1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
3 * bug fix: fixed input, output and error redirection in RUN-PROGRAM
4 for win32. (thanks to Mike Thomas and Yaroslav Kavenchuk)
5 * feature: implemented the READER-METHOD-CLASS and
6 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
8 * optimization: faster LOGCOUNT implementation on x86 and x86-64
10 * optimization: hashing of general arrays and vectors has been
11 improved. (reported by Any Fingerhut)
12 * enhancement: SB-INTROSPECT is now able to find definitions of
13 profiled functions. (thanks to Troels Henriksen)
14 * enhancement: compiler-macro expansion applies now to FUNCALL forms
16 * fixed bug #337: use of MAKE-METHOD in method combination now works
17 even in the presence of user-defined method classes. (reported by
18 Bruno Haible and Pascal Costanza)
19 * fixed bug #339(c): if there are applicable methods not part of any
20 long-form method-combination group, call INVALID-METHOD-ERROR.
21 (reported by Bruno Haible)
22 * bug fix: extensions of MAKE-METHOD-LAMBDA which wrap the
23 system-provided lambda expression no longer cause warnings about
24 unbound #:|pv-table| symbols.
25 * bug fix: improved the handling of type declarations and the
26 detection of violations for keyword arguments with non-constant
28 * bug fix: potentially erronous calls to PATHNAME and
29 MERGE-PATHNAMES were being flushed in some cases.
30 (reported by Richard Kreuter)
31 * bug fix: compiled calls to TYPEP were mishandling obsolete
32 instances. (reported by James Bielman and Attila Lendvai)
33 * bug fix: format strings with ~> without matching ~< no longer
34 trigger an AVER, but signal an understandable error instead.
35 (reported by Antonio Martinez)
36 * bug fix: specifying an output-file without a pathname-type for
37 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
38 of input-file instead of "fasl". (reported by Robert Dodier)
39 * bug fix: compiler-macro lambda-list parsing of FUNCALL forms.
40 (reported by James Y Knight).
41 * bug fix: compiler-macros-function did not consider the environment
42 argument for shadowing by local functions.
43 * bug fix: compiler-macros expansion was inhibited by local INLINE
45 * bug fix: inline expansions of known functions were subject to
46 step-instrumentation in high debug policies, leading to problems
49 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
50 * added support for the ucs-2 external format. (contributed by Ivan
52 * minor incompatible change: pretty printing of objects of type
53 (cons symbol) is, in the default pprint-dispatch-table, now
54 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
56 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
57 called later in a class's lifetime, possibly as late as when the
58 first instance of the class is created. Previously,
59 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
60 class became finalizeable.
61 * fixed bug: FILE-POSITION sometimes returned inconsistent results
62 for multibyte external-format streams. (thanks to "vbzoli")
63 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
64 with :ALLOCATION :CLASS inherited from superclasses of the
66 * fixed bug: anonymous classes can now be created using the :NAME
67 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
68 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
69 * fixed bug: core-files saved with :EXECUTABLE T can again be
70 executed when SBCL_HOME isn't set. (reported by James Knight)
71 * fixed bug: toplevel LOCALLY forms with declarations could
72 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
73 * fixed bug: printing from several different threads using different
74 values of *print-case* could cause invalid output, due to
75 some internal special variables of the printer not being bound
76 thread-locally (reported by Max Mikhanosha)
77 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
78 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
79 (reported by Pascal Costanza)
80 * fixed bug: FINALIZE-INHERITANCE is called from
81 REINITIALIZE-INSTANCE on classes when the class has previously
82 been finalized, as required by AMOP.
83 * minor code generation optimizations:
84 ** better register allocation in CLOS dispatching functions
85 ** overflow detection when coercing signed bytes to fixnums on x86-64
86 ** is now implemented with one IMUL instruction instead of three shifts
87 ** more efficient bit-vector access on x86 and x86-64
88 ** more efficient access to raw structure slots on x86 and x86-64
89 * fixed some bugs revealed by Paul Dietz' test suite:
90 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
93 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
94 * feature: thread support on Solaris/x86, and experimental thread support
96 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
98 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
99 default method for SLOT-UNBOUND.
100 * minor incompatible change: prevent the user from specializing the
101 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
102 somewhat counter to the intent of the protocol, I (CSR) think, and
103 additionally it just doesn't work in SBCL as currently
104 implemented, thanks to optimizations (that are always valid for
105 the other three SLOT-VALUEish functions, but not for the setter).
106 * minor incompatible change: the :SB-LDB feature is now enabled by
107 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
108 the low-level debugger.
109 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
110 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
111 down to an SB-SYS:FD-STREAM, allowing direct communication in
112 more cases. Temporary files and pipes are still used as a fallback
114 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
115 on the non-reentrant C functions localtime(3) and gmtime(3).
116 * bug fix: global optimization policy was not visible in LOCALLY and
118 * bug fix: class objects can be used as specializers in methods.
119 (reported by Pascal Costanza)
120 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
121 with a type error. (reported by blitz_ on #lisp)
122 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
123 failed with a type error. (reported by Pascal Bourguignon)
124 * bug fix: merging pathnames against defaults with :DIRECTORY
125 starting with '(:RELATIVE :BACK) should preserve the :BACK.
126 (reported by James Y Knight)
127 * bug fix: saving large (>2GB) cores on x86-64 now works
128 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
129 constant index and a (simple-array (signed-byte 32)) array
130 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
131 error (patch by Robert J. Macomber)
132 * feature: TIME now displays the amount of run-time spent in GC
133 * bug fix: The debugger now does a better job of respecting
134 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
135 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
136 workaround for bug 403.)
137 * bug fix: floating point exception handling now works on FreeBSD
138 (thanks to NIIMI Satoshi)
139 * fixed some bugs revealed by Paul Dietz' test suite:
140 ** MISC.641: LET-conversion were not supposed to work in late
142 * improvements to the Win32/x86 port:
143 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
144 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
146 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
147 * new feature: source path information is generated for macro-expansion
148 errors for use in IDE's like Slime (thanks to Helmut Eller)
149 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
150 * bug fix: compiler does not loop forever on an invalid type in
152 * improvement: compilation of most CLOS applications is significantly
154 * optimization: added a limited bytecode compiler for simple toplevel
155 forms, speeding up compilation and FASL loading
157 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
158 * minor incompatible change: in sbcl-0.9.11 (but not earlier
159 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
160 into subclasses of the finalized class. Now user calls to
161 FINALIZE-INHERITANCE finalize just the one class, and calls by the
162 system return before any subclasses are finalized.
163 * minor incompatible change: The reader no longer ignores errors
164 regarding non-existent packages in #+ and #- feature tests.
165 * new feature: command line options --no-sysinit, --no-userinit to
166 inhibit loading the corresponding init files
167 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
168 for sending data through UDP sockets (thanks to François-René Rideau)
169 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
170 error is signalled instead (thanks to François-René Rideau)
171 * minor incompatible change: Error signalling behaviour of lexical
172 operations violating package locks has changed slightly. Refer to
173 documentation on package locks for details.
174 * enhancement: EVAL can process IF-expressions without resorting to the
176 * bug fix: Some operations involving SIN, COS, or TAN weren't being
177 constant-folded properly due to missing out-of-line functions.
178 (reported by Mika Pihlajamäki)
179 * bug fix: LISTEN sometimes returned T even in cases where no data was
180 immediately available from the stream
181 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
182 were reversed. (reported by Levente Mészáros)
183 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
184 when given a vector displaced to an adjustable array. (reported
186 * bug fix: derivation of float boundaries from numbers outside the
187 appropriate float range (reported by John Wiseman)
188 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
190 * bug fix: precomputing cache entries for generic functions with
191 some subclasses of specializers as yet invalid does not attempt to
192 fill a cache line with a negative offset. (reported by Levente
194 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
195 allow condition class objects as arguments to DOCUMENTATION and
196 (SETF DOCUMENTATION); only find and set documentation for
197 structure names for the STRUCTURE doc-type. (suggested by Gary
199 * improvements to the Win32/x86 port:
200 ** (user-homedir-pathname) and default initialization file
201 locations now know about the user's "Documents and Settings"
202 directory (thanks to Yaroslav Kavenchuk)
203 ** run-program is implemented (thanks to Mike Thomas)
204 ** sockets support (thanks to Timothy Ritchey)
205 ** better backtrace support (thanks to Alastair Bridgewater)
206 ** sb-grovel supported
207 ** asdf-install and sb-posix work somewhat
208 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
209 * improvements to the Solaris/x86 port:
210 ** works on Solaris 11/Solaris Express
211 ** floating-point exception handling support
212 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
213 * fixed some bugs revealed by Paul Dietz' test suite:
214 ** REMOVE-METHOD returns its generic function argument even when
215 the method is not one of the generic functions' methods.
216 ** objects declared as MEMBER types can be admissible arguments to
218 ** printing characters should simply be printed by the FORMAT ~:C
220 ** compiler failure when compiling functions with hairy constant
221 defaults for optional parameters.
222 ** compiler produces wrong code when MAYBE-INLINE-expanding a
223 function, which is already optimized.
225 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
226 * new platform: experimental support for SBCL x86/Darwin, including
227 MacOS X 10.4.5 on Intel.
228 * new feature: Unicode character names are now known to the system
229 (through CHAR-NAME and NAME-CHAR).
230 * new feature: the filesystem location of SBCL's core file is
231 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
232 * minor incompatible change: the contrib modules SB-POSIX and
233 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
234 this change is to make it easier to distribute
235 location-independent binaries.
236 * bug fix: as implied by AMOP, standardized classes no longer have
237 slots named by external symbols of public packages. (reported by
239 * optimization: calling functions via constant symbols -- as in
240 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
241 via the function object as in (FUNCALL #'FOO).
242 * optimization: on x86 and x86-64, the calling convention for the
243 case of dealing with an unknown number of values has been altered
244 to be friendlier to the prediction heuristics implemented,
245 particularly with reference to CALL and RET pairing. (thanks to
246 Alastair Bridgewater)
247 * optimization: on x86, the code for access to thread-local symbol
248 values and binding/unbinding of thread-local symbols is smaller.
249 (thanks to Alastair Bridgewater)
250 * enhancement: CONSTANTP is now able to determine constantness of
251 more complex forms, including calls to constant-foldable standardized
252 functions and some special forms beyond QUOTE.
253 * fixed bug: occasional GC crashes on Solaris/x86
254 * optimization: x86-64 supports stack allocation of results of simple
255 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
256 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
257 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
258 (thanks to James Knight)
259 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
260 by the standard (thanks to Stephen Compall)
262 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
263 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
264 be used for bundling the runtime and the core file into one
265 executable binary. This feature is not currently supported on all SBCL
266 platforms. (thanks to James Bielman and NIIMI Satoshi)
267 * new feature: a generational or ephemeral garbage collector is now
268 the default on the PowerPC platform (both Linux and Darwin). The
269 old Cheney (stop and copy) collector is a build-time option.
270 (thanks to Cyrus Harmon, after Raymond Toy)
271 * minor incompatible change: the method by which SBCL finds its
272 contributed modules has changed; it no longer relies on symbolic
273 links from an $SBCL_HOME/systems directory, but searches directly
274 in the subdirectories of $SBCL_HOME.
275 * enhancement: the dynamic heap size on the Linux/PPC platform is
276 markedly larger, even using the older Cheney garbage collector.
277 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
278 certain complicated string types. (reported by Gary King)
279 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
280 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
281 * fixed bug: DOCUMENTATION on structures no longer signals an
282 error if no structure type of the right name exists. (reported by
284 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
285 applicable non-standard primary or :AROUND method on
286 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
287 no longer cause unbound variable errors. (reported by Kevin Reid)
288 * fixed bug: metacircle resolution in cases where methods have slots
289 added before the slots from STANDARD-METHOD. (reported by Jean
291 * fixed bug: the Power PC assembly code for calling into lisp
292 sometimes computed its return address wrongly (depending on the
293 15th bit of the address, fixed at link time). This used to
294 manifest itself in a segmentation violation while building PCL.
295 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
296 many others over the years)
297 * fixed bug: ROOM no longer reports silly numbers for stack usage if
298 the stack pointer (treated as a signed integer) is negative.
299 (thanks to Peter van Eynde)
301 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
302 * new platform: experimental support for the Windows operating
303 system has been added. (thanks to Alastair Bridgewater)
304 * fixed several bugs in and robustified the PPC FFI (including
305 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
306 * bug fix: allow non-simple string symbol names (reported by Paul
308 * bug fix: interrupt handling on NetBSD (thanks to Richard M
310 * bug fix: saving a core corrupted callbacks on x86/x86-64
311 * bug fix: closed a loophole in metacircularity detection and
312 grounding in the PCL implementation of CLOS.
313 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
314 (reported by tomppa on #lisp)
315 * bug fix: clear the direction flag on Lisp -> C transitions, as
316 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
317 (reported by Andrej Grozin and Hendrik Maryns)
318 * optimization: major improvements to GC efficiency on GENCGC platforms
319 * optimization: faster implementation of EQUAL
320 * optimization: emit more efficient opcodes for some common
321 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
322 * optimization: several other minor code-generation improvements on x86-64
323 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
325 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
326 * minor incompatible change: (SETF CLASS-NAME) and (SETF
327 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
328 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
329 by ANSI as a generic function, and for consistency (SETF
330 GENERIC-FUNCTION-NAME) is treated likewise.
331 * fixed bug #233.b: make constraint propagation notice when a variable
332 value is changed after it is referenced but before it is used
333 * fixed bug #296: no more arbitrary behaviour from filenames with
334 odd characters as --load arguments. (reported by Adam Warner)
335 * fixed bug #390: :CHARACTER-SET pathname components now work as
336 expected. (reported by Tim Daly Jr)
337 * fixed bug #391: complicated :TYPE intersections in slot
338 definitions no longer cause an error in PCL internals.
339 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
340 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
341 returning the number of octets which would be written to the
342 file-stream. (thanks to Robert J. Macomber)
343 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
344 arguments to the internal stream functions. (thanks to David
346 * bug fix: stack exhaustion is now handled on the x86/SunOS
347 platform. (thanks to Daisuke Homma)
348 * bug fix: more accurate ROOM results on GENCGC platforms
349 * optimization: improved type inference for arithmetic-for
350 index variables in LOOP
351 * optimization: faster floating-point SQRT on x86-64
352 * fixed some bugs revealed by Paul Dietz' test suite:
353 ** DOCUMENTATION returns NIL instead of "" for method combinations
354 that don't have a docstring
356 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
357 * minor incompatible change: (SETF CLASS-NAME) and (SETF
358 GENERIC-FUNCTION-NAME) are no longer generic functions, and
359 therefore are not customizeable by user code (as seems to be at
360 least permitted and maybe required by AMOP). As a consolation,
361 however, the SBCL implementation of these functions now calls
362 REINITIALIZE-INSTANCE as specified by AMOP.
363 * bug fix: slot-definition documentation is propagated as per ANSI
364 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
365 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
366 argument, as do eslotd initialization methods. (from Pascal
367 Costanza's "Closer" project)
368 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
369 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
371 * bug fix: it is now possible to have more than one subclass of
372 STANDARD-GENERIC-FUNCTION without causing stack overflow.
373 (reported by Bruno Haible, Pascal Costanza and others)
374 * bug fix: the dependent update protocol now works for generic
375 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
377 * bug fix: condition-class instances corresponding to
378 DEFINE-CONDITION forms are now created eagerly. (reported by
379 Kalle Olavi Niemitalo on comp.lang.lisp)
380 * bug fix: floating point printing is more accurate in some
381 circumstances. (thanks to Simon Alexander)
382 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
383 merged with *DEFAULT-PATHNAME-DEFAULTS*.
384 * bug fix: callbacks on OS X now preserve stack-alignment.
385 (thanks to Cyrus Harmon)
386 * enhancement: the x86-64 disassembler is much better at
387 disassembling SSE instructions. (thanks to Lutz Euler)
388 * enhancement: CHECK-TYPE on a variable now allows type-inference
389 (thanks to Vincent Arkesteijn)
390 * optimization: improved performance of EUC-JP external format.
391 (thanks to NIIMI Satoshi)
392 * optimization: performance improvements to IO on file streams of
393 :ELEMENT-TYPE CHARACTER
394 * optimization: much faster memory allocation on x86-64
395 * optimization: faster garbage collections (latency and throughput)
397 * optimization: faster bignum subtraction and division on x86 and x86-64
398 (thanks to Lutz Euler)
399 * optimization: more accurate type-inference for loops that use a
400 floating point index variable or a negative step.
402 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
403 * bug fix: add a workaround to SBCL looping infinitely at startup on
404 Linux kernels with apparently buggy implementations of personality().
405 (thanks to Svein Ove Aas)
406 * bug fix: Unicode symbols are correctly printed in LDB backtraces
407 (thanks to David Lichteblau)
408 * bug fix: local bindings shadow global symbol macros for
409 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
410 * optimization: non-open coded uses of numeric comparison operators
411 (e.g. >) no longer cons when called with more than one parameter
412 on platforms supporting dynamic-extent allocation.
413 * enhancement: saving cores with foreign code loaded is now supported
414 on MIPS/Linux in addition to the previously supported platforms.
415 * bug fix: FIND-RESTART now tests for activity, not applicability when given
416 a restart object as identifier. (reported by Helmut Eller for CMUCL)
417 * bug fix: division by zero in sb-sprof when no samples were collected
418 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
419 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
420 (thanks to Alexander Kjeldaas)
421 * new feature: ignoring the return values of destructive functions
422 like NREVERSE signals a compile-time style-warning.
423 (thanks to Kevin Reid)
424 * new feature: the alignment of alien structure fields can be
425 explicitly specified. (thanks to Cyrus Harmon)
426 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
427 and dump core on SIGQUIT
429 ** incompatible change: threads do not inherit values of specials
430 from their parents (see manual)
431 ** bug fix: threads stacks belonging to dead threads are freed by the
432 next exiting thread, no need to gc to collect thread stacks anymore
433 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
434 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
435 ** bug fix: dynamic variable and thread start related gc lossage
436 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
438 * fixed some bugs revealed by Paul Dietz' test suite:
439 ** SUBTYPEP is slightly more accurate on heinously complicated
440 CONS types where some of the members have uncertain (in the
441 NIL, NIL sense) type relationships to each other.
443 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
444 no longer purified unless :PURIFY T is explicitly specified.
445 ** Non-purified cores are significantly smaller than before
447 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
448 * new feature: timers based on Zach Beane's excellent timer package
449 * added support for the following external formats: koi8-u,
450 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
451 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
452 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
453 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
454 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
455 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
456 * incompatible change: a threaded SBCL will no longer revert to
457 non-threaded mode on non-NPTL systems, but refuse to start entirely.
458 * bug fix: interrupts are disabled until startup is complete; no
459 more sigsegvs when receiving a signal to soon
460 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
462 * bug fix: add a workaround for the memory randomization features in
463 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
464 workaround will only be in effect on systems with the proc filesystem
466 * bug fix: printing objects of type HASH-TABLE signals a
467 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
469 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
470 have been running for over 50 days. (reported by Gilbert Baumann)
471 * bug fix: the logic for getting names of functions gets less
472 confused when confronded with alternate-metaclass
473 funcallable-instances. (reported by Cyrus Harmon)
474 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
475 consistent, even on internal alternate-metaclass objects.
