually these
2884 changes were later undone, so we might very well be binary
2885 compatible with 0.7.7 after all, but leaving the version number
2886 incremented seemed like the simplest and most conservative
2889 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2890 * An alpha-quality port to the parisc architecture running Linux,
2891 based on the old CMUCL backend, has been made. This, even more so
2892 than the other backends, should be considered still a work in
2893 progress; known problems include that the Linux kernel in 64-bit
2894 mode does not propagate the correct sigcontext structure to
2895 userspace, and consequently SBCL on a parisc64 kernel will not
2897 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2898 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2899 suppressed.) Also now that the compiler is looking at declarations
2900 in the environment, it checks optimization declarations as well,
2901 and suppresses inlining when (> DEBUG SPEED).
2902 * More fixes have been made to treatment of floating point exception
2903 treatment and other Unix signals. In particular, floating point
2904 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2905 * The detection and handling of control stack exhaustion (infinite
2906 or very deeply nested recursion) has changed. Stack exhaustion
2907 detection is now done by write-protecting pages at the OS level
2908 and applies at all optimization settings; when found, a
2909 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2910 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2911 be caught using IGNORE-ERRORS.
2912 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2913 correctly when transforming recursive LABELS functions to LETs.
2914 (thanks to Alexey Dejneka)
2915 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2916 that are names of constants or global variables.
2917 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2918 alien routines with docstrings.
2919 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2920 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2922 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2923 (thanks to Eric Marsden)
2924 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2925 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2926 * Bug fix: The SPARC backend can now compile functions involving
2927 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2928 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2929 to a functions expecting a PATHNAME-DESIGNATOR.
2930 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2931 lambda lists. (thanks to Alexey Dejneka)
2932 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2933 functions so that EXPORTing the name of the class doesn't cause
2934 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2935 to Antonio Martinez for reporting this)
2936 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2937 constant-folded without failing an assertion. (thanks to Einar
2938 Floystad Dorum for reporting this)
2939 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2940 types are now dealt with more correctly by the compiler.
2941 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2942 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2943 behaviour analogous to RENAME-FILE. This puts its behaviour more
2944 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2946 * The fasl file version number has changed again. (because of the
2947 bug fix involving the names of PCL MAKE-INSTANCE functions)
2949 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2950 * bug fix: Floating point exceptions are treated much more
2951 consistently on the x86/Linux and PPC/Linux platforms.
2952 * Array initialization with :INITIAL-ELEMENT is now much faster for
2953 cases when the compiler cannot open code the array creation, but
2954 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2955 array accesses have also seen a speed increase.
2956 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2957 specified. (This was at the root of some bad interactions between
2958 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2960 * bug fix: Internal error arguments for undefined functions are now
2961 computed correctly on the PPC/Linux platform.
2962 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2963 Raymond Toy's patch for CMU CL)
2964 * Support for the Solaris 9 operating environment has been included
2965 (thanks to Daniel Merritt)
2966 * A very ugly but hopefully complete draft of the missing FFI chapter
2967 of the manual has been created by reformatting the corresponding
2968 CMU CL manual chapter into (currently very ugly and incoherent)
2969 DocBook and bringing it up to date for SBCL behavior. Thus, the
2970 manual is now essentially complete, at least by my extreme
2971 once-and-only-once standards, whereby it's acceptable to refer to
2972 the doc strings of SB-EXT functions as the primary documentation.
2973 * The fasl file version number has changed again, due to cleanup of
2974 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2975 is no longer a static symbol.)
2977 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2978 * SBCL now builds with OpenMCL (version 0.12) as the
2979 cross-compilation host; also, more progress has been made toward
2980 bootstrapping under CLISP.
2981 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2983 * bug 158 fixed: The compiler can now deal with integer loop
2984 increments different from 1; fixing this turned out also to fix
2986 * bug 169 fixed: no more bogus warnings about using lexical bindings
2987 despite the presence of perfectly good SPECIAL declarations (thanks
2988 to David Lichteblau)
2989 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2990 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2991 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2993 * bug 180 fixed: Method combination specifications no longer ignore
2994 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2995 * bug fix: Structure type predicate functions now check their argument
2996 count as they should.
2997 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2998 correctly. (thanks to Pierre Mai)
2999 * minor incompatible change: The --noprogrammer option is deprecated
3000 in favor of the new --disable-debugger option, which is very similar.
3001 (The major difference is that it takes effect at a slightly different
3002 time at startup, causing handling of errors in --sysinit and
3003 --userinit files will be affected differently.) The
3004 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
3005 been added to allow this functionality to be controlled from ordinary
3006 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
3007 maintainers, who might want to run non-interactive scripts to
3008 build SBCL cores which will later be used interactively.)
3009 * minor incompatible change: The LOAD function no longer, when given
3010 a wild pathname to load, loads all files matching that pathname.
3011 Instead, an error of type FILE-ERROR is signalled.
3013 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
3014 * bug 147 fixed: The compiler preserves its block link/count
3015 invariants more correctly now so that it doesn't crash. (thanks
3017 * Dynamic loading of object files in OpenBSD is now supported. (thanks
3019 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
3020 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
3021 no-value-for-this-character as NIL (as specified by ANSI).
3022 * HOST-NAMESTRING on physical pathnames now returns a string that is
3023 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
3024 (thanks to Christophe Rhodes)
3025 * The Alpha port handles icache flushing more correctly. (thanks to
3027 * More progress has been made toward bootstrapping under CLISP. (thanks
3028 to Christophe Rhodes)
3029 * The fasl file format has changed again, because dynamic loading
3030 on OpenBSD (which has non-ELF object files) motivated some cleanups
3031 in the way that foreign symbols are transformed and passed around.
3032 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
3033 is no longer treated as whitespace by the reader, but instead as
3034 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
3035 |A
\7fB|, instead of A as it used to.
3037 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
3038 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
3039 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
3040 * SBCL now runs on the PPC archtiecture under Linux. It actually did
3041 this as of 0.7.1.45, but was left out of the previous news section
3042 (thanks to Dan Barlow)
3043 * SBCL now runs on the Solaris operating system on SPARC architectures
3044 (thanks to Christophe Rhodes's port of the CMUCL runtime)
3045 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
3046 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
3047 * SPARC backend cleanups, allowing builds of cores optimized for V8
3048 and V9 SPARCS, and also emission of code targeted to a particular
3049 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
3051 * SBCL is closer to bootstrapping under CLISP, thanks to various
3052 fixes by Christophe Rhodes.
3053 * The fasl file format has changed again, to allow the compiler's
3054 INFO database to support symbol macros.
3055 * The user manual (in doc/) is formatted into HTML more nicely.
