1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.17 (0.9.99?) relative to sbcl-0.9.16:
3 * incompatible change: External-format support for FFI calls. The
4 SB-ALIEN:C-STRING no longer implies an ASCII
5 external-format. Instead the string is subject to external-format
6 conversion. Additionally return values of type SB-ALIEN:C-STRING
7 are by default of type (SIMPLE-ARRAY CHARACTER), not
8 SIMPLE-BASE-STRING. For an alien type that behaves like the old
9 SB-ALIEN:C-STRING type, use (SB-ALIEN:C-STRING :EXTERNAL-FORMAT
10 :ASCII :ELEMENT-TYPE BASE-CHAR). (thanks to Yaroslav Kavenchuk)
11 * incompatible change: SB-EXT package no longer contains the
12 following unused symbols: *GC-NOTIFY-AFTER*, *GC-NOTIFY-BEFORE*,
13 *GC-NOTIFY-STREAM*, *ERROR-PRINT-LENGTH*, *ERROR-PRINT-LEVEL*,
15 * minor incompatible change: the direct superclasses of
16 SB-MOP:FUNCALLABLE-STANDARD-OBJECT are (FUNCTION STANDARD-OBJECT),
17 not (STANDARD-OBJECT FUNCTION). This makes the
18 class-precedence-lists of GENERIC-FUNCTION and
19 STANDARD-GENERIC-FUNCTION comply with the requirement of ANSI
21 * bug fix: ENOUGH-NAMESTRING on pathnames with no name and a pattern
23 * bug fix: loading of default sysinit file works. (thanks to Leonid
25 * bug fix: better detection of circularities in the file-compiler.
26 (reported by Marco Monteiro)
27 * bug fix: the CL pathname functions now work with files that have
28 non-ASCII characters in their names (thanks to Yaroslav Kavenchuk)
29 * bug fix: The :PTY argument for RUN-PROGRAM will now work on
30 systems with Unix98 pty semantics.
31 * bug fix: ASDF-INSTALL will now work with bsd tar.
33 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
34 * feature: implemented the READER-METHOD-CLASS and
35 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
37 * incompatible change: variable SB-EXT:*USE-IMPLEMENTATION-TYPES*
39 * optimization: faster LOGCOUNT implementation on x86 and x86-64
40 (thanks to Lutz Euler)
41 * optimization: hashing of general arrays and vectors has been
42 improved. (reported by Any Fingerhut)
43 * enhancement: SB-INTROSPECT is now able to find definitions of
44 profiled functions. (thanks to Troels Henriksen)
45 * enhancement: compiler-macro expansion applies now to FUNCALL forms
46 as well, allowing compiler-macros for SETF-functions to expand.
47 * enhancement: step-instrumentation no longer wraps known
48 single-value functions with multiple-value context, allowing
49 better type inference.
50 * fixed bug #337: use of MAKE-METHOD in method combination now works
51 even in the presence of user-defined method classes. (reported by
52 Bruno Haible and Pascal Costanza)
53 * fixed bug #339(c): if there are applicable methods not part of any
54 long-form method-combination group, call INVALID-METHOD-ERROR.
55 (reported by Bruno Haible)
56 * fixed bug #361: the :FUNCTION initarg in the protocol for
57 initialization of methods can now be used to override
58 internally-produced optimized functions. (reported by Bruno
60 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
61 system-provided lambda expression no longer cause warnings about
62 unbound #:|pv-table| symbols.
63 * bug fix: improved the handling of type declarations and the
64 detection of violations for keyword arguments with non-constant
66 * bug fix: potentially erronous calls to PATHNAME and
67 MERGE-PATHNAMES were being flushed in some cases.
68 (reported by Richard Kreuter)
69 * bug fix: compiled calls to TYPEP were mishandling obsolete
70 instances. (reported by James Bielman and Attila Lendvai)
71 * bug fix: format strings with ~> without matching ~< no longer
72 trigger an AVER, but signal an understandable error instead.
73 (reported by Antonio Martinez)
74 * bug fix: specifying an output-file without a pathname-type for
75 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
76 of input-file instead of "fasl". (reported by Robert Dodier)
77 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
78 (reported by James Y Knight).
79 * bug fix: compiler-macros-function did not consider the environment
80 argument for shadowing by local functions.
81 * bug fix: compiler-macros expansion was inhibited by local INLINE
83 * bug fix: inline expansions of known functions were subject to
84 step-instrumentation in high debug policies, leading to problems
86 * bug fix: compiler failed to differentiate between different CONS
88 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
89 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
90 * bug fix: #368: incorrect use of expressed vs. upgraded array
92 * bug fix: #306a: more precise unions of array types.
93 * thread-safety improvements:
94 ** CONDITION-WAIT could return early on Linux, if the thread was
95 interrupted and subsequently continued with SIGCONT.
96 ** STABLE-SORT and ADJUST-ARRAY were not reentrant.
98 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
99 * added support for the ucs-2 external format. (contributed by Ivan
101 * minor incompatible change: pretty printing of objects of type
102 (cons symbol) is, in the default pprint-dispatch-table, now
103 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
105 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
106 called later in a class's lifetime, possibly as late as when the
107 first instance of the class is created. Previously,
108 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
109 class became finalizeable.
110 * fixed bug: FILE-POSITION sometimes returned inconsistent results
111 for multibyte external-format streams. (thanks to "vbzoli")
112 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
113 with :ALLOCATION :CLASS inherited from superclasses of the
115 * fixed bug: anonymous classes can now be created using the :NAME
116 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
117 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
118 * fixed bug: core-files saved with :EXECUTABLE T can again be
119 executed when SBCL_HOME isn't set. (reported by James Knight)
120 * fixed bug: toplevel LOCALLY forms with declarations could
121 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
122 * fixed bug: printing from several different threads using different
123 values of *print-case* could cause invalid output, due to
124 some internal special variables of the printer not being bound
125 thread-locally (reported by Max Mikhanosha)
126 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
127 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
128 (reported by Pascal Costanza)
129 * fixed bug: FINALIZE-INHERITANCE is called from
130 REINITIALIZE-INSTANCE on classes when the class has previously
131 been finalized, as required by AMOP.
132 * minor code generation optimizations:
133 ** better register allocation in CLOS dispatching functions
134 ** overflow detection when coercing signed bytes to fixnums on x86-64
135 ** is now implemented with one IMUL instruction instead of three shifts
136 ** more efficient bit-vector access on x86 and x86-64
137 ** more efficient access to raw structure slots on x86 and x86-64
138 * fixed some bugs revealed by Paul Dietz' test suite:
139 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
142 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
143 * feature: thread support on Solaris/x86, and experimental thread support
145 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
147 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
148 default method for SLOT-UNBOUND.
149 * minor incompatible change: prevent the user from specializing the
150 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
151 somewhat counter to the intent of the protocol, I (CSR) think, and
152 additionally it just doesn't work in SBCL as currently
153 implemented, thanks to optimizations (that are always valid for
154 the other three SLOT-VALUEish functions, but not for the setter).
155 * minor incompatible change: the :SB-LDB feature is now enabled by
156 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
157 the low-level debugger.
158 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
159 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
160 down to an SB-SYS:FD-STREAM, allowing direct communication in
161 more cases. Temporary files and pipes are still used as a fallback
163 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
164 on the non-reentrant C functions localtime(3) and gmtime(3).
165 * bug fix: global optimization policy was not visible in LOCALLY and
167 * bug fix: class objects can be used as specializers in methods.
168 (reported by Pascal Costanza)
169 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
170 with a type error. (reported by blitz_ on #lisp)
171 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
172 failed with a type error. (reported by Pascal Bourguignon)
173 * bug fix: merging pathnames against defaults with :DIRECTORY
174 starting with '(:RELATIVE :BACK) should preserve the :BACK.
175 (reported by James Y Knight)
176 * bug fix: saving large (>2GB) cores on x86-64 now works
177 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
178 constant index and a (simple-array (signed-byte 32)) array
179 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
180 error (patch by Robert J. Macomber)
181 * feature: TIME now displays the amount of run-time spent in GC
182 * bug fix: The debugger now does a better job of respecting
183 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
184 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
185 workaround for bug 403.)
186 * bug fix: floating point exception handling now works on FreeBSD
187 (thanks to NIIMI Satoshi)
188 * fixed some bugs revealed by Paul Dietz' test suite:
189 ** MISC.641: LET-conversion were not supposed to work in late
191 * improvements to the Win32/x86 port:
192 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
193 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
195 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
196 * new feature: source path information is generated for macro-expansion
197 errors for use in IDE's like Slime (thanks to Helmut Eller)
198 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
199 * bug fix: compiler does not loop forever on an invalid type in
201 * improvement: compilation of most CLOS applications is significantly
203 * optimization: added a limited bytecode compiler for simple toplevel
204 forms, speeding up compilation and FASL loading
205 * bug fix: the statistical profiler now properly distinguishes anonymous
208 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
209 * minor incompatible change: in sbcl-0.9.11 (but not earlier
210 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
211 into subclasses of the finalized class. Now user calls to
212 FINALIZE-INHERITANCE finalize just the one class, and calls by the
213 system return before any subclasses are finalized.
214 * minor incompatible change: The reader no longer ignores errors
215 regarding non-existent packages in #+ and #- feature tests.
216 * new feature: command line options --no-sysinit, --no-userinit to
217 inhibit loading the corresponding init files
218 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
219 for sending data through UDP sockets (thanks to François-René Rideau)
220 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
221 error is signalled instead (thanks to François-René Rideau)
222 * minor incompatible change: Error signalling behaviour of lexical
223 operations violating package locks has changed slightly. Refer to
224 documentation on package locks for details.
225 * enhancement: EVAL can process IF-expressions without resorting to the
227 * bug fix: Some operations involving SIN, COS, or TAN weren't being
228 constant-folded properly due to missing out-of-line functions.
229 (reported by Mika Pihlajamäki)
230 * bug fix: LISTEN sometimes returned T even in cases where no data was
231 immediately available from the stream
232 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
233 were reversed. (reported by Levente Mészáros)
234 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
235 when given a vector displaced to an adjustable array. (reported
237 * bug fix: derivation of float boundaries from numbers outside the
238 appropriate float range (reported by John Wiseman)
239 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
241 * bug fix: precomputing cache entries for generic functions with
242 some subclasses of specializers as yet invalid does not attempt to
243 fill a cache line with a negative offset. (reported by Levente
245 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
246 allow condition class objects as arguments to DOCUMENTATION and
247 (SETF DOCUMENTATION); only find and set documentation for
248 structure names for the STRUCTURE doc-type. (suggested by Gary
250 * improvements to the Win32/x86 port:
251 ** (user-homedir-pathname) and default initialization file
252 locations now know about the user's "Documents and Settings"
253 directory (thanks to Yaroslav Kavenchuk)
254 ** run-program is implemented (thanks to Mike Thomas)
255 ** sockets support (thanks to Timothy Ritchey)
256 ** better backtrace support (thanks to Alastair Bridgewater)
257 ** sb-grovel supported
258 ** asdf-install and sb-posix work somewhat
259 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
260 * improvements to the Solaris/x86 port:
261 ** works on Solaris 11/Solaris Express
262 ** floating-point exception handling support
263 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
264 * fixed some bugs revealed by Paul Dietz' test suite:
265 ** REMOVE-METHOD returns its generic function argument even when
266 the method is not one of the generic functions' methods.
267 ** objects declared as MEMBER types can be admissible arguments to
269 ** printing characters should simply be printed by the FORMAT ~:C
271 ** compiler failure when compiling functions with hairy constant
272 defaults for optional parameters.
273 ** compiler produces wrong code when MAYBE-INLINE-expanding a
274 function, which is already optimized.
276 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
277 * new platform: experimental support for SBCL x86/Darwin, including
278 MacOS X 10.4.5 on Intel.
279 * new feature: Unicode character names are now known to the system
280 (through CHAR-NAME and NAME-CHAR).
281 * new feature: the filesystem location of SBCL's core file is
282 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
283 * minor incompatible change: the contrib modules SB-POSIX and
284 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
285 this change is to make it easier to distribute
286 location-independent binaries.
287 * bug fix: as implied by AMOP, standardized classes no longer have
288 slots named by external symbols of public packages. (reported by
290 * optimization: calling functions via constant symbols -- as in
291 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
292 via the function object as in (FUNCALL #'FOO).
293 * optimization: on x86 and x86-64, the calling convention for the
294 case of dealing with an unknown number of values has been altered
295 to be friendlier to the prediction heuristics implemented,
296 particularly with reference to CALL and RET pairing. (thanks to
297 Alastair Bridgewater)
298 * optimization: on x86, the code for access to thread-local symbol
299 values and binding/unbinding of thread-local symbols is smaller.
300 (thanks to Alastair Bridgewater)
301 * enhancement: CONSTANTP is now able to determine constantness of
302 more complex forms, including calls to constant-foldable standardized
303 functions and some special forms beyond QUOTE.
304 * fixed bug: occasional GC crashes on Solaris/x86
305 * optimization: x86-64 supports stack allocation of results of simple
306 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
307 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
308 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
309 (thanks to James Knight)
310 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
311 by the standard (thanks to Stephen Compall)
313 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
314 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
315 be used for bundling the runtime and the core file into one
316 executable binary. This feature is not currently supported on all SBCL
317 platforms. (thanks to James Bielman and NIIMI Satoshi)
318 * new feature: a generational or ephemeral garbage collector is now
319 the default on the PowerPC platform (both Linux and Darwin). The
320 old Cheney (stop and copy) collector is a build-time option.
321 (thanks to Cyrus Harmon, after Raymond Toy)
322 * minor incompatible change: the method by which SBCL finds its
323 contributed modules has changed; it no longer relies on symbolic
324 links from an $SBCL_HOME/systems directory, but searches directly
325 in the subdirectories of $SBCL_HOME.
326 * enhancement: the dynamic heap size on the Linux/PPC platform is
327 markedly larger, even using the older Cheney garbage collector.
328 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
329 certain complicated string types. (reported by Gary King)
330 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
331 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
332 * fixed bug: DOCUMENTATION on structures no longer signals an
333 error if no structure type of the right name exists. (reported by
335 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
336 applicable non-standard primary or :AROUND method on
337 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
338 no longer cause unbound variable errors. (reported by Kevin Reid)
339 * fixed bug: metacircle resolution in cases where methods have slots
340 added before the slots from STANDARD-METHOD. (reported by Jean
342 * fixed bug: the Power PC assembly code for calling into lisp
343 sometimes computed its return address wrongly (depending on the
344 15th bit of the address, fixed at link time). This used to
345 manifest itself in a segmentation violation while building PCL.
346 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
347 many others over the years)
348 * fixed bug: ROOM no longer reports silly numbers for stack usage if
349 the stack pointer (treated as a signed integer) is negative.
350 (thanks to Peter van Eynde)
352 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
353 * new platform: experimental support for the Windows operating
354 system has been added. (thanks to Alastair Bridgewater)
355 * fixed several bugs in and robustified the PPC FFI (including
356 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
357 * bug fix: allow non-simple string symbol names (reported by Paul
359 * bug fix: interrupt handling on NetBSD (thanks to Richard M
361 * bug fix: saving a core corrupted callbacks on x86/x86-64
362 * bug fix: closed a loophole in metacircularity detection and
363 grounding in the PCL implementation of CLOS.
364 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
365 (reported by tomppa on #lisp)
366 * bug fix: clear the direction flag on Lisp -> C transitions, as
367 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
368 (reported by Andrej Grozin and Hendrik Maryns)
369 * optimization: major improvements to GC efficiency on GENCGC platforms
370 * optimization: faster implementation of EQUAL
371 * optimization: emit more efficient opcodes for some common
372 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
373 * optimization: several other minor code-generation improvements on x86-64
374 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
376 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
377 * minor incompatible change: (SETF CLASS-NAME) and (SETF
378 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
379 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
380 by ANSI as a generic function, and for consistency (SETF
381 GENERIC-FUNCTION-NAME) is treated likewise.
382 * fixed bug #233.b: make constraint propagation notice when a variable
383 value is changed after it is referenced but before it is used
384 * fixed bug #296: no more arbitrary behaviour from filenames with
385 odd characters as --load arguments. (reported by Adam Warner)
386 * fixed bug #390: :CHARACTER-SET pathname components now work as
387 expected. (reported by Tim Daly Jr)
388 * fixed bug #391: complicated :TYPE intersections in slot
389 definitions no longer cause an error in PCL internals.
390 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
391 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
392 returning the number of octets which would be written to the
393 file-stream. (thanks to Robert J. Macomber)
394 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
395 arguments to the internal stream functions. (thanks to David
397 * bug fix: stack exhaustion is now handled on the x86/SunOS
398 platform. (thanks to Daisuke Homma)
399 * bug fix: more accurate ROOM results on GENCGC platforms
400 * optimization: improved type inference for arithmetic-for
401 index variables in LOOP
402 * optimization: faster floating-point SQRT on x86-64
403 * fixed some bugs revealed by Paul Dietz' test suite:
404 ** DOCUMENTATION returns NIL instead of "" for method combinations
405 that don't have a docstring
407 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
408 * minor incompatible change: (SETF CLASS-NAME) and (SETF
409 GENERIC-FUNCTION-NAME) are no longer generic functions, and
410 therefore are not customizeable by user code (as seems to be at
411 least permitted and maybe required by AMOP). As a consolation,
412 however, the SBCL implementation of these functions now calls
413 REINITIALIZE-INSTANCE as specified by AMOP.
414 * bug fix: slot-definition documentation is propagated as per ANSI
415 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
416 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
417 argument, as do eslotd initialization methods. (from Pascal
418 Costanza's "Closer" project)
419 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
420 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
422 * bug fix: it is now possible to have more than one subclass of
423 STANDARD-GENERIC-FUNCTION without causing stack overflow.
424 (reported by Bruno Haible, Pascal Costanza and others)
425 * bug fix: the dependent update protocol now works for generic
426 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
428 * bug fix: condition-class instances corresponding to
429 DEFINE-CONDITION forms are now created eagerly. (reported by
430 Kalle Olavi Niemitalo on comp.lang.lisp)
431 * bug fix: floating point printing is more accurate in some
432 circumstances. (thanks to Simon Alexander)
433 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
434 merged with *DEFAULT-PATHNAME-DEFAULTS*.
435 * bug fix: callbacks on OS X now preserve stack-alignment.
436 (thanks to Cyrus Harmon)
437 * enhancement: the x86-64 disassembler is much better at
438 disassembling SSE instructions. (thanks to Lutz Euler)
439 * enhancement: CHECK-TYPE on a variable now allows type-inference
440 (thanks to Vincent Arkesteijn)
441 * optimization: improved performance of EUC-JP external format.
442 (thanks to NIIMI Satoshi)
443 * optimization: performance improvements to IO on file streams of
444 :ELEMENT-TYPE CHARACTER
445 * optimization: much faster memory allocation on x86-64
446 * optimization: faster garbage collections (latency and throughput)
448 * optimization: faster bignum subtraction and division on x86 and x86-64
449 (thanks to Lutz Euler)
450 * optimization: more accurate type-inference for loops that use a
451 floating point index variable or a negative step.
453 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
454 * bug fix: add a workaround to SBCL looping infinitely at startup on
455 Linux kernels with apparently buggy implementations of personality().
456 (thanks to Svein Ove Aas)
457 * bug fix: Unicode symbols are correctly printed in LDB backtraces
458 (thanks to David Lichteblau)
459 * bug fix: local bindings shadow global symbol macros for
460 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
461 * optimization: non-open coded uses of numeric comparison operators
462 (e.g. >) no longer cons when called with more than one parameter
463 on platforms supporting dynamic-extent allocation.