476 * bug fix: flush closure information collected by physical
477 environment analysis prepass before the main pass. (bug reported
479 * bug fix: compiler pack phase does not modify a hashtable, which is
480 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
481 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
482 may be expanded. (reported by Alexander <malishev> on c.l.l)
483 * bug fix: no more style-warnings when compiling code using LAST
484 under high-SPEED low-SPACE settings. (reported by David Wallin
486 * fixed some incompatibilities between SBCL's MOP and the MOP
488 ** the METAOBJECT class is now implemented;
489 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
490 STANDARD-OBJECT, as required;
491 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
492 now compatible (as required by VALIDATE-SUPERCLASS); there
493 remains a consistency requirement over the presence or absence
494 of the FUNCTION class in the superclass list of finalized
495 classes; see the manual for more details;
496 ** the :AROUND method for COMPUTE-SLOTS on
497 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
498 requested slot ordering.
500 ** bug fix: parent thread now can be gc'ed even with a live
502 ** bug fix: binding a special with PROGV to no value is not
503 the same as that symbol not having been bound (thanks to
505 ** bug fix: binding specials is thread safe (thanks to
507 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
508 and SB-SPROF do not die with 'no handler for signal XX in
509 interrupt_handle_now(..)' anymore
510 ** bug fix: WITH-TIMEOUT works with multiple running threads
511 * fixed some bugs revealed by Paul Dietz' test suite:
512 ** ENSURE-GENERIC-FUNCTION should take a method class object for
513 the :method-class keyword argument.
515 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
516 * new port: the Solaris operating system on x86 processors is now
517 mostly supported, though some rough edges in the environment
518 remain. (thanks to Daisuke Homma)
519 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
520 space for its heap. (thanks to Thiemo Seufer)
521 * optimization: SBCL on MIPS platforms now supports dynamic-extent
522 closures. (thanks to Thiemo Seufer)
523 * minor incompatible change: eof selects abort in the debugger.
524 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
525 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
527 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
528 deprecated, and will go away in a future revision of SBCL.
529 * minor incompatible change: GC-ON and GC-OFF are no longer
530 implemented with a counter, it does not matter how many times gc
531 is switched on or off
532 * bug fix: discriminating functions for generic function classes
533 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
534 make invalid assumptions about method precedence order. (reported
536 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
537 platforms. (thanks to Thiemo Seufer)
538 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
539 * bug fix: correct pathname printing with printer escaping is on.
540 (thanks to Kevin Reid)
541 * bug fix: complex VOP definitions in "user-space" no longer trigger
542 package locks. (reported by Zach Beane)
543 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
544 for funcallable-instance functions, and is no different from
546 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
547 characters in the current readtable. (reported by Nicholas Neuss)
548 * bug fix: SUBTYPEP on various CONS types returns more a more
549 accurate acknowledgment of its certainty.
550 * optimizations: REMOVE-DUPLICATES now runs in linear time on
551 lists in some cases. This partially fixes bug 384.
552 * flush all standard streams before prompting in the REPL and the
554 * bug fix: signal handling and triggering gc do not conflict
555 directly anymore, in particular a high frequency sb-sprof does
556 not prevent gc from running
557 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
558 approximation for timezone and DST information between the
559 universal time epoch and the smallest negative 32-bit time_t.
560 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
561 year 1899 so long as the encoded time is non-negative.
562 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
563 the correct answer for non-constant rotate amounts when performing
564 an inline 32-bit rotation.
566 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
567 there is only one thread in the session
568 ** bug fix: memory leak for streams created in one thread and
569 written to in another
570 ** bug fix: lockup when compiled with gcc4
571 ** bug fix: race that allows the gc to be triggered when gc is
573 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
574 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
576 ** bug fix: finished threads are gc'ed properly
577 * fixed some bugs revealed by Paul Dietz' test suite:
578 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
579 has arguments with a different set of applicable methods from
580 the orignal arguments.
581 ** The type error thrown by MAP now has a correct expected-type
583 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
584 name a compiled function.
585 ** Three MISC tests where a large bignum was improperly coerced to
586 a float that couldn't represent that bignum during type
587 derivation were fixed.
588 ** SUBTYPEP can now handle types involving the negation of a
589 list-form FUNCTION type.
590 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
591 as (COMPLEX (AND RATIO (NOT FIXNUM))).
592 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
594 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
595 * New feature: Experimental support for bivalent streams: streams
596 opened with :element-type :default now allow character and binary
597 (unsigned-byte 8) I/O
598 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
599 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
601 * bug fix: no more highly sporadic "couldn't check whether ~S is
602 readable" when reading a stream and an interrupt hits in the middle
603 of a select system call
604 * compiler better recognizes complex arrays (reported by Eduardo
606 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
607 for variables on SBCL built with linkage-tables. (reported by Luis
609 * various error reporting improvements.
610 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
611 (Thanks to Thiemo Seufer)
612 * bug fix: nested reader invokations work correctly
613 * bug fix: it is possible to have simultaneous references to foreign
614 code and foreign data with the same name.
616 ** added x86-64 support
617 ** incompatible change: the threading api now works with thread
618 objects instead of thread ids
619 ** bug fix: threads are protected from signals and interruption when
620 starting up or going down
621 ** bug fix: a race where an exiting thread could lose its stack to gc
622 ** bug fix: don't halt on infinite error in threads if possible
623 ** fixed numerous gc deadlocks introduced in the pthread merge
624 ** bug fix: fixed thread safety issues in read and print
625 ** bug fix: debugger doesn't hang on session lock if interrupted at
626 an inappropriate moment
627 ** bug fix: run-program is now thread safe(r)
628 ** bug fix: inner with-recursive-lock no longer releases the mutex
629 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
630 * fixed some bugs revealed by Paul Dietz' test suite:
631 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
632 EXPECTED-TYPE slots filled.
633 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
635 ** MAKE-CONDITION accepts classes as type-designators.
636 ** COMPILE may never return NIL.
637 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
638 range before calling Unix time functions
640 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
641 * numerous signal handling fixes to increase stability
642 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
643 * minor incompatible change: we now correctly canonize default
644 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
645 as per the MOP, rather than the historical (INITARG INITFUNCTION
646 INITFORM). (reported by Bruno Haible)
647 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
648 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
649 for more information.
650 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
651 pathname is a directory pathname.
652 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
653 * dynamic space size on PPC has been increased to 768Mb. (thanks to
655 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
656 :METACLASS argument in addition to a class name. (reported by
657 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
658 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
659 (reported by Bruno Haible; patch from Gerd Moellmann)
660 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
662 * bug fix: more cleanups to the floating point exception handling on
663 x86-64 (thanks to James Knight)
664 * bug fix: the compiler does not try to constant fold calls of
665 COERCE and BIT-* functions when they return freshly constructed
666 objects. (reported by Nikodemus Siivola and Paul Dietz)
667 * optimization: improved the MIPS versions of generic (in the
668 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
669 * optimization: direct conversion of (unsigned-byte 32) to floats on
670 the PowerPC platform.
671 * optimization: structure instances with raw slots now use less
672 memory, and probably show better memory locality. (thanks to
674 * optimization: DYNAMIC-EXTENT declarations for lists and closures
675 are treated as requests for stack allocation on the x86-64,
676 Alpha-32, PPC and SPARC platforms.
677 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
678 the wrong file descriptor; implementation of SOCKET-OPEN-P.
679 (thanks to Tony Martinez)
681 ** gcing a dead thread can no longer lead to lockups
682 ** threads block signals until they are set up properly
683 ** errno is no longer shared by threads
684 ** interrupt-thread restores the eflags register on x86
685 ** fixed some lockups due to gc/thread interaction
686 * fixed some bugs revealed by Paul Dietz' test suite:
687 ** invalid dotted lists no longer raise a read error when
689 ** signal an error if a symbol that names a declaration is used
690 as the name of a type, or vice versa
691 ** allow using the (declare (typespec var*)) abbreviation for
692 (declare (type typespec var*)) with all type specifiers
693 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
694 ** always return NIL from PPRINT-POP when OBJECT is NIL
695 ** don't signal errors when pretty-printing malformed LABELS,
696 FLET or MACROLET forms
697 ** declarations in a DOLIST body are also in scope for the
699 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
701 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
704 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
705 * fixed cross-compiler leakages that prevented building a 32-bit
706 target with a 64-bit host compiler.
707 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
708 opened with :IF-EXISTS :OVERWRITE.
709 * fixed bug 281: error for an invalid qualifier in a short-form method
710 combination method is not signalled until the faulty method is called.
711 * bug fix: iteration variable type inferrer failed to deal with open
712 intervals. (reported by Alan Shields)
713 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
714 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
715 or more runtime options were provided to the sbcl binary.
716 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
717 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
718 * optimization: there's now a fast-path for fixnum arguments in the
719 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
720 * optimization: the code generated on x86-64 is more compact thanks
721 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
722 * bug fix: floating-point exception handling now partly works on
723 x86-64. (Thanks to James Knight)
724 * improvement to the MIPS backend from Thiemo Seufer: C-style
725 64-bit long long arguments and return values to alien functions
727 * fixed some bugs revealed by Paul Dietz' test suite:
728 ** the type-error signalled from WARN has a filled-in DATUM slot.
729 ** the type-error required when a stream is not associated with
730 a file has the stream as its datum.
731 ** type-errors on single-floats on x86-64 no longer have
732 :INVALID-OBJECT as the datum
733 ** the type-errors signalled for invalid function names now have
734 a correct expected type
735 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
736 for structures defined with a :TYPE. Documentation strings for
737 typed structures are no longer immediately discarded
738 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
739 broadcast streams and synonym streams. FILE-LENGTH now also works
740 on broadcast streams.
742 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
743 * incompatible change: the --noprogrammer option, deprecated since
744 version 0.7.5, has been removed. Please use the equivalent
745 --disable-debugger option instead.
746 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
748 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
749 inoperational for a while now) has been completely removed.
750 * null lexical environments are now printed as #<NULL-LEXENV>,
751 significantly reducing the amount of clutter in typical backtraces.
752 * documentation on weak pointers, finalization, and after GC hooks
753 has been added to the manual.
754 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
755 as other specialized array types, is much faster. SUBSEQ and
756 COPY-SEQ on such arrays have also been sped up.
757 * optimization: EQL is now more efficient when at least other argument
758 is known to be of type (OR FIXNUM (NOT NUMBER)).
759 * fixed bug: the runtime is now less vulnerable to changes in the
760 size of the SBCL object on OS X, and virtual memory is reserved for
761 all spaces that need to be at a fixed address.
762 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
763 * fixed bug: finalizers and after GC hooks that cause consing are now
765 * fixed bug: compiler error messages and summaries are now printed to
766 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
767 * fixed inference of the upper bound of an iteration variable.
768 (reported by Rajat Datta).
769 * fixed bug 211e: calling local functions with duplicated constant
770 keyword argument no longer causes a bogus style warning about an
772 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
773 ftype declaration to be disregarded. (reported by Dave Roberts)
774 * fixed bug 373: caused by erronous compilation of references to alien
775 variables in the runtime on ppc/darwin.
776 * fixed bug 376: CONJUGATE type deriver.
777 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
778 (reported by Baughn on #lisp)
779 * a cross-compiler bug on non-x86 platforms has been identified and
780 fixed. (thanks to Bruno Haible)
781 * improvements to the MIPS runtime code for increased stability.
782 (thanks to Thiemo Seufer)
783 * increased the maximimum compact environment size to allow
784 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
785 * partial workaround for bug 135: don't name the function we're
786 calling for hairy cases of EVAL, so as not to accumulate one environment
787 entry per call to EVAL. (reported by Kevin Reid)
788 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
789 * optimization: type testing for non-vector arrays should be faster.
790 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
792 * bug fix: setting 31st element of a bit vector to zero did not work
794 * bug fix: redefining a class definition which failed due to a
795 previous accessor / function clash now works (but see BUGS entry
796 #380 for more problems in this area). (thanks to Zach Beane)
797 * the long form of DEFINE-METHOD-COMBINATION disables method group
798 checking when given a single method group with pattern *. (thanks
800 * on x86 compiler supports stack allocation of results of simple
801 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
802 * support for building and running on Mac OS X 10.4 "Tiger" has been added
803 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
804 * fixed some bugs related to Unicode integration:
805 ** the restarts for recovering from input and output encoding
806 errors only appear when there is in fact such an error to
808 * fixed some bugs revealed by Paul Dietz' test suite:
809 ** MISC.549 and similar: late transformation of unsafe type
810 assertions into derived types caused unexpected code
812 ** SCALE-FLOAT type deriver is less wrong.
813 ** type derivers for EXP, LOG and similar functions compute result
814 types for complex arguments better.
815 ** (MISC.563) CONJUGATE type deriver works for very restricted
817 ** out-of-line type testers for character strings are available.
818 ** EQUAL compiler transform understands specialness of objects
820 ** accessing double-floats stored on the stack now works on x86-64.
821 ** debugger internals could sometimes create invalid lispobjs,
822 resulting in GC crashes.
823 ** MISC.548: type check weakening can convert required type into
825 ** initialization forms for bindings are not in scope of free special
828 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
829 * incompatible change: thread support for non-NPTL systems has
830 been removed - locking is buggy and unreliable. A threaded
831 SBCL build will now warn at startup and refuse to create
832 new threads, unless futex support is detected in the kernel
833 * incompatible change: the top level REPL now has only an ABORT
834 restart associated with it, not TOPLEVEL and ABORT as it used to.
835 TOP and TOPLEVEL are now available as debugger commands for
836 returning to the top level.
837 * incompatible change: forms evaluated in the REPL now use the
838 global optimization policy.
839 * incompatible change: user- and system-initialization files are
840 no longer processed with LOAD, but by READ and EVAL; hence the
841 global optimization policy, startup package, readtable, etc,
843 * The .fasl file format number has been incremented because of
844 various incompatible changes.
845 * internal entry point details and argument counts no longer appear
846 in backtraces unless explicitly requested by setting
847 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
848 * built-in and standard functions no longer have names like "top
849 level local call to FOO".
850 * fixed bug 32: functions defined in non-null lexical environments
851 now have more legible printed representation
852 * fixed bug 33: functions defined in non-null lexical environemnts
853 are now more amenable to inspection by INSPECT.
854 * workaround for bug 354: XEPs no longer appear in backtraces unless
855 explicitly requested.
856 * fixed bug: receiving the signal which results from attempting to
857 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
858 SIGBUS on other *BSDs) on architectures where the C stack is also
859 the Lisp stack (x86 and x86-64) from foreign code no longer leads
860 to debugger confusion or wild instability. (reported by Cheuksan
862 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
863 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
864 (reported by Lutz Euler)
865 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
866 compiled code. (reported by Paul Dietz)
867 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
868 specialized parameter to a method to something that is not TYPEP
869 the specializer is now possible.
870 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
871 face of package deletion.
872 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
873 pathname. (reported by David Tolpin on c.l.l)
874 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
875 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
876 STRING-STREAM and FILE-STREAM, respectively.
877 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
878 than previously; in addition, its test suite is now run on build.
879 (thanks to Luke Gorrie)
880 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
881 * optimization: added a immediate representation for single-floats
883 * fixed some bugs revealed by Paul Dietz' test suite:
884 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
885 ** attempting to create a package with a colliding nickname causes
886 correctable errors to be signalled.
887 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
888 ** several x86-64 backend bugs related to sign-extension of immediate
891 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
892 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
893 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
895 * fixed loading of multiply forward-referenced layouts.
896 (thanks to Cheuksan Wang)
897 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
898 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
899 (see COMPILE-FILE documentation for details of :PRINT option)
900 * fixed bugs 19 and 317: fixed-format floating point printing is
901 more accurate. This also fixes a bug reported by Adam Warner
902 related to the ~@F format directive.
903 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
905 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
906 dispatch table if the from-char is a dispatch macro character.
907 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
908 by the wordsize no longer ignore the last word. (reported by Lutz
910 * fixed bug in type checking of dynamic-extent variables. (reported
912 * optimization: sequence traversal functions use their freedom to
913 coerce function designators to functions.
914 * optimization: code with many calls to CLOS methods specialized on
915 CLOS classes has had redundant type checks removed.
916 * optimization: added declarations to speed up operations that access
917 the internal character database (for example STRING-UPCASE)
918 * optimization: comparison operations between floats and sufficiently small
919 fixnums no longer create extra rationals
920 * fixed some bugs related to Unicode integration:
921 ** portions of multibyte characters at the end of buffers for
922 character-based file input are correctly transferred to the
923 start of the buffer at the next read.
924 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
925 passing it through to OPEN.
926 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
927 argument given to internal calls to OPEN.
928 ** the built-in comment readers (introduced by character sequences
929 ";" and "#|") are more forgiving to encoding errors; they will
930 STYLE-WARN and then attempt to resync the stream at a character
931 boundary. (thanks to Teemu Kalvas)
932 * fixed some bugs in the x86-64 port:
933 ** Negative short int return values from c-calls are sign-extended
935 ** The stack is aligned to 16-bytes for c-calls, as required by
936 the ABI. (thanks to Cheuksan Wang)
937 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
938 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
939 for enums. (thanks to Vincent Arkesteijn)
940 ** Multiple small optimizations and bugfixes for floating point
942 * fixed some bugs revealed by Paul Dietz' test suite:
943 ** Space, Tab, Linefeed, Return and Page have the invalid
944 secondary constituent character trait.
945 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
947 ** WITH-INPUT-FROM-STRING should only update the index place on
949 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
951 ** Bit-array operations (BIT-AND and similar) worked incorrectly
952 with one-dimensional arrays with fill pointers.
953 ** TYPE-OF failed on a complex with an integer realpart and a
955 ** compiler failure during type inference for the code of form
956 (IF (EQL X (THE ...)) ...) (MISC.535).
958 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
959 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
960 platform. The port supports 61-bit fixnums, large memory spaces
961 and reloading shared object files.
962 * enhancement: saving cores with foreign code loaded is now
963 supported on ppc/Darwin in addition to the previously supported
965 * enhancement: the statistical profiler now walks deeper into the
966 call stack for more meaningful call-graphs and accrued time
967 reports (x86/x86-64 only). It also now reports time spent in
969 * enhancement: it is now possible to trace most individual methods
970 of a generic function in addition to tracing the generic function
972 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
974 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
975 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
976 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
977 produces an error. (thanks to Vincent Arkesteijn)
978 * bug fix: NAMESTRING on pathnames with :WILD components in their
979 directories works correctly. (thanks to Artem V. Andreev)
980 * fixed bug 125: compiler preserves identity of closures. (reported
982 * bug fixed: functions with &REST arguments sometimes failed with
983 "Undefined function" when compiled with (DEBUG 3). (reported by
985 * bug fix: overflow during compiling of setting element of a bit
986 vector with constant index and value. (reported by Timmy Douglas)
987 * build fix: fixed the dependence on *LOAD-PATHNAME* and
988 *COMPILE-FILE-PATHNAME* being absolute pathnames.