3056 (thanks to coreythomas)
3057 * The system is smarter about SUBTYPEP relationships, especially
3058 those involving NOT types (including types such as ATOM which are
3059 represented internally using NOT types). Thus SUBTYPEP is less
3060 likely to return (VALUES NIL NIL) in general, and in particular
3061 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
3063 * The fasl file format has changed again, because the internal
3064 representation of types now includes a new slot to support the new
3065 SUBTYPEP-of-NOT-types logic.
3066 * (not a change in the main branch of SBCL, but a related prototype
3067 which can hopefully be merged into the main branch of SBCL in the
3068 future:) Brian Spilsbury has produced a Unicode-enabled variant of
3069 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
3070 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
3071 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
3072 dispatch macro character combinations. (thanks to Alexey Dejneka)
3073 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
3074 that superficially look like logical namestrings correctly.
3075 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
3077 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
3078 default supplied-p) for &optional and &key arguments. (thanks to
3081 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
3082 * incompatible change: The compiler is now less aggressive about
3083 tail call optimization, doing it only when (> SPACE DEBUG) or
3084 (> SPEED DEBUG). (This is an incompatible change because there are
3085 programs which relied on the old CMU-CL-style behavior to optimize
3086 away their unbounded recursion which will now die of stack overflow.)
3087 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
3088 for non-GENCGC systems has been increased to 20M (since that
3089 seems much closer to the likely performance optimum for modern
3090 systems than the old 4M value was)
3091 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
3092 and *DEBUG-PRINT-LEVEL*
3093 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
3094 of CMU CL's support for SPARC, and various endianness and other
3095 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
3096 * new syntactic sugar for the Unix command line: --load foo.bar is now
3097 an alternate notation for --eval '(load "foo.bar")'.
3099 ** The system now detects stack overflow and handles it gracefully,
3100 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
3101 optimization settings. (This is a good thing in general, and
3102 its introduction in this version should be particularly timely
3103 for anyone whose code fails because of suppression of tail
3105 ** The system now hunts for the C variable "environ" in a more
3106 devious way, to avoid segfaults when the C library version differs
3107 between compile time and run time. (thanks to Christophe Rhodes)
3108 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
3109 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
3110 ** The compiler no longer issues bogus style warnings for undefined
3111 classes in the same source file as the DEFCLASSes which defined
3112 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
3114 ** fixes in CONDITION class precedence list for undefined function
3115 errors (thanks to Alexei Dejneka)
3116 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
3117 correctly. (thanks to Dan Barlow)
3118 ** portability fixes aiming at bootstrapping under CLISP (thanks
3119 to Dave McDonald and Christophe Rhodes)
3120 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
3121 ** fixes in type translation and and type inference (thanks to
3123 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
3124 ** various fixes in the new ports (thanks to Dan Barlow)
3125 * several changes related to debugging:
3126 ** suppression of tail recursion, as noted above
3127 ** stack overflow detection, as noted above
3128 ** The default implementation of TRACE has changed. :ENCAPSULATE T
3129 is now the default. (For some time encapsulation has been more
3130 reliable than the breakpoint-based :ENCAPSULATE NIL
3131 implementation, at least on X86 systems; and I just noticed that
3132 encapsulation also seems closer to the spirit of the ANSI
3135 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
3137 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
3138 up properly again. (There was a packaging bug in 0.7.0 which
3139 left their definitions in SB-SYS::LOAD-FOREIGN and
3140 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
3141 vital for most things which interface to C-level interfaces,
3142 like extensions working with sockets or databases or
3143 Perl-compatible regexes or whatever, and the need to fix
3144 this bug is the main reason that 0.7.1 was released so
3146 ** DEFGENERIC is now choosier about the methods it redefines, so that
3147 reLOADing a previously-LOADed file containing DEFGENERICs does
3148 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
3149 cycle now works as it should. (thanks to Alexey Dejneka)
3150 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
3151 fixed by Christophe Rhodes. (He actually submitted this patch
3152 months ago, and I delayed until after 0.7.0.)
3153 ** Bug 111 (internal compiler confusion about runtime checks on
3154 FUNCTION types) was fixed by Alexey Dejneka.
3155 * Some internal cleanups (getting rid of variables which aren't
3156 needed now that the byte interpreter is gone) caused the fasl
3157 file format number to change again.
3159 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
3160 * major incompatible change: The default fasl file extension, i.e. the
3161 default extension for files produced by COMPILE-FILE, has changed
3162 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
3164 ** There are many changes in the implementation of the compiler.
3165 SBCL is now essentially a compiler-only implementation of ANSI
3166 Common Lisp. EVAL still "interprets" a few special cases, but
3167 almost all the interesting cases are handled by creating
3168 a LAMBDA expression, calling COMPILE on it, then calling
3169 FUNCALL on the result.
3170 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
3171 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
3172 Since the code is newer, there might still be some new bugs
3173 (though not as many as before Martin Atzmueller's fixes:-). But
3174 the new code is substantially simpler and clearer, and hopefully
3175 any remaining bugs will be simpler, less fundamental, and more
3176 fixable then the bugs in the old code.
3177 ** The revised compiler is still a little unsteady on its feet.
3179 *** The debugging information it produces (particularly the names
3180 of FUNCTION objects) is sometimes much less useful than what
3181 the old compiler produced.
3182 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
3183 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
3184 within a MACROLET) has been temporarily weakened.
3185 ** There are new compiler optimizations for various functions:
3186 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3187 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3188 *** the math functions TRUNCATE, FLOOR, and CEILING
3189 *** the function-of-all-trades COERCE
3190 Mostly these should be transparent, but there's one
3191 potentially-annoying problem (bug 117): when the compiler
3192 inline-expands a function and does type analysis on the result,
3193 it can create control paths which have type mismatches, and
3194 when it can't prove that those control paths aren't taken,
3195 it will issue WARNINGs about the type mismatches. This is
3196 a particular problem in practice for the new sequence functions.
3197 It's not clear how this should be fixed, and for now, a
3198 workaround is given in the entry for 117 in the BUGS file.
3199 ** (Because of the interaction between the two previous items --
3200 occasional inlining problems and new inline expansions -- some
3201 of the new sequence function optimizations won't really kick in
3202 completely until debugging information, and then inlining, are
3203 straightened out in some future version.)