464 * enhancement: saving cores with foreign code loaded is now supported
465 on MIPS/Linux in addition to the previously supported platforms.
466 * bug fix: FIND-RESTART now tests for activity, not applicability when given
467 a restart object as identifier. (reported by Helmut Eller for CMUCL)
468 * bug fix: division by zero in sb-sprof when no samples were collected
469 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
470 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
471 (thanks to Alexander Kjeldaas)
472 * new feature: ignoring the return values of destructive functions
473 like NREVERSE signals a compile-time style-warning.
474 (thanks to Kevin Reid)
475 * new feature: the alignment of alien structure fields can be
476 explicitly specified. (thanks to Cyrus Harmon)
477 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
478 and dump core on SIGQUIT
480 ** incompatible change: threads do not inherit values of specials
481 from their parents (see manual)
482 ** bug fix: threads stacks belonging to dead threads are freed by the
483 next exiting thread, no need to gc to collect thread stacks anymore
484 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
485 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
486 ** bug fix: dynamic variable and thread start related gc lossage
487 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
489 * fixed some bugs revealed by Paul Dietz' test suite:
490 ** SUBTYPEP is slightly more accurate on heinously complicated
491 CONS types where some of the members have uncertain (in the
492 NIL, NIL sense) type relationships to each other.
494 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
495 no longer purified unless :PURIFY T is explicitly specified.
496 ** Non-purified cores are significantly smaller than before
498 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
499 * new feature: timers based on Zach Beane's excellent timer package
500 * added support for the following external formats: koi8-u,
501 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
502 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
503 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
504 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
505 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
506 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
507 * incompatible change: a threaded SBCL will no longer revert to
508 non-threaded mode on non-NPTL systems, but refuse to start entirely.
509 * bug fix: interrupts are disabled until startup is complete; no
510 more sigsegvs when receiving a signal to soon
511 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
513 * bug fix: add a workaround for the memory randomization features in
514 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
515 workaround will only be in effect on systems with the proc filesystem
517 * bug fix: printing objects of type HASH-TABLE signals a
518 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
520 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
521 have been running for over 50 days. (reported by Gilbert Baumann)
522 * bug fix: the logic for getting names of functions gets less
523 confused when confronded with alternate-metaclass
524 funcallable-instances. (reported by Cyrus Harmon)
525 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
526 consistent, even on internal alternate-metaclass objects.
527 * bug fix: flush closure information collected by physical
528 environment analysis prepass before the main pass. (bug reported
530 * bug fix: compiler pack phase does not modify a hashtable, which is
531 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
532 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
533 may be expanded. (reported by Alexander <malishev> on c.l.l)
534 * bug fix: no more style-warnings when compiling code using LAST
535 under high-SPEED low-SPACE settings. (reported by David Wallin
537 * fixed some incompatibilities between SBCL's MOP and the MOP
539 ** the METAOBJECT class is now implemented;
540 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
541 STANDARD-OBJECT, as required;
542 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
543 now compatible (as required by VALIDATE-SUPERCLASS); there
544 remains a consistency requirement over the presence or absence
545 of the FUNCTION class in the superclass list of finalized
546 classes; see the manual for more details;
547 ** the :AROUND method for COMPUTE-SLOTS on
548 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
549 requested slot ordering.
551 ** bug fix: parent thread now can be gc'ed even with a live
553 ** bug fix: binding a special with PROGV to no value is not
554 the same as that symbol not having been bound (thanks to
556 ** bug fix: binding specials is thread safe (thanks to
558 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
559 and SB-SPROF do not die with 'no handler for signal XX in
560 interrupt_handle_now(..)' anymore
561 ** bug fix: WITH-TIMEOUT works with multiple running threads
562 * fixed some bugs revealed by Paul Dietz' test suite:
563 ** ENSURE-GENERIC-FUNCTION should take a method class object for
564 the :method-class keyword argument.
566 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
567 * new port: the Solaris operating system on x86 processors is now
568 mostly supported, though some rough edges in the environment
569 remain. (thanks to Daisuke Homma)
570 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
571 space for its heap. (thanks to Thiemo Seufer)
572 * optimization: SBCL on MIPS platforms now supports dynamic-extent
573 closures. (thanks to Thiemo Seufer)
574 * minor incompatible change: eof selects abort in the debugger.
575 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
576 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
578 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
579 deprecated, and will go away in a future revision of SBCL.
580 * minor incompatible change: GC-ON and GC-OFF are no longer
581 implemented with a counter, it does not matter how many times gc
582 is switched on or off
583 * bug fix: discriminating functions for generic function classes
584 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
585 make invalid assumptions about method precedence order. (reported
587 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
588 platforms. (thanks to Thiemo Seufer)
589 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
590 * bug fix: correct pathname printing with printer escaping is on.
591 (thanks to Kevin Reid)
592 * bug fix: complex VOP definitions in "user-space" no longer trigger
593 package locks. (reported by Zach Beane)
594 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
595 for funcallable-instance functions, and is no different from
597 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
598 characters in the current readtable. (reported by Nicholas Neuss)
599 * bug fix: SUBTYPEP on various CONS types returns more a more
600 accurate acknowledgment of its certainty.
601 * optimizations: REMOVE-DUPLICATES now runs in linear time on
602 lists in some cases. This partially fixes bug 384.
603 * flush all standard streams before prompting in the REPL and the
605 * bug fix: signal handling and triggering gc do not conflict
606 directly anymore, in particular a high frequency sb-sprof does
607 not prevent gc from running
608 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
609 approximation for timezone and DST information between the
610 universal time epoch and the smallest negative 32-bit time_t.
611 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
612 year 1899 so long as the encoded time is non-negative.
613 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
614 the correct answer for non-constant rotate amounts when performing
615 an inline 32-bit rotation.
617 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
618 there is only one thread in the session
619 ** bug fix: memory leak for streams created in one thread and
620 written to in another
621 ** bug fix: lockup when compiled with gcc4
622 ** bug fix: race that allows the gc to be triggered when gc is
624 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
625 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
627 ** bug fix: finished threads are gc'ed properly
628 * fixed some bugs revealed by Paul Dietz' test suite:
629 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
630 has arguments with a different set of applicable methods from
631 the orignal arguments.
632 ** The type error thrown by MAP now has a correct expected-type
634 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
635 name a compiled function.
636 ** Three MISC tests where a large bignum was improperly coerced to
637 a float that couldn't represent that bignum during type
638 derivation were fixed.
639 ** SUBTYPEP can now handle types involving the negation of a
640 list-form FUNCTION type.
641 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
642 as (COMPLEX (AND RATIO (NOT FIXNUM))).
643 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
645 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
646 * New feature: Experimental support for bivalent streams: streams
647 opened with :element-type :default now allow character and binary
648 (unsigned-byte 8) I/O
649 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
650 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
652 * bug fix: no more highly sporadic "couldn't check whether ~S is
653 readable" when reading a stream and an interrupt hits in the middle
654 of a select system call
655 * compiler better recognizes complex arrays (reported by Eduardo
657 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
658 for variables on SBCL built with linkage-tables. (reported by Luis
660 * various error reporting improvements.
661 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
662 (Thanks to Thiemo Seufer)
663 * bug fix: nested reader invokations work correctly
664 * bug fix: it is possible to have simultaneous references to foreign
665 code and foreign data with the same name.
667 ** added x86-64 support
668 ** incompatible change: the threading api now works with thread
669 objects instead of thread ids
670 ** bug fix: threads are protected from signals and interruption when
671 starting up or going down
672 ** bug fix: a race where an exiting thread could lose its stack to gc
673 ** bug fix: don't halt on infinite error in threads if possible
674 ** fixed numerous gc deadlocks introduced in the pthread merge
675 ** bug fix: fixed thread safety issues in read and print
676 ** bug fix: debugger doesn't hang on session lock if interrupted at
677 an inappropriate moment
678 ** bug fix: run-program is now thread safe(r)
679 ** bug fix: inner with-recursive-lock no longer releases the mutex
680 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
681 * fixed some bugs revealed by Paul Dietz' test suite:
682 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
683 EXPECTED-TYPE slots filled.
684 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
686 ** MAKE-CONDITION accepts classes as type-designators.
687 ** COMPILE may never return NIL.
688 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
689 range before calling Unix time functions
691 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
692 * numerous signal handling fixes to increase stability
693 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
694 * minor incompatible change: we now correctly canonize default
695 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
696 as per the MOP, rather than the historical (INITARG INITFUNCTION
697 INITFORM). (reported by Bruno Haible)
698 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
699 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
700 for more information.
701 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
702 pathname is a directory pathname.
703 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
704 * dynamic space size on PPC has been increased to 768Mb. (thanks to
706 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
707 :METACLASS argument in addition to a class name. (reported by
708 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
709 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
710 (reported by Bruno Haible; patch from Gerd Moellmann)
711 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
713 * bug fix: more cleanups to the floating point exception handling on
714 x86-64 (thanks to James Knight)
715 * bug fix: the compiler does not try to constant fold calls of
716 COERCE and BIT-* functions when they return freshly constructed
717 objects. (reported by Nikodemus Siivola and Paul Dietz)
718 * optimization: improved the MIPS versions of generic (in the
719 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
720 * optimization: direct conversion of (unsigned-byte 32) to floats on
721 the PowerPC platform.
722 * optimization: structure instances with raw slots now use less
723 memory, and probably show better memory locality. (thanks to
725 * optimization: DYNAMIC-EXTENT declarations for lists and closures
726 are treated as requests for stack allocation on the x86-64,
727 Alpha-32, PPC and SPARC platforms.
728 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
729 the wrong file descriptor; implementation of SOCKET-OPEN-P.
730 (thanks to Tony Martinez)
732 ** gcing a dead thread can no longer lead to lockups
733 ** threads block signals until they are set up properly
734 ** errno is no longer shared by threads
735 ** interrupt-thread restores the eflags register on x86
736 ** fixed some lockups due to gc/thread interaction
737 * fixed some bugs revealed by Paul Dietz' test suite:
738 ** invalid dotted lists no longer raise a read error when
740 ** signal an error if a symbol that names a declaration is used
741 as the name of a type, or vice versa
742 ** allow using the (declare (typespec var*)) abbreviation for
743 (declare (type typespec var*)) with all type specifiers
744 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
745 ** always return NIL from PPRINT-POP when OBJECT is NIL
746 ** don't signal errors when pretty-printing malformed LABELS,
747 FLET or MACROLET forms
748 ** declarations in a DOLIST body are also in scope for the
750 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
752 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
755 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
756 * fixed cross-compiler leakages that prevented building a 32-bit
757 target with a 64-bit host compiler.
758 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
759 opened with :IF-EXISTS :OVERWRITE.
760 * fixed bug 281: error for an invalid qualifier in a short-form method
761 combination method is not signalled until the faulty method is called.
762 * bug fix: iteration variable type inferrer failed to deal with open
763 intervals. (reported by Alan Shields)
764 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
765 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
766 or more runtime options were provided to the sbcl binary.
767 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
768 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
769 * optimization: there's now a fast-path for fixnum arguments in the
770 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
771 * optimization: the code generated on x86-64 is more compact thanks
772 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
773 * bug fix: floating-point exception handling now partly works on
774 x86-64. (Thanks to James Knight)
775 * improvement to the MIPS backend from Thiemo Seufer: C-style
776 64-bit long long arguments and return values to alien functions
778 * fixed some bugs revealed by Paul Dietz' test suite:
779 ** the type-error signalled from WARN has a filled-in DATUM slot.
780 ** the type-error required when a stream is not associated with
781 a file has the stream as its datum.
782 ** type-errors on single-floats on x86-64 no longer have
783 :INVALID-OBJECT as the datum
784 ** the type-errors signalled for invalid function names now have
785 a correct expected type
786 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
787 for structures defined with a :TYPE. Documentation strings for
788 typed structures are no longer immediately discarded
789 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
790 broadcast streams and synonym streams. FILE-LENGTH now also works
791 on broadcast streams.
793 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
794 * incompatible change: the --noprogrammer option, deprecated since
795 version 0.7.5, has been removed. Please use the equivalent
796 --disable-debugger option instead.
797 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
799 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
800 inoperational for a while now) has been completely removed.
801 * null lexical environments are now printed as #<NULL-LEXENV>,
802 significantly reducing the amount of clutter in typical backtraces.
803 * documentation on weak pointers, finalization, and after GC hooks
804 has been added to the manual.
805 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
806 as other specialized array types, is much faster. SUBSEQ and
807 COPY-SEQ on such arrays have also been sped up.
808 * optimization: EQL is now more efficient when at least other argument
809 is known to be of type (OR FIXNUM (NOT NUMBER)).
810 * fixed bug: the runtime is now less vulnerable to changes in the
811 size of the SBCL object on OS X, and virtual memory is reserved for
812 all spaces that need to be at a fixed address.
813 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
814 * fixed bug: finalizers and after GC hooks that cause consing are now
816 * fixed bug: compiler error messages and summaries are now printed to
817 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
818 * fixed inference of the upper bound of an iteration variable.
819 (reported by Rajat Datta).
820 * fixed bug 211e: calling local functions with duplicated constant
821 keyword argument no longer causes a bogus style warning about an
823 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
824 ftype declaration to be disregarded. (reported by Dave Roberts)
825 * fixed bug 373: caused by erronous compilation of references to alien
826 variables in the runtime on ppc/darwin.
827 * fixed bug 376: CONJUGATE type deriver.
828 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
829 (reported by Baughn on #lisp)
830 * a cross-compiler bug on non-x86 platforms has been identified and
831 fixed. (thanks to Bruno Haible)
832 * improvements to the MIPS runtime code for increased stability.
833 (thanks to Thiemo Seufer)
834 * increased the maximimum compact environment size to allow
835 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
836 * partial workaround for bug 135: don't name the function we're
837 calling for hairy cases of EVAL, so as not to accumulate one environment
838 entry per call to EVAL. (reported by Kevin Reid)
839 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
840 * optimization: type testing for non-vector arrays should be faster.
841 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
843 * bug fix: setting 31st element of a bit vector to zero did not work
845 * bug fix: redefining a class definition which failed due to a
846 previous accessor / function clash now works (but see BUGS entry
847 #380 for more problems in this area). (thanks to Zach Beane)
848 * the long form of DEFINE-METHOD-COMBINATION disables method group
849 checking when given a single method group with pattern *. (thanks
851 * on x86 compiler supports stack allocation of results of simple
852 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
853 * support for building and running on Mac OS X 10.4 "Tiger" has been added
854 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
855 * fixed some bugs related to Unicode integration:
856 ** the restarts for recovering from input and output encoding
857 errors only appear when there is in fact such an error to
859 * fixed some bugs revealed by Paul Dietz' test suite:
860 ** MISC.549 and similar: late transformation of unsafe type
861 assertions into derived types caused unexpected code
863 ** SCALE-FLOAT type deriver is less wrong.
864 ** type derivers for EXP, LOG and similar functions compute result
865 types for complex arguments better.
866 ** (MISC.563) CONJUGATE type deriver works for very restricted
868 ** out-of-line type testers for character strings are available.
869 ** EQUAL compiler transform understands specialness of objects
871 ** accessing double-floats stored on the stack now works on x86-64.
872 ** debugger internals could sometimes create invalid lispobjs,
873 resulting in GC crashes.
874 ** MISC.548: type check weakening can convert required type into
876 ** initialization forms for bindings are not in scope of free special
879 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
880 * incompatible change: thread support for non-NPTL systems has
881 been removed - locking is buggy and unreliable. A threaded
882 SBCL build will now warn at startup and refuse to create
883 new threads, unless futex support is detected in the kernel
884 * incompatible change: the top level REPL now has only an ABORT
885 restart associated with it, not TOPLEVEL and ABORT as it used to.
886 TOP and TOPLEVEL are now available as debugger commands for
887 returning to the top level.
888 * incompatible change: forms evaluated in the REPL now use the
889 global optimization policy.
890 * incompatible change: user- and system-initialization files are
891 no longer processed with LOAD, but by READ and EVAL; hence the
892 global optimization policy, startup package, readtable, etc,
894 * The .fasl file format number has been incremented because of
895 various incompatible changes.
896 * internal entry point details and argument counts no longer appear
897 in backtraces unless explicitly requested by setting
898 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
899 * built-in and standard functions no longer have names like "top
900 level local call to FOO".
901 * fixed bug 32: functions defined in non-null lexical environments
902 now have more legible printed representation
903 * fixed bug 33: functions defined in non-null lexical environemnts
904 are now more amenable to inspection by INSPECT.
905 * workaround for bug 354: XEPs no longer appear in backtraces unless
906 explicitly requested.
907 * fixed bug: receiving the signal which results from attempting to
908 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
909 SIGBUS on other *BSDs) on architectures where the C stack is also
910 the Lisp stack (x86 and x86-64) from foreign code no longer leads
911 to debugger confusion or wild instability. (reported by Cheuksan
913 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
914 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
915 (reported by Lutz Euler)
916 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
917 compiled code. (reported by Paul Dietz)
918 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
919 specialized parameter to a method to something that is not TYPEP
920 the specializer is now possible.
921 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
922 face of package deletion.
923 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
924 pathname. (reported by David Tolpin on c.l.l)
925 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
926 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
927 STRING-STREAM and FILE-STREAM, respectively.
928 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
929 than previously; in addition, its test suite is now run on build.
930 (thanks to Luke Gorrie)
931 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
932 * optimization: added a immediate representation for single-floats
934 * fixed some bugs revealed by Paul Dietz' test suite:
935 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
936 ** attempting to create a package with a colliding nickname causes
937 correctable errors to be signalled.
938 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
939 ** several x86-64 backend bugs related to sign-extension of immediate
942 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
943 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
944 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
946 * fixed loading of multiply forward-referenced layouts.
947 (thanks to Cheuksan Wang)
948 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
949 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
950 (see COMPILE-FILE documentation for details of :PRINT option)
951 * fixed bugs 19 and 317: fixed-format floating point printing is
952 more accurate. This also fixes a bug reported by Adam Warner
953 related to the ~@F format directive.
954 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
956 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
957 dispatch table if the from-char is a dispatch macro character.
958 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
959 by the wordsize no longer ignore the last word. (reported by Lutz
961 * fixed bug in type checking of dynamic-extent variables. (reported
963 * optimization: sequence traversal functions use their freedom to
964 coerce function designators to functions.
965 * optimization: code with many calls to CLOS methods specialized on
966 CLOS classes has had redundant type checks removed.
967 * optimization: added declarations to speed up operations that access
968 the internal character database (for example STRING-UPCASE)
969 * optimization: comparison operations between floats and sufficiently small
970 fixnums no longer create extra rationals
971 * fixed some bugs related to Unicode integration:
972 ** portions of multibyte characters at the end of buffers for
973 character-based file input are correctly transferred to the
974 start of the buffer at the next read.
975 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
976 passing it through to OPEN.
977 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
978 argument given to internal calls to OPEN.
979 ** the built-in comment readers (introduced by character sequences
980 ";" and "#|") are more forgiving to encoding errors; they will
981 STYLE-WARN and then attempt to resync the stream at a character
982 boundary. (thanks to Teemu Kalvas)
983 * fixed some bugs in the x86-64 port:
984 ** Negative short int return values from c-calls are sign-extended
986 ** The stack is aligned to 16-bytes for c-calls, as required by
987 the ABI. (thanks to Cheuksan Wang)
988 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
989 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
990 for enums. (thanks to Vincent Arkesteijn)
991 ** Multiple small optimizations and bugfixes for floating point
993 * fixed some bugs revealed by Paul Dietz' test suite:
994 ** Space, Tab, Linefeed, Return and Page have the invalid
995 secondary constituent character trait.