989 * on x86 compiler partially supports stack allocation of dynamic-extent
991 * GO and RETURN-FROM do not check the extent of their exit points
992 when compiled with SAFETY 0.
993 * fixed some bugs related to Unicode integration:
994 ** encoding and decoding errors are now much more robustly
995 handled; it should now be possible to recover even from invalid
996 input or output to the terminal. (thanks to Teemu Kalvas)
997 ** provided a first cut at implementing STRING-TO-OCTETS and
998 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
999 ** altered the SB-MD5 contributed module slightly, changing the
1000 interface just enough for it to be supportable for builds where
1001 lisp characters are not eight bits.
1002 * fixed some bugs revealed by Paul Dietz' test suite:
1003 ** the FORMATTER-generated functions for ~V[ conditionals require
1004 the correct number of arguments.
1005 ** READ-FROM-STRING returns the mandated second value when applied
1006 to displaced strings.
1007 ** the #\Rubout and #\Backspace characters are treated as invalid
1008 constituent characters by the tokenizer.
1010 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1011 * new feature: reloading changed shared object files with
1012 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1013 * new feature: references to foreign variables and functions
1014 can now be compiled and loaded before the corresponding shared
1015 object file is loaded, as long as the foreign definitions are
1016 available at runtime.
1017 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1018 supported. (thanks to Dan Debertin)
1019 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1020 just keywords. (thanks to Vincent Arkesteijn)
1021 * fixed bug #331: structure-class instances corresponding to
1022 DEFSTRUCT forms are now created eagerly.
1023 * fixed bug #345: backtraces from calls to undefined functions work
1024 on x86 as well. Related bug #61 is now also partially fixed on x86
1025 (backtraces from throws to unknown catch tags.)
1026 * bug fix: lambda-list parsing is now stricter vrt. order and number
1027 of lambda-list keywords.
1028 * bug fix: as specified by AMOP, an error is signalled if a
1029 class-option appears multiple times in a DEFCLASS form. (reported
1031 * bug fix: RANDOM can be compiled when the compiler derives the type
1032 of its numeric argument as a disjoint set of small integers.
1033 (reported by Paul Dietz)
1034 * bug fix: starting a core saved with shared objects loaded when
1035 those objects are not available no longer causes threaded SBCL to
1036 hang. (reported by Sean Ross)
1037 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1039 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1040 argument, which must always be NIL. (reported by Kalle Niemitalo)
1041 * bug fix: printing 1.0d+23 no longer results in an error.
1042 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1043 * bug fix: structure slot setters preserve evaluation order. (thanks
1045 * fixed some bugs related to Unicode integration:
1046 ** RUN-PROGRAM can allow its child to take input from a Lisp
1047 stream. (reported by Stefan Scholl)
1048 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1049 locales. (reported by Ken Causey)
1050 * fixed some bugs revealed by Paul Dietz' test suite:
1051 ** INCF, DECF and REMF evaluate their place form as specified in
1053 ** FORMATTER expands ~{ ~} iteration directives with V or #
1054 parameters correctly.
1055 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1056 consequent uses no arguments correctly.
1057 ** the system has a partial understanding of the (COMPLEX RATIO)
1059 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1062 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1063 * new feature: a build-time option (controlled by the :SB-UNICODE
1064 keyword feature, enabled by default) for building the system with
1065 support for the entire 21-bit character space defined by the
1067 * new feature: the system now has rudimentary external-format
1068 support; the primary user-visible change at this time is that
1069 characters with the high bit set (such as lower-case-e-acute) will
1070 print correctly to a terminal in a UTF-8 environment.
1071 * minor incompatible change: BASE-CHAR no longer names a class;
1072 however, CHARACTER continues to do so, as required by ANSI.
1073 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1074 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1075 has been moved to the SB-EXT package (temporarily re-exported from
1077 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1078 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1080 * fixed bug #351: better error-handlind and reporting for malformed
1082 * fixed bug #350: bignum-printing is now more memory-efficient,
1083 allowing printing of very large bignums, eg. (expt 2 10000000).
1084 (reported by Bruno Haible)
1085 * fixed bug #302: better primitive-type selection for intersection
1087 * fixed bug #308: non-graphic characters now all have names, as
1088 required. (reported by Bruno Haible)
1089 * bug fix: redefining a class with different superclasses now correctly
1090 removes it from the direct-subclasses of its previous superclasses.
1091 (reported by David Morse)
1092 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1093 new value now works. (reported by Bruno Haible)
1094 * bug fix: correct canonicalization of multiple non-standard slot
1095 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1096 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1097 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1098 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1099 now exists, an signals an error.
1100 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1101 during macroexpansion for non-iteration variables as well. (reported
1102 by Bruno Haible for CMUCL)
1103 * bug fix: Cyclic structures and unprintable objects in compiler
1104 messages no longer cause errors. (reported by Bruno Haible)
1105 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1106 and READ-FROM-STRING all now return a primary value of NIL if
1107 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1108 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1109 (reported by Bruno Haible for CMUCL)
1110 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1111 do not designate a condition. (reported by Bruno Haible for
1113 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1114 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1115 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1116 permitting resolution in favour of any of the conflicting symbols.
1117 (reported by Bruno Haible for CMUCL)
1118 * bug fix: EQUAL compiler optimizations is less aggressive on
1119 strings which can potentially compare true despite having distinct
1120 specialized array element types.
1121 * bug fix: unit enumerations can be defined without dividing by
1122 zero. (thanks to Vincent Arkesteijn)
1123 * FORMAT compile-time argument count checking has been enhanced.
1124 (report from Bruno Haible for CMUCL)
1125 * a partial workaround for the bug 262: the compiler does not try to
1126 inline-expand a local function doing RETURN-FROM from a deleted
1127 BLOCK. (thanks to Peter Denno for the bug report and to David
1128 Wragg for the simple test case)
1129 * fixed some bugs revealed by Paul Dietz' test suite:
1130 ** RENAME-PACKAGE allows all package designators as new package
1132 ** constraint propagation and lambda variable substitution are
1133 more cautious in dealing with partially deleted code.
1134 ** compiler sometimes lost reoptimization passes.
1135 ** CERROR, given a condition as condition designator, passes its
1136 remaining arguments to the continue format control without
1138 ** Case-altering FORMAT directives work correctly on non-ASCII
1140 ** The REAL type specifier handles bounds outside the floating
1141 point ranges without signalling FLOATING-POINT-OVERFLOW.
1142 ** Functions with IR1-transformations can create intercomponent
1143 references to global functions.
1144 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1146 ** FORMAT ~R treats a nil value for its first parameter correctly.
1148 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1149 * enhancement: saving cores with foreign code loaded is now
1150 supported on x86/NetBSD and sparc/Linux in addition to the previously
1151 supported platforms.
1152 * bug fix: on some platforms repeated installations caused multiple
1153 copies of HTML documentation to be installed -- should not happen
1154 any more. (reported by Stefan Scholl)
1155 * bug fix: parsing self-recursive alien record types multiple times
1156 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1157 original patch by Helmut Eller for CMUCL)
1158 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1159 (thanks to Richard Kreuter)
1160 * bug fix: defining classes whose accessors are methods on existing
1161 generic functions in other (locked) packages no longer signals
1162 bogus package lock violations. (reported by François-René Rideau)
1163 * bug fix: special variables as DEFMETHOD parameters no longer have
1164 associated bogus type declarations. (reported by David Wragg and
1166 * bug fix: read-write consistency on streams of element-type
1167 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1168 * bug fix: redefiniton of the only method of a generic function with
1169 no DEFGENERIC no longer emits a full WARNING. In addition,
1170 redefinition of generic functions with no DEFGENERIC to an
1171 incompatible lambda list now signals an error. (thanks to Zach
1173 * bug fix: DEFGENERIC now works even when there's a function of the
1174 same name in an enclosing lexical environment. (thanks to Zach
1176 * fixed compiler failure, caused by instrumenting code during
1177 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1178 * optimization: added loop analysis and improved register allocation
1179 * fixed some bugs revealed by Paul Dietz' test suite:
1180 ** POSITION on displaced vectors with non-zero displacement
1181 returns the right answer.
1182 ** (SIMPLE-STRING) is a valid type specifier for sequence
1184 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1186 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1187 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1189 ** The justification version of the FORMAT ~< directive treats
1190 non-zero minpad parameter correctly.
1192 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1193 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1194 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1195 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1196 the supported interface.
1197 * new feature: Single-stepping of code compiled with DEBUG 2 or
1198 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1199 * new feature: saving cores with foreign code loaded is now
1200 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1201 Timothy Moore's work for CMUCL)
1202 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1203 parameters to * instead of NIL if no initform is supplied.
1204 (reported by Johan Bockgaard)
1205 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1206 correctly when FUNCALL appears as the car of the form. Note:
1207 despite this FUNCALL forms are not currently subject to
1208 compiler-macro expansion. (port of Raymond Toy's fix for the
1209 same from CMUCL, reported by Johan Bockgaard)
1210 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1211 (thanks for Teemu Kalvas)
1212 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1213 case instead of terminating the iteration (thanks for Julian
1214 Squires, Sean Champ and Raymond Toy)
1215 * bug fix: incorrect expansion of defgeneric that caused a style
1216 warning. (thanks for Zach Beane)
1217 * bug fix: slot accessor effective method computation works properly
1218 for classes with multiple non-standard applicable methods on
1219 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1220 * on x86 compiler supports stack allocation of results of LIST and
1221 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1222 CMUCL implementation by Gerd Moellmann)
1223 * fixed some bugs revealed by Paul Dietz' test suite:
1224 ** FORMAT strings with both the ~<~:;~> form of the justification
1225 directive and pretty-printing directives cause an error.
1227 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1228 * incompatible change: the internal functions
1229 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1230 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1231 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1232 instead of the old functions.
1233 * new feature: on platforms where "dladdr" is available foreign
1234 function names now appear in backtraces. (based on Helmut Eller's
1236 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1237 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1239 * documentation: ASDF manual is now distributed with SBCL.
1240 * bug fix: SBCL can now load its contributed modules with REQUIRE
1241 even if the system-provided entries have been removed from
1242 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1244 * bug fix: backtraces involving undefined functions or assembly
1245 routines are more informative. (thanks to Brian Downing)
1246 * bug fix: mutually referent alien structures now work correctly.
1247 (reported by Rick Taube)
1248 * bug fix: structures defined by WITH-ALIEN can be referred to
1249 within other definitions in the same WITH-ALIEN.
1250 * bug fix: division operators (MOD, TRUNCATE and the like) with
1251 constant zero divisors and integer dividends no longer generate
1253 * bug fix: provide default methods for INPUT-STREAM-P and
1254 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1255 * bug fix: improve the detection and resolution of MOP metacycles
1256 (where effective method computation depends on the generic function
1257 having its effective method computed). (reported by Bruno Haible)
1258 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1259 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1260 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1261 * bug fix: pathnames with relative directory components can be
1262 represented relative to default pathnames.
1263 * optimization: in taking the GCD of bignums, reduce the two bignums
1264 to approximately the same size (using Euclid's algorithm) before
1265 applying the more sophisticated binary GCD. (thanks to Juho
1267 * optimization: COUNT on bitvectors now operates word-at-a-time.
1268 * optimization: ASH with a positive, but not necessarily constant,
1269 (leftwards) shift, when in a modular context, is compiled to a
1271 * fixed some bugs revealed by Paul Dietz' test suite:
1272 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1273 if the corresponding argument is NIL.
1274 ** FORMAT directives accepting a minpad parameter treat negative
1276 ** literal commas as character parameters to FORMAT directives are
1278 ** literal spaces directly after ~<Newline> directives within a
1279 format-logical-block (~:< ~@:>) do not induce :FILL-style
1280 conditional newlines.
1281 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1282 as specified: it no longer includes conditional newlines).
1283 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1285 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1286 :PER-LINE-PREFIX argument does not evaluate to a string.
1288 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1289 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1290 the manual for details; for now, package locks can be disabled by
1291 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1292 if no major problems are found then it is likely that they will be
1293 compiled in unconditionally.
1294 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1295 unsupported operators on all platforms. To load a shared library
1296 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1297 object file, link it into a shared library outside of SBCL and
1298 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1300 * bug workaround: ROOM T and the GC were not working together
1301 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1302 reporting facility aren't preserved. That reporting has been
1303 disabled, so now until and unless someone figures out how to make it
1304 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1305 * minor incompatible change: as threatened around sbcl-0.8.0, the
1306 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1307 The symbols are also exported from SB-PCL for backwards
1308 compatibility, but more so than before SB-PCL should be treated as
1309 an implementation-internal package.
1310 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1312 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1313 It is known as of this release not to work on the Alpha, however.
1314 * fixed bug #167: errors signalled due to illegal syntax in method
1315 bodies are now more legible.
1316 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1317 designators and can hence be used with TYPEP.
1318 * fixed bug #333: CHECK-TYPE now ensures that the type error
1319 signalled, if any, has the right object to be accessed by
1320 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1321 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1322 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1324 * fixed bug #334: programmatic addition of slots using specialized
1325 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1326 and :ALLOCATION :CLASS slots.
1327 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1328 float exponents. (rereported by Peter Seibel)
1329 * fixed a bug: #\Space (and other whitespace characters) are no
1330 longer considered to be macro characters in standard syntax by
1331 GET-MACRO-CHARACTER.
1332 * fixed bug: initialization of condition class metaobjects no longer
1333 causes an instance of the condition to be created. (reported by
1335 * fixed bug: it is now possible to have slots such that
1336 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1337 description is neither :INSTANCE nor :CLASS.
1338 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1339 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1340 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1341 non-standard SLOT-DEFINITION-ALLOCATION values.
1342 * fixed bugs: various race conditions handling exiting threads.
1343 CL-PPCRE's thread torture test now passes.
1344 * fixed arguably-a-bug: GC time proportional to number of threads in
1345 system even when most of them are idle
1346 * optimization: improved performance of BIT and SBIT on bit-vectors.
1347 * .fasl file incompatibility: The fasl file version number has
1348 been incremented because of changes associated with package locks.
1350 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1351 * minor incompatible change: the system no longer provides
1352 optimization hints (conditions which are TYPEP
1353 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1354 mode; these hints are still emitted when higher SPEED optimization
1356 * new contrib module: a sampling profiler (profiling by statistical
1357 sampling, rather than by instrumenting functions) is available as
1358 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1359 * the behaviour of the standard function ED is now customizeable by
1360 third parties through a hook variable: see ED's documentation
1361 string for information on the protocol.
1362 * the compiler no longer emits efficiency notes for (FUNCALL X)
1363 when the type of X is uncertain under default optimization
1365 * fixed bug 276: mutating a binding of a specialized parameter to a
1366 method to something that is not TYPEP the specializer is now
1368 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1369 exhibit the required behaviour on the x86 platform. (thanks to
1370 Peter van Eynde, Eric Marsden and Bruno Haible)
1371 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1372 even for difficult arguments. (reported by Peter Graves)
1373 * fixed bug 141a: the backquote printer now descends quoted
1375 * fixed another bug in backquote printing: no more destructive
1376 modification of the form's list structure. (reported by Brian
1378 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1379 move between its address being taken and the call to
1380 interrupt_thread, fixing a crashing race condition.
1381 * the SB-POSIX contrib implementation has been adjusted so that it
1382 no longer exhibits ridiculously poor performance when constructing
1383 instances corresponding to C structs.
1385 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1386 * minor incompatible change: the sb-grovel contrib now treats C
1387 structures as alien (in the sense of SB-ALIEN) objects rather than
1388 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1389 has implications for memory management of client code
1390 (sb-grovel-returned objects must now be manually managed) and for
1391 type safety (alien objects now have full types).
1392 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1393 used to control emission of compiler diagnostics, rather than the
1394 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1395 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1396 quality should be considered deprecated.
1397 * install.sh now installs the user manual as well
1398 * (not quite a new documentable feature, but worth considering in
1399 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1400 beginnings of a semantically meaningful condition hierarchy is
1401 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1403 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1404 designator as the defaults argument.
1405 * fixed bug: Displaced arrays whose displaced-to array has become
1406 too small now cause ARRAY-DIMENSION to signal an error, providing
1407 for safer bounds-checking. (reported by Bruno Haible)
1408 * fixed bug: DEFCLASS slot definitions with identical :READER and
1409 :WRITER names now signal a reasonable error. (reported by Thomas
1411 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1413 * fixed bug: Invalid slot specification errors now print correctly.
1414 (thanks to Zach Beane)
1415 * fixed bug 320: Shared to local slot value transfers in class
1416 redefinitions now happen corrently from superclasses as
1417 well. (reported by Bruno Haible)
1418 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1420 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1421 improper lists correctly. (reported by Bruno Haible)
1422 * fixed bug 313: source-transform for <fun-name> was erroneously
1423 applied to a call of a value of a variable with name <fun-name>.
1424 (reported by Antonio Menezes Leitao)
1425 * fixed bug 307: The obsolete instance protocol ensures that
1426 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1427 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1428 remove dead unknown-values globs from the middle of the stack.
1429 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1430 easier to resume long computations after using *BREAK-ON-SIGNALS*
1431 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1432 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1433 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1434 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1435 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1436 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1438 * fixed a (fairly theoretical) bug in string printing: if
1439 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1440 does not have array-element-type equal to the most general string
1442 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1443 function name. (thanks to Nikodemus Siivola)
1444 * fixed bug: compile-time format string checker failed on
1445 non-closed ~{. (reported by Thomas F Burdick)
1446 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1447 don't warn on legal loop code involving more than one aggregate
1448 boolean. (thanks to Nikodemus Siivola)
1449 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1450 in keyword position to MAKE-PACKAGE was not being reported as
1452 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1453 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1455 * fixed bugs 280 and 312: the checking for multiple definitions in a
1456 file is less likely to become confused by uses of inline
1458 * fixed bug: the #S reader macro performs the keyword coercion
1459 specified for slot names. (reported by Kalle Niemitalo)
1460 * fixed bug: lambda lists may contain symbols whose names start with
1461 & but are not lambda-list-keywords; their occurrence triggers a
1463 * fixed bug 321: define-method-combination argument lambda lists do
1464 not cause invalid code to be generated when &optional or &aux
1465 variables have default values. (reported by Bruno Haible)
1466 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1467 have CLOS classes; this appears also to have fixed reported
1468 instability in stack exhaustion detection.
1469 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1470 MUFFLE-WARNING when no associated restart is present is now
1472 * optimization: rearranged the expansion of various defining macros
1473 so that each expands into only one top-level form in a
1474 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1476 * optimization: used a previously little-used slot in symbols to
1477 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1479 * fixed some bugs revealed by Paul Dietz' test suite:
1480 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1482 ** Fixed an optimization bug related to inheritance of initforms
1483 from local to shared slots.
1484 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1485 ** CAST optimizer forgot to flush argument derived type.
1486 ** print/read consistency on floats is now orders of magnitude
1487 more likely. (thanks also to Bruno Haible for a similar report
1489 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1490 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1491 for objects of type REAL. Make it so.