3204 * minor incompatible changes:
3205 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3206 DIRECTORY no longer implicitly promotes NIL slots of its
3207 pathname argument to :WILD. In particular, when you ask for the
3208 contents of a directory (which you used to be able to do without
3209 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3210 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3211 ** changes in behavior that ANSI explicitly defines to be
3212 implementation dependent:
3213 *** The new compiler-only implementation still conforms with ANSI,
3214 but acts a little different than before. Besides the obvious
3215 changes in performance tradeoffs (that the cost per form passed
3216 to EVAL has gone up, and the cost per form executed by EVAL
3217 has gone down), the behavior of the system changes a little
3218 because there are no longer any interpreted function objects.
3219 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3220 e.g. doing COMPILE on the output of interactive DEFUN is
3222 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3224 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3225 has changed from (:CL) to NIL.
3226 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3227 e.g. #\Newline and #\Tab, don't have names specified in the
3228 ANSI Common Lisp standard, is now based on their ASCII symbolic
3229 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3230 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3231 are no longer used for output.
3232 ** changes in internal implementation constants:
3233 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3234 4 million. (If your application spends a lot of time GCing and
3235 you have a lot of RAM, you might want to experiment with
3236 increasing it even more.)
3237 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3238 However, almost all old code should still continue to work without
3239 immediate update, as SB-C-CALL is now a (deprecated) nickname
3241 ** Old operator names in the style DEF-FOO are now deprecated in
3242 favor of new corresponding names DEFINE-FOO, for consistency with
3243 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3244 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3245 internal symbols, but a few supported extensions like
3246 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3247 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3248 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3249 etc. as you get deeper into recursive calls to the debugger
3250 command loop, instead of the old "5]", "5]]", "5]]]"
3251 sequence. (I was motivated to do this when squabbles between
3252 ILISP and SBCL left me very deeply nested in the debugger. In the
3253 short term, this change will probably provoke more ILISP/SBCL
3254 squabbles, but hopefully it will be an improvement in the long run.)
3255 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3256 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3257 function, since declaiming return types involving aliens is
3258 (1) annoyingly messy to do by hand and (2) vital to efficient
3259 compilation of code which calls such functions.
3260 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3261 longer reexported by the SB-EXT package. They're solely useful
3262 for alien code, so it seems more logical that you should get
3263 them from the SB-ALIEN package, not in SB-EXT.
3264 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3265 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3266 features. Instead, the code that they used to control is always
3267 built into the system.
3268 * many other bug fixes
3269 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3270 advantage of the new EVAL-WHEN stuff and to clean them up in
3271 general, and they are now more ANSI-compliant in a number of
3272 ways. Martin Atzmueller is responsible for a lot of this.
3273 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3275 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3276 *** correct ERROR type for various file operations
3277 *** some fixes for Lisp streams
3278 *** DEFMETHOD syntax checking
3279 *** changing old weird representation of debug information as
3280 strings (which, among their other deficiencies, don't transform
3281 correctly when you rename packages, and don't change their
3282 print representation when you change things like *PACKAGE*
3283 and *PRINT-LENGTH*) to symbols and lists of symbols
3284 He also made several improvements and fixed several bugs in DESCRIBE.
3285 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3287 *** misbehavior of WRITE-STRING/WRITE-LINE
3288 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3289 and several other LOOP problems as well
3290 *** DIRECTORY when similar filenames are present
3291 *** DEFGENERIC with :METHOD options
3292 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3293 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3294 *** argument ordering in FIND with :TEST option
3295 *** mishandled package designator argument in APROPOS-LIST
3296 *** various problems in the backquote readmacro
3297 *** a bug in APROPOS
3298 *** probably some others that I'm not describing very well here,
3299 since the CVS log documents them by reference to sbcl-devel
3300 messages, and the SourceForge archives aren't working well.:-(
3301 ** Dan Barlow improved the Alpha port (and is making progress on the
3302 PPC port, for those of you who think different).
3303 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3304 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3305 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3306 continues to work on the SPARC port (for those of you in a position
3307 to look down upon our little PC-compatible boxes from a great height).
3308 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3309 and uses that copy, rather than the current dynamic value, when
3310 it's trying to decide whether to truncate output. Thus e.g.
3311 (let ((*print-lines* 50))
3312 (pprint-logical-block (stream nil)
3314 (let ((*print-lines* 8))
3315 (print (aref possiblybigthings i) stream)))))
3316 should now truncate the logical block only at 50 lines, instead of
3317 often truncating it at 8 lines, as it did before.
3318 * The doc/cmucl/ directory, containing old CMU CL documentation
3319 from the time of the fork, is no longer part of the base system.
3320 SourceForge has shut down its anonymous FTP service, and with it
3321 my original plan for distributing the old CMU CL documentation
3322 there. For now, if you need these files you can download an old
3323 SBCL source release and extract them from it.
3324 * The fasl file version number changed again, for dozens of reasons,
3325 some of which are apparent above.
3327 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3328 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3329 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3330 patch, so that DISASSEMBLE output is much nicer.
3331 * The code in the SB-PROFILE package now seems reasonably stable.
3332 I still haven't decided what the final interface should look like
3333 (I'd like PROFILE to interact cleanly with TRACE, since both
3334 facilities use function encapsulation) but if you have a need
3335 for profiling now, you can probably use it successfully with
3336 the current CMU-CL-style interface.
3337 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3338 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3339 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3340 link now returns the dangling link itself, and for similar
3341 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3342 link itself. (In these cases the old code signalled an error and
3343 looped endlessly, respectively.) Thus, DIRECTORY now works even
3344 in the presence of dangling and cyclic symbolic links.
3345 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3346 is now a supported extension again, since the consensus on
3347 sbcl-devel was that it can be useful for ordinary development
3348 work, not just for debugging SBCL itself.
3349 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3350 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3351 that currently-defined functions might be redefined later with
3352 different return types.
3353 * Hash tables can be printed readably, as inspired by CMU CL code
3354 of Eric Marsden and SBCL code of Martin Atzmueller.
3355 * better error handling in CLOS method combination, thanks to
3356 Martin Atzmueller porting Pierre Mai's CMU CL patches
3357 * more overflow fixes for >16Mbyte I/O buffers
3358 * A bug in READ has been fixed, so that now a single Ctrl-D
3359 character suffices to cause end-of-file on character streams.
3360 In particular, now you only need one Ctrl-D at the command
3361 line (not two) to exit SBCL.
3362 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3364 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3365 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3366 does the right thing.
3367 * The implementation of some type tests, especially for CONDITION
3368 types, is now tidier and maybe faster, due to CMU CL code
3369 originally by Douglas Crosher, ported by Martin Atzmueller.
3370 * Some math functions have been fixed, and there are new
3371 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3372 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3373 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3374 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3375 likely to be built into sbcl-0.7.0.)