996 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
998 ** WITH-INPUT-FROM-STRING should only update the index place on
1000 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
1001 works more reliably.
1002 ** Bit-array operations (BIT-AND and similar) worked incorrectly
1003 with one-dimensional arrays with fill pointers.
1004 ** TYPE-OF failed on a complex with an integer realpart and a
1006 ** compiler failure during type inference for the code of form
1007 (IF (EQL X (THE ...)) ...) (MISC.535).
1009 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
1010 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
1011 platform. The port supports 61-bit fixnums, large memory spaces
1012 and reloading shared object files.
1013 * enhancement: saving cores with foreign code loaded is now
1014 supported on ppc/Darwin in addition to the previously supported
1016 * enhancement: the statistical profiler now walks deeper into the
1017 call stack for more meaningful call-graphs and accrued time
1018 reports (x86/x86-64 only). It also now reports time spent in
1020 * enhancement: it is now possible to trace most individual methods
1021 of a generic function in addition to tracing the generic function
1023 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
1025 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
1026 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
1027 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
1028 produces an error. (thanks to Vincent Arkesteijn)
1029 * bug fix: NAMESTRING on pathnames with :WILD components in their
1030 directories works correctly. (thanks to Artem V. Andreev)
1031 * fixed bug 125: compiler preserves identity of closures. (reported
1033 * bug fixed: functions with &REST arguments sometimes failed with
1034 "Undefined function" when compiled with (DEBUG 3). (reported by
1036 * bug fix: overflow during compiling of setting element of a bit
1037 vector with constant index and value. (reported by Timmy Douglas)
1038 * build fix: fixed the dependence on *LOAD-PATHNAME* and
1039 *COMPILE-FILE-PATHNAME* being absolute pathnames.
1040 * on x86 compiler partially supports stack allocation of dynamic-extent
1042 * GO and RETURN-FROM do not check the extent of their exit points
1043 when compiled with SAFETY 0.
1044 * fixed some bugs related to Unicode integration:
1045 ** encoding and decoding errors are now much more robustly
1046 handled; it should now be possible to recover even from invalid
1047 input or output to the terminal. (thanks to Teemu Kalvas)
1048 ** provided a first cut at implementing STRING-TO-OCTETS and
1049 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1050 ** altered the SB-MD5 contributed module slightly, changing the
1051 interface just enough for it to be supportable for builds where
1052 lisp characters are not eight bits.
1053 * fixed some bugs revealed by Paul Dietz' test suite:
1054 ** the FORMATTER-generated functions for ~V[ conditionals require
1055 the correct number of arguments.
1056 ** READ-FROM-STRING returns the mandated second value when applied
1057 to displaced strings.
1058 ** the #\Rubout and #\Backspace characters are treated as invalid
1059 constituent characters by the tokenizer.
1061 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1062 * new feature: reloading changed shared object files with
1063 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1064 * new feature: references to foreign variables and functions
1065 can now be compiled and loaded before the corresponding shared
1066 object file is loaded, as long as the foreign definitions are
1067 available at runtime.
1068 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1069 supported. (thanks to Dan Debertin)
1070 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1071 just keywords. (thanks to Vincent Arkesteijn)
1072 * fixed bug #331: structure-class instances corresponding to
1073 DEFSTRUCT forms are now created eagerly.
1074 * fixed bug #345: backtraces from calls to undefined functions work
1075 on x86 as well. Related bug #61 is now also partially fixed on x86
1076 (backtraces from throws to unknown catch tags.)
1077 * bug fix: lambda-list parsing is now stricter vrt. order and number
1078 of lambda-list keywords.
1079 * bug fix: as specified by AMOP, an error is signalled if a
1080 class-option appears multiple times in a DEFCLASS form. (reported
1082 * bug fix: RANDOM can be compiled when the compiler derives the type
1083 of its numeric argument as a disjoint set of small integers.
1084 (reported by Paul Dietz)
1085 * bug fix: starting a core saved with shared objects loaded when
1086 those objects are not available no longer causes threaded SBCL to
1087 hang. (reported by Sean Ross)
1088 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1090 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1091 argument, which must always be NIL. (reported by Kalle Niemitalo)
1092 * bug fix: printing 1.0d+23 no longer results in an error.
1093 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1094 * bug fix: structure slot setters preserve evaluation order. (thanks
1096 * fixed some bugs related to Unicode integration:
1097 ** RUN-PROGRAM can allow its child to take input from a Lisp
1098 stream. (reported by Stefan Scholl)
1099 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1100 locales. (reported by Ken Causey)
1101 * fixed some bugs revealed by Paul Dietz' test suite:
1102 ** INCF, DECF and REMF evaluate their place form as specified in
1104 ** FORMATTER expands ~{ ~} iteration directives with V or #
1105 parameters correctly.
1106 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1107 consequent uses no arguments correctly.
1108 ** the system has a partial understanding of the (COMPLEX RATIO)
1110 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1113 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1114 * new feature: a build-time option (controlled by the :SB-UNICODE
1115 keyword feature, enabled by default) for building the system with
1116 support for the entire 21-bit character space defined by the
1118 * new feature: the system now has rudimentary external-format
1119 support; the primary user-visible change at this time is that
1120 characters with the high bit set (such as lower-case-e-acute) will
1121 print correctly to a terminal in a UTF-8 environment.
1122 * minor incompatible change: BASE-CHAR no longer names a class;
1123 however, CHARACTER continues to do so, as required by ANSI.
1124 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1125 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1126 has been moved to the SB-EXT package (temporarily re-exported from
1128 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1129 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1131 * fixed bug #351: better error-handlind and reporting for malformed
1133 * fixed bug #350: bignum-printing is now more memory-efficient,
1134 allowing printing of very large bignums, eg. (expt 2 10000000).
1135 (reported by Bruno Haible)
1136 * fixed bug #302: better primitive-type selection for intersection
1138 * fixed bug #308: non-graphic characters now all have names, as
1139 required. (reported by Bruno Haible)
1140 * bug fix: redefining a class with different superclasses now correctly
1141 removes it from the direct-subclasses of its previous superclasses.
1142 (reported by David Morse)
1143 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1144 new value now works. (reported by Bruno Haible)
1145 * bug fix: correct canonicalization of multiple non-standard slot
1146 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1147 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1148 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1149 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1150 now exists, an signals an error.
1151 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1152 during macroexpansion for non-iteration variables as well. (reported
1153 by Bruno Haible for CMUCL)
1154 * bug fix: Cyclic structures and unprintable objects in compiler
1155 messages no longer cause errors. (reported by Bruno Haible)
1156 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1157 and READ-FROM-STRING all now return a primary value of NIL if
1158 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1159 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1160 (reported by Bruno Haible for CMUCL)
1161 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1162 do not designate a condition. (reported by Bruno Haible for
1164 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1165 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1166 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1167 permitting resolution in favour of any of the conflicting symbols.
1168 (reported by Bruno Haible for CMUCL)
1169 * bug fix: EQUAL compiler optimizations is less aggressive on
1170 strings which can potentially compare true despite having distinct
1171 specialized array element types.
1172 * bug fix: unit enumerations can be defined without dividing by
1173 zero. (thanks to Vincent Arkesteijn)
1174 * FORMAT compile-time argument count checking has been enhanced.
1175 (report from Bruno Haible for CMUCL)
1176 * a partial workaround for the bug 262: the compiler does not try to
1177 inline-expand a local function doing RETURN-FROM from a deleted
1178 BLOCK. (thanks to Peter Denno for the bug report and to David
1179 Wragg for the simple test case)
1180 * fixed some bugs revealed by Paul Dietz' test suite:
1181 ** RENAME-PACKAGE allows all package designators as new package
1183 ** constraint propagation and lambda variable substitution are
1184 more cautious in dealing with partially deleted code.
1185 ** compiler sometimes lost reoptimization passes.
1186 ** CERROR, given a condition as condition designator, passes its
1187 remaining arguments to the continue format control without
1189 ** Case-altering FORMAT directives work correctly on non-ASCII
1191 ** The REAL type specifier handles bounds outside the floating
1192 point ranges without signalling FLOATING-POINT-OVERFLOW.
1193 ** Functions with IR1-transformations can create intercomponent
1194 references to global functions.
1195 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1197 ** FORMAT ~R treats a nil value for its first parameter correctly.
1199 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1200 * enhancement: saving cores with foreign code loaded is now
1201 supported on x86/NetBSD and sparc/Linux in addition to the previously
1202 supported platforms.
1203 * bug fix: on some platforms repeated installations caused multiple
1204 copies of HTML documentation to be installed -- should not happen
1205 any more. (reported by Stefan Scholl)
1206 * bug fix: parsing self-recursive alien record types multiple times
1207 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1208 original patch by Helmut Eller for CMUCL)
1209 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1210 (thanks to Richard Kreuter)
1211 * bug fix: defining classes whose accessors are methods on existing
1212 generic functions in other (locked) packages no longer signals
1213 bogus package lock violations. (reported by François-René Rideau)
1214 * bug fix: special variables as DEFMETHOD parameters no longer have
1215 associated bogus type declarations. (reported by David Wragg and
1217 * bug fix: read-write consistency on streams of element-type
1218 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1219 * bug fix: redefiniton of the only method of a generic function with
1220 no DEFGENERIC no longer emits a full WARNING. In addition,
1221 redefinition of generic functions with no DEFGENERIC to an
1222 incompatible lambda list now signals an error. (thanks to Zach
1224 * bug fix: DEFGENERIC now works even when there's a function of the
1225 same name in an enclosing lexical environment. (thanks to Zach
1227 * fixed compiler failure, caused by instrumenting code during
1228 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1229 * optimization: added loop analysis and improved register allocation
1230 * fixed some bugs revealed by Paul Dietz' test suite:
1231 ** POSITION on displaced vectors with non-zero displacement
1232 returns the right answer.
1233 ** (SIMPLE-STRING) is a valid type specifier for sequence
1235 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1237 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1238 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1240 ** The justification version of the FORMAT ~< directive treats
1241 non-zero minpad parameter correctly.
1243 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1244 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1245 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1246 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1247 the supported interface.
1248 * new feature: Single-stepping of code compiled with DEBUG 2 or
1249 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1250 * new feature: saving cores with foreign code loaded is now
1251 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1252 Timothy Moore's work for CMUCL)
1253 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1254 parameters to * instead of NIL if no initform is supplied.
1255 (reported by Johan Bockgaard)
1256 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1257 correctly when FUNCALL appears as the car of the form. Note:
1258 despite this FUNCALL forms are not currently subject to
1259 compiler-macro expansion. (port of Raymond Toy's fix for the
1260 same from CMUCL, reported by Johan Bockgaard)
1261 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1262 (thanks for Teemu Kalvas)
1263 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1264 case instead of terminating the iteration (thanks for Julian
1265 Squires, Sean Champ and Raymond Toy)
1266 * bug fix: incorrect expansion of defgeneric that caused a style
1267 warning. (thanks for Zach Beane)
1268 * bug fix: slot accessor effective method computation works properly
1269 for classes with multiple non-standard applicable methods on
1270 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1271 * on x86 compiler supports stack allocation of results of LIST and
1272 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1273 CMUCL implementation by Gerd Moellmann)
1274 * fixed some bugs revealed by Paul Dietz' test suite:
1275 ** FORMAT strings with both the ~<~:;~> form of the justification
1276 directive and pretty-printing directives cause an error.
1278 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1279 * incompatible change: the internal functions
1280 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1281 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1282 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1283 instead of the old functions.
1284 * new feature: on platforms where "dladdr" is available foreign
1285 function names now appear in backtraces. (based on Helmut Eller's
1287 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1288 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1290 * documentation: ASDF manual is now distributed with SBCL.
1291 * bug fix: SBCL can now load its contributed modules with REQUIRE
1292 even if the system-provided entries have been removed from
1293 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1295 * bug fix: backtraces involving undefined functions or assembly
1296 routines are more informative. (thanks to Brian Downing)
1297 * bug fix: mutually referent alien structures now work correctly.
1298 (reported by Rick Taube)
1299 * bug fix: structures defined by WITH-ALIEN can be referred to
1300 within other definitions in the same WITH-ALIEN.
1301 * bug fix: division operators (MOD, TRUNCATE and the like) with
1302 constant zero divisors and integer dividends no longer generate
1304 * bug fix: provide default methods for INPUT-STREAM-P and
1305 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1306 * bug fix: improve the detection and resolution of MOP metacycles
1307 (where effective method computation depends on the generic function
1308 having its effective method computed). (reported by Bruno Haible)
1309 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1310 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1311 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1312 * bug fix: pathnames with relative directory components can be
1313 represented relative to default pathnames.
1314 * optimization: in taking the GCD of bignums, reduce the two bignums
1315 to approximately the same size (using Euclid's algorithm) before
1316 applying the more sophisticated binary GCD. (thanks to Juho
1318 * optimization: COUNT on bitvectors now operates word-at-a-time.
1319 * optimization: ASH with a positive, but not necessarily constant,
1320 (leftwards) shift, when in a modular context, is compiled to a
1322 * fixed some bugs revealed by Paul Dietz' test suite:
1323 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1324 if the corresponding argument is NIL.
1325 ** FORMAT directives accepting a minpad parameter treat negative
1327 ** literal commas as character parameters to FORMAT directives are
1329 ** literal spaces directly after ~<Newline> directives within a
1330 format-logical-block (~:< ~@:>) do not induce :FILL-style
1331 conditional newlines.
1332 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1333 as specified: it no longer includes conditional newlines).
1334 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1336 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1337 :PER-LINE-PREFIX argument does not evaluate to a string.
1339 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1340 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1341 the manual for details; for now, package locks can be disabled by
1342 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1343 if no major problems are found then it is likely that they will be
1344 compiled in unconditionally.
1345 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1346 unsupported operators on all platforms. To load a shared library
1347 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1348 object file, link it into a shared library outside of SBCL and
1349 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1351 * bug workaround: ROOM T and the GC were not working together
1352 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1353 reporting facility aren't preserved. That reporting has been
1354 disabled, so now until and unless someone figures out how to make it
1355 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1356 * minor incompatible change: as threatened around sbcl-0.8.0, the
1357 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1358 The symbols are also exported from SB-PCL for backwards
1359 compatibility, but more so than before SB-PCL should be treated as
1360 an implementation-internal package.
1361 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1363 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1364 It is known as of this release not to work on the Alpha, however.
1365 * fixed bug #167: errors signalled due to illegal syntax in method
1366 bodies are now more legible.
1367 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1368 designators and can hence be used with TYPEP.
1369 * fixed bug #333: CHECK-TYPE now ensures that the type error
1370 signalled, if any, has the right object to be accessed by
1371 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1372 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1373 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1375 * fixed bug #334: programmatic addition of slots using specialized
1376 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1377 and :ALLOCATION :CLASS slots.
1378 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1379 float exponents. (rereported by Peter Seibel)
1380 * fixed a bug: #\Space (and other whitespace characters) are no
1381 longer considered to be macro characters in standard syntax by
1382 GET-MACRO-CHARACTER.
1383 * fixed bug: initialization of condition class metaobjects no longer
1384 causes an instance of the condition to be created. (reported by
1386 * fixed bug: it is now possible to have slots such that
1387 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1388 description is neither :INSTANCE nor :CLASS.
1389 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1390 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1391 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1392 non-standard SLOT-DEFINITION-ALLOCATION values.
1393 * fixed bugs: various race conditions handling exiting threads.
1394 CL-PPCRE's thread torture test now passes.
1395 * fixed arguably-a-bug: GC time proportional to number of threads in
1396 system even when most of them are idle
1397 * optimization: improved performance of BIT and SBIT on bit-vectors.
1398 * .fasl file incompatibility: The fasl file version number has
1399 been incremented because of changes associated with package locks.
1401 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1402 * minor incompatible change: the system no longer provides
1403 optimization hints (conditions which are TYPEP
1404 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1405 mode; these hints are still emitted when higher SPEED optimization
1407 * new contrib module: a sampling profiler (profiling by statistical
1408 sampling, rather than by instrumenting functions) is available as
1409 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1410 * the behaviour of the standard function ED is now customizeable by
1411 third parties through a hook variable: see ED's documentation
1412 string for information on the protocol.
1413 * the compiler no longer emits efficiency notes for (FUNCALL X)
1414 when the type of X is uncertain under default optimization
1416 * fixed bug 276: mutating a binding of a specialized parameter to a
1417 method to something that is not TYPEP the specializer is now
1419 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1420 exhibit the required behaviour on the x86 platform. (thanks to
1421 Peter van Eynde, Eric Marsden and Bruno Haible)
1422 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1423 even for difficult arguments. (reported by Peter Graves)
1424 * fixed bug 141a: the backquote printer now descends quoted
1426 * fixed another bug in backquote printing: no more destructive
1427 modification of the form's list structure. (reported by Brian
1429 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1430 move between its address being taken and the call to
1431 interrupt_thread, fixing a crashing race condition.
1432 * the SB-POSIX contrib implementation has been adjusted so that it
1433 no longer exhibits ridiculously poor performance when constructing
1434 instances corresponding to C structs.
1436 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1437 * minor incompatible change: the sb-grovel contrib now treats C
1438 structures as alien (in the sense of SB-ALIEN) objects rather than
1439 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1440 has implications for memory management of client code
1441 (sb-grovel-returned objects must now be manually managed) and for
1442 type safety (alien objects now have full types).
1443 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1444 used to control emission of compiler diagnostics, rather than the
1445 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1446 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1447 quality should be considered deprecated.
1448 * install.sh now installs the user manual as well
1449 * (not quite a new documentable feature, but worth considering in
1450 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1451 beginnings of a semantically meaningful condition hierarchy is
1452 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1454 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1455 designator as the defaults argument.
1456 * fixed bug: Displaced arrays whose displaced-to array has become
1457 too small now cause ARRAY-DIMENSION to signal an error, providing
1458 for safer bounds-checking. (reported by Bruno Haible)
1459 * fixed bug: DEFCLASS slot definitions with identical :READER and
1460 :WRITER names now signal a reasonable error. (reported by Thomas
1462 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1464 * fixed bug: Invalid slot specification errors now print correctly.
1465 (thanks to Zach Beane)
1466 * fixed bug 320: Shared to local slot value transfers in class
1467 redefinitions now happen corrently from superclasses as
1468 well. (reported by Bruno Haible)
1469 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1471 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1472 improper lists correctly. (reported by Bruno Haible)
1473 * fixed bug 313: source-transform for <fun-name> was erroneously
1474 applied to a call of a value of a variable with name <fun-name>.
1475 (reported by Antonio Menezes Leitao)
1476 * fixed bug 307: The obsolete instance protocol ensures that
1477 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1478 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1479 remove dead unknown-values globs from the middle of the stack.