1492 ** SXHASH is specified (infelicitously) to respect similarity,
1493 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1494 it so. (thanks to Markus Ziegler)
1495 ** on the Alpha, the compiler succeeds in compiling functions
1496 returning a known number of arguments greater than 63.
1497 ** fixed handling of invalid NIL arguments in keyword position
1499 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1501 ** provided a readably-printable representation for RANDOM-STATE
1503 ** ensured that pathnames, where they have a namestring, always
1504 print using #P"..." syntax.
1506 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1507 * Support for the forthcoming 2.0 version of the NetBSD kernel
1508 running on x86 hardware has been added. (thanks to Perry
1509 E. Metzger most immediately, and others for their past work)
1510 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1511 current release version; SBCL's OpenBSD support had been broken
1512 since about the time of OpenBSD's switch to ELF binary format.)
1513 * [placeholder for DX summary]
1514 ** user code with &REST lists declared dynamic-extent, under high
1515 speed or space and low safety and debug optimization policy.
1516 * The manual has been converted to Texinfo format and the debugger
1517 chapter from the cmucl manual has been added.
1518 * A facility has been added to extract documentation strings from
1519 sbcl and store them as Texinfo-formatted snippets for inclusion in
1520 the manual (via Texinfo's @include directive)
1521 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1522 the test case to Dave Roberts)
1523 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1524 pointers. (reported by Sean Ross)
1525 * bug fix: PROFILE output is printed nicely even for large numerical
1526 values. (thanks to Zach Beane)
1527 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1528 greater than 32 handle EOF correctly.
1529 * bug fix: on X86 an immediate argument of the IMUL instruction is
1530 correctly printed in disassembly. (thanks to Lutz Euler)
1531 * bug fix: class slots in redefined classes preserve their old
1532 values. (thanks to Bruno Haible and Nikodemus Siivola)
1533 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1534 succeeds. (reported by Marco Baringer)
1535 * fixed some bugs revealed by Paul Dietz' test suite:
1536 ** READ-SEQUENCE now works on ECHO-STREAMs.
1537 ** RATIONALIZE works more according to its specification. (thanks
1540 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1541 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1542 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1543 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1544 should matter to you only if you rebind the printer control
1545 variables and then find you want different bindings in the
1546 debugger than in the ordinary execution of your program.)
1547 * The runtime build system has been tweaked to support building
1548 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1549 assembler and linker. (thanks to Nikodemus Siivola)
1550 * Unbound, undefined, undeclared variables now trigger full
1551 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1552 more often programmer error than deliberate exploitation of undefined
1554 * optimization: the hash algorithm for strings has changed to one
1555 that is less vulnerable to spurious collisions. (thanks to Juho
1557 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1558 needless bounds checking. (thanks to Juho Snellman)
1559 * optimization: implemented multiplication as a modular
1560 (UNSIGNED-BYTE 32) operation on the PPC backend.
1561 * fixed some bugs revealed by Paul Dietz' test suite:
1562 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1564 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1565 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1566 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1568 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1570 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1572 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1573 their output stream on EOF from read.
1574 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1575 have been read to end-of-file.
1576 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1578 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1579 description of determination of which consecutive characters
1581 ** Printing the "Space" character with escaping on now yields "#\\ ",
1582 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1583 ** Reading floating-point numbers with *READ-BASE* set to a number
1584 less than 10 works correctly.
1585 ** Reading floating-point numbers with *READ-BASE* set to a number
1586 more than 10 works correctly.
1587 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1588 the readtable currently in effect.
1590 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1591 * minor incompatible change: parsing of namestrings on a physical
1592 (Unix) host has changed; numbers after the final #\. in a
1593 namestring are no longer interpreted as a version field. This is
1594 intented to be largely invisible to the user, except that the
1595 meaning of the namestring "*.*.*" has changed: it now refers to a
1596 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1597 should usually be replaced by
1598 (make-pathname :name :wild :type :wild :version :wild)
1599 with the added benefit that this is more likely to be portable.
1600 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1601 option to OPEN now signals an error if the file being opened
1602 exists; this may have an impact on existing code.
1603 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1604 signals received, on the PowerPC platforms (both Linux and
1605 Darwin). (thanks to Pierre Mai for pointing out the location of
1607 * several fixes on OS X: The system now builds and runs cleanly on
1608 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1609 (thanks to Brian Mastenbrook)
1610 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1611 second-resolution: integer multiples of 1/3600 between -24 and 24.
1612 (thanks to Vincent Arkesteijn)
1613 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1614 type when called with 1 argument; PEEK-CHAR checked type of
1615 PEEK-TYPE only after having read first character from a
1616 stream. (reported by Peter Graves)
1617 * bug fix: the garbage collector now has much better locality
1618 behaviour, and in particular no longer treats all memory as being
1619 exhausted when many small objects point to each other in a deeply
1621 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1622 recognized as being TYPEP their class.
1623 * bug fix: the PUSHNEW documentation string has been corrected.
1624 (thanks to Vincent Arkesteijn)
1625 * bug fix: defaulting of the value for the last of an atypically
1626 large number of multiple values being bound was not being
1627 performed correctly on the Alpha or PPC platforms
1628 * optimization: implemented multiplication as a modular
1629 (UNSIGNED-BYTE 32) operation on the x86 backend.
1630 * optimization: SEARCH on simple-base-strings can now be open-coded.
1631 (see also contrib/compiler-extras.lisp for inspiration for
1632 teaching the compiler about the Boyer-Moore algorithm).
1633 * value, returned by MAX (and MIN) called with several EQUALP, but
1634 not EQL, arguments now does not depend on compiler settings.
1635 * fixed some bugs revealed by Paul Dietz' test suite:
1636 ** in stack analysis liveness information is propagated from
1637 non-local entry points.
1638 ** pathwise CAST removing failed when the CAST node did not start
1640 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1641 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1643 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1644 host is already defined.
1645 ** RENAME-FILE works on streams instead of signalling an internal
1647 ** PEEK-CHAR uses the current readtable when determining whether
1648 or not a character is whitespace.
1649 ** MERGE-PATHNAMES handles the case when the pathname does not
1650 specify a name while the default-pathname specifies a version.
1651 ** Pathnames now stand a better chance of respecting print/read
1653 ** Attempting to use standardized file system operators with a
1654 pathname with invalid :DIRECTORY components signals a
1656 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1658 ** DIRECTORY on logical pathnames is more correct.
1659 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1660 signal a TYPE-ERROR if their argument is not a stream.
1661 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1662 designator argument does not designate a stream.
1663 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1664 examining the synonym.
1665 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1667 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1668 element-type larger than ([UN]SIGNED-BYTE 32).
1670 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1671 * When built with the :SB-FUTEX feature, threaded builds now take
1672 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1673 for faster/more reliable mutex and condition variable support.
1674 * Incompatible change (but one you probably shouldn't have been using
1675 anyway): the interface and code for arbitrating between multiple
1676 threads in the same user session has been redesigned.
1677 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1678 error when called without an explicit environment argument.
1679 (thanks to Nikodemus Siivola)
1680 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1681 confused about their position. (thanks to Adam Warner and Gerd
1683 * bug fix: Pretty printing backquoted forms with unquotations in the
1684 argument list position of various code constructs such as LAMBDA
1685 now works correctly. (reported by Paul Dietz)
1686 * bug fix: Pretty printing unquotations no longer loses all
1687 stream position information.
1688 * optimization: performance of string output streams is now less
1689 poor for multiple small sequence writes.
1690 * optimization: performance of CSUBTYPEP in the presence of complex
1691 expressions involving CONS and NOT many times has been improved.
1692 (reported by Paul Dietz)
1693 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1695 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1696 streams for the REPL, for the debugger, and so on. (thanks to
1698 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1699 * fixed some bugs revealed by Paul Dietz' test suite:
1700 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1701 is ignored if it is too small, rather than propagating through
1702 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1703 ** extremely complex negations of CONS types were not being
1704 sufficiently canonicalized, leading to inconsistencies in
1706 ** VALUES tranformer lost derived type.
1708 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1709 * fixed a bootstrapping bug: the build process no longer assumes
1710 that the various BOOLE-related constants have the same value in
1711 host and target lisps. (noted by Paul Dietz' test suite on an
1712 SBCL binary built from CLISP)
1713 * The system can now be dynamically linked on the MIPS platform,
1714 which enables dynamic loading of foreign code from Lisp. (thanks
1715 to Ralf Baechle for discussions on the MIPS ABI)
1716 * The system now records debugging information for its own source
1717 files in a filesystem-position-independent manner, relative to
1718 the "SYS" logical host.
1719 * fixed a compiler bug: MV-LET convertion did not check references
1720 to the "max args" entry point. (reported by Brian Downing)
1721 * tweaked disassembly notes to be less confident about proclaiming
1722 some instruction as an LRA. (thanks to Brian Downing)
1723 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1724 now each have their own history, command character, and other
1725 characteristics. (thanks to David Lichteblau)
1726 * fixed some bugs revealed by Paul Dietz' test suite:
1727 ** compiler failure in compiling LOGAND expressions including a
1729 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1730 shift greater than 32.
1731 ** FLUSH-DEST did not mark blocks for type check regeneration.
1732 ** HANDLER-CASE failed to accept declarations in handler clauses
1733 in some circumstances.
1735 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1736 * New code in contrib/sb-introspect (still probably not entirely
1737 stable yet) provides some support for smart Lisp development
1738 environments like SLIME.
1739 * The conditions signalled for errors occurring when loading .fasl
1740 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1741 in a way which should help ASDF recover gracefully.
1742 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1743 stuff has been cleaned up. If you code contrib/ stuff, this might
1744 affect you, and you can look at contrib/README, contrib/STANDARDS,
1745 and/or the 0.8.4.27 diff to check.
1746 * In full calls the compiler now does not generate checks for declared
1747 argument types for all arguments.
1748 * various threading fixes
1749 ** and some experimental patches which didn't make it into
1750 the main tree for this release, but which are shipped in
1751 contrib/experimental-thread.patch as a possible fix for some
1752 failures (deadlock, spinning...) in GC-intensive multithreaded
1754 * fixed PPC build problem (source code incompatibility of different
1755 library versions): added offsetof() hackery which attempts to divine
1756 where glibc maintainers put uc_mcontext today
1757 * fixed bug 282: compiler does not trust type assertions while passing
1758 arguments to a full call.
1759 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1760 &OPTIONAL VALUES type parameter.
1761 * fix bug 214: algorithm for noting rejected templates is now more
1762 similar to that of template seletion. (also reported by rydis on
1764 * fixed bug 141b: printing backquoted information readably and prettily
1765 inserts a space where necessary.
1766 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1767 (CHARACTER 10) are now reported as errors, rather than propagated
1768 as unknown types. (reported by piso on #lisp)
1769 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1770 with values NIL and :ERROR. (thanks to Milan Zamazal)
1771 * fixed bug 191c: CLOS now does proper keyword argument checking as
1772 described in CLHS 7.6.5 and 7.6.5.1.
1773 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1774 longer raise an error; further, using a list as a for-as-arithmetic
1775 counter now raises a meaningful error.
1776 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1777 now understood by sequence creation functions such as MAKE-SEQUENCE
1779 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1780 a STRING-INPUT-STREAM.
1781 * compiler enhancement: SIGNUM is now better able to derive the type
1783 * type declarations inside WITH-SLOTS are checked. (reported by
1785 * fixed some bugs revealed by Paul Dietz' test suite:
1786 ** incorrect optimization of TRUNCATE for a positive first
1787 argument and negative second.
1788 ** compiler failure in let-convertion during flushing dead code.
1789 ** compiler failure while deriving type of TRUNCATE on an
1790 interval, containing 0.
1791 ** ASH of a negative bignum by a negative bignum count now returns
1793 ** intersection of CONS types now canonicalizes properly, fixing
1794 inconsistencies in SUBTYPEP.
1796 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1797 * incompatible change: The --disable-debugger command line
1798 option now clobbers the debugger at a more fundamental
1799 level, by redefining #'INVOKE-DEBUGGER instead of by
1800 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1801 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1802 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1803 enter the debugger REPL and then suffer recursive errors
1804 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1805 while under the new system, BREAK in --disable-debugger mode
1806 terminates the system just as an unhandled error would.
1807 * fixed compiler performance when processing loops with a step >1;
1808 * bug fix: DOCUMENTATION now retrieves generic function
1809 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1810 support has been systematized, and now supports the methods
1811 specified by ANSI, along with a default method and a method for
1812 slot documentation. (reported by Nathan Froyd)
1813 * bug fix: effective methods associated with a generic function are
1814 no longer cached over a change of that generic function's method
1815 combination. (reported by Andreas Fuchs)
1816 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1817 names a non-existent directory. (thanks to Andreas Fuchs)
1818 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1819 the PPC platform, lead to stack corruption; this has been fixed.
1820 (reported by Rainer Joswig)
1821 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1822 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1823 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1824 arithmetic in general, is now much more reliable on the PPC
1826 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1827 platform now returns the right answer.
1828 * optimization: restored some effective method precomputation in
1829 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1830 precomputation is now tunable.
1831 * optimization: compiler-internal data structure use has been
1832 reviewed, and changes have been made that should improve the
1833 performance of the compiler by about 20%.
1834 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1835 simple-base-strings and simple-bit-vectors is improved.
1836 * optimization: the optimization of 32-bit logical and arithmetic
1837 functions introduced in version 0.8.3 on the x86 has been
1838 implemented on the mips, ppc and sparc platforms; an
1839 implementation of the same facility, but for 64-bit arithmetic,
1840 has been added for the alpha.
1841 * microoptimization: the compiler is better able to make use of the
1842 x86 LEA instruction for multiplication by constants.
1843 * bug fix: in some situations compiler did not report usage of
1844 generic arithmetic in (SPEED 3) policy.
1845 * bug 145b fix: compiler used wrong type specifier while converting
1846 MEMBER-types to numeric.
1847 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1849 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1850 index argument now works properly on the Alpha platform.
1851 * bug fix: floating point exception treatment on the Alpha platform
1853 * bug fix: FILE-POSITION works much better on string input and
1854 output streams. (thanks to Nikodemus Siivola)
1855 * bug fix: many threading/garbage collection symptoms sorted.
1856 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1857 might be pseudo-atomic.
1858 * internal change: Stopping for GC is now done with signals not
1859 ptrace. GC is now done in whichever thread wanted it, instead of
1861 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1863 * bug fix: result form in DO is not contained in the implicit
1865 * incompatible change: ICR structure is changed; the value part of
1866 CONTINUATION is now called LVAR; corresponding functions are
1867 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1868 * added type deriver for ISQRT (thanks to Robert E. Brown).
1869 * bug fix: better support for loading from the command line when an
1870 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1872 * fixed some bugs revealed by Paul Dietz' test suite:
1873 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1874 ** ROUND and FROUND now give the right answer when given very
1875 small float arguments.
1876 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1878 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1879 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1880 ** LCM should return a non-negative integer.
1881 ** PARSE-INTEGER returned the index of a terminator instead of the
1882 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1883 ** PARSE-INTEGER returned an incorrect index being applied to a
1885 ** LCM with two arguments of 0 returns 0 rather than signalling
1887 ** unsigned addition of a 32-bit constant with the high bit set no
1888 longer causes an internal compiler error on the x86.
1889 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1890 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1891 with negative last argument.
1892 ** byte specifiers with zero size and position no longer cause
1893 an error during type derivation.
1894 ** bignum multiplication on the Alpha platform now returns the
1896 * porting: The system now builds on SuSE AMD64, although it still
1897 generates a 32-bit binary.
1898 * .fasl file incompatibility: The fasl file version number has
1899 been incremented (because of the changes to internal compiler
1900 data structures referred to above).
1902 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1903 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1904 more accurately, on the Darwin kernel running on PowerPC hardware.
1905 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1906 * Compiler code deletion notes now signal a condition of type
1907 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1908 an associated MUFFLE-WARNING restart.
1909 * The compiler now performs limited argument count validation of
1910 constant format strings in FORMAT, and where appropriate in ERROR,
1911 CERROR and WARN. (thanks to Gerd Moellmann)
1912 * New ASDF-INSTALL contrib can be used for automatic download and
1913 installation of third-party Lisp code from CCLAN or other sites
1915 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1916 which forces another thread to execute a function supplied by the
1918 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1919 now accept and act upon their :ELEMENT-TYPE keyword argument.
1920 (reported by Martin Atzmueller, Edi Weitz)
1921 * bug fix: FILE-POSITION now accepts position designators up to
1922 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1923 is the greater. (thanks to Patrik Nordebo)
1924 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1925 Kalle Olavi Niemitalo)
1926 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1927 printable. (reported by Eric Marsden)
1928 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1929 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1930 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1931 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1932 (thanks to Stig Sandoe)
1933 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1934 UNREAD-CHAR now correctly includes the unread character in the
1935 target sequence. (thanks to Gerd Moellmann)
1936 * bug fix in threaded builds: the system can now be suspended and
1937 resumed by shell job control with minimal disruption.
1938 * bug fixes in times and timezones >2038AD
1939 * better handling of "where is GNU make?" problem in build scripts
1940 (thanks to Nikodemus Siivola)
1941 * new optimization: inside a named function any reference to a
1942 function with the same name is considered to be a self-reference;
1943 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1944 optimization quality.
1945 * new optimization on x86: logical functions and + now have
1946 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1947 used when the result is truncated to 32 bits.
1948 * VALUES declaration is partially enabled.
1949 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1950 * fixed some bugs revealed by Paul Dietz' test suite:
1951 ** The system now obeys the constraint imposed by
1952 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1953 types form a lattice under type intersection.
1954 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1955 ** ASSOC now ignores NIL elements in an alist.
1956 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1957 and (1+ MOST-POSITIVE-FIXNUM) answers.
1958 ** The addition of a method with invalid qualifiers to a generic
1959 function does not cause an error to be signalled immediately;
1960 a warning is signalled, and the error is generated only on
1961 calling the generic function.
1962 * changed .fasl file version number, in part to add type codes for
1963 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1964 obscure ANSI requirements
1966 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1967 * fixed bug 148: failure to inline-expand a local function left
1968 garbage, confusing the compiler.
1969 * fixed bugs 3cd: structure slot readers perform type check if the
1970 slot can have an invalid value (i.e. it is either not initialized
1971 or can be written with a less specific slot writer).
1972 * bug fix: the compiler now traps array references to elements off
1973 the end of an array; previously, the bounds checking in some
1974 circumstances could go off-by-one.
1975 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1977 * type declarations for array element types now obey the description
1978 on the CLHS page "Declaration TYPE", as per discussions on
1979 sbcl-help around 2003-05-08. This means that a declaration
1980 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1981 declaration, all references to BAR will be asserted or assumed
1982 (with THE, so dependent on compiler policy) to involve objects of
1983 type FOO. Note that no such declaration is implied in
1984 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1985 * declared types of functions from the "Conditions"
1986 chapter. (reported by Paul Dietz)
1987 * bug fix: CERROR accepts a function as its first argument.
1988 * bug fix: NTH an NTHCDR accept a bignum as index
1989 arguments. (reported by Adam Warner)
1990 * optimization: character compare routines now optimize comparing
1991 against a constant character. (reported by Gilbert Baumann)
1992 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1993 argument now works properly on the MIPS platform.
