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
48 * bug fix: compiler failed to differentiate between different CONS
51 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
52 * added support for the ucs-2 external format. (contributed by Ivan
54 * minor incompatible change: pretty printing of objects of type
55 (cons symbol) is, in the default pprint-dispatch-table, now
56 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
58 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
59 called later in a class's lifetime, possibly as late as when the
60 first instance of the class is created. Previously,
61 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
62 class became finalizeable.
63 * fixed bug: FILE-POSITION sometimes returned inconsistent results
64 for multibyte external-format streams. (thanks to "vbzoli")
65 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
66 with :ALLOCATION :CLASS inherited from superclasses of the
68 * fixed bug: anonymous classes can now be created using the :NAME
69 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
70 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
71 * fixed bug: core-files saved with :EXECUTABLE T can again be
72 executed when SBCL_HOME isn't set. (reported by James Knight)
73 * fixed bug: toplevel LOCALLY forms with declarations could
74 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
75 * fixed bug: printing from several different threads using different
76 values of *print-case* could cause invalid output, due to
77 some internal special variables of the printer not being bound
78 thread-locally (reported by Max Mikhanosha)
79 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
80 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
81 (reported by Pascal Costanza)
82 * fixed bug: FINALIZE-INHERITANCE is called from
83 REINITIALIZE-INSTANCE on classes when the class has previously
84 been finalized, as required by AMOP.
85 * minor code generation optimizations:
86 ** better register allocation in CLOS dispatching functions
87 ** overflow detection when coercing signed bytes to fixnums on x86-64
88 ** is now implemented with one IMUL instruction instead of three shifts
89 ** more efficient bit-vector access on x86 and x86-64
90 ** more efficient access to raw structure slots on x86 and x86-64
91 * fixed some bugs revealed by Paul Dietz' test suite:
92 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
95 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
96 * feature: thread support on Solaris/x86, and experimental thread support
98 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
100 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
101 default method for SLOT-UNBOUND.
102 * minor incompatible change: prevent the user from specializing the
103 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
104 somewhat counter to the intent of the protocol, I (CSR) think, and
105 additionally it just doesn't work in SBCL as currently
106 implemented, thanks to optimizations (that are always valid for
107 the other three SLOT-VALUEish functions, but not for the setter).
108 * minor incompatible change: the :SB-LDB feature is now enabled by
109 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
110 the low-level debugger.
111 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
112 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
113 down to an SB-SYS:FD-STREAM, allowing direct communication in
114 more cases. Temporary files and pipes are still used as a fallback
116 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
117 on the non-reentrant C functions localtime(3) and gmtime(3).
118 * bug fix: global optimization policy was not visible in LOCALLY and
120 * bug fix: class objects can be used as specializers in methods.
121 (reported by Pascal Costanza)
122 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
123 with a type error. (reported by blitz_ on #lisp)
124 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
125 failed with a type error. (reported by Pascal Bourguignon)
126 * bug fix: merging pathnames against defaults with :DIRECTORY
127 starting with '(:RELATIVE :BACK) should preserve the :BACK.
128 (reported by James Y Knight)
129 * bug fix: saving large (>2GB) cores on x86-64 now works
130 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
131 constant index and a (simple-array (signed-byte 32)) array
132 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
133 error (patch by Robert J. Macomber)
134 * feature: TIME now displays the amount of run-time spent in GC
135 * bug fix: The debugger now does a better job of respecting
136 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
137 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
138 workaround for bug 403.)
139 * bug fix: floating point exception handling now works on FreeBSD
140 (thanks to NIIMI Satoshi)
141 * fixed some bugs revealed by Paul Dietz' test suite:
142 ** MISC.641: LET-conversion were not supposed to work in late
144 * improvements to the Win32/x86 port:
145 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
146 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
148 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
149 * new feature: source path information is generated for macro-expansion
150 errors for use in IDE's like Slime (thanks to Helmut Eller)
151 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
152 * bug fix: compiler does not loop forever on an invalid type in
154 * improvement: compilation of most CLOS applications is significantly
156 * optimization: added a limited bytecode compiler for simple toplevel
157 forms, speeding up compilation and FASL loading
159 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
160 * minor incompatible change: in sbcl-0.9.11 (but not earlier
161 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
162 into subclasses of the finalized class. Now user calls to
163 FINALIZE-INHERITANCE finalize just the one class, and calls by the
164 system return before any subclasses are finalized.
165 * minor incompatible change: The reader no longer ignores errors
166 regarding non-existent packages in #+ and #- feature tests.
167 * new feature: command line options --no-sysinit, --no-userinit to
168 inhibit loading the corresponding init files
169 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
170 for sending data through UDP sockets (thanks to François-René Rideau)
171 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
172 error is signalled instead (thanks to François-René Rideau)
173 * minor incompatible change: Error signalling behaviour of lexical
174 operations violating package locks has changed slightly. Refer to
175 documentation on package locks for details.