3376 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3377 CL, ported to SBCL by Martin Atzmueller)
3378 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3379 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3380 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3381 * There's a new slam.sh hack to shorten the edit/compile/debug
3382 cycle for low-level changes to SBCL itself, and a new
3383 :SB-AFTER-XC-CORE target feature to control the generation of
3384 the after-xc.core file needed by slam.sh.
3385 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3386 COMPILE-FILE is no longer supported, so that now every function
3387 gets an entry point, so that block compilation looks a little
3388 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3389 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3390 returns the number of bytes consed since the system started,
3391 rather than the number consed since the first time the function
3392 was called. (The new definition parallels ANSI functions like
3393 CL:GET-INTERNAL-RUN-TIME.)
3394 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3395 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3396 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3397 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3398 * Fasl file version numbers are now independent of the target CPU,
3399 since historically most system changes which required version
3400 number changes have affected all CPUs equally. Similarly,
3401 the byte fasl file version is now equal to the ordinary
3404 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3405 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3406 is no longer recognized. I apologize for this, because it was
3407 listed in SB-EXT as a supported extension, but I found that
3408 its existing behavior was poorly specified, as well as incorrectly
3409 specified, and it looked like too much of a mess to straighten it
3410 out. I have enough on my hands trying to get ANSI stuff to work..
3411 * many patches ported from CMU CL by Martin Atzmueller, with
3412 half a dozen bug fixes in pretty-printing and the debugger, and
3413 half a dozen others elsewhere
3414 * fixed bug 13: Floating point infinities are now supported again.
3415 They might still be a little bit flaky, but thanks to bug reports
3416 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3417 as flaky as they were.
3418 * The --noprogrammer command line option is now supported. (Its
3419 behavior is slightly different in detail from what the old man
3420 page claimed it would do, but it's still appropriate under the
3421 same circumstances that the man page talks about.)
3422 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3423 are now supported, and enabled by default. Thus, the compiler can
3424 handle many floating point and complex operations much less
3425 inefficiently. (Thus e.g. you can implement a complex FFT
3427 * The compiler now detects type mismatches between DECLAIM FTYPE
3428 and DEFUN better, and implements CHECK-TYPE more correctly, and
3429 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3430 thanks to patches from Martin Atzmueller.
3431 * various fixes to make the cross-compiler more portable to
3432 ANSI-conforming-but-different cross-compilation hosts (notably
3433 Lispworks for Windows, following bug reports from Arthur Lemmens)
3434 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3435 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3436 fixed thanks to Pierre Mai's CMU CL patches.
3437 * fixes to keep the system from overflowing internal counters when
3438 it tries to use i/o buffers larger than 16M bytes
3439 * fixed bug 45a: Various internal functions required to support
3440 complex special functions have been merged from CMU CL sources.
3441 (When I was first setting up SBCL, I misunderstood a compile-time
3442 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3443 * improved support for type intersection and union, fixing bug 12
3444 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3445 more obscure bugs as well
3446 * some steps toward byte-compiling non-performance-critical
3447 parts of the system, courtesy of patches from Martin Atzmueller
3448 * Christophe Rhodes has made some debian packages of sbcl at
3449 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3450 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3451 stable, but are nonetheless usable. When he's ready, I'd be happy
3452 to add them to the SourceForge "File Releases" section. (And if
3453 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3454 * new fasl file format version number (because of changes in
3455 internal representation of (OR ..) types to accommodate the new
3456 support for (AND ..) types, among other things)
3458 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3459 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3461 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3463 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3464 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3465 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3466 * DESCRIBE now works on structure objects again.
3467 * Most function call argument type mismatches are now handled as
3468 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3469 know whether the function will be redefined before the call is
3470 executed. (The compiler could flag local calls with full WARNINGs,
3471 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3472 it doesn't keep track of enough information to know whether calls
3473 are local in this sense.)
3474 * Compiler output is now more verbose, with messages truncated
3475 later than before. (There should be some supported way for users
3476 to override the default verbosity, but I haven't decided how to
3477 provide it yet, so this behavior is still controlled by the internal
3478 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3479 src/compiler/ir1util.lisp.)
3480 * Fasl file format version numbers have increased again, because
3481 support for the Gray streams extension changes the layout of the
3482 system's STREAM objects.
3483 * The Gray subclassable streams extension now works, thanks to a
3484 patch from Martin Atzmueller.
3485 * The full LOAD-FOREIGN extension (not just the primitive
3486 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3487 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3488 but like most other programs, it defaults to copying the Unix
3489 environment from the original process instead of starting the
3490 new process in an empty environment.
3491 * Extensions which manipulate the Unix environment now support
3492 an :ENVIRONMENT keyword option which doesn't smash case or
3493 do other bad things. The CMU-CL-style :ENV option is retained
3494 for porting convenience.
3495 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3496 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3498 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3500 * A patch from Martin Atzmueller seems to have solved the SIGINT
3501 problem, and as far as we know, signal-handling now works cleanly.
3502 (If you find any new bugs, please report them!)
3503 * The system no longer defaults Lisp source file names to types
3504 ".l", ".cl", or ".lsp", but only to ".lisp".
3505 * The compiler no longer uses special default file extensions for
3506 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3507 seems to expect a single default extension for all compiled code,
3508 and there's no compelling reason to try to stretch the standard
3509 to allow two different extensions.) Instead, byte-compiled files
3510 default to the same extension as native-compiled files.
3511 * Fasl file format version numbers have increased again, because
3512 a rearrangement of internal implementation packages made some
3513 dumped symbols in old fasl files unreadable in new cores.
3514 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3515 many fewer weird special cases.
3516 * Bug #17 (differing COMPILE-FILE behavior between logical and
3517 physical pathnames) has been fixed, and some related misbehavior too,
3518 thanks to a patch from Martin Atzmueller.
3519 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3520 by Tim Moore, ported to SBCL by Martin Atzmueller.
3521 * Martin Atzmueller fixed several filesystem-related problems,
3522 including bug #36, in part by porting CMU CL patches, which were
3523 written in part by Paul Werkowski.
3524 * More compiler warnings in src/runtime/ are gone, thanks to
3525 more patches from Martin Atzmueller.
3526 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3529 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3531 * DESCRIBE now works on CONDITION objects.
3532 * The debugger now handles errors which arise when trying to print
3533 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3535 * The build system now uses an additional file, customize-target-features.lisp,
3536 to allow local modifications to the target *FEATURES* list. (The point of
3537 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3538 debugging features enabled, without having to worry about propagating it
3539 into everyone's system when I do a "cvs update".) When no
3540 customize-target-features.lisp file exists, the target *FEATURES* list
3541 should be constructed the same way as before.