1994 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1995 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1996 declared to have a functional type, does not cause an error
1998 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1999 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2000 * fixed bug: if last continuation of a deleted block has a
2001 destination, this destination should be deleted too. (reported by
2003 * fixed a bug in the bootstrap process: the host compiler's values
2004 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2005 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2006 test case from Patrik Nordebo)
2007 * improved the ability of the disassembler on the PPC platform to
2008 provide helpful disassembly notes.
2009 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2010 the class in more cases than previously.
2011 * bug fix: FILE-POSITION now understands :START and :END for
2012 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2013 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2014 8) data. (thanks to David Lichteblau)
2015 * bug fix: it is possible to add a method to a generic function
2016 without lambda list.
2017 * bug fix: reader failed to signal END-OF-FILE inside an
2018 object representation. (reported by Nikodemus Siivola)
2019 * fixed some bugs revealed by Paul Dietz' test suite:
2020 ** LAST and [N]BUTLAST should accept a bignum.
2021 ** condition slot accessors are methods.
2022 ** (VECTOR NIL) is a subtype of STRING.
2024 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2025 * minor incompatible change: some nonsensical specialized lambda
2026 lists (used in DEFMETHOD) which were previously ignored now signal
2028 * minor incompatible change: the system is now aware of the types of
2029 variables in the COMMON-LISP package, and will signal errors for
2030 most violations of these type constraints (where previously they
2031 were silently accepted).
2032 * minor incompatible change: COMPILE-FILE now uses the freedom
2033 afforded (ANSI 3.2.2.3) to use derived function types for
2034 functions defined in the same file. This also permits the system
2035 to warn on static type mismatches and function
2036 redefinition. (Currently it does not work with high DEBUG level.)
2037 * minor incompatible change: VALUES declaration is disabled.
2038 * When issuing notes, the compiler now signals a condition of type
2039 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2040 restart for use in user handlers. It is expected that the
2041 COMPILER-NOTE condition will eventually become a condition
2042 supertype to a hierarchy of note types, which will then be
2043 handleable in a similar fashion. However, at the moment, no such
2044 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2045 but it's an implementation detail, not a classification for the
2047 * Changes in type checking closed the following bugs:
2048 ** type checking of unused values (192b, 194d, 203);
2049 ** template selection based on unsafe type assertions (192c, 236);
2050 ** type checking in branches (194bc).
2051 * A short form of VALUES type specifier has ANSI meaning (it has
2052 increased the number of situations when SBCL cannot perform type
2054 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2055 keywords or constants is permissible.
2056 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2057 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2058 classes. (thanks to Antonio Martinez)
2059 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2060 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2061 argument to a pretty-print stream. Instead, it leaves any such
2062 operations to DESCRIBE-OBJECT methods.
2063 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2065 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2066 lambda lists. (thanks to David Lichteblau)
2067 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2068 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2069 (thanks to Gerd Moellmann)
2070 * fixed bug 246: increased compilation speed of long
2071 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2073 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2075 * DEFCONSTANT now throws a condition of type
2076 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2077 constant to a non-EQL value; CONTINUE and ABORT restarts
2078 respectively change and preserve the value.
2079 * fixed bug 63: The code walker, part of the implementation of CLOS,
2080 is now better at handling symbol macros.
2081 * bug fix: There is no longer an internal implementation type named
2082 CL:LENGTH. (reported by Raymond Toy)
2083 * bug fix: In macro-like defining macros/special operators the
2084 implicit block does not enclose the lambda list.
2085 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2086 enclosed in parentheses) are not suitable as type specifiers, and
2087 their use properly signals an error now.
2088 * bug fix: An argument count mismatch for a type specifier in code
2089 being compiled no longer causes an unhandled error at compile
2090 time, but signals a compile-time warning.
2091 * fixed simple vector readable printing
2092 * bug fix: DESCRIBE takes more care over whether the class
2093 precedence list slot of a class is bound before accessing it.
2094 (reported by Markus Krummenacker)
2095 * bug fix: FORMATTER can successfully compile pretty-printer format
2096 strings which use variants of the ~* directive inside.
2097 * bug fix: SEARCH now applies its TEST predicate to the elements of
2098 the arguments in the correct order. (thanks to Wolfhard Buss)
2099 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2100 in type inference. (thanks to Robert E. Brown)
2101 * bug fix: Reading in symbols with an explicit package name of ""
2102 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2103 * fixed some bugs revealed by Paul Dietz' test suite:
2104 ** NIL is now allowed as a structure slot name.
2105 ** Arbitrary numbers, not just REALs, are allowed in certain
2106 circumstances in LOOP for-as-arithmetic clauses.
2107 ** Multiple class redefinitions before slot access no longer
2108 causes a type error.
2109 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2110 association between the name and a class.
2111 ** Generic functions with non-standard method-combination and over
2112 five methods all of which return constants no longer return NIL
2113 after the first few invocations. (thanks to Gerd Moellmann)
2114 ** CALL-NEXT-METHOD with no arguments now passes the original
2115 values of the arguments, even in the presence of assignment.
2116 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2118 ** DISASSEMBLE works with closures and funcallable instances.
2119 ** ADD-METHOD now returns the generic function, not the new method.
2120 ** FIND-METHOD signals an error if the lengths of the specializers
2121 is incompatible with the generic function, even if the ERRORP
2123 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2124 which its argument is a member.
2125 ** DEFCLASS only redefines the class named by its class-name
2126 argument if that name is the proper name of the class;
2127 otherwise, it creates a new class.
2128 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2129 of the UNBOUND-SLOT condition to the name of the slot.
2130 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2131 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2132 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2133 SLOT-MAKUNBOUND in the specified fashion.
2135 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2136 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2137 cross-compilation host. As a consequence, we can now bootstrap our
2138 way up to SBCL starting with a bare gcc toolchain and human-readable
2139 source code (first the source to CLISP, then the source to SBCL).
2140 * A contributed module containing a partial implementation of the
2141 simple-streams interface has been included. (thanks to Rudi
2143 * A contributed module implementing the RFC1321 Message Digest
2144 Algorithm, known as MD5, has been included.
2145 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2146 no longer has any effect, as the code controlled by this feature
2147 has been deleted. (As far as we know, no-one has ever built using
2148 this feature, and its semantics were confused in any case).
2149 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2150 work on conditions (as required by the ANSI specification),
2151 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2152 expected behaviour on conditions. Users should note, however,
2153 that such behaviour is not required by the ANSI specification,
2154 and so use of this behaviour may render their code unportable.
2155 * fixed some bugs revealed by Paul Dietz' test suite:
2156 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2158 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2159 ** SXHASH on condition objects no longer returns NIL.
2160 ** :ALLOCATION :CLASS slots are better treated; their values are
2161 updated on class redefinition, and initforms inherited from
2162 superclasses are applied.
2163 ** REMOVE-METHOD returns its generic function argument even when
2164 no method was removed.
2165 ** SHARED-INITIALIZE now initializes the values of the requested
2166 slots, including those with :ALLOCATION :CLASS.
2167 ** ALLOCATE-INSTANCE now works on structure classes defined via
2168 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2170 ** SLOT-EXISTS-P now works on conditions, as well as structures
2172 ** MAKE-LOAD-FORM now has the required methods on
2173 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2174 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2175 keyword, and now implements the SLOT-NAMES argument.
2176 ** methods with &OPTIONAL arguments no longer allow too many
2177 arguments to be passed in the call without error.
2178 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2179 option is consistent with the required arguments of the generic
2180 function lambda list.
2181 * bug fix: REQUIRE accepts a string designator. (Thanks to
2183 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2184 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2185 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2186 * bug fix: The compiler now checks for duplicated variables in macro
2188 * bug fix: SETQ on globals returns the correct value.
2189 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2190 not a COMPILER-ERROR (followed by some other strange error on
2191 choosing the CONTINUE restart).
2192 * bug fix: make.sh and friends are now more consistent in the way that
2193 they look for GNU "make".
2195 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2196 * experimental native threads support (on x86 Linux >=2.4 only).
2197 This is not compiled in by default: you need to add :SB-THREAD to
2198 the target features. See the "Beyond ANSI" chapter of the manual
2200 * fix for longstanding nonANSIism: The old distinction between
2201 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2202 The return value from CL:FIND-CLASS is now a CLOS class, and
2203 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2204 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2205 * An interface to the MetaObject Protocol, as described in Kiczales,
2206 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2207 MIT Press, 1991, is available from the SB-MOP package.
2208 * incompatible change: the SB-PCL package should now be considered
2209 a private implementation detail, and no longer a semi-private MOP
2211 * minor incompatible change: due to rearrangement for threads, the
2212 control stack and binding stack are now allocated at arbitrary
2213 addresses instead of being hardcoded per-port. Users affected by
2214 this probably have to be doing advanced things with shared
2215 libraries, and will know who they are.
2216 * minor incompatible change: Previously, all --eval forms used were
2217 processed with READ before any of them were processed with EVAL.
2218 Now each --eval form is processed with both READ and EVAL before
2219 the next --eval form is processed. (Thus package operations like
2220 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2221 work as the user might reasonably expect.)
2222 * minor incompatible change: *STANDARD-INPUT* is now only an
2223 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2225 * minor incompatible change: Y-OR-N-P is now character-oriented,
2226 not line oriented. Also, YES-OR-NO-P now works without errors.
2227 (thanks to Antonio Martinez)
2228 * sb-aclrepl module improvements: an integrated inspector, added
2229 repl features, and a bug fix to :trace command.
2230 * Known functions, which cannot be open coded by the backend, are
2231 considered to be able to check types of their arguments. (fixing
2232 a bug report by Nathan J. Froyd)
2233 * fixed a bug in computing method discriminating functions: It is
2234 now possible to define methods specialized on classes which have
2235 forward-referenced superclasses. (thanks to Gerd Moellmann)
2236 * fixed evaluation order in optional entries (reported by Gilbert
2238 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2239 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2240 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2241 required-by-AMOP NAME argument, as well as CLASS and
2242 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2243 * fixed bug 20: DEFMETHOD can define methods using names that are
2244 not the proper names of classes to designate class specializers.
2245 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2246 Unix file descriptors, instead of blowing up. (thanks to Antonio
2248 * Garbage collection refactoring: user-visible change is that a
2249 call to the GC function during WITHOUT-GCING will not do garbage
2250 collection until the end of the WITHOUT-GCING. If you were doing
2251 this you were probably losing anyway.
2252 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2253 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2254 TYPEP the latter but not the former.
2255 * The compiler issues a full WARNING for calls to undefined functions
2256 with names from the CL package.
2257 * MAP-INTO for a vector destination is open coded. (reported by
2258 Brian Downing on c.l.l)
2259 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2260 documentation string.
2261 * fixed some bugs revealed by Paul Dietz' test suite:
2262 ** COPY-ALIST now signals an error if its argument is a dotted
2264 ** Condition slots are now accessed more correctly in the presence
2265 of multiple initargs for a given slot.
2266 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2267 exclude restarts of the same name associated with a different
2269 ** DEFCLASS of forward-referenced classes with another
2270 forward-referenced class in the superclasses list no longer
2272 ** Condition slots are now initialized once each, not multiple
2273 times. (thanks to Gerd Moellmann)
2274 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2276 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2277 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2278 arguments contain duplicated elements.
2279 ** RESTART-CASE understands local macros.
2280 ** RESTART-CASE associates exactly its own restarts with a condition.
2281 ** ENDP in safe mode checks its argument to be of type LIST.
2282 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2283 in question is unbound.
2284 ** Optimized MAKE-INSTANCE functions no longer cause internal
2285 assertion failures in the presence of duplicate initargs.
2286 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2287 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2288 propagated even in the ctor optimized implementation of
2290 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2292 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2293 * a better implementation of SXHASH on (simple) bit vectors,
2294 measured both in execution speed and in distribution of results
2295 over the positive fixnums, has been installed. Likewise, a better
2296 implementation of EQUAL for simple bit vectors is now available.
2297 * fixed CEILING optimization for a divisor of form 2^k.
2298 * fixed bug 240 (emitting extra style warnings "using the lexical
2299 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2300 by Antonio Martinez)
2301 * fixed SXHASH, giving different results for NIL depending on type
2302 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2303 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2304 variable. (found by Rolf Wester)
2305 * SBCL does not ignore type declarations for special
2306 variables. (reported by rif on c.l.l 2003-03-05)
2307 * some bug fixes in contrib/sb-aclrepl/
2308 * fixed some bugs revealed by Paul Dietz' test suite:
2309 ** a bug in the CONS type specifier, whereby the CAR and CDR
2310 types got intertwined, has been fixed;
2311 ** the type system is now able to reason about the interaction
2312 between INTEGER and RATIO types more completely;
2313 ** APPEND, [N]REVERSE and NRECONC check that those their
2314 arguments, which must be proper lists, are really so;
2315 ** An array specialized to be unable to hold elements has been
2316 implemented, as required -- yes, really -- by ANSI;
2317 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2318 on malformed property lists;
2320 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2321 * incompatible packaging change: in line with Unix convention,
2322 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2323 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2324 * REQUIRE and PROVIDE are now optionally capable of doing something
2325 useful. See the documentation string for REQUIRE.
2326 * infrastructure for a managed SBCL contrib system: contributed
2327 modules in this release include:
2328 ** the ASDF system definition facility;
2329 ** an interface to the BSD Sockets API;
2330 ** an ACL-like convenience interface to the repl;
2331 (thanks to Kevin Rosenberg)
2332 ** an implementation of ROTATE-BYTE, with efficient implementation
2334 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2335 gives a closer approximation to the right answer than previously.
2336 (thanks to Raymond Toy)
2337 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2338 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2339 optional environment arguments, as required by ANSI.
2340 * fixed bugs in other functions taking environment objects, allowing
2341 calls with an explicit NIL environment argument to be compiled
2343 * fixed bug 228: primary return values from
2344 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2345 COMPILE or FUNCTION.
2346 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2347 :TYPE will no longer signal errors on innocuous objects.
2348 * fixed bug 231b: SETQ is better at respecting type declarations in
2349 the lexical environment.
2350 * fixed a bug in DEFCLASS: classes named by symbols with no or
2351 unprintable packages can now be defined.
2352 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2353 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2354 * fixed compiler failure related to checking types of functions.
2355 (reported by Robert E. Brown)
2356 * the compiler is now much more consistent in its error-checking
2357 treatment of bounding index arguments to sequence functions: in
2358 (SAFETY 3) code, errors will be signalled in almost all cases if
2359 invalid sequence bounding indices are passed to functions defined
2360 by ANSI to operate on sequences.
2361 * fixed a bug in the build procedure: documentation of SBCL-specific
2362 packages is now preserved and available in the final Lisp image.
2363 * lifted FDEFINITION lookup out of loops in the implementation of
2364 many list operations. (thanks to Robert E. Brown)
2365 * fixed a bug in the reader: the #n# reader macro now works for
2366 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2367 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2368 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2369 E. Brown and Tony Martinez respectively)
2370 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2371 SLOT-BOUNDP: in particular, it is now able to optimize them much
2372 better, and is now not vulnerable to having packages renamed.
2373 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2374 always called when a slot is not present in an instance. (thanks
2376 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2377 longer signal an error on unreading a character following EOF on
2378 the previous constituent stream. (thanks to Tony Martinez)
2379 * fixed some bugs revealed by Paul Dietz' test suite:
2380 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2381 not just nonnegative fixnums;
2382 ** the logical bit-array operators such as BIT-AND now accept an
2383 explicit NIL for their "opt-arg" argument (to indicate a
2384 freshly-consed result bit-array);
2385 ** ELT now signals an error on an invalid sequence index in safe
2387 ** the type system is now cleverer about negations of numeric
2388 types, and consequently understands the BIGNUM and RATIO types
2390 ** the type system is now cleverer about the interaction between
2391 INTEGER and RATIO types: while bugs still remain, many more
2392 cases are accurately computed;
2393 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2394 if it is in the last clause;
2395 ** CONSTANTLY now correctly returns a side-effect-free function in
2397 ** DECLARE is no longer treated as a special-operator; in
2398 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2399 * incremented fasl file version number due to the change in the
2400 DEFSTRUCT-SLOT-DESCRIPTION structure.
2402 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2403 * minor incompatible change: code processed by the "interpreter" or
2404 EVAL now has a compilation optimization policy of (DEBUG 2)
2405 (changed from (DEBUG 1)) to improve debuggability of interactive
2406 development, and to allow the use of the debug RETURN command in
2408 * an experimental implementation of the RETURN command for the
2409 debugger has been included. (thanks to Frederik Kuivinen)
2410 * fixed bug 62: constraints were not propagated into a loop.
2411 * fixed bug in embedded calls of SORT (reported and investigated by
2413 * fixed some bugs revealed by Paul Dietz' test suite:
2414 ** printing and reading of arrays with some dimensions having
2415 length 0 (thanks to Gerd Moellmann);
2416 ** BOA constructor with &AUX argument without a default value does
2417 not cause a type error;
2418 ** CONSTANTP now returns true for all self-evaluating objects.
2420 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2421 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2422 accessors that are related by inheritance, as specified in the
2423 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2424 Valtteri Vuorikoski)
2425 * The compiler is now able to inline functions that were defined in
2426 a complex lexical environment (e.g. inside a MACROLET).
2427 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2429 * Support for the upcoming FreeBSD-5.0 release has been included.
2430 (thanks to Dag-Erling Smorgrav)
2431 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2432 effect when it is not in a toplevel context.
2433 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2434 stand a better chance of being correct. (thanks to Gerd
2436 * fixed bug in COERCE, which now signals an error on coercing a
2437 rational to a bounded real type which excludes the expected
2439 * The compiler is now able to derive types more accurately from the
2440 COERCE and COMPILE functions.
2441 * fixed bug 223: functional binding is considered to be constant
2442 only for symbols in the CL package.
2443 * fixed bug 231: SETQ did not check the type of a variable being set
2444 (reported by Robert E. Brown)
2445 * A new optimization for MAKE-INSTANCE has been included, fixing
2446 various bugs (including relating to :ALLOCATION :CLASS slots and
2447 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2449 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2450 ** As required by ANSI, LOOP now disallows anonymous collection
2451 clauses such as COLLECT I in conjunction with aggregate boolean
2452 clauses such as THEREIS (= I 1);
2453 ** LOOP now signals an error when any variable is reused in the
2454 same loop (including the potentially useful construct analogous
2455 to WITH A = 1 WITH A = (1+ A);
2456 ** IT is only a special loop symbol within the first clause of a
2457 conditional loop clause;
2458 ** LOOP with a typed iteration variable over a hashtable now
2459 signals a type error iff it should.
2460 * fixed some other bugs revealed by Paul Dietz' test suite:
2461 ** FILE-STREAM now names the class previously known as FD-STREAM;
2462 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2463 argument) no longer signals an error;
2464 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2465 of the slot symbol, rather than using the current package
2466 ((:CONC-NAME "") continues to intern the slot's name in the
2468 * incremented fasl file version number, because of the incompatible
2469 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2470 of the new implementation of DEFINE-COMPILER-MACRO.