1480 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1481 easier to resume long computations after using *BREAK-ON-SIGNALS*
1482 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1483 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1484 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1485 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1486 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1487 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1489 * fixed a (fairly theoretical) bug in string printing: if
1490 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1491 does not have array-element-type equal to the most general string
1493 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1494 function name. (thanks to Nikodemus Siivola)
1495 * fixed bug: compile-time format string checker failed on
1496 non-closed ~{. (reported by Thomas F Burdick)
1497 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1498 don't warn on legal loop code involving more than one aggregate
1499 boolean. (thanks to Nikodemus Siivola)
1500 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1501 in keyword position to MAKE-PACKAGE was not being reported as
1503 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1504 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1506 * fixed bugs 280 and 312: the checking for multiple definitions in a
1507 file is less likely to become confused by uses of inline
1509 * fixed bug: the #S reader macro performs the keyword coercion
1510 specified for slot names. (reported by Kalle Niemitalo)
1511 * fixed bug: lambda lists may contain symbols whose names start with
1512 & but are not lambda-list-keywords; their occurrence triggers a
1514 * fixed bug 321: define-method-combination argument lambda lists do
1515 not cause invalid code to be generated when &optional or &aux
1516 variables have default values. (reported by Bruno Haible)
1517 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1518 have CLOS classes; this appears also to have fixed reported
1519 instability in stack exhaustion detection.
1520 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1521 MUFFLE-WARNING when no associated restart is present is now
1523 * optimization: rearranged the expansion of various defining macros
1524 so that each expands into only one top-level form in a
1525 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1527 * optimization: used a previously little-used slot in symbols to
1528 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1530 * fixed some bugs revealed by Paul Dietz' test suite:
1531 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1533 ** Fixed an optimization bug related to inheritance of initforms
1534 from local to shared slots.
1535 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1536 ** CAST optimizer forgot to flush argument derived type.
1537 ** print/read consistency on floats is now orders of magnitude
1538 more likely. (thanks also to Bruno Haible for a similar report
1540 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1541 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1542 for objects of type REAL. Make it so.
1543 ** SXHASH is specified (infelicitously) to respect similarity,
1544 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1545 it so. (thanks to Markus Ziegler)
1546 ** on the Alpha, the compiler succeeds in compiling functions
1547 returning a known number of arguments greater than 63.
1548 ** fixed handling of invalid NIL arguments in keyword position
1550 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1552 ** provided a readably-printable representation for RANDOM-STATE
1554 ** ensured that pathnames, where they have a namestring, always
1555 print using #P"..." syntax.
1557 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1558 * Support for the forthcoming 2.0 version of the NetBSD kernel
1559 running on x86 hardware has been added. (thanks to Perry
1560 E. Metzger most immediately, and others for their past work)
1561 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1562 current release version; SBCL's OpenBSD support had been broken
1563 since about the time of OpenBSD's switch to ELF binary format.)
1564 * [placeholder for DX summary]
1565 ** user code with &REST lists declared dynamic-extent, under high
1566 speed or space and low safety and debug optimization policy.
1567 * The manual has been converted to Texinfo format and the debugger
1568 chapter from the cmucl manual has been added.
1569 * A facility has been added to extract documentation strings from
1570 sbcl and store them as Texinfo-formatted snippets for inclusion in
1571 the manual (via Texinfo's @include directive)
1572 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1573 the test case to Dave Roberts)
1574 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1575 pointers. (reported by Sean Ross)
1576 * bug fix: PROFILE output is printed nicely even for large numerical
1577 values. (thanks to Zach Beane)
1578 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1579 greater than 32 handle EOF correctly.
1580 * bug fix: on X86 an immediate argument of the IMUL instruction is
1581 correctly printed in disassembly. (thanks to Lutz Euler)
1582 * bug fix: class slots in redefined classes preserve their old
1583 values. (thanks to Bruno Haible and Nikodemus Siivola)
1584 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1585 succeeds. (reported by Marco Baringer)
1586 * fixed some bugs revealed by Paul Dietz' test suite:
1587 ** READ-SEQUENCE now works on ECHO-STREAMs.
1588 ** RATIONALIZE works more according to its specification. (thanks
1591 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1592 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1593 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1594 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1595 should matter to you only if you rebind the printer control
1596 variables and then find you want different bindings in the
1597 debugger than in the ordinary execution of your program.)
1598 * The runtime build system has been tweaked to support building
1599 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1600 assembler and linker. (thanks to Nikodemus Siivola)
1601 * Unbound, undefined, undeclared variables now trigger full
1602 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1603 more often programmer error than deliberate exploitation of undefined
1605 * optimization: the hash algorithm for strings has changed to one
1606 that is less vulnerable to spurious collisions. (thanks to Juho
1608 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1609 needless bounds checking. (thanks to Juho Snellman)
1610 * optimization: implemented multiplication as a modular
1611 (UNSIGNED-BYTE 32) operation on the PPC backend.
1612 * fixed some bugs revealed by Paul Dietz' test suite:
1613 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1615 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1616 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1617 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1619 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1621 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1623 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1624 their output stream on EOF from read.
1625 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1626 have been read to end-of-file.
1627 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1629 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1630 description of determination of which consecutive characters
1632 ** Printing the "Space" character with escaping on now yields "#\\ ",
1633 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1634 ** Reading floating-point numbers with *READ-BASE* set to a number
1635 less than 10 works correctly.
1636 ** Reading floating-point numbers with *READ-BASE* set to a number
1637 more than 10 works correctly.
1638 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1639 the readtable currently in effect.
1641 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1642 * minor incompatible change: parsing of namestrings on a physical
1643 (Unix) host has changed; numbers after the final #\. in a
1644 namestring are no longer interpreted as a version field. This is
1645 intented to be largely invisible to the user, except that the
1646 meaning of the namestring "*.*.*" has changed: it now refers to a
1647 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1648 should usually be replaced by
1649 (make-pathname :name :wild :type :wild :version :wild)
1650 with the added benefit that this is more likely to be portable.
1651 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1652 option to OPEN now signals an error if the file being opened
1653 exists; this may have an impact on existing code.
1654 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1655 signals received, on the PowerPC platforms (both Linux and
1656 Darwin). (thanks to Pierre Mai for pointing out the location of
1658 * several fixes on OS X: The system now builds and runs cleanly on
1659 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1660 (thanks to Brian Mastenbrook)
1661 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1662 second-resolution: integer multiples of 1/3600 between -24 and 24.
1663 (thanks to Vincent Arkesteijn)
1664 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1665 type when called with 1 argument; PEEK-CHAR checked type of
1666 PEEK-TYPE only after having read first character from a
1667 stream. (reported by Peter Graves)
1668 * bug fix: the garbage collector now has much better locality
1669 behaviour, and in particular no longer treats all memory as being
1670 exhausted when many small objects point to each other in a deeply
1672 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1673 recognized as being TYPEP their class.
1674 * bug fix: the PUSHNEW documentation string has been corrected.
1675 (thanks to Vincent Arkesteijn)
1676 * bug fix: defaulting of the value for the last of an atypically
1677 large number of multiple values being bound was not being
1678 performed correctly on the Alpha or PPC platforms
1679 * optimization: implemented multiplication as a modular
1680 (UNSIGNED-BYTE 32) operation on the x86 backend.
1681 * optimization: SEARCH on simple-base-strings can now be open-coded.
1682 (see also contrib/compiler-extras.lisp for inspiration for
1683 teaching the compiler about the Boyer-Moore algorithm).
1684 * value, returned by MAX (and MIN) called with several EQUALP, but
1685 not EQL, arguments now does not depend on compiler settings.
1686 * fixed some bugs revealed by Paul Dietz' test suite:
1687 ** in stack analysis liveness information is propagated from
1688 non-local entry points.
1689 ** pathwise CAST removing failed when the CAST node did not start
1691 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1692 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1694 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1695 host is already defined.
1696 ** RENAME-FILE works on streams instead of signalling an internal
1698 ** PEEK-CHAR uses the current readtable when determining whether
1699 or not a character is whitespace.
1700 ** MERGE-PATHNAMES handles the case when the pathname does not
1701 specify a name while the default-pathname specifies a version.
1702 ** Pathnames now stand a better chance of respecting print/read
1704 ** Attempting to use standardized file system operators with a
1705 pathname with invalid :DIRECTORY components signals a
1707 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1709 ** DIRECTORY on logical pathnames is more correct.
1710 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1711 signal a TYPE-ERROR if their argument is not a stream.
1712 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1713 designator argument does not designate a stream.
1714 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1715 examining the synonym.
1716 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1718 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1719 element-type larger than ([UN]SIGNED-BYTE 32).
1721 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1722 * When built with the :SB-FUTEX feature, threaded builds now take
1723 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1724 for faster/more reliable mutex and condition variable support.
1725 * Incompatible change (but one you probably shouldn't have been using
1726 anyway): the interface and code for arbitrating between multiple
1727 threads in the same user session has been redesigned.
1728 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1729 error when called without an explicit environment argument.
1730 (thanks to Nikodemus Siivola)
1731 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1732 confused about their position. (thanks to Adam Warner and Gerd
1734 * bug fix: Pretty printing backquoted forms with unquotations in the
1735 argument list position of various code constructs such as LAMBDA
1736 now works correctly. (reported by Paul Dietz)
1737 * bug fix: Pretty printing unquotations no longer loses all
1738 stream position information.
1739 * optimization: performance of string output streams is now less
1740 poor for multiple small sequence writes.
1741 * optimization: performance of CSUBTYPEP in the presence of complex
1742 expressions involving CONS and NOT many times has been improved.
1743 (reported by Paul Dietz)
1744 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1746 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1747 streams for the REPL, for the debugger, and so on. (thanks to
1749 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1750 * fixed some bugs revealed by Paul Dietz' test suite:
1751 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1752 is ignored if it is too small, rather than propagating through
1753 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1754 ** extremely complex negations of CONS types were not being
1755 sufficiently canonicalized, leading to inconsistencies in
1757 ** VALUES tranformer lost derived type.
1759 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1760 * fixed a bootstrapping bug: the build process no longer assumes
1761 that the various BOOLE-related constants have the same value in
1762 host and target lisps. (noted by Paul Dietz' test suite on an
1763 SBCL binary built from CLISP)
1764 * The system can now be dynamically linked on the MIPS platform,
1765 which enables dynamic loading of foreign code from Lisp. (thanks
1766 to Ralf Baechle for discussions on the MIPS ABI)
1767 * The system now records debugging information for its own source
1768 files in a filesystem-position-independent manner, relative to
1769 the "SYS" logical host.
1770 * fixed a compiler bug: MV-LET convertion did not check references
1771 to the "max args" entry point. (reported by Brian Downing)
1772 * tweaked disassembly notes to be less confident about proclaiming
1773 some instruction as an LRA. (thanks to Brian Downing)
1774 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1775 now each have their own history, command character, and other
1776 characteristics. (thanks to David Lichteblau)
1777 * fixed some bugs revealed by Paul Dietz' test suite:
1778 ** compiler failure in compiling LOGAND expressions including a
1780 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1781 shift greater than 32.
1782 ** FLUSH-DEST did not mark blocks for type check regeneration.
1783 ** HANDLER-CASE failed to accept declarations in handler clauses
1784 in some circumstances.
1786 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1787 * New code in contrib/sb-introspect (still probably not entirely
1788 stable yet) provides some support for smart Lisp development
1789 environments like SLIME.
1790 * The conditions signalled for errors occurring when loading .fasl
1791 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1792 in a way which should help ASDF recover gracefully.
1793 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1794 stuff has been cleaned up. If you code contrib/ stuff, this might
1795 affect you, and you can look at contrib/README, contrib/STANDARDS,
1796 and/or the 0.8.4.27 diff to check.
1797 * In full calls the compiler now does not generate checks for declared
1798 argument types for all arguments.
1799 * various threading fixes
1800 ** and some experimental patches which didn't make it into
1801 the main tree for this release, but which are shipped in
1802 contrib/experimental-thread.patch as a possible fix for some
1803 failures (deadlock, spinning...) in GC-intensive multithreaded
1805 * fixed PPC build problem (source code incompatibility of different
1806 library versions): added offsetof() hackery which attempts to divine
1807 where glibc maintainers put uc_mcontext today
1808 * fixed bug 282: compiler does not trust type assertions while passing
1809 arguments to a full call.
1810 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1811 &OPTIONAL VALUES type parameter.
1812 * fix bug 214: algorithm for noting rejected templates is now more
1813 similar to that of template seletion. (also reported by rydis on
1815 * fixed bug 141b: printing backquoted information readably and prettily
1816 inserts a space where necessary.
1817 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1818 (CHARACTER 10) are now reported as errors, rather than propagated
1819 as unknown types. (reported by piso on #lisp)
1820 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1821 with values NIL and :ERROR. (thanks to Milan Zamazal)
1822 * fixed bug 191c: CLOS now does proper keyword argument checking as
1823 described in CLHS 7.6.5 and 7.6.5.1.
1824 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1825 longer raise an error; further, using a list as a for-as-arithmetic
1826 counter now raises a meaningful error.
1827 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1828 now understood by sequence creation functions such as MAKE-SEQUENCE
1830 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1831 a STRING-INPUT-STREAM.
1832 * compiler enhancement: SIGNUM is now better able to derive the type
1834 * type declarations inside WITH-SLOTS are checked. (reported by
1836 * fixed some bugs revealed by Paul Dietz' test suite:
1837 ** incorrect optimization of TRUNCATE for a positive first
1838 argument and negative second.
1839 ** compiler failure in let-convertion during flushing dead code.
1840 ** compiler failure while deriving type of TRUNCATE on an
1841 interval, containing 0.
1842 ** ASH of a negative bignum by a negative bignum count now returns
1844 ** intersection of CONS types now canonicalizes properly, fixing
1845 inconsistencies in SUBTYPEP.
1847 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1848 * incompatible change: The --disable-debugger command line
1849 option now clobbers the debugger at a more fundamental
1850 level, by redefining #'INVOKE-DEBUGGER instead of by
1851 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1852 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1853 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1854 enter the debugger REPL and then suffer recursive errors
1855 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1856 while under the new system, BREAK in --disable-debugger mode
1857 terminates the system just as an unhandled error would.
1858 * fixed compiler performance when processing loops with a step >1;
1859 * bug fix: DOCUMENTATION now retrieves generic function
1860 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1861 support has been systematized, and now supports the methods
1862 specified by ANSI, along with a default method and a method for
1863 slot documentation. (reported by Nathan Froyd)
1864 * bug fix: effective methods associated with a generic function are
1865 no longer cached over a change of that generic function's method
1866 combination. (reported by Andreas Fuchs)
1867 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1868 names a non-existent directory. (thanks to Andreas Fuchs)
1869 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1870 the PPC platform, lead to stack corruption; this has been fixed.
1871 (reported by Rainer Joswig)
1872 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1873 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1874 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1875 arithmetic in general, is now much more reliable on the PPC
1877 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1878 platform now returns the right answer.
1879 * optimization: restored some effective method precomputation in
1880 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1881 precomputation is now tunable.
1882 * optimization: compiler-internal data structure use has been
1883 reviewed, and changes have been made that should improve the
1884 performance of the compiler by about 20%.
1885 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1886 simple-base-strings and simple-bit-vectors is improved.
1887 * optimization: the optimization of 32-bit logical and arithmetic
1888 functions introduced in version 0.8.3 on the x86 has been
1889 implemented on the mips, ppc and sparc platforms; an
1890 implementation of the same facility, but for 64-bit arithmetic,
1891 has been added for the alpha.
1892 * microoptimization: the compiler is better able to make use of the
1893 x86 LEA instruction for multiplication by constants.
1894 * bug fix: in some situations compiler did not report usage of
1895 generic arithmetic in (SPEED 3) policy.
1896 * bug 145b fix: compiler used wrong type specifier while converting
1897 MEMBER-types to numeric.
1898 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1900 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1901 index argument now works properly on the Alpha platform.
1902 * bug fix: floating point exception treatment on the Alpha platform
1904 * bug fix: FILE-POSITION works much better on string input and
1905 output streams. (thanks to Nikodemus Siivola)
1906 * bug fix: many threading/garbage collection symptoms sorted.
1907 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1908 might be pseudo-atomic.
1909 * internal change: Stopping for GC is now done with signals not
1910 ptrace. GC is now done in whichever thread wanted it, instead of
1912 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1914 * bug fix: result form in DO is not contained in the implicit
1916 * incompatible change: ICR structure is changed; the value part of
1917 CONTINUATION is now called LVAR; corresponding functions are
1918 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1919 * added type deriver for ISQRT (thanks to Robert E. Brown).
1920 * bug fix: better support for loading from the command line when an
1921 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1923 * fixed some bugs revealed by Paul Dietz' test suite:
1924 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1925 ** ROUND and FROUND now give the right answer when given very
1926 small float arguments.
1927 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1929 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1930 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1931 ** LCM should return a non-negative integer.
1932 ** PARSE-INTEGER returned the index of a terminator instead of the
1933 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1934 ** PARSE-INTEGER returned an incorrect index being applied to a
1936 ** LCM with two arguments of 0 returns 0 rather than signalling
1938 ** unsigned addition of a 32-bit constant with the high bit set no
1939 longer causes an internal compiler error on the x86.
1940 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1941 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1942 with negative last argument.
1943 ** byte specifiers with zero size and position no longer cause
1944 an error during type derivation.
1945 ** bignum multiplication on the Alpha platform now returns the
1947 * porting: The system now builds on SuSE AMD64, although it still
1948 generates a 32-bit binary.
1949 * .fasl file incompatibility: The fasl file version number has
1950 been incremented (because of the changes to internal compiler
1951 data structures referred to above).
1953 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1954 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1955 more accurately, on the Darwin kernel running on PowerPC hardware.
1956 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1957 * Compiler code deletion notes now signal a condition of type
1958 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1959 an associated MUFFLE-WARNING restart.
1960 * The compiler now performs limited argument count validation of
1961 constant format strings in FORMAT, and where appropriate in ERROR,
1962 CERROR and WARN. (thanks to Gerd Moellmann)
1963 * New ASDF-INSTALL contrib can be used for automatic download and
1964 installation of third-party Lisp code from CCLAN or other sites
1966 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1967 which forces another thread to execute a function supplied by the
1969 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1970 now accept and act upon their :ELEMENT-TYPE keyword argument.
1971 (reported by Martin Atzmueller, Edi Weitz)
1972 * bug fix: FILE-POSITION now accepts position designators up to
1973 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1974 is the greater. (thanks to Patrik Nordebo)
1975 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1976 Kalle Olavi Niemitalo)
1977 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1978 printable. (reported by Eric Marsden)
1979 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1980 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1981 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1982 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1983 (thanks to Stig Sandoe)
1984 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1985 UNREAD-CHAR now correctly includes the unread character in the
1986 target sequence. (thanks to Gerd Moellmann)
1987 * bug fix in threaded builds: the system can now be suspended and
1988 resumed by shell job control with minimal disruption.
1989 * bug fixes in times and timezones >2038AD
1990 * better handling of "where is GNU make?" problem in build scripts
1991 (thanks to Nikodemus Siivola)
1992 * new optimization: inside a named function any reference to a
1993 function with the same name is considered to be a self-reference;
1994 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1995 optimization quality.
1996 * new optimization on x86: logical functions and + now have
1997 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1998 used when the result is truncated to 32 bits.
1999 * VALUES declaration is partially enabled.
2000 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
2001 * fixed some bugs revealed by Paul Dietz' test suite:
2002 ** The system now obeys the constraint imposed by
2003 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
2004 types form a lattice under type intersection.
2005 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
2006 ** ASSOC now ignores NIL elements in an alist.
2007 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
2008 and (1+ MOST-POSITIVE-FIXNUM) answers.
2009 ** The addition of a method with invalid qualifiers to a generic
2010 function does not cause an error to be signalled immediately;
2011 a warning is signalled, and the error is generated only on
2012 calling the generic function.