3542 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3543 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3544 warn about behavior which is undefined under the ANSI spec. Note: This
3545 is specified by ANSI, but it's not very popular with programmers.
3546 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3547 macro in the SBCL sources for an example of a workaround which you
3548 might use to make portable ANSI-standard code which does what you want.
3549 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3550 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3551 when it's used as a non-toplevel form. (This is required in order
3552 to implement the DEFCONSTANT-EQX macro.)
3553 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3554 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3555 funny things, due to bugs in the implementation of EVAL-WHEN
3556 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3557 non-toplevel DEFCONSTANTs are uncommon.)
3558 * The core file version number and fasl file version number have been
3559 incremented, because the old noncompliant DEFCONSTANT behavior involved
3560 calling functions which no longer exist, and because I also took the
3561 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3562 * fixed bug 1 (error handling before read-eval-print loop starts), and
3563 redid debugger restarts and related debugger commands somewhat while
3565 ** The QUIT debugger command is gone, since it did something
3566 rather different than the SB-EXT:QUIT command, and since it never
3567 worked properly outside the main toplevel read/eval/print loop.
3568 Invoking the new TOPLEVEL restart provides the same functionality.
3569 ** The GO debugger command is also gone, since you can just invoke
3570 the CONTINUE restart directly instead.
3571 ** The TOP debugger command is also gone, since it's redundant with the
3572 FRAME 0 command, and since it interfered with abbreviations for the
3574 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3576 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3577 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3578 some porting work by Martin Atzmueller)
3579 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3580 with a boatload of patches which he ported to SBCL. Now that those
3582 ** The system tries to make sure that its low-priority messages
3583 are prefixed by semicolons, to help people who like to use
3584 syntax highlighting in their ILISP buffer. (This patch
3585 was originally due to Raymond Toy.)
3586 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3587 patches originally written by Raymond Toy.
3588 ** The compiler understands coercion between single-value and
3589 multiple-VALUES type expressions better, getting rid of some very
3590 weird behavior, thanks to patches originally by Robert MacLachlan
3591 and Douglas Crosher.
3592 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3593 lambda lists, thanks to a patch originally by Pierre Mai.
3594 ** The system no longer bogusly warns about "abbreviated type
3596 ** The compiler gets less confused by inlining and RETURN-FROM,
3597 thanks to some patches originally by Tim Moore.
3598 ** The system no longer hangs when dumping circular lists to fasl
3599 files, thanks to a patch originally from Douglas Crosher.
3600 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3601 undefined function error.
3602 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3603 in structure slot initforms) for now, documented workaround instead:-|
3604 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3605 * fixed bug 5: added stubs for various Gray stream functions called
3606 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3607 installed, at least appropriate type errors are generated
3608 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3609 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3610 consistently in DEFMETHOD forms.
3611 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3612 it doesn't seem to affect SBCL after all
3613 * The C runtime system now builds with better optimization and many
3614 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3616 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3618 * The system is now under CVS at SourceForge (instead of the
3619 CVS repository on my home machine).
3620 * The new signal handling code has been tweaked to treat register
3621 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3622 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3623 downstream, e.g. in debug-int.lisp, has implicit dependencies
3624 on the unsignedness of integer representation of machine words,
3625 and that caused the system to bomb out with infinite regress
3626 when trying to recover from type errors involving signed values,
3627 e.g. (BUTLAST '(1 2 3) -1).)
3628 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3629 (This was one of the bugs Peter Van Eynde reported back in July.)
3630 * The system now uses code inspired by Colin Walters' O(N)
3631 implementation of MAP (from the cmucl-imp@cons.org mailing
3632 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3633 inline the MAP operation, and there is more than one
3634 sequence argument to the MAP call (so that it can't just
3635 do ETYPECASE once and for all based on the type of the
3636 single sequence argument). (The old non-inline implementation
3637 of the general M-argument sequence-of-length-N case required
3638 O(M*N*N) time when any of the sequence arguments were LISTs.)
3639 * The QUIT :UNIX-CODE keyword argument has been renamed to
3640 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3641 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3643 * Raymond Wiker's patches to port dynamic loading from Linux to
3644 FreeBSD have been added.
3645 * The BUGS file is now more nearly up to date, thanks in large part
3646 to Martin Atzmueller's review of it.
3647 * The debugger now flushes standard output streams before it begins
3648 its output ("debugger invoked" and so forth).
3649 * The core version number and fasl file version number have both
3650 been incremented, because of incompatible changes in the layout
3652 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3653 used to not be called for a saved Lisp image.)
3654 * Martin Atzmueller's version of a patch to fix a compiler crash,
3655 as posted on sbcl-devel 13 September 2000, has been installed.
3656 * Instead of installing Martin Atzmueller's patch for the
3657 compiler transform for SUBSEQ, I deleted the compiler transform,
3658 and transforms for some similar consing operations.
3659 * A bug in signal handling which kept TRACE from working on OpenBSD
3661 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3662 away by constant folding
3663 * The system now defines its address space constants in one place
3664 (in the Lisp sources), and propagates them automatically elsewhere
3665 (through GENESIS and the sbcl.h file). Therefore, patching the
3666 address map is less unnecessarily tedious and error-prone. The
3667 Lisp names of address space constants have also been systematized.
3668 * CVS tags like dollar-Header-dollar have been removed from
3669 the sources, because they have never saved me trouble and
3670 they've been source of trouble working with patches and other
3671 diff-related operations.
3672 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3673 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3675 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3677 * The system has been ported to OpenBSD.
3678 * The system now compiles with a simple "sh make.sh" on the systems
3679 that it's supported on. I.e., now you no longer need to tweak
3680 text in the target-features.lisp-expr and symlinks in src/runtime/
3681 by hand, the make.sh takes care of it for you.
3682 * The system is no longer so grossly inefficient when compiling code
3683 involving vectors implemented as general (not simple) vectors (VECTOR T),
3684 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3685 which dares to use the various sequence functions on non-simple
3686 vectors, takes less of a performance hit.
3687 * There is now a primitive type predicate VECTOR-T-P
3688 to test for the (VECTOR T) type, so that e.g.
3689 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3690 can now be compiled with some semblance of efficiency. (The old code
3691 turned the type declaration into a full call to %TYPEP at runtime!)
3692 * AREF on (VECTOR T) is still not fast, since it's still compiled
3693 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3694 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3695 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3696 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3697 without making full calls to SUBTYPEP at runtime.
3698 (Some analogous efficiency issues for non-simple vectors specialized to
3699 element types other than T, or for non-simple multidimensional arrays,
3700 have not been addressed. They could almost certainly be handled the
3701 same way if anyone is motivated to do so.)