2472 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2473 * Support for building SBCL for MIPS platforms running in
2474 little-endian mode has now been checked in, and basic
2475 functionality on said platforms verified.
2476 * minor incompatible change: PCL now records the pathname of a file
2477 in which methods and the like are defined, rather than its
2479 * minor incompatible change: TRUENAME now considers the truename of
2480 a file naming a directory to be the pathname with :DIRECTORY
2481 component indicating that directory.
2482 * minor incompatible change: a NAMED clause in the extended form of
2483 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2484 reason for the previous behaviour is unclear.
2485 * more systematization and improvement of CLOS and MOP conformance
2486 in PCL (thanks to Gerd Moellman and Pierre Mai):
2487 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2489 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2490 primary methods with no specializers;
2491 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2493 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2494 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2495 FORWARD-REFERENCED-CLASSes; error reporting on
2496 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2498 ** SXHASH on CLOS instances now uses a slot internal to the
2499 instance to return different numbers on distinct instances,
2500 while preserving the same return value through invocations of
2502 ** DEFMETHOD signals errors when methods with longer incongruent
2503 lambda lists are added to generic functions;
2504 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2505 CLASS, as specified in AMOP;
2506 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2507 on the order returned by the primary method for classes of
2508 class STANDARD-CLASS;
2509 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2510 * fixed some bugs shown by Paul Dietz' test suite:
2511 ** DOLIST puts its body in TAGBODY;
2512 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2514 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2515 value producing form;
2516 ** if more variables are given to PROGV than values, extra
2517 variables are bound and made to have no value;
2518 ** NSUBSTITUTE on list arguments gets the right answer with
2520 ** ELT signals an error of type TYPE-ERROR when the index argument
2521 is not a valid sequence index;
2522 ** LOOP signals (at macroexpansion time) an error of type
2523 PROGRAM-ERROR when duplicate variable names are found;
2524 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2525 ** FUNCALL of special-operators now cause an error of type
2527 ** PSETQ now works as required in the presence of side-effecting
2528 symbol-macro places;
2529 ** NCONC accepts any object as its last argument;
2530 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2532 ** loop-for-as-package does not require a package to be explicitely
2534 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2536 * fixed bug 166: compiler preserves "there is a way to go"
2537 invariant when deleting code.
2538 * fixed bug 172: macro lambda lists with required arguments after
2539 &REST arguments now cause an error to be signalled. (thanks to
2541 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2542 bug. (thanks to Matthew Danish)
2543 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2545 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2546 arguments are lexically rebound. (thanks to Gerd Moellmann and
2548 * fixed bug 194: error messages are now more informative when there
2549 is no primary method applicable in a call to a generic
2550 function. (thanks to Gerd Moellmann)
2551 * fixed bug in command line argument checking (thanks to Julian
2553 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2554 (thanks to Matthew Danish)
2555 * incremented fasl file version number, because of the
2556 SXHASH-related changes in the layout of CLOS data structures
2558 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2559 * minor incompatible change: The runtime (the Unix executable named
2560 "sbcl") is now much pickier about the .core files it will load.
2561 Essentially it now requires .core files to descend from the same
2562 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2563 as the runtime does. (The intent is to prevent the crashes which
2564 can occur, and which can even be reported as mysterious failures,
2565 when people patch the sources or change the build parameters
2566 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2567 sbcl and .core files.)
2568 * fixed bug: VALUES-LIST is no longer optimized away.
2569 * fixed bug 142: The FFI conversion of C string values to Lisp
2570 string values no longer conses excessively. (thanks to Nathan
2571 Froyd porting Raymond Toy's fix to CMU CL)
2572 * began to systematize and improve MOP conformance in PCL (thanks to
2573 Nathan Froyd, Gerd Moellman and Pierre Mai):
2574 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2576 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2577 ** FINALIZE-INHERITANCE is now called on class finalization;
2578 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2579 argument precedence order.
2580 * fixed bug 202: The compiler no longer fails on functions whose
2581 derived types contradict their declared type.
2582 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2583 so it can be non-toplevel.
2584 * The fasl file version number has changed (because of the new
2585 implementation of DEFMACRO).
2586 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2587 safe code, that any length requirement by their type-specifier
2588 argument is valid. The exceptions to this are described in bug
2590 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2591 functions now check the types of their inputs as required by ANSI.
2592 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2593 introduced symbol is DECLAREd to be SPECIAL.
2594 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2595 * fixed bug: :COUNT argument to sequence functions may be negative
2596 * fixed bug: body of DO-SYMBOLS may contain declarations
2597 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2598 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2599 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2600 symbol macro only once
2601 * fixed printing of call frame when argument list is unavailable
2602 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2603 * compiler no longer signals WARNING on unknown keyword
2606 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2607 * A beta-quality port to the mips architecture running Linux,
2608 based on the old CMUCL backend, has been made. It has been tested
2609 on a big-endian kernel, and works sufficiently well to be able to
2610 rebuild itself; it has not been tested in little-endian mode.
2611 * fixed an inconsistency between gencgc.c and purify.c which made
2612 dumping/loading .core files unreliable
2613 * fixed bug 120a: The compiler now deals correctly with IFs where
2614 the consequent is the same as the alternative, instead of
2615 misderiving the return type. (thanks to Alexey Dejneka)
2616 * fixed bug 113: Logical pathnames are now dumpable (the logical
2617 host is resolved at load-time, throwing an error if it is not
2619 * fixed bug 174: FORMAT's error message is slightly clearer when a
2620 non-printing character is used in a format directive.
2621 * fixed several bugs in compiler checking of type declarations, i.e.
2622 violations of the Python "declarations are assertions" principle
2623 (thanks to Alexey Dejneka)
2624 * fixed several bugs in PCL's error checking (thanks to Gerd
2626 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2628 * fixed bug in compilation of functions as first class values
2629 (thanks to Antonio Martinez-Shotton)
2630 * The compiler's handling TYPE-ERRORs which it can prove will
2631 inevitably happen at runtime has been cleaned up and corrected
2632 in several ways. (thanks to Alexey Dejneka)
2633 * improved argument type checking for various basic arithmetic
2634 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2635 lavished on them in the past that they can be compiled in many
2636 ways in different special cases
2637 * fixed bug 181: compiler checks validity of user supplied type
2639 * cleaned up code flushing in optimization: Function calls which
2640 should signal errors for safety purposes (e.g. which ANSI says
2641 should signal errors when their arguments are of incorrect type)
2642 are no longer optimized away.
2643 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2644 * incremented fasl file version number, because changes in the
2645 implementation of sequence functions like COERCE caused
2646 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2647 in old inline expansions) to become undefined. (Actually these
2648 changes were later undone, so we might very well be binary
2649 compatible with 0.7.7 after all, but leaving the version number
2650 incremented seemed like the simplest and most conservative
2653 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2654 * An alpha-quality port to the parisc architecture running Linux,
2655 based on the old CMUCL backend, has been made. This, even more so
2656 than the other backends, should be considered still a work in
2657 progress; known problems include that the Linux kernel in 64-bit
2658 mode does not propagate the correct sigcontext structure to
2659 userspace, and consequently SBCL on a parisc64 kernel will not
2661 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2662 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2663 suppressed.) Also now that the compiler is looking at declarations
2664 in the environment, it checks optimization declarations as well,
2665 and suppresses inlining when (> DEBUG SPEED).
2666 * More fixes have been made to treatment of floating point exception
2667 treatment and other Unix signals. In particular, floating point
2668 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2669 * The detection and handling of control stack exhaustion (infinite
2670 or very deeply nested recursion) has changed. Stack exhaustion
2671 detection is now done by write-protecting pages at the OS level
2672 and applies at all optimization settings; when found, a
2673 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2674 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2675 be caught using IGNORE-ERRORS.
2676 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2677 correctly when transforming recursive LABELS functions to LETs.
2678 (thanks to Alexey Dejneka)
2679 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2680 that are names of constants or global variables.
2681 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2682 alien routines with docstrings.
2683 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2684 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2686 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2687 (thanks to Eric Marsden)
2688 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2689 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2690 * Bug fix: The SPARC backend can now compile functions involving
2691 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2692 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2693 to a functions expecting a PATHNAME-DESIGNATOR.
2694 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2695 lambda lists. (thanks to Alexey Dejneka)
2696 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2697 functions so that EXPORTing the name of the class doesn't cause
2698 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2699 to Antonio Martinez for reporting this)
2700 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2701 constant-folded without failing an assertion. (thanks to Einar
2702 Floystad Dorum for reporting this)
2703 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2704 types are now dealt with more correctly by the compiler.
2705 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2706 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2707 behaviour analogous to RENAME-FILE. This puts its behaviour more
2708 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2710 * The fasl file version number has changed again. (because of the
2711 bug fix involving the names of PCL MAKE-INSTANCE functions)
2713 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2714 * bug fix: Floating point exceptions are treated much more
2715 consistently on the x86/Linux and PPC/Linux platforms.
2716 * Array initialization with :INITIAL-ELEMENT is now much faster for
2717 cases when the compiler cannot open code the array creation, but
2718 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2719 array accesses have also seen a speed increase.
2720 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2721 specified. (This was at the root of some bad interactions between
2722 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2724 * bug fix: Internal error arguments for undefined functions are now
2725 computed correctly on the PPC/Linux platform.
2726 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2727 Raymond Toy's patch for CMU CL)
2728 * Support for the Solaris 9 operating environment has been included
2729 (thanks to Daniel Merritt)
2730 * A very ugly but hopefully complete draft of the missing FFI chapter
2731 of the manual has been created by reformatting the corresponding
2732 CMU CL manual chapter into (currently very ugly and incoherent)
2733 DocBook and bringing it up to date for SBCL behavior. Thus, the
2734 manual is now essentially complete, at least by my extreme
2735 once-and-only-once standards, whereby it's acceptable to refer to
2736 the doc strings of SB-EXT functions as the primary documentation.
2737 * The fasl file version number has changed again, due to cleanup of
2738 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2739 is no longer a static symbol.)
2741 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2742 * SBCL now builds with OpenMCL (version 0.12) as the
2743 cross-compilation host; also, more progress has been made toward
2744 bootstrapping under CLISP.
2745 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2747 * bug 158 fixed: The compiler can now deal with integer loop
2748 increments different from 1; fixing this turned out also to fix
2750 * bug 169 fixed: no more bogus warnings about using lexical bindings
2751 despite the presence of perfectly good SPECIAL declarations (thanks
2752 to David Lichteblau)
2753 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2754 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2755 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2757 * bug 180 fixed: Method combination specifications no longer ignore
2758 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2759 * bug fix: Structure type predicate functions now check their argument
2760 count as they should.
2761 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2762 correctly. (thanks to Pierre Mai)
2763 * minor incompatible change: The --noprogrammer option is deprecated
2764 in favor of the new --disable-debugger option, which is very similar.
2765 (The major difference is that it takes effect at a slightly different
2766 time at startup, causing handling of errors in --sysinit and
2767 --userinit files will be affected differently.) The
2768 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2769 been added to allow this functionality to be controlled from ordinary
2770 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2771 maintainers, who might want to run non-interactive scripts to
2772 build SBCL cores which will later be used interactively.)
2773 * minor incompatible change: The LOAD function no longer, when given
2774 a wild pathname to load, loads all files matching that pathname.
2775 Instead, an error of type FILE-ERROR is signalled.
2777 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2778 * bug 147 fixed: The compiler preserves its block link/count
2779 invariants more correctly now so that it doesn't crash. (thanks
2781 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2783 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2784 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2785 no-value-for-this-character as NIL (as specified by ANSI).
2786 * HOST-NAMESTRING on physical pathnames now returns a string that is
2787 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2788 (thanks to Christophe Rhodes)
2789 * The Alpha port handles icache flushing more correctly. (thanks to
2791 * More progress has been made toward bootstrapping under CLISP. (thanks
2792 to Christophe Rhodes)
2793 * The fasl file format has changed again, because dynamic loading
2794 on OpenBSD (which has non-ELF object files) motivated some cleanups
2795 in the way that foreign symbols are transformed and passed around.
2796 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2797 is no longer treated as whitespace by the reader, but instead as
2798 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2799 |A
\7fB|, instead of A as it used to.
2801 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2802 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2803 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2804 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2805 this as of 0.7.1.45, but was left out of the previous news section
2806 (thanks to Dan Barlow)
2807 * SBCL now runs on the Solaris operating system on SPARC architectures
2808 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2809 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2810 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2811 * SPARC backend cleanups, allowing builds of cores optimized for V8
2812 and V9 SPARCS, and also emission of code targeted to a particular
2813 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2815 * SBCL is closer to bootstrapping under CLISP, thanks to various
2816 fixes by Christophe Rhodes.
2817 * The fasl file format has changed again, to allow the compiler's
2818 INFO database to support symbol macros.
2819 * The user manual (in doc/) is formatted into HTML more nicely.
2820 (thanks to coreythomas)
2821 * The system is smarter about SUBTYPEP relationships, especially
2822 those involving NOT types (including types such as ATOM which are
2823 represented internally using NOT types). Thus SUBTYPEP is less
2824 likely to return (VALUES NIL NIL) in general, and in particular
2825 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2827 * The fasl file format has changed again, because the internal
2828 representation of types now includes a new slot to support the new
2829 SUBTYPEP-of-NOT-types logic.
2830 * (not a change in the main branch of SBCL, but a related prototype
2831 which can hopefully be merged into the main branch of SBCL in the
2832 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2833 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2834 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2835 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2836 dispatch macro character combinations. (thanks to Alexey Dejneka)
2837 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2838 that superficially look like logical namestrings correctly.
2839 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2841 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2842 default supplied-p) for &optional and &key arguments. (thanks to
2845 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2846 * incompatible change: The compiler is now less aggressive about
2847 tail call optimization, doing it only when (> SPACE DEBUG) or
2848 (> SPEED DEBUG). (This is an incompatible change because there are
2849 programs which relied on the old CMU-CL-style behavior to optimize
2850 away their unbounded recursion which will now die of stack overflow.)
2851 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2852 for non-GENCGC systems has been increased to 20M (since that
2853 seems much closer to the likely performance optimum for modern
2854 systems than the old 4M value was)
2855 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2856 and *DEBUG-PRINT-LEVEL*
2857 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2858 of CMU CL's support for SPARC, and various endianness and other
2859 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2860 * new syntactic sugar for the Unix command line: --load foo.bar is now
2861 an alternate notation for --eval '(load "foo.bar")'.
2863 ** The system now detects stack overflow and handles it gracefully,
2864 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2865 optimization settings. (This is a good thing in general, and
2866 its introduction in this version should be particularly timely
2867 for anyone whose code fails because of suppression of tail
2869 ** The system now hunts for the C variable "environ" in a more
2870 devious way, to avoid segfaults when the C library version differs
2871 between compile time and run time. (thanks to Christophe Rhodes)
2872 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2873 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2874 ** The compiler no longer issues bogus style warnings for undefined
2875 classes in the same source file as the DEFCLASSes which defined
2876 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2878 ** fixes in CONDITION class precedence list for undefined function
2879 errors (thanks to Alexei Dejneka)
2880 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2881 correctly. (thanks to Dan Barlow)
2882 ** portability fixes aiming at bootstrapping under CLISP (thanks
2883 to Dave McDonald and Christophe Rhodes)
2884 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2885 ** fixes in type translation and and type inference (thanks to
2887 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2888 ** various fixes in the new ports (thanks to Dan Barlow)
2889 * several changes related to debugging:
2890 ** suppression of tail recursion, as noted above
2891 ** stack overflow detection, as noted above
2892 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2893 is now the default. (For some time encapsulation has been more
2894 reliable than the breakpoint-based :ENCAPSULATE NIL
2895 implementation, at least on X86 systems; and I just noticed that
2896 encapsulation also seems closer to the spirit of the ANSI
2899 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2901 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2902 up properly again. (There was a packaging bug in 0.7.0 which
2903 left their definitions in SB-SYS::LOAD-FOREIGN and
2904 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2905 vital for most things which interface to C-level interfaces,
2906 like extensions working with sockets or databases or
2907 Perl-compatible regexes or whatever, and the need to fix
2908 this bug is the main reason that 0.7.1 was released so
2910 ** DEFGENERIC is now choosier about the methods it redefines, so that
2911 reLOADing a previously-LOADed file containing DEFGENERICs does
2912 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2913 cycle now works as it should. (thanks to Alexey Dejneka)
2914 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2915 fixed by Christophe Rhodes. (He actually submitted this patch
2916 months ago, and I delayed until after 0.7.0.)
2917 ** Bug 111 (internal compiler confusion about runtime checks on
2918 FUNCTION types) was fixed by Alexey Dejneka.
2919 * Some internal cleanups (getting rid of variables which aren't
2920 needed now that the byte interpreter is gone) caused the fasl
2921 file format number to change again.
2923 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2924 * major incompatible change: The default fasl file extension, i.e. the
2925 default extension for files produced by COMPILE-FILE, has changed
2926 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2928 ** There are many changes in the implementation of the compiler.
2929 SBCL is now essentially a compiler-only implementation of ANSI
2930 Common Lisp. EVAL still "interprets" a few special cases, but
2931 almost all the interesting cases are handled by creating
2932 a LAMBDA expression, calling COMPILE on it, then calling
2933 FUNCALL on the result.
2934 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2935 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2936 Since the code is newer, there might still be some new bugs
2937 (though not as many as before Martin Atzmueller's fixes:-). But
2938 the new code is substantially simpler and clearer, and hopefully
2939 any remaining bugs will be simpler, less fundamental, and more
2940 fixable then the bugs in the old code.
2941 ** The revised compiler is still a little unsteady on its feet.
2943 *** The debugging information it produces (particularly the names
2944 of FUNCTION objects) is sometimes much less useful than what
2945 the old compiler produced.
2946 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2947 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2948 within a MACROLET) has been temporarily weakened.
2949 ** There are new compiler optimizations for various functions:
2950 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2951 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2952 *** the math functions TRUNCATE, FLOOR, and CEILING
2953 *** the function-of-all-trades COERCE
2954 Mostly these should be transparent, but there's one
2955 potentially-annoying problem (bug 117): when the compiler
2956 inline-expands a function and does type analysis on the result,
2957 it can create control paths which have type mismatches, and
2958 when it can't prove that those control paths aren't taken,
2959 it will issue WARNINGs about the type mismatches. This is
2960 a particular problem in practice for the new sequence functions.
2961 It's not clear how this should be fixed, and for now, a
2962 workaround is given in the entry for 117 in the BUGS file.
2963 ** (Because of the interaction between the two previous items --
2964 occasional inlining problems and new inline expansions -- some
2965 of the new sequence function optimizations won't really kick in
2966 completely until debugging information, and then inlining, are
2967 straightened out in some future version.)
2968 * minor incompatible changes:
2969 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2970 DIRECTORY no longer implicitly promotes NIL slots of its
2971 pathname argument to :WILD. In particular, when you ask for the
2972 contents of a directory (which you used to be able to do without
2973 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2974 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2975 ** changes in behavior that ANSI explicitly defines to be
2976 implementation dependent:
2977 *** The new compiler-only implementation still conforms with ANSI,
2978 but acts a little different than before. Besides the obvious
2979 changes in performance tradeoffs (that the cost per form passed
2980 to EVAL has gone up, and the cost per form executed by EVAL
2981 has gone down), the behavior of the system changes a little
2982 because there are no longer any interpreted function objects.