2013 * changed .fasl file version number, in part to add type codes for
2014 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
2015 obscure ANSI requirements
2017 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
2018 * fixed bug 148: failure to inline-expand a local function left
2019 garbage, confusing the compiler.
2020 * fixed bugs 3cd: structure slot readers perform type check if the
2021 slot can have an invalid value (i.e. it is either not initialized
2022 or can be written with a less specific slot writer).
2023 * bug fix: the compiler now traps array references to elements off
2024 the end of an array; previously, the bounds checking in some
2025 circumstances could go off-by-one.
2026 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
2028 * type declarations for array element types now obey the description
2029 on the CLHS page "Declaration TYPE", as per discussions on
2030 sbcl-help around 2003-05-08. This means that a declaration
2031 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
2032 declaration, all references to BAR will be asserted or assumed
2033 (with THE, so dependent on compiler policy) to involve objects of
2034 type FOO. Note that no such declaration is implied in
2035 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
2036 * declared types of functions from the "Conditions"
2037 chapter. (reported by Paul Dietz)
2038 * bug fix: CERROR accepts a function as its first argument.
2039 * bug fix: NTH an NTHCDR accept a bignum as index
2040 arguments. (reported by Adam Warner)
2041 * optimization: character compare routines now optimize comparing
2042 against a constant character. (reported by Gilbert Baumann)
2043 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
2044 argument now works properly on the MIPS platform.
2045 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
2046 * repeated evaluation of the same DEFSTRUCT, a slot of which is
2047 declared to have a functional type, does not cause an error
2049 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2050 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2051 * fixed bug: if last continuation of a deleted block has a
2052 destination, this destination should be deleted too. (reported by
2054 * fixed a bug in the bootstrap process: the host compiler's values
2055 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2056 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2057 test case from Patrik Nordebo)
2058 * improved the ability of the disassembler on the PPC platform to
2059 provide helpful disassembly notes.
2060 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2061 the class in more cases than previously.
2062 * bug fix: FILE-POSITION now understands :START and :END for
2063 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2064 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2065 8) data. (thanks to David Lichteblau)
2066 * bug fix: it is possible to add a method to a generic function
2067 without lambda list.
2068 * bug fix: reader failed to signal END-OF-FILE inside an
2069 object representation. (reported by Nikodemus Siivola)
2070 * fixed some bugs revealed by Paul Dietz' test suite:
2071 ** LAST and [N]BUTLAST should accept a bignum.
2072 ** condition slot accessors are methods.
2073 ** (VECTOR NIL) is a subtype of STRING.
2075 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2076 * minor incompatible change: some nonsensical specialized lambda
2077 lists (used in DEFMETHOD) which were previously ignored now signal
2079 * minor incompatible change: the system is now aware of the types of
2080 variables in the COMMON-LISP package, and will signal errors for
2081 most violations of these type constraints (where previously they
2082 were silently accepted).
2083 * minor incompatible change: COMPILE-FILE now uses the freedom
2084 afforded (ANSI 3.2.2.3) to use derived function types for
2085 functions defined in the same file. This also permits the system
2086 to warn on static type mismatches and function
2087 redefinition. (Currently it does not work with high DEBUG level.)
2088 * minor incompatible change: VALUES declaration is disabled.
2089 * When issuing notes, the compiler now signals a condition of type
2090 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2091 restart for use in user handlers. It is expected that the
2092 COMPILER-NOTE condition will eventually become a condition
2093 supertype to a hierarchy of note types, which will then be
2094 handleable in a similar fashion. However, at the moment, no such
2095 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2096 but it's an implementation detail, not a classification for the
2098 * Changes in type checking closed the following bugs:
2099 ** type checking of unused values (192b, 194d, 203);
2100 ** template selection based on unsafe type assertions (192c, 236);
2101 ** type checking in branches (194bc).
2102 * A short form of VALUES type specifier has ANSI meaning (it has
2103 increased the number of situations when SBCL cannot perform type
2105 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2106 keywords or constants is permissible.
2107 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2108 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2109 classes. (thanks to Antonio Martinez)
2110 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2111 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2112 argument to a pretty-print stream. Instead, it leaves any such
2113 operations to DESCRIBE-OBJECT methods.
2114 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2116 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2117 lambda lists. (thanks to David Lichteblau)
2118 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2119 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2120 (thanks to Gerd Moellmann)
2121 * fixed bug 246: increased compilation speed of long
2122 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2124 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2126 * DEFCONSTANT now throws a condition of type
2127 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2128 constant to a non-EQL value; CONTINUE and ABORT restarts
2129 respectively change and preserve the value.
2130 * fixed bug 63: The code walker, part of the implementation of CLOS,
2131 is now better at handling symbol macros.
2132 * bug fix: There is no longer an internal implementation type named
2133 CL:LENGTH. (reported by Raymond Toy)
2134 * bug fix: In macro-like defining macros/special operators the
2135 implicit block does not enclose the lambda list.
2136 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2137 enclosed in parentheses) are not suitable as type specifiers, and
2138 their use properly signals an error now.
2139 * bug fix: An argument count mismatch for a type specifier in code
2140 being compiled no longer causes an unhandled error at compile
2141 time, but signals a compile-time warning.
2142 * fixed simple vector readable printing
2143 * bug fix: DESCRIBE takes more care over whether the class
2144 precedence list slot of a class is bound before accessing it.
2145 (reported by Markus Krummenacker)
2146 * bug fix: FORMATTER can successfully compile pretty-printer format
2147 strings which use variants of the ~* directive inside.
2148 * bug fix: SEARCH now applies its TEST predicate to the elements of
2149 the arguments in the correct order. (thanks to Wolfhard Buss)
2150 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2151 in type inference. (thanks to Robert E. Brown)
2152 * bug fix: Reading in symbols with an explicit package name of ""
2153 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2154 * fixed some bugs revealed by Paul Dietz' test suite:
2155 ** NIL is now allowed as a structure slot name.
2156 ** Arbitrary numbers, not just REALs, are allowed in certain
2157 circumstances in LOOP for-as-arithmetic clauses.
2158 ** Multiple class redefinitions before slot access no longer
2159 causes a type error.
2160 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2161 association between the name and a class.
2162 ** Generic functions with non-standard method-combination and over
2163 five methods all of which return constants no longer return NIL
2164 after the first few invocations. (thanks to Gerd Moellmann)
2165 ** CALL-NEXT-METHOD with no arguments now passes the original
2166 values of the arguments, even in the presence of assignment.
2167 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2169 ** DISASSEMBLE works with closures and funcallable instances.
2170 ** ADD-METHOD now returns the generic function, not the new method.
2171 ** FIND-METHOD signals an error if the lengths of the specializers
2172 is incompatible with the generic function, even if the ERRORP
2174 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2175 which its argument is a member.
2176 ** DEFCLASS only redefines the class named by its class-name
2177 argument if that name is the proper name of the class;
2178 otherwise, it creates a new class.
2179 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2180 of the UNBOUND-SLOT condition to the name of the slot.
2181 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2182 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2183 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2184 SLOT-MAKUNBOUND in the specified fashion.
2186 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2187 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2188 cross-compilation host. As a consequence, we can now bootstrap our
2189 way up to SBCL starting with a bare gcc toolchain and human-readable
2190 source code (first the source to CLISP, then the source to SBCL).
2191 * A contributed module containing a partial implementation of the
2192 simple-streams interface has been included. (thanks to Rudi
2194 * A contributed module implementing the RFC1321 Message Digest
2195 Algorithm, known as MD5, has been included.
2196 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2197 no longer has any effect, as the code controlled by this feature
2198 has been deleted. (As far as we know, no-one has ever built using
2199 this feature, and its semantics were confused in any case).
2200 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2201 work on conditions (as required by the ANSI specification),
2202 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2203 expected behaviour on conditions. Users should note, however,
2204 that such behaviour is not required by the ANSI specification,
2205 and so use of this behaviour may render their code unportable.
2206 * fixed some bugs revealed by Paul Dietz' test suite:
2207 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2209 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2210 ** SXHASH on condition objects no longer returns NIL.
2211 ** :ALLOCATION :CLASS slots are better treated; their values are
2212 updated on class redefinition, and initforms inherited from
2213 superclasses are applied.
2214 ** REMOVE-METHOD returns its generic function argument even when
2215 no method was removed.
2216 ** SHARED-INITIALIZE now initializes the values of the requested
2217 slots, including those with :ALLOCATION :CLASS.
2218 ** ALLOCATE-INSTANCE now works on structure classes defined via
2219 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2221 ** SLOT-EXISTS-P now works on conditions, as well as structures
2223 ** MAKE-LOAD-FORM now has the required methods on
2224 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2225 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2226 keyword, and now implements the SLOT-NAMES argument.
2227 ** methods with &OPTIONAL arguments no longer allow too many
2228 arguments to be passed in the call without error.
2229 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2230 option is consistent with the required arguments of the generic
2231 function lambda list.
2232 * bug fix: REQUIRE accepts a string designator. (Thanks to
2234 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2235 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2236 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2237 * bug fix: The compiler now checks for duplicated variables in macro
2239 * bug fix: SETQ on globals returns the correct value.
2240 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2241 not a COMPILER-ERROR (followed by some other strange error on
2242 choosing the CONTINUE restart).
2243 * bug fix: make.sh and friends are now more consistent in the way that
2244 they look for GNU "make".
2246 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2247 * experimental native threads support (on x86 Linux >=2.4 only).
2248 This is not compiled in by default: you need to add :SB-THREAD to
2249 the target features. See the "Beyond ANSI" chapter of the manual
2251 * fix for longstanding nonANSIism: The old distinction between
2252 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2253 The return value from CL:FIND-CLASS is now a CLOS class, and
2254 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2255 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2256 * An interface to the MetaObject Protocol, as described in Kiczales,
2257 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2258 MIT Press, 1991, is available from the SB-MOP package.
2259 * incompatible change: the SB-PCL package should now be considered
2260 a private implementation detail, and no longer a semi-private MOP
2262 * minor incompatible change: due to rearrangement for threads, the
2263 control stack and binding stack are now allocated at arbitrary
2264 addresses instead of being hardcoded per-port. Users affected by
2265 this probably have to be doing advanced things with shared
2266 libraries, and will know who they are.
2267 * minor incompatible change: Previously, all --eval forms used were
2268 processed with READ before any of them were processed with EVAL.
2269 Now each --eval form is processed with both READ and EVAL before
2270 the next --eval form is processed. (Thus package operations like
2271 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2272 work as the user might reasonably expect.)
2273 * minor incompatible change: *STANDARD-INPUT* is now only an
2274 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2276 * minor incompatible change: Y-OR-N-P is now character-oriented,
2277 not line oriented. Also, YES-OR-NO-P now works without errors.
2278 (thanks to Antonio Martinez)
2279 * sb-aclrepl module improvements: an integrated inspector, added
2280 repl features, and a bug fix to :trace command.
2281 * Known functions, which cannot be open coded by the backend, are
2282 considered to be able to check types of their arguments. (fixing
2283 a bug report by Nathan J. Froyd)
2284 * fixed a bug in computing method discriminating functions: It is
2285 now possible to define methods specialized on classes which have
2286 forward-referenced superclasses. (thanks to Gerd Moellmann)
2287 * fixed evaluation order in optional entries (reported by Gilbert
2289 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2290 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2291 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2292 required-by-AMOP NAME argument, as well as CLASS and
2293 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2294 * fixed bug 20: DEFMETHOD can define methods using names that are
2295 not the proper names of classes to designate class specializers.
2296 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2297 Unix file descriptors, instead of blowing up. (thanks to Antonio
2299 * Garbage collection refactoring: user-visible change is that a
2300 call to the GC function during WITHOUT-GCING will not do garbage
2301 collection until the end of the WITHOUT-GCING. If you were doing
2302 this you were probably losing anyway.
2303 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2304 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2305 TYPEP the latter but not the former.
2306 * The compiler issues a full WARNING for calls to undefined functions
2307 with names from the CL package.
2308 * MAP-INTO for a vector destination is open coded. (reported by
2309 Brian Downing on c.l.l)
2310 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2311 documentation string.
2312 * fixed some bugs revealed by Paul Dietz' test suite:
2313 ** COPY-ALIST now signals an error if its argument is a dotted
2315 ** Condition slots are now accessed more correctly in the presence
2316 of multiple initargs for a given slot.
2317 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2318 exclude restarts of the same name associated with a different
2320 ** DEFCLASS of forward-referenced classes with another
2321 forward-referenced class in the superclasses list no longer
2323 ** Condition slots are now initialized once each, not multiple
2324 times. (thanks to Gerd Moellmann)
2325 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2327 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2328 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2329 arguments contain duplicated elements.
2330 ** RESTART-CASE understands local macros.
2331 ** RESTART-CASE associates exactly its own restarts with a condition.
2332 ** ENDP in safe mode checks its argument to be of type LIST.
2333 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2334 in question is unbound.
2335 ** Optimized MAKE-INSTANCE functions no longer cause internal
2336 assertion failures in the presence of duplicate initargs.
2337 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2338 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2339 propagated even in the ctor optimized implementation of
2341 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2343 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2344 * a better implementation of SXHASH on (simple) bit vectors,
2345 measured both in execution speed and in distribution of results
2346 over the positive fixnums, has been installed. Likewise, a better
2347 implementation of EQUAL for simple bit vectors is now available.
2348 * fixed CEILING optimization for a divisor of form 2^k.
2349 * fixed bug 240 (emitting extra style warnings "using the lexical
2350 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2351 by Antonio Martinez)
2352 * fixed SXHASH, giving different results for NIL depending on type
2353 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2354 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2355 variable. (found by Rolf Wester)
2356 * SBCL does not ignore type declarations for special
2357 variables. (reported by rif on c.l.l 2003-03-05)
2358 * some bug fixes in contrib/sb-aclrepl/
2359 * fixed some bugs revealed by Paul Dietz' test suite:
2360 ** a bug in the CONS type specifier, whereby the CAR and CDR
2361 types got intertwined, has been fixed;
2362 ** the type system is now able to reason about the interaction
2363 between INTEGER and RATIO types more completely;
2364 ** APPEND, [N]REVERSE and NRECONC check that those their
2365 arguments, which must be proper lists, are really so;
2366 ** An array specialized to be unable to hold elements has been
2367 implemented, as required -- yes, really -- by ANSI;
2368 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2369 on malformed property lists;
2371 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2372 * incompatible packaging change: in line with Unix convention,
2373 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2374 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2375 * REQUIRE and PROVIDE are now optionally capable of doing something
2376 useful. See the documentation string for REQUIRE.
2377 * infrastructure for a managed SBCL contrib system: contributed
2378 modules in this release include:
2379 ** the ASDF system definition facility;
2380 ** an interface to the BSD Sockets API;
2381 ** an ACL-like convenience interface to the repl;
2382 (thanks to Kevin Rosenberg)
2383 ** an implementation of ROTATE-BYTE, with efficient implementation
2385 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2386 gives a closer approximation to the right answer than previously.
2387 (thanks to Raymond Toy)
2388 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2389 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2390 optional environment arguments, as required by ANSI.
2391 * fixed bugs in other functions taking environment objects, allowing
2392 calls with an explicit NIL environment argument to be compiled
2394 * fixed bug 228: primary return values from
2395 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2396 COMPILE or FUNCTION.
2397 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2398 :TYPE will no longer signal errors on innocuous objects.
2399 * fixed bug 231b: SETQ is better at respecting type declarations in
2400 the lexical environment.
2401 * fixed a bug in DEFCLASS: classes named by symbols with no or
2402 unprintable packages can now be defined.
2403 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2404 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2405 * fixed compiler failure related to checking types of functions.
2406 (reported by Robert E. Brown)
2407 * the compiler is now much more consistent in its error-checking
2408 treatment of bounding index arguments to sequence functions: in
2409 (SAFETY 3) code, errors will be signalled in almost all cases if
2410 invalid sequence bounding indices are passed to functions defined
2411 by ANSI to operate on sequences.
2412 * fixed a bug in the build procedure: documentation of SBCL-specific
2413 packages is now preserved and available in the final Lisp image.
2414 * lifted FDEFINITION lookup out of loops in the implementation of
2415 many list operations. (thanks to Robert E. Brown)
2416 * fixed a bug in the reader: the #n# reader macro now works for
2417 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2418 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2419 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2420 E. Brown and Tony Martinez respectively)
2421 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2422 SLOT-BOUNDP: in particular, it is now able to optimize them much
2423 better, and is now not vulnerable to having packages renamed.
2424 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2425 always called when a slot is not present in an instance. (thanks
2427 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2428 longer signal an error on unreading a character following EOF on
2429 the previous constituent stream. (thanks to Tony Martinez)
2430 * fixed some bugs revealed by Paul Dietz' test suite:
2431 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2432 not just nonnegative fixnums;
2433 ** the logical bit-array operators such as BIT-AND now accept an
2434 explicit NIL for their "opt-arg" argument (to indicate a
2435 freshly-consed result bit-array);
2436 ** ELT now signals an error on an invalid sequence index in safe
2438 ** the type system is now cleverer about negations of numeric
2439 types, and consequently understands the BIGNUM and RATIO types
2441 ** the type system is now cleverer about the interaction between
2442 INTEGER and RATIO types: while bugs still remain, many more
2443 cases are accurately computed;
2444 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2445 if it is in the last clause;
2446 ** CONSTANTLY now correctly returns a side-effect-free function in
2448 ** DECLARE is no longer treated as a special-operator; in
2449 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2450 * incremented fasl file version number due to the change in the
2451 DEFSTRUCT-SLOT-DESCRIPTION structure.
2453 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2454 * minor incompatible change: code processed by the "interpreter" or
2455 EVAL now has a compilation optimization policy of (DEBUG 2)
2456 (changed from (DEBUG 1)) to improve debuggability of interactive
2457 development, and to allow the use of the debug RETURN command in
2459 * an experimental implementation of the RETURN command for the
2460 debugger has been included. (thanks to Frederik Kuivinen)
2461 * fixed bug 62: constraints were not propagated into a loop.
2462 * fixed bug in embedded calls of SORT (reported and investigated by
2464 * fixed some bugs revealed by Paul Dietz' test suite:
2465 ** printing and reading of arrays with some dimensions having
2466 length 0 (thanks to Gerd Moellmann);
2467 ** BOA constructor with &AUX argument without a default value does
2468 not cause a type error;
2469 ** CONSTANTP now returns true for all self-evaluating objects.
2471 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2472 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2473 accessors that are related by inheritance, as specified in the
2474 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2475 Valtteri Vuorikoski)
2476 * The compiler is now able to inline functions that were defined in
2477 a complex lexical environment (e.g. inside a MACROLET).
2478 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2480 * Support for the upcoming FreeBSD-5.0 release has been included.
2481 (thanks to Dag-Erling Smorgrav)
2482 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2483 effect when it is not in a toplevel context.
2484 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2485 stand a better chance of being correct. (thanks to Gerd
2487 * fixed bug in COERCE, which now signals an error on coercing a
2488 rational to a bounded real type which excludes the expected
2490 * The compiler is now able to derive types more accurately from the
2491 COERCE and COMPILE functions.
2492 * fixed bug 223: functional binding is considered to be constant
2493 only for symbols in the CL package.