3702 * The changes in array handling break binary compatibility, so
3703 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3704 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3705 instead of (VALUES T T).
3706 * By following the instructions that Dan Barlow posted to sbcl-devel
3707 on 2 July 2000, I was able to enable primitive dynamic object
3708 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3709 functionality is not implemented (since it calls ld to resolve
3710 library references automatically, requiring RUN-PROGRAM for its
3711 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3712 try to resolve library references) is now supported.
3713 * The system now flushes the standard output streams when it terminates,
3714 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3715 them at several other probably-convenient times, e.g. in each pass of
3716 the toplevel read-eval-print loop, and after evaluating a form given
3717 as an "--eval" command-line option. (These changes were motivated by a
3718 discussion of stream flushing issues on cmucl-imp in August 2000.)
3719 * The source transform for TYPEP of array types no longer assumes
3720 that an array whose element type is a not-yet-defined type
3721 is implemented as an array of T, but instead punts, so that the
3722 type will be interpreted at runtime.
3723 * There is now some support for cross-compiling in make.sh: each of
3724 the phases of make.sh has its own script. (This should be transparent
3725 to people doing ordinary, non-cross-compile builds.)
3726 * Since my laptop doesn't have hundreds of megabytes of memory like
3727 my desktop machine, I became more motivated to do some items on
3728 my to-do list in order to reduce the size of the system a little:
3729 ** Arrange for various needed-only-at-cold-init things to be
3730 uninterned after cold init. To support this, those things have
3731 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3732 symbols with such names are now uninterned after cold init).
3733 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3734 fasl files for cold load.
3735 ** Remove the old compiler structure pooling code (which used to
3736 be conditional on the target feature :SB-ALLOC) completely.
3737 ** Redo the representation of some data in cold init to be more compact.
3738 (I also looked into supporting byte compiled code at bootstrap time,
3739 which would probably reduce the size of the system a lot, but that
3740 looked too complicated, so I punted for now.)
3741 * The maximum signal nesting depth in the src/runtime/ support code has
3742 been reduced from 4096 to 256. (I don't know any reason for the very
3743 large old value. If the new smaller value turns out to break something,
3744 I'll probably just bump it back up.)
3745 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3747 * Many, many bugs reported by Peter Van Eynde have been added to
3748 the BUGS list; some have even been fixed.
3749 * While enabling dynamic object file loading, I tried to make the
3750 code easier to understand, renaming various functions and variables
3751 with less ambiguous names, and changing some function calling
3752 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3753 * While trying to figure out how to do the OpenBSD port, I tried to
3754 clean up some of the code in src/runtime/. In particular, I dropped
3755 support for non-POSIX signal handling, added various comments,
3756 tweaked the code to reduce the number of compilation warnings, and
3757 renamed some files to increase consistency.
3758 * To support the new automatic configuration functionality in make.sh,
3759 the source file target-features.lisp-expr has been replaced with the
3760 source file base-target-features.lisp-expr and the machine-generated
3761 file local-target-features.lisp-expr.
3762 * fixed a stupid quoting error in make.sh so that using CMU CL
3763 "lisp -batch" as cross-compilation host works again
3765 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3767 * DESCRIBE no longer tries to call itself recursively to describe
3768 bound/fbound values, so that it no longer fails on symbols which are
3769 bound to themselves (like keywords, T, and NIL).
3770 * DESCRIBE now works on generic functions.
3771 * The printer now prints less-screwed-up representations of closures
3772 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3773 * A private symbol is used instead of the :EMPTY keyword previously
3774 used to mark empty slots in hash tables. Thus
3775 (DEFVAR *HT* (MAKE-HASH-TABLE))
3776 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3777 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3778 now does what ANSI says that it should. (You can still get
3779 similar noncompliant behavior if bang on the hash table
3780 implementation with all the symbols you get back from
3781 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3782 This breaks binary compatibility, since tests for equality to
3783 :EMPTY are wired into things like the macroexpansion of
3784 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3786 * There's now a minimal placeholder implementation for CL:STEP,
3787 as required by ANSI.
3788 * An obscure bug in the interaction of the normal compiler, the byte
3789 compiler, inlining, and structure predicates has been patched
3790 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3791 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3793 * Missing ordinary arguments in a macro call are now detected even
3794 when the macro lambda list contains &KEY or &REST.
3795 * The debugger no longer complains about encountering the top of the
3796 stack when you type "FRAME 0" to explicitly instruct it to go to
3797 the top of the stack. And it now prints the frame you request even
3798 if it's the current frame (instead of saying "You are here.").
3799 * As specified by ANSI, the system now always prints keywords
3800 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3802 * The default initial SIZE of HASH-TABLEs is now smaller.
3803 * Type information from CLOS class dispatch is now propagated
3804 into DEFMETHOD bodies, so that e.g.
3805 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3807 is now basically equivalent to
3808 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3809 (DECLARE (TYPE SINGLE-FLOAT X))
3811 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3812 operation, without having to do run-time type dispatch.
3813 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3814 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3815 * Since I don't seem to be making big file reorganizations very often
3816 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3817 is overflowing my ability to conveniently back them up), I've finally
3818 checked the system into CVS. (The CVS repository is on my home system,
3819 not at SourceForge -- putting it on SourceForge might come later.)
3820 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3821 high-level GC-NOTIFY-FOO functions send their output. (There's
3822 still very little control of where low-level verbose GC functions
3823 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3824 less than it used to -- the GC-NOTIFY-FOO functions are now under
3825 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3826 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3827 in only one place in the source code, and propagates it automatically
3828 everywhere that it's needed. Thus e.g. when I bump the version from
3829 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3830 * The C source files now include boilerplate legalese and documentation
3831 at the head of each file (just as the Lisp source files already did).
3832 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3833 to his page will be replaced with a link to his new CLiki service.
3835 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3837 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3838 * The build process now looks for GNU make under the default name "gmake",
3839 instead of "make" as it used to. If GNU make is not available as "gmake"
3840 on your system, you can change this default behavior by setting the
3841 GNUMAKE environment variable.
3842 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3843 can build without error under CMU CL.
3845 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3847 * There is now a partial SBCL user manual (with some new text and some
3848 text cribbed from the CMU CL manual).
3849 * The beginnings of a profiler have been added (starting with the
3850 CMU CL profiler and simplifying and cleaning up). Eventually the
3851 main interface should be through the TRACE macro, but for now,
3852 it's still accessed through vaguely CMU-CL-style functions and macros
3853 exported from the package SB-PROFILE.