2983 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2984 e.g. doing COMPILE on the output of interactive DEFUN is
2986 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2988 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2989 has changed from (:CL) to NIL.
2990 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2991 e.g. #\Newline and #\Tab, don't have names specified in the
2992 ANSI Common Lisp standard, is now based on their ASCII symbolic
2993 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2994 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2995 are no longer used for output.
2996 ** changes in internal implementation constants:
2997 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2998 4 million. (If your application spends a lot of time GCing and
2999 you have a lot of RAM, you might want to experiment with
3000 increasing it even more.)
3001 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3002 However, almost all old code should still continue to work without
3003 immediate update, as SB-C-CALL is now a (deprecated) nickname
3005 ** Old operator names in the style DEF-FOO are now deprecated in
3006 favor of new corresponding names DEFINE-FOO, for consistency with
3007 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3008 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3009 internal symbols, but a few supported extensions like
3010 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3011 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3012 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3013 etc. as you get deeper into recursive calls to the debugger
3014 command loop, instead of the old "5]", "5]]", "5]]]"
3015 sequence. (I was motivated to do this when squabbles between
3016 ILISP and SBCL left me very deeply nested in the debugger. In the
3017 short term, this change will probably provoke more ILISP/SBCL
3018 squabbles, but hopefully it will be an improvement in the long run.)
3019 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3020 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3021 function, since declaiming return types involving aliens is
3022 (1) annoyingly messy to do by hand and (2) vital to efficient
3023 compilation of code which calls such functions.
3024 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3025 longer reexported by the SB-EXT package. They're solely useful
3026 for alien code, so it seems more logical that you should get
3027 them from the SB-ALIEN package, not in SB-EXT.
3028 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3029 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3030 features. Instead, the code that they used to control is always
3031 built into the system.
3032 * many other bug fixes
3033 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3034 advantage of the new EVAL-WHEN stuff and to clean them up in
3035 general, and they are now more ANSI-compliant in a number of
3036 ways. Martin Atzmueller is responsible for a lot of this.
3037 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3039 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3040 *** correct ERROR type for various file operations
3041 *** some fixes for Lisp streams
3042 *** DEFMETHOD syntax checking
3043 *** changing old weird representation of debug information as
3044 strings (which, among their other deficiencies, don't transform
3045 correctly when you rename packages, and don't change their
3046 print representation when you change things like *PACKAGE*
3047 and *PRINT-LENGTH*) to symbols and lists of symbols
3048 He also made several improvements and fixed several bugs in DESCRIBE.
3049 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3051 *** misbehavior of WRITE-STRING/WRITE-LINE
3052 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3053 and several other LOOP problems as well
3054 *** DIRECTORY when similar filenames are present
3055 *** DEFGENERIC with :METHOD options
3056 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3057 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3058 *** argument ordering in FIND with :TEST option
3059 *** mishandled package designator argument in APROPOS-LIST
3060 *** various problems in the backquote readmacro
3061 *** a bug in APROPOS
3062 *** probably some others that I'm not describing very well here,
3063 since the CVS log documents them by reference to sbcl-devel
3064 messages, and the SourceForge archives aren't working well.:-(
3065 ** Dan Barlow improved the Alpha port (and is making progress on the
3066 PPC port, for those of you who think different).
3067 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3068 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3069 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3070 continues to work on the SPARC port (for those of you in a position
3071 to look down upon our little PC-compatible boxes from a great height).
3072 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3073 and uses that copy, rather than the current dynamic value, when
3074 it's trying to decide whether to truncate output. Thus e.g.
3075 (let ((*print-lines* 50))
3076 (pprint-logical-block (stream nil)
3078 (let ((*print-lines* 8))
3079 (print (aref possiblybigthings i) stream)))))
3080 should now truncate the logical block only at 50 lines, instead of
3081 often truncating it at 8 lines, as it did before.
3082 * The doc/cmucl/ directory, containing old CMU CL documentation
3083 from the time of the fork, is no longer part of the base system.
3084 SourceForge has shut down its anonymous FTP service, and with it
3085 my original plan for distributing the old CMU CL documentation
3086 there. For now, if you need these files you can download an old
3087 SBCL source release and extract them from it.
3088 * The fasl file version number changed again, for dozens of reasons,
3089 some of which are apparent above.
3091 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3092 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3093 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3094 patch, so that DISASSEMBLE output is much nicer.
3095 * The code in the SB-PROFILE package now seems reasonably stable.
3096 I still haven't decided what the final interface should look like
3097 (I'd like PROFILE to interact cleanly with TRACE, since both
3098 facilities use function encapsulation) but if you have a need
3099 for profiling now, you can probably use it successfully with
3100 the current CMU-CL-style interface.
3101 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3102 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3103 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3104 link now returns the dangling link itself, and for similar
3105 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3106 link itself. (In these cases the old code signalled an error and
3107 looped endlessly, respectively.) Thus, DIRECTORY now works even
3108 in the presence of dangling and cyclic symbolic links.
3109 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3110 is now a supported extension again, since the consensus on
3111 sbcl-devel was that it can be useful for ordinary development
3112 work, not just for debugging SBCL itself.
3113 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3114 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3115 that currently-defined functions might be redefined later with
3116 different return types.
3117 * Hash tables can be printed readably, as inspired by CMU CL code
3118 of Eric Marsden and SBCL code of Martin Atzmueller.
3119 * better error handling in CLOS method combination, thanks to
3120 Martin Atzmueller porting Pierre Mai's CMU CL patches
3121 * more overflow fixes for >16Mbyte I/O buffers
3122 * A bug in READ has been fixed, so that now a single Ctrl-D
3123 character suffices to cause end-of-file on character streams.
3124 In particular, now you only need one Ctrl-D at the command
3125 line (not two) to exit SBCL.
3126 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3128 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3129 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3130 does the right thing.
3131 * The implementation of some type tests, especially for CONDITION
3132 types, is now tidier and maybe faster, due to CMU CL code
3133 originally by Douglas Crosher, ported by Martin Atzmueller.
3134 * Some math functions have been fixed, and there are new
3135 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3136 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3137 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3138 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3139 likely to be built into sbcl-0.7.0.)
3140 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3141 CL, ported to SBCL by Martin Atzmueller)
3142 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3143 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3144 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3145 * There's a new slam.sh hack to shorten the edit/compile/debug
3146 cycle for low-level changes to SBCL itself, and a new
3147 :SB-AFTER-XC-CORE target feature to control the generation of
3148 the after-xc.core file needed by slam.sh.
3149 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3150 COMPILE-FILE is no longer supported, so that now every function
3151 gets an entry point, so that block compilation looks a little
3152 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3153 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3154 returns the number of bytes consed since the system started,
3155 rather than the number consed since the first time the function
3156 was called. (The new definition parallels ANSI functions like
3157 CL:GET-INTERNAL-RUN-TIME.)
3158 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3159 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3160 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3161 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3162 * Fasl file version numbers are now independent of the target CPU,
3163 since historically most system changes which required version
3164 number changes have affected all CPUs equally. Similarly,
3165 the byte fasl file version is now equal to the ordinary
3168 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3169 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3170 is no longer recognized. I apologize for this, because it was
3171 listed in SB-EXT as a supported extension, but I found that
3172 its existing behavior was poorly specified, as well as incorrectly
3173 specified, and it looked like too much of a mess to straighten it
3174 out. I have enough on my hands trying to get ANSI stuff to work..
3175 * many patches ported from CMU CL by Martin Atzmueller, with
3176 half a dozen bug fixes in pretty-printing and the debugger, and
3177 half a dozen others elsewhere
3178 * fixed bug 13: Floating point infinities are now supported again.
3179 They might still be a little bit flaky, but thanks to bug reports
3180 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3181 as flaky as they were.
3182 * The --noprogrammer command line option is now supported. (Its
3183 behavior is slightly different in detail from what the old man
3184 page claimed it would do, but it's still appropriate under the
3185 same circumstances that the man page talks about.)
3186 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3187 are now supported, and enabled by default. Thus, the compiler can
3188 handle many floating point and complex operations much less
3189 inefficiently. (Thus e.g. you can implement a complex FFT
3191 * The compiler now detects type mismatches between DECLAIM FTYPE
3192 and DEFUN better, and implements CHECK-TYPE more correctly, and
3193 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3194 thanks to patches from Martin Atzmueller.
3195 * various fixes to make the cross-compiler more portable to
3196 ANSI-conforming-but-different cross-compilation hosts (notably
3197 Lispworks for Windows, following bug reports from Arthur Lemmens)
3198 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3199 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3200 fixed thanks to Pierre Mai's CMU CL patches.
3201 * fixes to keep the system from overflowing internal counters when
3202 it tries to use i/o buffers larger than 16M bytes
3203 * fixed bug 45a: Various internal functions required to support
3204 complex special functions have been merged from CMU CL sources.
3205 (When I was first setting up SBCL, I misunderstood a compile-time
3206 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3207 * improved support for type intersection and union, fixing bug 12
3208 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3209 more obscure bugs as well
3210 * some steps toward byte-compiling non-performance-critical
3211 parts of the system, courtesy of patches from Martin Atzmueller
3212 * Christophe Rhodes has made some debian packages of sbcl at
3213 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3214 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3215 stable, but are nonetheless usable. When he's ready, I'd be happy
3216 to add them to the SourceForge "File Releases" section. (And if
3217 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3218 * new fasl file format version number (because of changes in
3219 internal representation of (OR ..) types to accommodate the new
3220 support for (AND ..) types, among other things)
3222 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3223 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3225 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3227 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3228 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3229 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3230 * DESCRIBE now works on structure objects again.
3231 * Most function call argument type mismatches are now handled as
3232 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3233 know whether the function will be redefined before the call is
3234 executed. (The compiler could flag local calls with full WARNINGs,
3235 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3236 it doesn't keep track of enough information to know whether calls
3237 are local in this sense.)
3238 * Compiler output is now more verbose, with messages truncated
3239 later than before. (There should be some supported way for users
3240 to override the default verbosity, but I haven't decided how to
3241 provide it yet, so this behavior is still controlled by the internal
3242 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3243 src/compiler/ir1util.lisp.)
3244 * Fasl file format version numbers have increased again, because
3245 support for the Gray streams extension changes the layout of the
3246 system's STREAM objects.
3247 * The Gray subclassable streams extension now works, thanks to a
3248 patch from Martin Atzmueller.
3249 * The full LOAD-FOREIGN extension (not just the primitive
3250 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3251 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3252 but like most other programs, it defaults to copying the Unix
3253 environment from the original process instead of starting the
3254 new process in an empty environment.
3255 * Extensions which manipulate the Unix environment now support
3256 an :ENVIRONMENT keyword option which doesn't smash case or
3257 do other bad things. The CMU-CL-style :ENV option is retained
3258 for porting convenience.
3259 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3260 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3262 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3264 * A patch from Martin Atzmueller seems to have solved the SIGINT
3265 problem, and as far as we know, signal-handling now works cleanly.
3266 (If you find any new bugs, please report them!)
3267 * The system no longer defaults Lisp source file names to types
3268 ".l", ".cl", or ".lsp", but only to ".lisp".
3269 * The compiler no longer uses special default file extensions for
3270 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3271 seems to expect a single default extension for all compiled code,
3272 and there's no compelling reason to try to stretch the standard
3273 to allow two different extensions.) Instead, byte-compiled files
3274 default to the same extension as native-compiled files.
3275 * Fasl file format version numbers have increased again, because
3276 a rearrangement of internal implementation packages made some
3277 dumped symbols in old fasl files unreadable in new cores.
3278 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3279 many fewer weird special cases.
3280 * Bug #17 (differing COMPILE-FILE behavior between logical and
3281 physical pathnames) has been fixed, and some related misbehavior too,
3282 thanks to a patch from Martin Atzmueller.
3283 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3284 by Tim Moore, ported to SBCL by Martin Atzmueller.
3285 * Martin Atzmueller fixed several filesystem-related problems,
3286 including bug #36, in part by porting CMU CL patches, which were
3287 written in part by Paul Werkowski.
3288 * More compiler warnings in src/runtime/ are gone, thanks to
3289 more patches from Martin Atzmueller.
3290 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3293 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3295 * DESCRIBE now works on CONDITION objects.
3296 * The debugger now handles errors which arise when trying to print
3297 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3299 * The build system now uses an additional file, customize-target-features.lisp,
3300 to allow local modifications to the target *FEATURES* list. (The point of
3301 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3302 debugging features enabled, without having to worry about propagating it
3303 into everyone's system when I do a "cvs update".) When no
3304 customize-target-features.lisp file exists, the target *FEATURES* list
3305 should be constructed the same way as before.
3306 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3307 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3308 warn about behavior which is undefined under the ANSI spec. Note: This
3309 is specified by ANSI, but it's not very popular with programmers.
3310 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3311 macro in the SBCL sources for an example of a workaround which you
3312 might use to make portable ANSI-standard code which does what you want.
3313 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3314 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3315 when it's used as a non-toplevel form. (This is required in order
3316 to implement the DEFCONSTANT-EQX macro.)
3317 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3318 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3319 funny things, due to bugs in the implementation of EVAL-WHEN
3320 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3321 non-toplevel DEFCONSTANTs are uncommon.)
3322 * The core file version number and fasl file version number have been
3323 incremented, because the old noncompliant DEFCONSTANT behavior involved
3324 calling functions which no longer exist, and because I also took the
3325 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3326 * fixed bug 1 (error handling before read-eval-print loop starts), and
3327 redid debugger restarts and related debugger commands somewhat while
3329 ** The QUIT debugger command is gone, since it did something
3330 rather different than the SB-EXT:QUIT command, and since it never
3331 worked properly outside the main toplevel read/eval/print loop.
3332 Invoking the new TOPLEVEL restart provides the same functionality.
3333 ** The GO debugger command is also gone, since you can just invoke
3334 the CONTINUE restart directly instead.
3335 ** The TOP debugger command is also gone, since it's redundant with the
3336 FRAME 0 command, and since it interfered with abbreviations for the
3338 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3340 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3341 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3342 some porting work by Martin Atzmueller)
3343 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3344 with a boatload of patches which he ported to SBCL. Now that those
3346 ** The system tries to make sure that its low-priority messages
3347 are prefixed by semicolons, to help people who like to use
3348 syntax highlighting in their ILISP buffer. (This patch
3349 was originally due to Raymond Toy.)
3350 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3351 patches originally written by Raymond Toy.
3352 ** The compiler understands coercion between single-value and
3353 multiple-VALUES type expressions better, getting rid of some very
3354 weird behavior, thanks to patches originally by Robert MacLachlan
3355 and Douglas Crosher.
3356 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3357 lambda lists, thanks to a patch originally by Pierre Mai.
3358 ** The system no longer bogusly warns about "abbreviated type
3360 ** The compiler gets less confused by inlining and RETURN-FROM,
3361 thanks to some patches originally by Tim Moore.
3362 ** The system no longer hangs when dumping circular lists to fasl
3363 files, thanks to a patch originally from Douglas Crosher.
3364 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3365 undefined function error.
3366 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3367 in structure slot initforms) for now, documented workaround instead:-|
3368 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3369 * fixed bug 5: added stubs for various Gray stream functions called
3370 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3371 installed, at least appropriate type errors are generated
3372 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3373 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3374 consistently in DEFMETHOD forms.
3375 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3376 it doesn't seem to affect SBCL after all
3377 * The C runtime system now builds with better optimization and many
3378 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3380 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3382 * The system is now under CVS at SourceForge (instead of the
3383 CVS repository on my home machine).
3384 * The new signal handling code has been tweaked to treat register
3385 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3386 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3387 downstream, e.g. in debug-int.lisp, has implicit dependencies
3388 on the unsignedness of integer representation of machine words,
3389 and that caused the system to bomb out with infinite regress
3390 when trying to recover from type errors involving signed values,
3391 e.g. (BUTLAST '(1 2 3) -1).)
3392 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3393 (This was one of the bugs Peter Van Eynde reported back in July.)
3394 * The system now uses code inspired by Colin Walters' O(N)
3395 implementation of MAP (from the cmucl-imp@cons.org mailing
3396 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3397 inline the MAP operation, and there is more than one
3398 sequence argument to the MAP call (so that it can't just
3399 do ETYPECASE once and for all based on the type of the
3400 single sequence argument). (The old non-inline implementation
3401 of the general M-argument sequence-of-length-N case required
3402 O(M*N*N) time when any of the sequence arguments were LISTs.)
3403 * The QUIT :UNIX-CODE keyword argument has been renamed to
3404 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3405 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3407 * Raymond Wiker's patches to port dynamic loading from Linux to
3408 FreeBSD have been added.
3409 * The BUGS file is now more nearly up to date, thanks in large part
3410 to Martin Atzmueller's review of it.
3411 * The debugger now flushes standard output streams before it begins
3412 its output ("debugger invoked" and so forth).
3413 * The core version number and fasl file version number have both
3414 been incremented, because of incompatible changes in the layout
3416 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3417 used to not be called for a saved Lisp image.)
3418 * Martin Atzmueller's version of a patch to fix a compiler crash,
3419 as posted on sbcl-devel 13 September 2000, has been installed.
3420 * Instead of installing Martin Atzmueller's patch for the
3421 compiler transform for SUBSEQ, I deleted the compiler transform,
3422 and transforms for some similar consing operations.
3423 * A bug in signal handling which kept TRACE from working on OpenBSD
3425 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3426 away by constant folding
3427 * The system now defines its address space constants in one place
3428 (in the Lisp sources), and propagates them automatically elsewhere
3429 (through GENESIS and the sbcl.h file). Therefore, patching the
3430 address map is less unnecessarily tedious and error-prone. The
3431 Lisp names of address space constants have also been systematized.
3432 * CVS tags like dollar-Header-dollar have been removed from
3433 the sources, because they have never saved me trouble and
3434 they've been source of trouble working with patches and other
3435 diff-related operations.
3436 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3437 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3439 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3441 * The system has been ported to OpenBSD.
3442 * The system now compiles with a simple "sh make.sh" on the systems
3443 that it's supported on. I.e., now you no longer need to tweak
3444 text in the target-features.lisp-expr and symlinks in src/runtime/
3445 by hand, the make.sh takes care of it for you.
3446 * The system is no longer so grossly inefficient when compiling code
3447 involving vectors implemented as general (not simple) vectors (VECTOR T),
3448 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3449 which dares to use the various sequence functions on non-simple
3450 vectors, takes less of a performance hit.
3451 * There is now a primitive type predicate VECTOR-T-P
3452 to test for the (VECTOR T) type, so that e.g.
3453 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3454 can now be compiled with some semblance of efficiency. (The old code
3455 turned the type declaration into a full call to %TYPEP at runtime!)
3456 * AREF on (VECTOR T) is still not fast, since it's still compiled
3457 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3458 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3459 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3460 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3461 without making full calls to SUBTYPEP at runtime.