2494 * fixed bug 231: SETQ did not check the type of a variable being set
2495 (reported by Robert E. Brown)
2496 * A new optimization for MAKE-INSTANCE has been included, fixing
2497 various bugs (including relating to :ALLOCATION :CLASS slots and
2498 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2500 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2501 ** As required by ANSI, LOOP now disallows anonymous collection
2502 clauses such as COLLECT I in conjunction with aggregate boolean
2503 clauses such as THEREIS (= I 1);
2504 ** LOOP now signals an error when any variable is reused in the
2505 same loop (including the potentially useful construct analogous
2506 to WITH A = 1 WITH A = (1+ A);
2507 ** IT is only a special loop symbol within the first clause of a
2508 conditional loop clause;
2509 ** LOOP with a typed iteration variable over a hashtable now
2510 signals a type error iff it should.
2511 * fixed some other bugs revealed by Paul Dietz' test suite:
2512 ** FILE-STREAM now names the class previously known as FD-STREAM;
2513 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2514 argument) no longer signals an error;
2515 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2516 of the slot symbol, rather than using the current package
2517 ((:CONC-NAME "") continues to intern the slot's name in the
2519 * incremented fasl file version number, because of the incompatible
2520 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2521 of the new implementation of DEFINE-COMPILER-MACRO.
2523 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2524 * Support for building SBCL for MIPS platforms running in
2525 little-endian mode has now been checked in, and basic
2526 functionality on said platforms verified.
2527 * minor incompatible change: PCL now records the pathname of a file
2528 in which methods and the like are defined, rather than its
2530 * minor incompatible change: TRUENAME now considers the truename of
2531 a file naming a directory to be the pathname with :DIRECTORY
2532 component indicating that directory.
2533 * minor incompatible change: a NAMED clause in the extended form of
2534 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2535 reason for the previous behaviour is unclear.
2536 * more systematization and improvement of CLOS and MOP conformance
2537 in PCL (thanks to Gerd Moellman and Pierre Mai):
2538 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2540 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2541 primary methods with no specializers;
2542 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2544 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2545 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2546 FORWARD-REFERENCED-CLASSes; error reporting on
2547 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2549 ** SXHASH on CLOS instances now uses a slot internal to the
2550 instance to return different numbers on distinct instances,
2551 while preserving the same return value through invocations of
2553 ** DEFMETHOD signals errors when methods with longer incongruent
2554 lambda lists are added to generic functions;
2555 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2556 CLASS, as specified in AMOP;
2557 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2558 on the order returned by the primary method for classes of
2559 class STANDARD-CLASS;
2560 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2561 * fixed some bugs shown by Paul Dietz' test suite:
2562 ** DOLIST puts its body in TAGBODY;
2563 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2565 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2566 value producing form;
2567 ** if more variables are given to PROGV than values, extra
2568 variables are bound and made to have no value;
2569 ** NSUBSTITUTE on list arguments gets the right answer with
2571 ** ELT signals an error of type TYPE-ERROR when the index argument
2572 is not a valid sequence index;
2573 ** LOOP signals (at macroexpansion time) an error of type
2574 PROGRAM-ERROR when duplicate variable names are found;
2575 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2576 ** FUNCALL of special-operators now cause an error of type
2578 ** PSETQ now works as required in the presence of side-effecting
2579 symbol-macro places;
2580 ** NCONC accepts any object as its last argument;
2581 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2583 ** loop-for-as-package does not require a package to be explicitely
2585 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2587 * fixed bug 166: compiler preserves "there is a way to go"
2588 invariant when deleting code.
2589 * fixed bug 172: macro lambda lists with required arguments after
2590 &REST arguments now cause an error to be signalled. (thanks to
2592 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2593 bug. (thanks to Matthew Danish)
2594 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2596 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2597 arguments are lexically rebound. (thanks to Gerd Moellmann and
2599 * fixed bug 194: error messages are now more informative when there
2600 is no primary method applicable in a call to a generic
2601 function. (thanks to Gerd Moellmann)
2602 * fixed bug in command line argument checking (thanks to Julian
2604 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2605 (thanks to Matthew Danish)
2606 * incremented fasl file version number, because of the
2607 SXHASH-related changes in the layout of CLOS data structures
2609 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2610 * minor incompatible change: The runtime (the Unix executable named
2611 "sbcl") is now much pickier about the .core files it will load.
2612 Essentially it now requires .core files to descend from the same
2613 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2614 as the runtime does. (The intent is to prevent the crashes which
2615 can occur, and which can even be reported as mysterious failures,
2616 when people patch the sources or change the build parameters
2617 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2618 sbcl and .core files.)
2619 * fixed bug: VALUES-LIST is no longer optimized away.
2620 * fixed bug 142: The FFI conversion of C string values to Lisp
2621 string values no longer conses excessively. (thanks to Nathan
2622 Froyd porting Raymond Toy's fix to CMU CL)
2623 * began to systematize and improve MOP conformance in PCL (thanks to
2624 Nathan Froyd, Gerd Moellman and Pierre Mai):
2625 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2627 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2628 ** FINALIZE-INHERITANCE is now called on class finalization;
2629 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2630 argument precedence order.
2631 * fixed bug 202: The compiler no longer fails on functions whose
2632 derived types contradict their declared type.
2633 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2634 so it can be non-toplevel.
2635 * The fasl file version number has changed (because of the new
2636 implementation of DEFMACRO).
2637 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2638 safe code, that any length requirement by their type-specifier
2639 argument is valid. The exceptions to this are described in bug
2641 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2642 functions now check the types of their inputs as required by ANSI.
2643 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2644 introduced symbol is DECLAREd to be SPECIAL.
2645 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2646 * fixed bug: :COUNT argument to sequence functions may be negative
2647 * fixed bug: body of DO-SYMBOLS may contain declarations
2648 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2649 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2650 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2651 symbol macro only once
2652 * fixed printing of call frame when argument list is unavailable
2653 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2654 * compiler no longer signals WARNING on unknown keyword
2657 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2658 * A beta-quality port to the mips architecture running Linux,
2659 based on the old CMUCL backend, has been made. It has been tested
2660 on a big-endian kernel, and works sufficiently well to be able to
2661 rebuild itself; it has not been tested in little-endian mode.
2662 * fixed an inconsistency between gencgc.c and purify.c which made
2663 dumping/loading .core files unreliable
2664 * fixed bug 120a: The compiler now deals correctly with IFs where
2665 the consequent is the same as the alternative, instead of
2666 misderiving the return type. (thanks to Alexey Dejneka)
2667 * fixed bug 113: Logical pathnames are now dumpable (the logical
2668 host is resolved at load-time, throwing an error if it is not
2670 * fixed bug 174: FORMAT's error message is slightly clearer when a
2671 non-printing character is used in a format directive.
2672 * fixed several bugs in compiler checking of type declarations, i.e.
2673 violations of the Python "declarations are assertions" principle
2674 (thanks to Alexey Dejneka)
2675 * fixed several bugs in PCL's error checking (thanks to Gerd
2677 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2679 * fixed bug in compilation of functions as first class values
2680 (thanks to Antonio Martinez-Shotton)
2681 * The compiler's handling TYPE-ERRORs which it can prove will
2682 inevitably happen at runtime has been cleaned up and corrected
2683 in several ways. (thanks to Alexey Dejneka)
2684 * improved argument type checking for various basic arithmetic
2685 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2686 lavished on them in the past that they can be compiled in many
2687 ways in different special cases
2688 * fixed bug 181: compiler checks validity of user supplied type
2690 * cleaned up code flushing in optimization: Function calls which
2691 should signal errors for safety purposes (e.g. which ANSI says
2692 should signal errors when their arguments are of incorrect type)
2693 are no longer optimized away.
2694 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2695 * incremented fasl file version number, because changes in the
2696 implementation of sequence functions like COERCE caused
2697 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2698 in old inline expansions) to become undefined. (Actually these
2699 changes were later undone, so we might very well be binary
2700 compatible with 0.7.7 after all, but leaving the version number
2701 incremented seemed like the simplest and most conservative
2704 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2705 * An alpha-quality port to the parisc architecture running Linux,
2706 based on the old CMUCL backend, has been made. This, even more so
2707 than the other backends, should be considered still a work in
2708 progress; known problems include that the Linux kernel in 64-bit
2709 mode does not propagate the correct sigcontext structure to
2710 userspace, and consequently SBCL on a parisc64 kernel will not
2712 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2713 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2714 suppressed.) Also now that the compiler is looking at declarations
2715 in the environment, it checks optimization declarations as well,
2716 and suppresses inlining when (> DEBUG SPEED).
2717 * More fixes have been made to treatment of floating point exception
2718 treatment and other Unix signals. In particular, floating point
2719 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2720 * The detection and handling of control stack exhaustion (infinite
2721 or very deeply nested recursion) has changed. Stack exhaustion
2722 detection is now done by write-protecting pages at the OS level
2723 and applies at all optimization settings; when found, a
2724 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2725 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2726 be caught using IGNORE-ERRORS.
2727 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2728 correctly when transforming recursive LABELS functions to LETs.
2729 (thanks to Alexey Dejneka)
2730 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2731 that are names of constants or global variables.
2732 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2733 alien routines with docstrings.
2734 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2735 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2737 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2738 (thanks to Eric Marsden)
2739 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2740 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2741 * Bug fix: The SPARC backend can now compile functions involving
2742 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2743 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2744 to a functions expecting a PATHNAME-DESIGNATOR.
2745 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2746 lambda lists. (thanks to Alexey Dejneka)
2747 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2748 functions so that EXPORTing the name of the class doesn't cause
2749 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2750 to Antonio Martinez for reporting this)
2751 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2752 constant-folded without failing an assertion. (thanks to Einar
2753 Floystad Dorum for reporting this)
2754 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2755 types are now dealt with more correctly by the compiler.
2756 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2757 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2758 behaviour analogous to RENAME-FILE. This puts its behaviour more
2759 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2761 * The fasl file version number has changed again. (because of the
2762 bug fix involving the names of PCL MAKE-INSTANCE functions)
2764 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2765 * bug fix: Floating point exceptions are treated much more
2766 consistently on the x86/Linux and PPC/Linux platforms.
2767 * Array initialization with :INITIAL-ELEMENT is now much faster for
2768 cases when the compiler cannot open code the array creation, but
2769 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2770 array accesses have also seen a speed increase.
2771 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2772 specified. (This was at the root of some bad interactions between
2773 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2775 * bug fix: Internal error arguments for undefined functions are now
2776 computed correctly on the PPC/Linux platform.
2777 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2778 Raymond Toy's patch for CMU CL)
2779 * Support for the Solaris 9 operating environment has been included
2780 (thanks to Daniel Merritt)
2781 * A very ugly but hopefully complete draft of the missing FFI chapter
2782 of the manual has been created by reformatting the corresponding
2783 CMU CL manual chapter into (currently very ugly and incoherent)
2784 DocBook and bringing it up to date for SBCL behavior. Thus, the
2785 manual is now essentially complete, at least by my extreme
2786 once-and-only-once standards, whereby it's acceptable to refer to
2787 the doc strings of SB-EXT functions as the primary documentation.
2788 * The fasl file version number has changed again, due to cleanup of
2789 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2790 is no longer a static symbol.)
2792 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2793 * SBCL now builds with OpenMCL (version 0.12) as the
2794 cross-compilation host; also, more progress has been made toward
2795 bootstrapping under CLISP.
2796 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2798 * bug 158 fixed: The compiler can now deal with integer loop
2799 increments different from 1; fixing this turned out also to fix
2801 * bug 169 fixed: no more bogus warnings about using lexical bindings
2802 despite the presence of perfectly good SPECIAL declarations (thanks
2803 to David Lichteblau)
2804 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2805 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2806 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2808 * bug 180 fixed: Method combination specifications no longer ignore
2809 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2810 * bug fix: Structure type predicate functions now check their argument
2811 count as they should.
2812 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2813 correctly. (thanks to Pierre Mai)
2814 * minor incompatible change: The --noprogrammer option is deprecated
2815 in favor of the new --disable-debugger option, which is very similar.
2816 (The major difference is that it takes effect at a slightly different
2817 time at startup, causing handling of errors in --sysinit and
2818 --userinit files will be affected differently.) The
2819 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2820 been added to allow this functionality to be controlled from ordinary
2821 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2822 maintainers, who might want to run non-interactive scripts to
2823 build SBCL cores which will later be used interactively.)
2824 * minor incompatible change: The LOAD function no longer, when given
2825 a wild pathname to load, loads all files matching that pathname.
2826 Instead, an error of type FILE-ERROR is signalled.
2828 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2829 * bug 147 fixed: The compiler preserves its block link/count
2830 invariants more correctly now so that it doesn't crash. (thanks
2832 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2834 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2835 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2836 no-value-for-this-character as NIL (as specified by ANSI).
2837 * HOST-NAMESTRING on physical pathnames now returns a string that is
2838 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2839 (thanks to Christophe Rhodes)
2840 * The Alpha port handles icache flushing more correctly. (thanks to
2842 * More progress has been made toward bootstrapping under CLISP. (thanks
2843 to Christophe Rhodes)
2844 * The fasl file format has changed again, because dynamic loading
2845 on OpenBSD (which has non-ELF object files) motivated some cleanups
2846 in the way that foreign symbols are transformed and passed around.
2847 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2848 is no longer treated as whitespace by the reader, but instead as
2849 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2850 |A
\7fB|, instead of A as it used to.
2852 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2853 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2854 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2855 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2856 this as of 0.7.1.45, but was left out of the previous news section
2857 (thanks to Dan Barlow)
2858 * SBCL now runs on the Solaris operating system on SPARC architectures
2859 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2860 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2861 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2862 * SPARC backend cleanups, allowing builds of cores optimized for V8
2863 and V9 SPARCS, and also emission of code targeted to a particular
2864 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2866 * SBCL is closer to bootstrapping under CLISP, thanks to various
2867 fixes by Christophe Rhodes.
2868 * The fasl file format has changed again, to allow the compiler's
2869 INFO database to support symbol macros.
2870 * The user manual (in doc/) is formatted into HTML more nicely.
2871 (thanks to coreythomas)
2872 * The system is smarter about SUBTYPEP relationships, especially
2873 those involving NOT types (including types such as ATOM which are
2874 represented internally using NOT types). Thus SUBTYPEP is less
2875 likely to return (VALUES NIL NIL) in general, and in particular
2876 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2878 * The fasl file format has changed again, because the internal
2879 representation of types now includes a new slot to support the new
2880 SUBTYPEP-of-NOT-types logic.
2881 * (not a change in the main branch of SBCL, but a related prototype
2882 which can hopefully be merged into the main branch of SBCL in the
2883 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2884 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2885 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2886 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2887 dispatch macro character combinations. (thanks to Alexey Dejneka)
2888 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2889 that superficially look like logical namestrings correctly.
2890 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2892 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2893 default supplied-p) for &optional and &key arguments. (thanks to
2896 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2897 * incompatible change: The compiler is now less aggressive about
2898 tail call optimization, doing it only when (> SPACE DEBUG) or
2899 (> SPEED DEBUG). (This is an incompatible change because there are
2900 programs which relied on the old CMU-CL-style behavior to optimize
2901 away their unbounded recursion which will now die of stack overflow.)
2902 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2903 for non-GENCGC systems has been increased to 20M (since that
2904 seems much closer to the likely performance optimum for modern
2905 systems than the old 4M value was)
2906 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2907 and *DEBUG-PRINT-LEVEL*
2908 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2909 of CMU CL's support for SPARC, and various endianness and other
2910 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2911 * new syntactic sugar for the Unix command line: --load foo.bar is now
2912 an alternate notation for --eval '(load "foo.bar")'.
2914 ** The system now detects stack overflow and handles it gracefully,
2915 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2916 optimization settings. (This is a good thing in general, and
2917 its introduction in this version should be particularly timely
2918 for anyone whose code fails because of suppression of tail
2920 ** The system now hunts for the C variable "environ" in a more
2921 devious way, to avoid segfaults when the C library version differs
2922 between compile time and run time. (thanks to Christophe Rhodes)
2923 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2924 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2925 ** The compiler no longer issues bogus style warnings for undefined
2926 classes in the same source file as the DEFCLASSes which defined
2927 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2929 ** fixes in CONDITION class precedence list for undefined function
2930 errors (thanks to Alexei Dejneka)
2931 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2932 correctly. (thanks to Dan Barlow)
2933 ** portability fixes aiming at bootstrapping under CLISP (thanks
2934 to Dave McDonald and Christophe Rhodes)
2935 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2936 ** fixes in type translation and and type inference (thanks to
2938 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2939 ** various fixes in the new ports (thanks to Dan Barlow)
2940 * several changes related to debugging:
2941 ** suppression of tail recursion, as noted above
2942 ** stack overflow detection, as noted above
2943 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2944 is now the default. (For some time encapsulation has been more
2945 reliable than the breakpoint-based :ENCAPSULATE NIL
2946 implementation, at least on X86 systems; and I just noticed that
2947 encapsulation also seems closer to the spirit of the ANSI
2950 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2952 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2953 up properly again. (There was a packaging bug in 0.7.0 which
2954 left their definitions in SB-SYS::LOAD-FOREIGN and
2955 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2956 vital for most things which interface to C-level interfaces,
2957 like extensions working with sockets or databases or
2958 Perl-compatible regexes or whatever, and the need to fix
2959 this bug is the main reason that 0.7.1 was released so
2961 ** DEFGENERIC is now choosier about the methods it redefines, so that
2962 reLOADing a previously-LOADed file containing DEFGENERICs does
2963 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2964 cycle now works as it should. (thanks to Alexey Dejneka)
2965 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2966 fixed by Christophe Rhodes. (He actually submitted this patch
2967 months ago, and I delayed until after 0.7.0.)
2968 ** Bug 111 (internal compiler confusion about runtime checks on
2969 FUNCTION types) was fixed by Alexey Dejneka.
2970 * Some internal cleanups (getting rid of variables which aren't
2971 needed now that the byte interpreter is gone) caused the fasl
2972 file format number to change again.
2974 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2975 * major incompatible change: The default fasl file extension, i.e. the
2976 default extension for files produced by COMPILE-FILE, has changed
2977 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2979 ** There are many changes in the implementation of the compiler.
2980 SBCL is now essentially a compiler-only implementation of ANSI
2981 Common Lisp. EVAL still "interprets" a few special cases, but
2982 almost all the interesting cases are handled by creating
2983 a LAMBDA expression, calling COMPILE on it, then calling
2984 FUNCALL on the result.
2985 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2986 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2987 Since the code is newer, there might still be some new bugs
2988 (though not as many as before Martin Atzmueller's fixes:-). But
2989 the new code is substantially simpler and clearer, and hopefully
2990 any remaining bugs will be simpler, less fundamental, and more
2991 fixable then the bugs in the old code.
2992 ** The revised compiler is still a little unsteady on its feet.
2994 *** The debugging information it produces (particularly the names
2995 of FUNCTION objects) is sometimes much less useful than what
2996 the old compiler produced.
2997 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2998 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2999 within a MACROLET) has been temporarily weakened.
3000 ** There are new compiler optimizations for various functions:
3001 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
3002 FIND-IF-NOT, POSITION-IF-NOT, and FILL
3003 *** the math functions TRUNCATE, FLOOR, and CEILING
3004 *** the function-of-all-trades COERCE
3005 Mostly these should be transparent, but there's one
3006 potentially-annoying problem (bug 117): when the compiler
3007 inline-expands a function and does type analysis on the result,
3008 it can create control paths which have type mismatches, and
3009 when it can't prove that those control paths aren't taken,
3010 it will issue WARNINGs about the type mismatches. This is
3011 a particular problem in practice for the new sequence functions.
3012 It's not clear how this should be fixed, and for now, a
3013 workaround is given in the entry for 117 in the BUGS file.