3854 * Some problems left over from porting CMU CL to the new
3855 cross-compilation bootstrap process have been cleaned up:
3856 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3857 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3858 part when getting rid of the compiler *BACKEND* stuff.)
3859 ** The value of *NULL-TYPE* was screwed up, because it was
3860 being initialized before the type system knew the final
3861 definition of the 'NULL type. This screwed up several key
3862 optimizations in the compiler, causing inefficiency in all sorts
3863 of places. (I found it because I wanted to understand why
3864 GET-INTERNAL-RUN-TIME was consing.)
3865 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3866 PROCLAIM FTYPE information. Unfortunately this broke binary
3867 compatibility again, since now the forms output by DEFGENERIC
3868 to refer to functions which didn't exist in 0.6.3.
3869 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3870 can use the new (as of 0.6.3) transform for SOME into MAP into
3872 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3873 values to (INTEGER 0 1000000), so that the time code will no longer
3874 occasionally get blown up by Linux returning 1000000 microseconds
3875 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3876 its output conform to the ANSI spec. (Alas, this makes its output
3877 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3878 * A full call to MAP NIL with a single sequence argument no longer conses.
3879 * fixes to problems pointed out by Martin Atzmueller:
3880 * The manual page no longer talks about multiprocessing as though
3881 it were currently supported.
3882 * The ILISP support patches have been removed from the distribution,
3883 because as of version 5.10.1, ILISP now supports SBCL without us
3884 having to maintain patches.
3885 * added a modified version of Raymond Toy's recent CMU CL patch for
3886 EQUALP comparison of HASH-TABLE
3888 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3890 * The system still can't cross-compile itself with
3891 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3892 entails), but at least it can compile more of itself that way
3893 than it used to be able to, and various buglets which were uncovered
3894 by trying to cross-compile itself that way have now been fixed.
3895 * This release breaks binary compatibility again. This time
3896 at least I've incremented the FASL file format version to 2, so that the
3897 problem can be detected reliably instead of just causing weird errors.
3898 * various new style warnings:
3899 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3900 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3901 ** using the lexical binding of a variable named in the *FOO* style
3902 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3903 as specified by ANSI.
3904 * *RANDOM-STATE* is no longer automatically initialized from
3905 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3906 default behavior of the system is to repeat its behavior every time
3907 it's run. If you'd like to change this behavior, you can always
3908 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3909 old convention there was no comparably easy way to get the system to
3910 repeat its behavior every time it was run.
3911 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3912 FTYPE declarations has been removed, in favor of their ANSI
3913 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3914 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3915 the types of their sequence arguments can be determined at compile time.
3916 This is done through a new open code expansion for MAP which eliminates
3917 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3918 argument types can be determined at compile time.
3919 * The optimizer now transforms COERCE into an identity operation when it
3920 can prove that the coerced object is already of the correct type. (This
3921 can be a win for machine generated code, including the output of other
3922 optimization transforms, such as the MAP transform above.)
3923 * Credit information has been moved from source file headers into CREDITS.
3924 * Source file headers have been made more standard.
3925 * The CASE macro now compiles without complaining even when it has
3928 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3930 * (Note that the way that the PCL macroexpansions were rewritten
3931 to accommodate the change in DEFGENERIC below breaks binary
3932 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3933 not run under sbcl-0.6.2. Once we get out of alpha releases,
3934 i.e. hit release 1.0.0, we'll probably try to maintain binary
3935 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3936 and sbcl-1.4.4. Until then, however, it might be fairly common
3937 for maintenance releases to break binary compatibility.)
3938 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3940 * The reporting of unhandled signals has been changed to print some
3941 explanatory text as well as the report form. (Previously only
3942 the report form was printed.)
3943 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3944 it defines, so that the compiler no longer issues undefined function
3945 warnings for compiled-but-not-yet-loaded generic functions.
3946 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3947 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3948 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3949 section of the ANSI spec are supported.
3950 * The "" nickname for the "KEYWORD" package has been removed.
3951 The reader still handles symbol tokens which begin with a package marker
3952 as keywords, but it doesn't expose its mechanism for doing so in the
3953 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3954 * The system now issues STYLE-WARNINGs for contradictory TYPE
3955 proclamations. (Warnings for contradictory FTYPE proclamations would
3956 be nice too, but those can't be done usefully unless the type system
3957 is made smarter about FUNCTION types.)
3958 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3959 systematized, so that "*target-*.lisp is supposed to exist only on the
3960 target and imply that there's a related file which exists on the
3961 host, and *host-*.lisp is supposed to exist only on the host and imply
3962 that there's a related file which exists on the target. This involves a
3963 lot of renaming. Hopefully the acute confusion caused by the renaming
3964 will be justified by the reduction in chronic confusion..
3965 ** runtime-type.lisp -> early-target-type.lisp
3966 ** target-type.lisp -> late-target-type.lisp
3967 ** early-host-format.lisp -> early-format.lisp
3968 ** late-host-format.lisp -> late-format.lisp
3969 ** host-error.lisp -> misc-error.lisp
3970 ** early-error.lisp -> early-target-error.lisp
3971 ** late-error.lisp -> late-target-error.lisp
3972 ** host-defboot.lisp -> early-defboot.lisp
3973 ** code/misc.lisp -> code/target-misc.lisp
3974 ** code/host-misc.lisp -> code/misc.lisp
3975 ** code/numbers.lisp -> code/target-numbers.lisp
3976 ** code/early-numbers.lisp -> numbers.lisp
3977 ** early-host-type.lisp -> early-type.lisp
3978 ** late-host-type.lisp -> late-type.lisp
3979 ** host-typep.lisp -> typep.lisp
3980 ** load.lisp -> target-load.lisp
3981 ** host-load.lisp -> load.lisp
3982 ** host-disassem.lisp -> disassem.lisp
3983 ** host-insts.lisp -> insts.lisp
3984 ** byte-comp.lisp -> target-byte-comp.lisp
3985 ** host-byte-comp.lisp -> byte-comp.lisp
3986 ** host-signal.lisp -> signal.lisp
3987 ** host-defstruct.lisp -> defstruct.lisp
3988 ** late-target-type.lisp -> deftypes-for-target.lisp
3989 Furthermore, several other previously target-only files foo.lisp (e.g.
3990 hash-table.lisp and random.lisp) have been split into a target-and-host
3991 foo.lisp file and a target-only target-foo.lisp file, with their key type
3992 definitions in the target-and-host part, so that the cross-compiler will
3993 know more about target types.
3994 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3995 gone away. In their place are various *BACKEND-FOO* variables
3996 corresponding to the slots of the old structure.