176 * enhancement: EVAL can process IF-expressions without resorting to the
178 * bug fix: Some operations involving SIN, COS, or TAN weren't being
179 constant-folded properly due to missing out-of-line functions.
180 (reported by Mika Pihlajamäki)
181 * bug fix: LISTEN sometimes returned T even in cases where no data was
182 immediately available from the stream
183 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
184 were reversed. (reported by Levente Mészáros)
185 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
186 when given a vector displaced to an adjustable array. (reported
188 * bug fix: derivation of float boundaries from numbers outside the
189 appropriate float range (reported by John Wiseman)
190 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
192 * bug fix: precomputing cache entries for generic functions with
193 some subclasses of specializers as yet invalid does not attempt to
194 fill a cache line with a negative offset. (reported by Levente
196 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
197 allow condition class objects as arguments to DOCUMENTATION and
198 (SETF DOCUMENTATION); only find and set documentation for
199 structure names for the STRUCTURE doc-type. (suggested by Gary
201 * improvements to the Win32/x86 port:
202 ** (user-homedir-pathname) and default initialization file
203 locations now know about the user's "Documents and Settings"
204 directory (thanks to Yaroslav Kavenchuk)
205 ** run-program is implemented (thanks to Mike Thomas)
206 ** sockets support (thanks to Timothy Ritchey)
207 ** better backtrace support (thanks to Alastair Bridgewater)
208 ** sb-grovel supported
209 ** asdf-install and sb-posix work somewhat
210 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
211 * improvements to the Solaris/x86 port:
212 ** works on Solaris 11/Solaris Express
213 ** floating-point exception handling support
214 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
215 * fixed some bugs revealed by Paul Dietz' test suite:
216 ** REMOVE-METHOD returns its generic function argument even when
217 the method is not one of the generic functions' methods.
218 ** objects declared as MEMBER types can be admissible arguments to
220 ** printing characters should simply be printed by the FORMAT ~:C
222 ** compiler failure when compiling functions with hairy constant
223 defaults for optional parameters.
224 ** compiler produces wrong code when MAYBE-INLINE-expanding a
225 function, which is already optimized.
227 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
228 * new platform: experimental support for SBCL x86/Darwin, including
229 MacOS X 10.4.5 on Intel.
230 * new feature: Unicode character names are now known to the system
231 (through CHAR-NAME and NAME-CHAR).
232 * new feature: the filesystem location of SBCL's core file is
233 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
234 * minor incompatible change: the contrib modules SB-POSIX and
235 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
236 this change is to make it easier to distribute
237 location-independent binaries.
238 * bug fix: as implied by AMOP, standardized classes no longer have
239 slots named by external symbols of public packages. (reported by
241 * optimization: calling functions via constant symbols -- as in
242 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
243 via the function object as in (FUNCALL #'FOO).
244 * optimization: on x86 and x86-64, the calling convention for the
245 case of dealing with an unknown number of values has been altered
246 to be friendlier to the prediction heuristics implemented,
247 particularly with reference to CALL and RET pairing. (thanks to
248 Alastair Bridgewater)
249 * optimization: on x86, the code for access to thread-local symbol
250 values and binding/unbinding of thread-local symbols is smaller.
251 (thanks to Alastair Bridgewater)
252 * enhancement: CONSTANTP is now able to determine constantness of
253 more complex forms, including calls to constant-foldable standardized
254 functions and some special forms beyond QUOTE.
255 * fixed bug: occasional GC crashes on Solaris/x86
256 * optimization: x86-64 supports stack allocation of results of simple
257 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
258 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
259 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
260 (thanks to James Knight)
261 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
262 by the standard (thanks to Stephen Compall)
264 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
265 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
266 be used for bundling the runtime and the core file into one
267 executable binary. This feature is not currently supported on all SBCL
268 platforms. (thanks to James Bielman and NIIMI Satoshi)
269 * new feature: a generational or ephemeral garbage collector is now
270 the default on the PowerPC platform (both Linux and Darwin). The
271 old Cheney (stop and copy) collector is a build-time option.
272 (thanks to Cyrus Harmon, after Raymond Toy)
273 * minor incompatible change: the method by which SBCL finds its
274 contributed modules has changed; it no longer relies on symbolic
275 links from an $SBCL_HOME/systems directory, but searches directly
276 in the subdirectories of $SBCL_HOME.
277 * enhancement: the dynamic heap size on the Linux/PPC platform is
278 markedly larger, even using the older Cheney garbage collector.
279 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
280 certain complicated string types. (reported by Gary King)
281 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
282 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
283 * fixed bug: DOCUMENTATION on structures no longer signals an
284 error if no structure type of the right name exists. (reported by
286 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
287 applicable non-standard primary or :AROUND method on
288 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
289 no longer cause unbound variable errors. (reported by Kevin Reid)
290 * fixed bug: metacircle resolution in cases where methods have slots
291 added before the slots from STANDARD-METHOD. (reported by Jean
293 * fixed bug: the Power PC assembly code for calling into lisp
294 sometimes computed its return address wrongly (depending on the
295 15th bit of the address, fixed at link time). This used to
296 manifest itself in a segmentation violation while building PCL.