3014 ** (Because of the interaction between the two previous items --
3015 occasional inlining problems and new inline expansions -- some
3016 of the new sequence function optimizations won't really kick in
3017 completely until debugging information, and then inlining, are
3018 straightened out in some future version.)
3019 * minor incompatible changes:
3020 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
3021 DIRECTORY no longer implicitly promotes NIL slots of its
3022 pathname argument to :WILD. In particular, when you ask for the
3023 contents of a directory (which you used to be able to do without
3024 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
3025 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
3026 ** changes in behavior that ANSI explicitly defines to be
3027 implementation dependent:
3028 *** The new compiler-only implementation still conforms with ANSI,
3029 but acts a little different than before. Besides the obvious
3030 changes in performance tradeoffs (that the cost per form passed
3031 to EVAL has gone up, and the cost per form executed by EVAL
3032 has gone down), the behavior of the system changes a little
3033 because there are no longer any interpreted function objects.
3034 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
3035 e.g. doing COMPILE on the output of interactive DEFUN is
3037 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
3039 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
3040 has changed from (:CL) to NIL.
3041 *** The CHAR-NAME of unprintable ASCII characters which, unlike
3042 e.g. #\Newline and #\Tab, don't have names specified in the
3043 ANSI Common Lisp standard, is now based on their ASCII symbolic
3044 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
3045 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
3046 are no longer used for output.
3047 ** changes in internal implementation constants:
3048 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3049 4 million. (If your application spends a lot of time GCing and
3050 you have a lot of RAM, you might want to experiment with
3051 increasing it even more.)
3052 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3053 However, almost all old code should still continue to work without
3054 immediate update, as SB-C-CALL is now a (deprecated) nickname
3056 ** Old operator names in the style DEF-FOO are now deprecated in
3057 favor of new corresponding names DEFINE-FOO, for consistency with
3058 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3059 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3060 internal symbols, but a few supported extensions like
3061 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3062 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3063 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3064 etc. as you get deeper into recursive calls to the debugger
3065 command loop, instead of the old "5]", "5]]", "5]]]"
3066 sequence. (I was motivated to do this when squabbles between
3067 ILISP and SBCL left me very deeply nested in the debugger. In the
3068 short term, this change will probably provoke more ILISP/SBCL
3069 squabbles, but hopefully it will be an improvement in the long run.)
3070 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3071 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3072 function, since declaiming return types involving aliens is
3073 (1) annoyingly messy to do by hand and (2) vital to efficient
3074 compilation of code which calls such functions.
3075 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3076 longer reexported by the SB-EXT package. They're solely useful
3077 for alien code, so it seems more logical that you should get
3078 them from the SB-ALIEN package, not in SB-EXT.
3079 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3080 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3081 features. Instead, the code that they used to control is always
3082 built into the system.
3083 * many other bug fixes
3084 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3085 advantage of the new EVAL-WHEN stuff and to clean them up in
3086 general, and they are now more ANSI-compliant in a number of
3087 ways. Martin Atzmueller is responsible for a lot of this.
3088 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3090 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3091 *** correct ERROR type for various file operations
3092 *** some fixes for Lisp streams
3093 *** DEFMETHOD syntax checking
3094 *** changing old weird representation of debug information as
3095 strings (which, among their other deficiencies, don't transform
3096 correctly when you rename packages, and don't change their
3097 print representation when you change things like *PACKAGE*
3098 and *PRINT-LENGTH*) to symbols and lists of symbols
3099 He also made several improvements and fixed several bugs in DESCRIBE.
3100 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3102 *** misbehavior of WRITE-STRING/WRITE-LINE
3103 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3104 and several other LOOP problems as well
3105 *** DIRECTORY when similar filenames are present
3106 *** DEFGENERIC with :METHOD options
3107 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3108 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3109 *** argument ordering in FIND with :TEST option
3110 *** mishandled package designator argument in APROPOS-LIST
3111 *** various problems in the backquote readmacro
3112 *** a bug in APROPOS
3113 *** probably some others that I'm not describing very well here,
3114 since the CVS log documents them by reference to sbcl-devel
3115 messages, and the SourceForge archives aren't working well.:-(
3116 ** Dan Barlow improved the Alpha port (and is making progress on the
3117 PPC port, for those of you who think different).
3118 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3119 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3120 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3121 continues to work on the SPARC port (for those of you in a position
3122 to look down upon our little PC-compatible boxes from a great height).
3123 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3124 and uses that copy, rather than the current dynamic value, when
3125 it's trying to decide whether to truncate output. Thus e.g.
3126 (let ((*print-lines* 50))
3127 (pprint-logical-block (stream nil)
3129 (let ((*print-lines* 8))
3130 (print (aref possiblybigthings i) stream)))))
3131 should now truncate the logical block only at 50 lines, instead of
3132 often truncating it at 8 lines, as it did before.
3133 * The doc/cmucl/ directory, containing old CMU CL documentation
3134 from the time of the fork, is no longer part of the base system.
3135 SourceForge has shut down its anonymous FTP service, and with it
3136 my original plan for distributing the old CMU CL documentation
3137 there. For now, if you need these files you can download an old
3138 SBCL source release and extract them from it.
3139 * The fasl file version number changed again, for dozens of reasons,
3140 some of which are apparent above.
3142 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3143 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3144 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3145 patch, so that DISASSEMBLE output is much nicer.
3146 * The code in the SB-PROFILE package now seems reasonably stable.
3147 I still haven't decided what the final interface should look like
3148 (I'd like PROFILE to interact cleanly with TRACE, since both
3149 facilities use function encapsulation) but if you have a need
3150 for profiling now, you can probably use it successfully with
3151 the current CMU-CL-style interface.
3152 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3153 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3154 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3155 link now returns the dangling link itself, and for similar
3156 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3157 link itself. (In these cases the old code signalled an error and
3158 looped endlessly, respectively.) Thus, DIRECTORY now works even
3159 in the presence of dangling and cyclic symbolic links.
3160 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3161 is now a supported extension again, since the consensus on
3162 sbcl-devel was that it can be useful for ordinary development
3163 work, not just for debugging SBCL itself.
3164 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3165 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3166 that currently-defined functions might be redefined later with
3167 different return types.
3168 * Hash tables can be printed readably, as inspired by CMU CL code
3169 of Eric Marsden and SBCL code of Martin Atzmueller.
3170 * better error handling in CLOS method combination, thanks to
3171 Martin Atzmueller porting Pierre Mai's CMU CL patches
3172 * more overflow fixes for >16Mbyte I/O buffers
3173 * A bug in READ has been fixed, so that now a single Ctrl-D
3174 character suffices to cause end-of-file on character streams.
3175 In particular, now you only need one Ctrl-D at the command
3176 line (not two) to exit SBCL.
3177 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3179 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3180 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3181 does the right thing.
3182 * The implementation of some type tests, especially for CONDITION
3183 types, is now tidier and maybe faster, due to CMU CL code
3184 originally by Douglas Crosher, ported by Martin Atzmueller.
3185 * Some math functions have been fixed, and there are new
3186 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3187 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3188 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3189 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3190 likely to be built into sbcl-0.7.0.)
3191 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3192 CL, ported to SBCL by Martin Atzmueller)
3193 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3194 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3195 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3196 * There's a new slam.sh hack to shorten the edit/compile/debug
3197 cycle for low-level changes to SBCL itself, and a new
3198 :SB-AFTER-XC-CORE target feature to control the generation of
3199 the after-xc.core file needed by slam.sh.
3200 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3201 COMPILE-FILE is no longer supported, so that now every function
3202 gets an entry point, so that block compilation looks a little
3203 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3204 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3205 returns the number of bytes consed since the system started,
3206 rather than the number consed since the first time the function
3207 was called. (The new definition parallels ANSI functions like
3208 CL:GET-INTERNAL-RUN-TIME.)
3209 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3210 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3211 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3212 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3213 * Fasl file version numbers are now independent of the target CPU,
3214 since historically most system changes which required version
3215 number changes have affected all CPUs equally. Similarly,
3216 the byte fasl file version is now equal to the ordinary
3219 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3220 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3221 is no longer recognized. I apologize for this, because it was
3222 listed in SB-EXT as a supported extension, but I found that
3223 its existing behavior was poorly specified, as well as incorrectly
3224 specified, and it looked like too much of a mess to straighten it
3225 out. I have enough on my hands trying to get ANSI stuff to work..
3226 * many patches ported from CMU CL by Martin Atzmueller, with
3227 half a dozen bug fixes in pretty-printing and the debugger, and
3228 half a dozen others elsewhere
3229 * fixed bug 13: Floating point infinities are now supported again.
3230 They might still be a little bit flaky, but thanks to bug reports
3231 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3232 as flaky as they were.
3233 * The --noprogrammer command line option is now supported. (Its
3234 behavior is slightly different in detail from what the old man
3235 page claimed it would do, but it's still appropriate under the
3236 same circumstances that the man page talks about.)
3237 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3238 are now supported, and enabled by default. Thus, the compiler can
3239 handle many floating point and complex operations much less
3240 inefficiently. (Thus e.g. you can implement a complex FFT
3242 * The compiler now detects type mismatches between DECLAIM FTYPE
3243 and DEFUN better, and implements CHECK-TYPE more correctly, and
3244 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3245 thanks to patches from Martin Atzmueller.
3246 * various fixes to make the cross-compiler more portable to
3247 ANSI-conforming-but-different cross-compilation hosts (notably
3248 Lispworks for Windows, following bug reports from Arthur Lemmens)
3249 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3250 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3251 fixed thanks to Pierre Mai's CMU CL patches.
3252 * fixes to keep the system from overflowing internal counters when
3253 it tries to use i/o buffers larger than 16M bytes
3254 * fixed bug 45a: Various internal functions required to support
3255 complex special functions have been merged from CMU CL sources.
3256 (When I was first setting up SBCL, I misunderstood a compile-time
3257 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3258 * improved support for type intersection and union, fixing bug 12
3259 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3260 more obscure bugs as well
3261 * some steps toward byte-compiling non-performance-critical
3262 parts of the system, courtesy of patches from Martin Atzmueller
3263 * Christophe Rhodes has made some debian packages of sbcl at
3264 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3265 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3266 stable, but are nonetheless usable. When he's ready, I'd be happy
3267 to add them to the SourceForge "File Releases" section. (And if
3268 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3269 * new fasl file format version number (because of changes in
3270 internal representation of (OR ..) types to accommodate the new
3271 support for (AND ..) types, among other things)
3273 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3274 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3276 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3278 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3279 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3280 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3281 * DESCRIBE now works on structure objects again.
3282 * Most function call argument type mismatches are now handled as
3283 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3284 know whether the function will be redefined before the call is
3285 executed. (The compiler could flag local calls with full WARNINGs,
3286 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3287 it doesn't keep track of enough information to know whether calls
3288 are local in this sense.)
3289 * Compiler output is now more verbose, with messages truncated
3290 later than before. (There should be some supported way for users
3291 to override the default verbosity, but I haven't decided how to
3292 provide it yet, so this behavior is still controlled by the internal
3293 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3294 src/compiler/ir1util.lisp.)
3295 * Fasl file format version numbers have increased again, because
3296 support for the Gray streams extension changes the layout of the
3297 system's STREAM objects.
3298 * The Gray subclassable streams extension now works, thanks to a
3299 patch from Martin Atzmueller.
3300 * The full LOAD-FOREIGN extension (not just the primitive
3301 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3302 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3303 but like most other programs, it defaults to copying the Unix
3304 environment from the original process instead of starting the
3305 new process in an empty environment.
3306 * Extensions which manipulate the Unix environment now support
3307 an :ENVIRONMENT keyword option which doesn't smash case or
3308 do other bad things. The CMU-CL-style :ENV option is retained
3309 for porting convenience.
3310 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3311 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3313 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3315 * A patch from Martin Atzmueller seems to have solved the SIGINT
3316 problem, and as far as we know, signal-handling now works cleanly.
3317 (If you find any new bugs, please report them!)
3318 * The system no longer defaults Lisp source file names to types
3319 ".l", ".cl", or ".lsp", but only to ".lisp".
3320 * The compiler no longer uses special default file extensions for
3321 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3322 seems to expect a single default extension for all compiled code,
3323 and there's no compelling reason to try to stretch the standard
3324 to allow two different extensions.) Instead, byte-compiled files
3325 default to the same extension as native-compiled files.
3326 * Fasl file format version numbers have increased again, because
3327 a rearrangement of internal implementation packages made some
3328 dumped symbols in old fasl files unreadable in new cores.
3329 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3330 many fewer weird special cases.
3331 * Bug #17 (differing COMPILE-FILE behavior between logical and
3332 physical pathnames) has been fixed, and some related misbehavior too,
3333 thanks to a patch from Martin Atzmueller.
3334 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3335 by Tim Moore, ported to SBCL by Martin Atzmueller.
3336 * Martin Atzmueller fixed several filesystem-related problems,
3337 including bug #36, in part by porting CMU CL patches, which were
3338 written in part by Paul Werkowski.
3339 * More compiler warnings in src/runtime/ are gone, thanks to
3340 more patches from Martin Atzmueller.
3341 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3344 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3346 * DESCRIBE now works on CONDITION objects.
3347 * The debugger now handles errors which arise when trying to print
3348 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3350 * The build system now uses an additional file, customize-target-features.lisp,
3351 to allow local modifications to the target *FEATURES* list. (The point of
3352 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3353 debugging features enabled, without having to worry about propagating it
3354 into everyone's system when I do a "cvs update".) When no
3355 customize-target-features.lisp file exists, the target *FEATURES* list
3356 should be constructed the same way as before.
3357 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3358 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3359 warn about behavior which is undefined under the ANSI spec. Note: This
3360 is specified by ANSI, but it's not very popular with programmers.
3361 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3362 macro in the SBCL sources for an example of a workaround which you
3363 might use to make portable ANSI-standard code which does what you want.
3364 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3365 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3366 when it's used as a non-toplevel form. (This is required in order
3367 to implement the DEFCONSTANT-EQX macro.)
3368 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3369 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3370 funny things, due to bugs in the implementation of EVAL-WHEN
3371 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3372 non-toplevel DEFCONSTANTs are uncommon.)
3373 * The core file version number and fasl file version number have been
3374 incremented, because the old noncompliant DEFCONSTANT behavior involved
3375 calling functions which no longer exist, and because I also took the
3376 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3377 * fixed bug 1 (error handling before read-eval-print loop starts), and
3378 redid debugger restarts and related debugger commands somewhat while
3380 ** The QUIT debugger command is gone, since it did something
3381 rather different than the SB-EXT:QUIT command, and since it never
3382 worked properly outside the main toplevel read/eval/print loop.
3383 Invoking the new TOPLEVEL restart provides the same functionality.
3384 ** The GO debugger command is also gone, since you can just invoke
3385 the CONTINUE restart directly instead.
3386 ** The TOP debugger command is also gone, since it's redundant with the
3387 FRAME 0 command, and since it interfered with abbreviations for the
3389 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3391 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3392 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3393 some porting work by Martin Atzmueller)
3394 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3395 with a boatload of patches which he ported to SBCL. Now that those
3397 ** The system tries to make sure that its low-priority messages
3398 are prefixed by semicolons, to help people who like to use
3399 syntax highlighting in their ILISP buffer. (This patch
3400 was originally due to Raymond Toy.)
3401 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3402 patches originally written by Raymond Toy.
3403 ** The compiler understands coercion between single-value and
3404 multiple-VALUES type expressions better, getting rid of some very
3405 weird behavior, thanks to patches originally by Robert MacLachlan
3406 and Douglas Crosher.
3407 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3408 lambda lists, thanks to a patch originally by Pierre Mai.
3409 ** The system no longer bogusly warns about "abbreviated type
3411 ** The compiler gets less confused by inlining and RETURN-FROM,
3412 thanks to some patches originally by Tim Moore.
3413 ** The system no longer hangs when dumping circular lists to fasl
3414 files, thanks to a patch originally from Douglas Crosher.
3415 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3416 undefined function error.
3417 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3418 in structure slot initforms) for now, documented workaround instead:-|
3419 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3420 * fixed bug 5: added stubs for various Gray stream functions called
3421 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3422 installed, at least appropriate type errors are generated
3423 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3424 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3425 consistently in DEFMETHOD forms.
3426 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3427 it doesn't seem to affect SBCL after all
3428 * The C runtime system now builds with better optimization and many
3429 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3431 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3433 * The system is now under CVS at SourceForge (instead of the
3434 CVS repository on my home machine).
3435 * The new signal handling code has been tweaked to treat register
3436 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3437 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3438 downstream, e.g. in debug-int.lisp, has implicit dependencies
3439 on the unsignedness of integer representation of machine words,
3440 and that caused the system to bomb out with infinite regress
3441 when trying to recover from type errors involving signed values,
3442 e.g. (BUTLAST '(1 2 3) -1).)
3443 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3444 (This was one of the bugs Peter Van Eynde reported back in July.)
3445 * The system now uses code inspired by Colin Walters' O(N)
3446 implementation of MAP (from the cmucl-imp@cons.org mailing
3447 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3448 inline the MAP operation, and there is more than one
3449 sequence argument to the MAP call (so that it can't just
3450 do ETYPECASE once and for all based on the type of the
3451 single sequence argument). (The old non-inline implementation
3452 of the general M-argument sequence-of-length-N case required
3453 O(M*N*N) time when any of the sequence arguments were LISTs.)
3454 * The QUIT :UNIX-CODE keyword argument has been renamed to
3455 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3456 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3458 * Raymond Wiker's patches to port dynamic loading from Linux to
3459 FreeBSD have been added.
3460 * The BUGS file is now more nearly up to date, thanks in large part
3461 to Martin Atzmueller's review of it.
3462 * The debugger now flushes standard output streams before it begins
3463 its output ("debugger invoked" and so forth).
3464 * The core version number and fasl file version number have both
3465 been incremented, because of incompatible changes in the layout
3467 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3468 used to not be called for a saved Lisp image.)
3469 * Martin Atzmueller's version of a patch to fix a compiler crash,
3470 as posted on sbcl-devel 13 September 2000, has been installed.
3471 * Instead of installing Martin Atzmueller's patch for the
3472 compiler transform for SUBSEQ, I deleted the compiler transform,
3473 and transforms for some similar consing operations.
3474 * A bug in signal handling which kept TRACE from working on OpenBSD
3476 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3477 away by constant folding
3478 * The system now defines its address space constants in one place
3479 (in the Lisp sources), and propagates them automatically elsewhere
3480 (through GENESIS and the sbcl.h file). Therefore, patching the
3481 address map is less unnecessarily tedious and error-prone. The
3482 Lisp names of address space constants have also been systematized.
3483 * CVS tags like dollar-Header-dollar have been removed from
3484 the sources, because they have never saved me trouble and
3485 they've been source of trouble working with patches and other
3486 diff-related operations.
3487 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3488 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3490 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3492 * The system has been ported to OpenBSD.
3493 * The system now compiles with a simple "sh make.sh" on the systems
3494 that it's supported on. I.e., now you no longer need to tweak
3495 text in the target-features.lisp-expr and symlinks in src/runtime/
3496 by hand, the make.sh takes care of it for you.
3497 * The system is no longer so grossly inefficient when compiling code
3498 involving vectors implemented as general (not simple) vectors (VECTOR T),
3499 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3500 which dares to use the various sequence functions on non-simple
3501 vectors, takes less of a performance hit.
3502 * There is now a primitive type predicate VECTOR-T-P
3503 to test for the (VECTOR T) type, so that e.g.