3997 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3998 into the target SBCL has been fixed.
3999 * The chill.lisp system for loading cold code into a running SBCL
4001 * Support for the CMU CL "scavenger hook" extension has been removed.
4002 (It was undocumented and unused in the CMU CL sources that SBCL was
4003 derived from, and stale in sbcl-0.6.1.)
4004 * Various errors in the cross-compiler type system were detected
4005 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
4006 (enabling various consistency checks). Many of them were fixed,
4007 but some hard problems remain, so the compiler is back to
4008 running without *TYPE-SYSTEM-INITIALIZED* for now.
4009 * As part of the cross-compiler type system cleanup, I implemented
4010 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
4011 * I have started adding UNCROSS calls throughout the type system
4012 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
4013 on cross-compiler input files will be able to go away, and various
4015 * CONSTANTP now returns true for quoted forms (as explicitly required
4018 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
4020 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
4021 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
4022 (MAKE-LIST -1). In the longer term, it ought to become true
4023 that declarations are assertions even at SAFETY 1. For now, it's not
4024 quite true even at SAFETY 3, but it's at least more nearly true..
4025 (Note that this change seems to increases the size of the system by
4026 O(5%) and to decrease the speed of the compiler by 20% or more.)
4027 * changed ALIEN printing to be much more abbreviated, as a short-term fix
4028 for the problem of printing dozens of lines of distracting information
4029 about low-level system machinery as part of the top stack frame
4030 on entry to the debugger when an undefined function was called.
4031 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
4032 is not reset to COMMON-LISP-USER.
4033 * Compilation of stuff related to dyncount.lisp has been made conditional
4034 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
4035 smaller. The various dyncount-related symbols have been moved into
4036 a new "SB-DYNCOUNT" package.
4037 * tty-inspect.lisp has been renamed to inspect.lisp.
4038 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
4039 feature has gone away. (When we eventually port to other flavors of
4040 libc and/or Unix, we'll try to make the differences between flavors
4041 invisible at the user level.)
4042 * Various other *FEATURES* tags, and/or their associated conditionals,
4043 have been removed if obsolescent, or given better documentation, or
4044 sometimes given more-mnemonic names.
4046 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
4048 * tidied up "make.sh" script
4049 * tidied up system directory structure
4050 * better "clean.sh" behavior
4051 * added doc/FOR-CMUCL-DEVELOPERS
4052 * many many small tweaks to output format, e.g. removing possibly-confusing
4053 trailing #\. character in DESCRIBE-INSTANCE
4054 * (EQUALP #\A 'A) no longer signals an error.
4055 * new hashing code, including EQUALP hashing
4056 * tidied up Lisp initialization and toplevel
4057 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
4058 * command line argument processing
4059 * added POSIX-GETENV function to deal with Unix-ish environment variables
4060 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
4061 terminating SBCL on EOF
4062 * non-verbose GC by default
4063 * There is no more "sbcl" shell script; the sbcl file is now the C
4064 runtime executable (just like CMU CL).
4065 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
4067 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
4068 symbol and package information as Lisp native symbol and package objects
4069 instead of strings naming symbols and strings naming packages. This way,
4070 whenever packages are renamed (as in warm init), debug information is
4071 transformed along with everything else.
4072 * tweaked the optimization policy declarations which control the building
4073 of SBCL itself. Now, among other things, the system no longer saves
4074 source location debugging information. (This helps two problems at once
4075 by reducing SBCL size and by keeping SBCL from trying to look for its
4076 sources -- which may not exist -- when reporting errors.)
4077 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
4078 debugging and testing purposes
4079 * cleaned up printing, making the printer call PRINT-OBJECT for
4080 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
4081 methods, giving nearly-ANSI behavior
4082 * converted almost all special variables to use *FOO* naming convention
4083 * deleted PARSE-TIME functionality, since it can be done portably
4084 * moved some files out of cold init into warm init
4085 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
4087 * regularized formatting of source files
4088 * added an install.sh script
4089 * fixed ridiculous memory usage of cross-compiler by making
4090 compiler/alloc.lisp not try to do pooling unless it can hook
4091 itself into the GC of the cross-compilation host. Now the system
4092 builds nicely on my old laptop.
4093 * added :SB-ALLOC in target-features.lisp-expr
4094 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
4095 was not implemented)
4096 * re-did condition handling and note reporting in the compiler. Notes
4097 are no longer handled by signalling conditions. Style warnings
4098 and warnings are handled more correctly and reported in such a way
4099 that it's easy to find one or the other in your output (so that you
4100 can e.g. figure out which of many problems caused COMPILE-FILE to
4102 * changed the severity of several compiler warnings from full WARNING
4103 to STYLE-WARNING in order to conform with the ANSI spec; also changed
4104 compiler note reporting so that it doesn't use the condition system
4105 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
4106 COMPILE-FILE command)
4107 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
4108 function. As a consequence, START-BLOCK and END-BLOCK declarations are
4109 no longer supported, since their implementation was deeply intertwingled
4110 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
4111 * removed bogus "support" for compiler macros named (SETF FOO), and
4112 removed the compiler macro for SETF INFO (but only after making a fool
4113 of myself on the cmucl-imp mailing list by posting a bogus patch for
4114 DEFINE-COMPILER-MACRO..)
4115 * Compiled files containing forms which have side effects on the Lisp
4116 reader (such as DEFPACKAGE forms) are now handled more correctly.
4117 (Compiler queuing of top level lambdas has been suppressed by setting
4118 *TOP-LEVEL-LAMBDA-MAX* to 0. )
4119 * deleted various currently-unused source files, e.g. gengc.lisp. They
4120 may be added back at some point e.g. when porting to other architectures,
4121 but until they are it's distracting to distribute them and to try to
4123 * deleted "UNCROSS couldn't recurse through.." style warnings, since
4124 there were so many of them they're just distractions, and UNCROSS is
4125 known to be able to handle the current sources
4126 * moved PROFILE functionality into TRACE, so that it will be clear
4127 how the wrapping and unwrapping of functions when you profile them
4128 interacts with the wrapping and unwrapping of functions when you
4129 trace them. (Actually, the functionality isn't there yet, but at least
4130 the interface specification is there. Hopefully, the functionality will
4131 arrive with some future maintenance release.)
4132 * removed host-oops.lisp
4133 * changed signature of QUIT function to allow UNIX-CODE argument
4134 * fixed READ-SEQUENCE bug
4135 * tweaked verbose GC output so that it looks more like the progress
4136 output that ANSI specifies for functions like LOAD
4137 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
4138 * added <http://sbcl.sourceforge.com> to the banner information printed by