3462 (Some analogous efficiency issues for non-simple vectors specialized to
3463 element types other than T, or for non-simple multidimensional arrays,
3464 have not been addressed. They could almost certainly be handled the
3465 same way if anyone is motivated to do so.)
3466 * The changes in array handling break binary compatibility, so
3467 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3468 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3469 instead of (VALUES T T).
3470 * By following the instructions that Dan Barlow posted to sbcl-devel
3471 on 2 July 2000, I was able to enable primitive dynamic object
3472 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3473 functionality is not implemented (since it calls ld to resolve
3474 library references automatically, requiring RUN-PROGRAM for its
3475 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3476 try to resolve library references) is now supported.
3477 * The system now flushes the standard output streams when it terminates,
3478 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3479 them at several other probably-convenient times, e.g. in each pass of
3480 the toplevel read-eval-print loop, and after evaluating a form given
3481 as an "--eval" command-line option. (These changes were motivated by a
3482 discussion of stream flushing issues on cmucl-imp in August 2000.)
3483 * The source transform for TYPEP of array types no longer assumes
3484 that an array whose element type is a not-yet-defined type
3485 is implemented as an array of T, but instead punts, so that the
3486 type will be interpreted at runtime.
3487 * There is now some support for cross-compiling in make.sh: each of
3488 the phases of make.sh has its own script. (This should be transparent
3489 to people doing ordinary, non-cross-compile builds.)
3490 * Since my laptop doesn't have hundreds of megabytes of memory like
3491 my desktop machine, I became more motivated to do some items on
3492 my to-do list in order to reduce the size of the system a little:
3493 ** Arrange for various needed-only-at-cold-init things to be
3494 uninterned after cold init. To support this, those things have
3495 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3496 symbols with such names are now uninterned after cold init).
3497 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3498 fasl files for cold load.
3499 ** Remove the old compiler structure pooling code (which used to
3500 be conditional on the target feature :SB-ALLOC) completely.
3501 ** Redo the representation of some data in cold init to be more compact.
3502 (I also looked into supporting byte compiled code at bootstrap time,
3503 which would probably reduce the size of the system a lot, but that
3504 looked too complicated, so I punted for now.)
3505 * The maximum signal nesting depth in the src/runtime/ support code has
3506 been reduced from 4096 to 256. (I don't know any reason for the very
3507 large old value. If the new smaller value turns out to break something,
3508 I'll probably just bump it back up.)
3509 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3511 * Many, many bugs reported by Peter Van Eynde have been added to
3512 the BUGS list; some have even been fixed.
3513 * While enabling dynamic object file loading, I tried to make the
3514 code easier to understand, renaming various functions and variables
3515 with less ambiguous names, and changing some function calling
3516 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3517 * While trying to figure out how to do the OpenBSD port, I tried to
3518 clean up some of the code in src/runtime/. In particular, I dropped
3519 support for non-POSIX signal handling, added various comments,
3520 tweaked the code to reduce the number of compilation warnings, and
3521 renamed some files to increase consistency.
3522 * To support the new automatic configuration functionality in make.sh,
3523 the source file target-features.lisp-expr has been replaced with the
3524 source file base-target-features.lisp-expr and the machine-generated
3525 file local-target-features.lisp-expr.
3526 * fixed a stupid quoting error in make.sh so that using CMU CL
3527 "lisp -batch" as cross-compilation host works again
3529 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3531 * DESCRIBE no longer tries to call itself recursively to describe
3532 bound/fbound values, so that it no longer fails on symbols which are
3533 bound to themselves (like keywords, T, and NIL).
3534 * DESCRIBE now works on generic functions.
3535 * The printer now prints less-screwed-up representations of closures
3536 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3537 * A private symbol is used instead of the :EMPTY keyword previously
3538 used to mark empty slots in hash tables. Thus
3539 (DEFVAR *HT* (MAKE-HASH-TABLE))
3540 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3541 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3542 now does what ANSI says that it should. (You can still get
3543 similar noncompliant behavior if bang on the hash table
3544 implementation with all the symbols you get back from
3545 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3546 This breaks binary compatibility, since tests for equality to
3547 :EMPTY are wired into things like the macroexpansion of
3548 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3550 * There's now a minimal placeholder implementation for CL:STEP,
3551 as required by ANSI.
3552 * An obscure bug in the interaction of the normal compiler, the byte
3553 compiler, inlining, and structure predicates has been patched
3554 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3555 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3557 * Missing ordinary arguments in a macro call are now detected even
3558 when the macro lambda list contains &KEY or &REST.
3559 * The debugger no longer complains about encountering the top of the
3560 stack when you type "FRAME 0" to explicitly instruct it to go to
3561 the top of the stack. And it now prints the frame you request even
3562 if it's the current frame (instead of saying "You are here.").
3563 * As specified by ANSI, the system now always prints keywords
3564 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3566 * The default initial SIZE of HASH-TABLEs is now smaller.
3567 * Type information from CLOS class dispatch is now propagated
3568 into DEFMETHOD bodies, so that e.g.
3569 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3571 is now basically equivalent to
3572 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3573 (DECLARE (TYPE SINGLE-FLOAT X))
3575 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3576 operation, without having to do run-time type dispatch.
3577 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3578 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3579 * Since I don't seem to be making big file reorganizations very often
3580 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3581 is overflowing my ability to conveniently back them up), I've finally
3582 checked the system into CVS. (The CVS repository is on my home system,
3583 not at SourceForge -- putting it on SourceForge might come later.)
3584 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3585 high-level GC-NOTIFY-FOO functions send their output. (There's
3586 still very little control of where low-level verbose GC functions
3587 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3588 less than it used to -- the GC-NOTIFY-FOO functions are now under
3589 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3590 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3591 in only one place in the source code, and propagates it automatically
3592 everywhere that it's needed. Thus e.g. when I bump the version from
3593 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3594 * The C source files now include boilerplate legalese and documentation
3595 at the head of each file (just as the Lisp source files already did).
3596 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3597 to his page will be replaced with a link to his new CLiki service.
3599 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3601 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3602 * The build process now looks for GNU make under the default name "gmake",
3603 instead of "make" as it used to. If GNU make is not available as "gmake"
3604 on your system, you can change this default behavior by setting the
3605 GNUMAKE environment variable.
3606 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3607 can build without error under CMU CL.
3609 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3611 * There is now a partial SBCL user manual (with some new text and some
3612 text cribbed from the CMU CL manual).
3613 * The beginnings of a profiler have been added (starting with the
3614 CMU CL profiler and simplifying and cleaning up). Eventually the
3615 main interface should be through the TRACE macro, but for now,
3616 it's still accessed through vaguely CMU-CL-style functions and macros
3617 exported from the package SB-PROFILE.
3618 * Some problems left over from porting CMU CL to the new
3619 cross-compilation bootstrap process have been cleaned up:
3620 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3621 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3622 part when getting rid of the compiler *BACKEND* stuff.)
3623 ** The value of *NULL-TYPE* was screwed up, because it was
3624 being initialized before the type system knew the final
3625 definition of the 'NULL type. This screwed up several key
3626 optimizations in the compiler, causing inefficiency in all sorts
3627 of places. (I found it because I wanted to understand why
3628 GET-INTERNAL-RUN-TIME was consing.)
3629 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3630 PROCLAIM FTYPE information. Unfortunately this broke binary
3631 compatibility again, since now the forms output by DEFGENERIC
3632 to refer to functions which didn't exist in 0.6.3.
3633 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3634 can use the new (as of 0.6.3) transform for SOME into MAP into
3636 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3637 values to (INTEGER 0 1000000), so that the time code will no longer
3638 occasionally get blown up by Linux returning 1000000 microseconds
3639 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3640 its output conform to the ANSI spec. (Alas, this makes its output
3641 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3642 * A full call to MAP NIL with a single sequence argument no longer conses.
3643 * fixes to problems pointed out by Martin Atzmueller:
3644 * The manual page no longer talks about multiprocessing as though
3645 it were currently supported.
3646 * The ILISP support patches have been removed from the distribution,
3647 because as of version 5.10.1, ILISP now supports SBCL without us
3648 having to maintain patches.
3649 * added a modified version of Raymond Toy's recent CMU CL patch for
3650 EQUALP comparison of HASH-TABLE
3652 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3654 * The system still can't cross-compile itself with
3655 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3656 entails), but at least it can compile more of itself that way
3657 than it used to be able to, and various buglets which were uncovered
3658 by trying to cross-compile itself that way have now been fixed.
3659 * This release breaks binary compatibility again. This time
3660 at least I've incremented the FASL file format version to 2, so that the
3661 problem can be detected reliably instead of just causing weird errors.
3662 * various new style warnings:
3663 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3664 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3665 ** using the lexical binding of a variable named in the *FOO* style
3666 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3667 as specified by ANSI.
3668 * *RANDOM-STATE* is no longer automatically initialized from
3669 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3670 default behavior of the system is to repeat its behavior every time
3671 it's run. If you'd like to change this behavior, you can always
3672 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3673 old convention there was no comparably easy way to get the system to
3674 repeat its behavior every time it was run.
3675 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3676 FTYPE declarations has been removed, in favor of their ANSI
3677 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3678 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3679 the types of their sequence arguments can be determined at compile time.
3680 This is done through a new open code expansion for MAP which eliminates
3681 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3682 argument types can be determined at compile time.
3683 * The optimizer now transforms COERCE into an identity operation when it
3684 can prove that the coerced object is already of the correct type. (This
3685 can be a win for machine generated code, including the output of other
3686 optimization transforms, such as the MAP transform above.)
3687 * Credit information has been moved from source file headers into CREDITS.
3688 * Source file headers have been made more standard.
3689 * The CASE macro now compiles without complaining even when it has
3692 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3694 * (Note that the way that the PCL macroexpansions were rewritten
3695 to accommodate the change in DEFGENERIC below breaks binary
3696 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3697 not run under sbcl-0.6.2. Once we get out of alpha releases,
3698 i.e. hit release 1.0.0, we'll probably try to maintain binary
3699 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3700 and sbcl-1.4.4. Until then, however, it might be fairly common
3701 for maintenance releases to break binary compatibility.)
3702 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3704 * The reporting of unhandled signals has been changed to print some
3705 explanatory text as well as the report form. (Previously only
3706 the report form was printed.)
3707 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3708 it defines, so that the compiler no longer issues undefined function
3709 warnings for compiled-but-not-yet-loaded generic functions.
3710 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3711 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3712 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3713 section of the ANSI spec are supported.
3714 * The "" nickname for the "KEYWORD" package has been removed.
3715 The reader still handles symbol tokens which begin with a package marker
3716 as keywords, but it doesn't expose its mechanism for doing so in the
3717 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3718 * The system now issues STYLE-WARNINGs for contradictory TYPE
3719 proclamations. (Warnings for contradictory FTYPE proclamations would
3720 be nice too, but those can't be done usefully unless the type system
3721 is made smarter about FUNCTION types.)
3722 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3723 systematized, so that "*target-*.lisp is supposed to exist only on the
3724 target and imply that there's a related file which exists on the
3725 host, and *host-*.lisp is supposed to exist only on the host and imply
3726 that there's a related file which exists on the target. This involves a
3727 lot of renaming. Hopefully the acute confusion caused by the renaming
3728 will be justified by the reduction in chronic confusion..
3729 ** runtime-type.lisp -> early-target-type.lisp
3730 ** target-type.lisp -> late-target-type.lisp
3731 ** early-host-format.lisp -> early-format.lisp
3732 ** late-host-format.lisp -> late-format.lisp
3733 ** host-error.lisp -> misc-error.lisp
3734 ** early-error.lisp -> early-target-error.lisp
3735 ** late-error.lisp -> late-target-error.lisp
3736 ** host-defboot.lisp -> early-defboot.lisp
3737 ** code/misc.lisp -> code/target-misc.lisp
3738 ** code/host-misc.lisp -> code/misc.lisp
3739 ** code/numbers.lisp -> code/target-numbers.lisp
3740 ** code/early-numbers.lisp -> numbers.lisp
3741 ** early-host-type.lisp -> early-type.lisp
3742 ** late-host-type.lisp -> late-type.lisp
3743 ** host-typep.lisp -> typep.lisp
3744 ** load.lisp -> target-load.lisp
3745 ** host-load.lisp -> load.lisp
3746 ** host-disassem.lisp -> disassem.lisp
3747 ** host-insts.lisp -> insts.lisp
3748 ** byte-comp.lisp -> target-byte-comp.lisp
3749 ** host-byte-comp.lisp -> byte-comp.lisp
3750 ** host-signal.lisp -> signal.lisp
3751 ** host-defstruct.lisp -> defstruct.lisp
3752 ** late-target-type.lisp -> deftypes-for-target.lisp
3753 Furthermore, several other previously target-only files foo.lisp (e.g.
3754 hash-table.lisp and random.lisp) have been split into a target-and-host
3755 foo.lisp file and a target-only target-foo.lisp file, with their key type
3756 definitions in the target-and-host part, so that the cross-compiler will
3757 know more about target types.
3758 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3759 gone away. In their place are various *BACKEND-FOO* variables
3760 corresponding to the slots of the old structure.
3761 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3762 into the target SBCL has been fixed.
3763 * The chill.lisp system for loading cold code into a running SBCL
3765 * Support for the CMU CL "scavenger hook" extension has been removed.
3766 (It was undocumented and unused in the CMU CL sources that SBCL was
3767 derived from, and stale in sbcl-0.6.1.)
3768 * Various errors in the cross-compiler type system were detected
3769 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3770 (enabling various consistency checks). Many of them were fixed,
3771 but some hard problems remain, so the compiler is back to
3772 running without *TYPE-SYSTEM-INITIALIZED* for now.
3773 * As part of the cross-compiler type system cleanup, I implemented
3774 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3775 * I have started adding UNCROSS calls throughout the type system
3776 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3777 on cross-compiler input files will be able to go away, and various
3779 * CONSTANTP now returns true for quoted forms (as explicitly required
3782 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3784 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3785 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3786 (MAKE-LIST -1). In the longer term, it ought to become true
3787 that declarations are assertions even at SAFETY 1. For now, it's not
3788 quite true even at SAFETY 3, but it's at least more nearly true..
3789 (Note that this change seems to increases the size of the system by
3790 O(5%) and to decrease the speed of the compiler by 20% or more.)
3791 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3792 for the problem of printing dozens of lines of distracting information
3793 about low-level system machinery as part of the top stack frame
3794 on entry to the debugger when an undefined function was called.
3795 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3796 is not reset to COMMON-LISP-USER.
3797 * Compilation of stuff related to dyncount.lisp has been made conditional
3798 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3799 smaller. The various dyncount-related symbols have been moved into
3800 a new "SB-DYNCOUNT" package.
3801 * tty-inspect.lisp has been renamed to inspect.lisp.
3802 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3803 feature has gone away. (When we eventually port to other flavors of
3804 libc and/or Unix, we'll try to make the differences between flavors
3805 invisible at the user level.)
3806 * Various other *FEATURES* tags, and/or their associated conditionals,
3807 have been removed if obsolescent, or given better documentation, or
3808 sometimes given more-mnemonic names.
3810 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3812 * tidied up "make.sh" script
3813 * tidied up system directory structure
3814 * better "clean.sh" behavior
3815 * added doc/FOR-CMUCL-DEVELOPERS
3816 * many many small tweaks to output format, e.g. removing possibly-confusing
3817 trailing #\. character in DESCRIBE-INSTANCE
3818 * (EQUALP #\A 'A) no longer signals an error.
3819 * new hashing code, including EQUALP hashing
3820 * tidied up Lisp initialization and toplevel
3821 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3822 * command line argument processing
3823 * added POSIX-GETENV function to deal with Unix-ish environment variables
3824 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3825 terminating SBCL on EOF
3826 * non-verbose GC by default
3827 * There is no more "sbcl" shell script; the sbcl file is now the C
3828 runtime executable (just like CMU CL).
3829 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3831 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3832 symbol and package information as Lisp native symbol and package objects
3833 instead of strings naming symbols and strings naming packages. This way,
3834 whenever packages are renamed (as in warm init), debug information is
3835 transformed along with everything else.
3836 * tweaked the optimization policy declarations which control the building
3837 of SBCL itself. Now, among other things, the system no longer saves
3838 source location debugging information. (This helps two problems at once
3839 by reducing SBCL size and by keeping SBCL from trying to look for its
3840 sources -- which may not exist -- when reporting errors.)
3841 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3842 debugging and testing purposes
3843 * cleaned up printing, making the printer call PRINT-OBJECT for
3844 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3845 methods, giving nearly-ANSI behavior
3846 * converted almost all special variables to use *FOO* naming convention
3847 * deleted PARSE-TIME functionality, since it can be done portably
3848 * moved some files out of cold init into warm init
3849 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3851 * regularized formatting of source files
3852 * added an install.sh script
3853 * fixed ridiculous memory usage of cross-compiler by making
3854 compiler/alloc.lisp not try to do pooling unless it can hook
3855 itself into the GC of the cross-compilation host. Now the system
3856 builds nicely on my old laptop.
3857 * added :SB-ALLOC in target-features.lisp-expr
3858 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3859 was not implemented)
3860 * re-did condition handling and note reporting in the compiler. Notes
3861 are no longer handled by signalling conditions. Style warnings
3862 and warnings are handled more correctly and reported in such a way
3863 that it's easy to find one or the other in your output (so that you
3864 can e.g. figure out which of many problems caused COMPILE-FILE to
3866 * changed the severity of several compiler warnings from full WARNING
3867 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3868 compiler note reporting so that it doesn't use the condition system
3869 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3870 COMPILE-FILE command)
3871 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3872 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3873 no longer supported, since their implementation was deeply intertwingled
3874 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3875 * removed bogus "support" for compiler macros named (SETF FOO), and
3876 removed the compiler macro for SETF INFO (but only after making a fool
3877 of myself on the cmucl-imp mailing list by posting a bogus patch for
3878 DEFINE-COMPILER-MACRO..)
3879 * Compiled files containing forms which have side effects on the Lisp
3880 reader (such as DEFPACKAGE forms) are now handled more correctly.
3881 (Compiler queuing of top level lambdas has been suppressed by setting
3882 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3883 * deleted various currently-unused source files, e.g. gengc.lisp. They
3884 may be added back at some point e.g. when porting to other architectures,
3885 but until they are it's distracting to distribute them and to try to
3887 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3888 there were so many of them they're just distractions, and UNCROSS is
3889 known to be able to handle the current sources
3890 * moved PROFILE functionality into TRACE, so that it will be clear
3891 how the wrapping and unwrapping of functions when you profile them
3892 interacts with the wrapping and unwrapping of functions when you
3893 trace them. (Actually, the functionality isn't there yet, but at least
3894 the interface specification is there. Hopefully, the functionality will
3895 arrive with some future maintenance release.)
3896 * removed host-oops.lisp
3897 * changed signature of QUIT function to allow UNIX-CODE argument
3898 * fixed READ-SEQUENCE bug
3899 * tweaked verbose GC output so that it looks more like the progress
3900 output that ANSI specifies for functions like LOAD
3901 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3902 * added <http://sbcl.sourceforge.com> to the banner information printed by