3504 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3505 can now be compiled with some semblance of efficiency. (The old code
3506 turned the type declaration into a full call to %TYPEP at runtime!)
3507 * AREF on (VECTOR T) is still not fast, since it's still compiled
3508 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3509 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3510 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3511 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3512 without making full calls to SUBTYPEP at runtime.
3513 (Some analogous efficiency issues for non-simple vectors specialized to
3514 element types other than T, or for non-simple multidimensional arrays,
3515 have not been addressed. They could almost certainly be handled the
3516 same way if anyone is motivated to do so.)
3517 * The changes in array handling break binary compatibility, so
3518 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3519 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3520 instead of (VALUES T T).
3521 * By following the instructions that Dan Barlow posted to sbcl-devel
3522 on 2 July 2000, I was able to enable primitive dynamic object
3523 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3524 functionality is not implemented (since it calls ld to resolve
3525 library references automatically, requiring RUN-PROGRAM for its
3526 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3527 try to resolve library references) is now supported.
3528 * The system now flushes the standard output streams when it terminates,
3529 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3530 them at several other probably-convenient times, e.g. in each pass of
3531 the toplevel read-eval-print loop, and after evaluating a form given
3532 as an "--eval" command-line option. (These changes were motivated by a
3533 discussion of stream flushing issues on cmucl-imp in August 2000.)
3534 * The source transform for TYPEP of array types no longer assumes
3535 that an array whose element type is a not-yet-defined type
3536 is implemented as an array of T, but instead punts, so that the
3537 type will be interpreted at runtime.
3538 * There is now some support for cross-compiling in make.sh: each of
3539 the phases of make.sh has its own script. (This should be transparent
3540 to people doing ordinary, non-cross-compile builds.)
3541 * Since my laptop doesn't have hundreds of megabytes of memory like
3542 my desktop machine, I became more motivated to do some items on
3543 my to-do list in order to reduce the size of the system a little:
3544 ** Arrange for various needed-only-at-cold-init things to be
3545 uninterned after cold init. To support this, those things have
3546 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3547 symbols with such names are now uninterned after cold init).
3548 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3549 fasl files for cold load.
3550 ** Remove the old compiler structure pooling code (which used to
3551 be conditional on the target feature :SB-ALLOC) completely.
3552 ** Redo the representation of some data in cold init to be more compact.
3553 (I also looked into supporting byte compiled code at bootstrap time,
3554 which would probably reduce the size of the system a lot, but that
3555 looked too complicated, so I punted for now.)
3556 * The maximum signal nesting depth in the src/runtime/ support code has
3557 been reduced from 4096 to 256. (I don't know any reason for the very
3558 large old value. If the new smaller value turns out to break something,
3559 I'll probably just bump it back up.)
3560 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3562 * Many, many bugs reported by Peter Van Eynde have been added to
3563 the BUGS list; some have even been fixed.
3564 * While enabling dynamic object file loading, I tried to make the
3565 code easier to understand, renaming various functions and variables
3566 with less ambiguous names, and changing some function calling
3567 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3568 * While trying to figure out how to do the OpenBSD port, I tried to
3569 clean up some of the code in src/runtime/. In particular, I dropped
3570 support for non-POSIX signal handling, added various comments,
3571 tweaked the code to reduce the number of compilation warnings, and
3572 renamed some files to increase consistency.
3573 * To support the new automatic configuration functionality in make.sh,
3574 the source file target-features.lisp-expr has been replaced with the
3575 source file base-target-features.lisp-expr and the machine-generated
3576 file local-target-features.lisp-expr.
3577 * fixed a stupid quoting error in make.sh so that using CMU CL
3578 "lisp -batch" as cross-compilation host works again
3580 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3582 * DESCRIBE no longer tries to call itself recursively to describe
3583 bound/fbound values, so that it no longer fails on symbols which are
3584 bound to themselves (like keywords, T, and NIL).
3585 * DESCRIBE now works on generic functions.
3586 * The printer now prints less-screwed-up representations of closures
3587 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3588 * A private symbol is used instead of the :EMPTY keyword previously
3589 used to mark empty slots in hash tables. Thus
3590 (DEFVAR *HT* (MAKE-HASH-TABLE))
3591 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3592 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3593 now does what ANSI says that it should. (You can still get
3594 similar noncompliant behavior if bang on the hash table
3595 implementation with all the symbols you get back from
3596 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3597 This breaks binary compatibility, since tests for equality to
3598 :EMPTY are wired into things like the macroexpansion of
3599 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3601 * There's now a minimal placeholder implementation for CL:STEP,
3602 as required by ANSI.
3603 * An obscure bug in the interaction of the normal compiler, the byte
3604 compiler, inlining, and structure predicates has been patched
3605 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3606 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3608 * Missing ordinary arguments in a macro call are now detected even
3609 when the macro lambda list contains &KEY or &REST.
3610 * The debugger no longer complains about encountering the top of the
3611 stack when you type "FRAME 0" to explicitly instruct it to go to
3612 the top of the stack. And it now prints the frame you request even
3613 if it's the current frame (instead of saying "You are here.").
3614 * As specified by ANSI, the system now always prints keywords
3615 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3617 * The default initial SIZE of HASH-TABLEs is now smaller.
3618 * Type information from CLOS class dispatch is now propagated
3619 into DEFMETHOD bodies, so that e.g.
3620 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3622 is now basically equivalent to
3623 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3624 (DECLARE (TYPE SINGLE-FLOAT X))
3626 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3627 operation, without having to do run-time type dispatch.
3628 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3629 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3630 * Since I don't seem to be making big file reorganizations very often
3631 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3632 is overflowing my ability to conveniently back them up), I've finally
3633 checked the system into CVS. (The CVS repository is on my home system,
3634 not at SourceForge -- putting it on SourceForge might come later.)
3635 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3636 high-level GC-NOTIFY-FOO functions send their output. (There's
3637 still very little control of where low-level verbose GC functions
3638 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3639 less than it used to -- the GC-NOTIFY-FOO functions are now under
3640 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3641 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3642 in only one place in the source code, and propagates it automatically
3643 everywhere that it's needed. Thus e.g. when I bump the version from
3644 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3645 * The C source files now include boilerplate legalese and documentation
3646 at the head of each file (just as the Lisp source files already did).
3647 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3648 to his page will be replaced with a link to his new CLiki service.
3650 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3652 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3653 * The build process now looks for GNU make under the default name "gmake",
3654 instead of "make" as it used to. If GNU make is not available as "gmake"
3655 on your system, you can change this default behavior by setting the
3656 GNUMAKE environment variable.
3657 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3658 can build without error under CMU CL.
3660 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3662 * There is now a partial SBCL user manual (with some new text and some
3663 text cribbed from the CMU CL manual).
3664 * The beginnings of a profiler have been added (starting with the
3665 CMU CL profiler and simplifying and cleaning up). Eventually the
3666 main interface should be through the TRACE macro, but for now,
3667 it's still accessed through vaguely CMU-CL-style functions and macros
3668 exported from the package SB-PROFILE.
3669 * Some problems left over from porting CMU CL to the new
3670 cross-compilation bootstrap process have been cleaned up:
3671 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3672 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3673 part when getting rid of the compiler *BACKEND* stuff.)
3674 ** The value of *NULL-TYPE* was screwed up, because it was
3675 being initialized before the type system knew the final
3676 definition of the 'NULL type. This screwed up several key
3677 optimizations in the compiler, causing inefficiency in all sorts
3678 of places. (I found it because I wanted to understand why
3679 GET-INTERNAL-RUN-TIME was consing.)
3680 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3681 PROCLAIM FTYPE information. Unfortunately this broke binary
3682 compatibility again, since now the forms output by DEFGENERIC
3683 to refer to functions which didn't exist in 0.6.3.
3684 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3685 can use the new (as of 0.6.3) transform for SOME into MAP into
3687 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3688 values to (INTEGER 0 1000000), so that the time code will no longer
3689 occasionally get blown up by Linux returning 1000000 microseconds
3690 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3691 its output conform to the ANSI spec. (Alas, this makes its output
3692 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3693 * A full call to MAP NIL with a single sequence argument no longer conses.
3694 * fixes to problems pointed out by Martin Atzmueller:
3695 * The manual page no longer talks about multiprocessing as though
3696 it were currently supported.
3697 * The ILISP support patches have been removed from the distribution,
3698 because as of version 5.10.1, ILISP now supports SBCL without us
3699 having to maintain patches.
3700 * added a modified version of Raymond Toy's recent CMU CL patch for
3701 EQUALP comparison of HASH-TABLE
3703 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3705 * The system still can't cross-compile itself with
3706 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3707 entails), but at least it can compile more of itself that way
3708 than it used to be able to, and various buglets which were uncovered
3709 by trying to cross-compile itself that way have now been fixed.
3710 * This release breaks binary compatibility again. This time
3711 at least I've incremented the FASL file format version to 2, so that the
3712 problem can be detected reliably instead of just causing weird errors.
3713 * various new style warnings:
3714 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3715 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3716 ** using the lexical binding of a variable named in the *FOO* style
3717 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3718 as specified by ANSI.
3719 * *RANDOM-STATE* is no longer automatically initialized from
3720 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3721 default behavior of the system is to repeat its behavior every time
3722 it's run. If you'd like to change this behavior, you can always
3723 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3724 old convention there was no comparably easy way to get the system to
3725 repeat its behavior every time it was run.
3726 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3727 FTYPE declarations has been removed, in favor of their ANSI
3728 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3729 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3730 the types of their sequence arguments can be determined at compile time.
3731 This is done through a new open code expansion for MAP which eliminates
3732 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3733 argument types can be determined at compile time.
3734 * The optimizer now transforms COERCE into an identity operation when it
3735 can prove that the coerced object is already of the correct type. (This
3736 can be a win for machine generated code, including the output of other
3737 optimization transforms, such as the MAP transform above.)
3738 * Credit information has been moved from source file headers into CREDITS.
3739 * Source file headers have been made more standard.
3740 * The CASE macro now compiles without complaining even when it has
3743 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3745 * (Note that the way that the PCL macroexpansions were rewritten
3746 to accommodate the change in DEFGENERIC below breaks binary
3747 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3748 not run under sbcl-0.6.2. Once we get out of alpha releases,
3749 i.e. hit release 1.0.0, we'll probably try to maintain binary
3750 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3751 and sbcl-1.4.4. Until then, however, it might be fairly common
3752 for maintenance releases to break binary compatibility.)
3753 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3755 * The reporting of unhandled signals has been changed to print some
3756 explanatory text as well as the report form. (Previously only
3757 the report form was printed.)
3758 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3759 it defines, so that the compiler no longer issues undefined function
3760 warnings for compiled-but-not-yet-loaded generic functions.
3761 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3762 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3763 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3764 section of the ANSI spec are supported.
3765 * The "" nickname for the "KEYWORD" package has been removed.
3766 The reader still handles symbol tokens which begin with a package marker
3767 as keywords, but it doesn't expose its mechanism for doing so in the
3768 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3769 * The system now issues STYLE-WARNINGs for contradictory TYPE
3770 proclamations. (Warnings for contradictory FTYPE proclamations would
3771 be nice too, but those can't be done usefully unless the type system
3772 is made smarter about FUNCTION types.)
3773 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3774 systematized, so that "*target-*.lisp is supposed to exist only on the
3775 target and imply that there's a related file which exists on the
3776 host, and *host-*.lisp is supposed to exist only on the host and imply
3777 that there's a related file which exists on the target. This involves a
3778 lot of renaming. Hopefully the acute confusion caused by the renaming
3779 will be justified by the reduction in chronic confusion..
3780 ** runtime-type.lisp -> early-target-type.lisp
3781 ** target-type.lisp -> late-target-type.lisp
3782 ** early-host-format.lisp -> early-format.lisp
3783 ** late-host-format.lisp -> late-format.lisp
3784 ** host-error.lisp -> misc-error.lisp
3785 ** early-error.lisp -> early-target-error.lisp
3786 ** late-error.lisp -> late-target-error.lisp
3787 ** host-defboot.lisp -> early-defboot.lisp
3788 ** code/misc.lisp -> code/target-misc.lisp
3789 ** code/host-misc.lisp -> code/misc.lisp
3790 ** code/numbers.lisp -> code/target-numbers.lisp
3791 ** code/early-numbers.lisp -> numbers.lisp
3792 ** early-host-type.lisp -> early-type.lisp
3793 ** late-host-type.lisp -> late-type.lisp
3794 ** host-typep.lisp -> typep.lisp
3795 ** load.lisp -> target-load.lisp
3796 ** host-load.lisp -> load.lisp
3797 ** host-disassem.lisp -> disassem.lisp
3798 ** host-insts.lisp -> insts.lisp
3799 ** byte-comp.lisp -> target-byte-comp.lisp
3800 ** host-byte-comp.lisp -> byte-comp.lisp
3801 ** host-signal.lisp -> signal.lisp
3802 ** host-defstruct.lisp -> defstruct.lisp
3803 ** late-target-type.lisp -> deftypes-for-target.lisp
3804 Furthermore, several other previously target-only files foo.lisp (e.g.
3805 hash-table.lisp and random.lisp) have been split into a target-and-host
3806 foo.lisp file and a target-only target-foo.lisp file, with their key type
3807 definitions in the target-and-host part, so that the cross-compiler will
3808 know more about target types.
3809 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3810 gone away. In their place are various *BACKEND-FOO* variables
3811 corresponding to the slots of the old structure.
3812 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3813 into the target SBCL has been fixed.
3814 * The chill.lisp system for loading cold code into a running SBCL
3816 * Support for the CMU CL "scavenger hook" extension has been removed.
3817 (It was undocumented and unused in the CMU CL sources that SBCL was
3818 derived from, and stale in sbcl-0.6.1.)
3819 * Various errors in the cross-compiler type system were detected
3820 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3821 (enabling various consistency checks). Many of them were fixed,
3822 but some hard problems remain, so the compiler is back to
3823 running without *TYPE-SYSTEM-INITIALIZED* for now.
3824 * As part of the cross-compiler type system cleanup, I implemented
3825 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3826 * I have started adding UNCROSS calls throughout the type system
3827 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3828 on cross-compiler input files will be able to go away, and various
3830 * CONSTANTP now returns true for quoted forms (as explicitly required
3833 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3835 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3836 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3837 (MAKE-LIST -1). In the longer term, it ought to become true
3838 that declarations are assertions even at SAFETY 1. For now, it's not
3839 quite true even at SAFETY 3, but it's at least more nearly true..
3840 (Note that this change seems to increases the size of the system by
3841 O(5%) and to decrease the speed of the compiler by 20% or more.)
3842 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3843 for the problem of printing dozens of lines of distracting information
3844 about low-level system machinery as part of the top stack frame
3845 on entry to the debugger when an undefined function was called.
3846 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3847 is not reset to COMMON-LISP-USER.
3848 * Compilation of stuff related to dyncount.lisp has been made conditional
3849 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3850 smaller. The various dyncount-related symbols have been moved into
3851 a new "SB-DYNCOUNT" package.
3852 * tty-inspect.lisp has been renamed to inspect.lisp.
3853 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3854 feature has gone away. (When we eventually port to other flavors of
3855 libc and/or Unix, we'll try to make the differences between flavors
3856 invisible at the user level.)
3857 * Various other *FEATURES* tags, and/or their associated conditionals,
3858 have been removed if obsolescent, or given better documentation, or
3859 sometimes given more-mnemonic names.
3861 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3863 * tidied up "make.sh" script
3864 * tidied up system directory structure
3865 * better "clean.sh" behavior
3866 * added doc/FOR-CMUCL-DEVELOPERS
3867 * many many small tweaks to output format, e.g. removing possibly-confusing
3868 trailing #\. character in DESCRIBE-INSTANCE
3869 * (EQUALP #\A 'A) no longer signals an error.
3870 * new hashing code, including EQUALP hashing
3871 * tidied up Lisp initialization and toplevel
3872 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3873 * command line argument processing
3874 * added POSIX-GETENV function to deal with Unix-ish environment variables
3875 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3876 terminating SBCL on EOF
3877 * non-verbose GC by default
3878 * There is no more "sbcl" shell script; the sbcl file is now the C
3879 runtime executable (just like CMU CL).
3880 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3882 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3883 symbol and package information as Lisp native symbol and package objects
3884 instead of strings naming symbols and strings naming packages. This way,
3885 whenever packages are renamed (as in warm init), debug information is
3886 transformed along with everything else.
3887 * tweaked the optimization policy declarations which control the building
3888 of SBCL itself. Now, among other things, the system no longer saves
3889 source location debugging information. (This helps two problems at once
3890 by reducing SBCL size and by keeping SBCL from trying to look for its
3891 sources -- which may not exist -- when reporting errors.)
3892 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3893 debugging and testing purposes
3894 * cleaned up printing, making the printer call PRINT-OBJECT for
3895 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3896 methods, giving nearly-ANSI behavior
3897 * converted almost all special variables to use *FOO* naming convention
3898 * deleted PARSE-TIME functionality, since it can be done portably
3899 * moved some files out of cold init into warm init
3900 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3902 * regularized formatting of source files
3903 * added an install.sh script
3904 * fixed ridiculous memory usage of cross-compiler by making
3905 compiler/alloc.lisp not try to do pooling unless it can hook
3906 itself into the GC of the cross-compilation host. Now the system
3907 builds nicely on my old laptop.
3908 * added :SB-ALLOC in target-features.lisp-expr
3909 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3910 was not implemented)
3911 * re-did condition handling and note reporting in the compiler. Notes
3912 are no longer handled by signalling conditions. Style warnings
3913 and warnings are handled more correctly and reported in such a way
3914 that it's easy to find one or the other in your output (so that you
3915 can e.g. figure out which of many problems caused COMPILE-FILE to
3917 * changed the severity of several compiler warnings from full WARNING
3918 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3919 compiler note reporting so that it doesn't use the condition system
3920 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3921 COMPILE-FILE command)
3922 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3923 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3924 no longer supported, since their implementation was deeply intertwingled
3925 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3926 * removed bogus "support" for compiler macros named (SETF FOO), and
3927 removed the compiler macro for SETF INFO (but only after making a fool
3928 of myself on the cmucl-imp mailing list by posting a bogus patch for
3929 DEFINE-COMPILER-MACRO..)
3930 * Compiled files containing forms which have side effects on the Lisp
3931 reader (such as DEFPACKAGE forms) are now handled more correctly.
3932 (Compiler queuing of top level lambdas has been suppressed by setting
3933 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3934 * deleted various currently-unused source files, e.g. gengc.lisp. They
3935 may be added back at some point e.g. when porting to other architectures,
3936 but until they are it's distracting to distribute them and to try to
3938 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3939 there were so many of them they're just distractions, and UNCROSS is
3940 known to be able to handle the current sources
3941 * moved PROFILE functionality into TRACE, so that it will be clear
3942 how the wrapping and unwrapping of functions when you profile them
3943 interacts with the wrapping and unwrapping of functions when you
3944 trace them. (Actually, the functionality isn't there yet, but at least
3945 the interface specification is there. Hopefully, the functionality will
3946 arrive with some future maintenance release.)
3947 * removed host-oops.lisp
3948 * changed signature of QUIT function to allow UNIX-CODE argument
3949 * fixed READ-SEQUENCE bug
3950 * tweaked verbose GC output so that it looks more like the progress
3951 output that ANSI specifies for functions like LOAD
3952 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3953 * added <http://sbcl.sourceforge.com> to the banner information printed by