297 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
298 many others over the years)
299 * fixed bug: ROOM no longer reports silly numbers for stack usage if
300 the stack pointer (treated as a signed integer) is negative.
301 (thanks to Peter van Eynde)
303 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
304 * new platform: experimental support for the Windows operating
305 system has been added. (thanks to Alastair Bridgewater)
306 * fixed several bugs in and robustified the PPC FFI (including
307 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
308 * bug fix: allow non-simple string symbol names (reported by Paul
310 * bug fix: interrupt handling on NetBSD (thanks to Richard M
312 * bug fix: saving a core corrupted callbacks on x86/x86-64
313 * bug fix: closed a loophole in metacircularity detection and
314 grounding in the PCL implementation of CLOS.
315 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
316 (reported by tomppa on #lisp)
317 * bug fix: clear the direction flag on Lisp -> C transitions, as
318 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
319 (reported by Andrej Grozin and Hendrik Maryns)
320 * optimization: major improvements to GC efficiency on GENCGC platforms
321 * optimization: faster implementation of EQUAL
322 * optimization: emit more efficient opcodes for some common
323 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
324 * optimization: several other minor code-generation improvements on x86-64
325 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
327 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
328 * minor incompatible change: (SETF CLASS-NAME) and (SETF
329 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
330 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
331 by ANSI as a generic function, and for consistency (SETF
332 GENERIC-FUNCTION-NAME) is treated likewise.
333 * fixed bug #233.b: make constraint propagation notice when a variable
334 value is changed after it is referenced but before it is used
335 * fixed bug #296: no more arbitrary behaviour from filenames with
336 odd characters as --load arguments. (reported by Adam Warner)
337 * fixed bug #390: :CHARACTER-SET pathname components now work as
338 expected. (reported by Tim Daly Jr)
339 * fixed bug #391: complicated :TYPE intersections in slot
340 definitions no longer cause an error in PCL internals.
341 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
342 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
343 returning the number of octets which would be written to the
344 file-stream. (thanks to Robert J. Macomber)
345 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
346 arguments to the internal stream functions. (thanks to David
348 * bug fix: stack exhaustion is now handled on the x86/SunOS
349 platform. (thanks to Daisuke Homma)
350 * bug fix: more accurate ROOM results on GENCGC platforms
351 * optimization: improved type inference for arithmetic-for
352 index variables in LOOP
353 * optimization: faster floating-point SQRT on x86-64
354 * fixed some bugs revealed by Paul Dietz' test suite:
355 ** DOCUMENTATION returns NIL instead of "" for method combinations
356 that don't have a docstring
358 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
359 * minor incompatible change: (SETF CLASS-NAME) and (SETF
360 GENERIC-FUNCTION-NAME) are no longer generic functions, and
361 therefore are not customizeable by user code (as seems to be at
362 least permitted and maybe required by AMOP). As a consolation,
363 however, the SBCL implementation of these functions now calls
364 REINITIALIZE-INSTANCE as specified by AMOP.
365 * bug fix: slot-definition documentation is propagated as per ANSI
366 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
367 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
368 argument, as do eslotd initialization methods. (from Pascal
369 Costanza's "Closer" project)
370 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
371 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
373 * bug fix: it is now possible to have more than one subclass of
374 STANDARD-GENERIC-FUNCTION without causing stack overflow.
375 (reported by Bruno Haible, Pascal Costanza and others)
376 * bug fix: the dependent update protocol now works for generic
377 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
379 * bug fix: condition-class instances corresponding to
380 DEFINE-CONDITION forms are now created eagerly. (reported by
381 Kalle Olavi Niemitalo on comp.lang.lisp)
382 * bug fix: floating point printing is more accurate in some
383 circumstances. (thanks to Simon Alexander)
384 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
385 merged with *DEFAULT-PATHNAME-DEFAULTS*.
386 * bug fix: callbacks on OS X now preserve stack-alignment.
387 (thanks to Cyrus Harmon)
388 * enhancement: the x86-64 disassembler is much better at
389 disassembling SSE instructions. (thanks to Lutz Euler)
390 * enhancement: CHECK-TYPE on a variable now allows type-inference
391 (thanks to Vincent Arkesteijn)
392 * optimization: improved performance of EUC-JP external format.
393 (thanks to NIIMI Satoshi)
394 * optimization: performance improvements to IO on file streams of
395 :ELEMENT-TYPE CHARACTER
396 * optimization: much faster memory allocation on x86-64
397 * optimization: faster garbage collections (latency and throughput)
399 * optimization: faster bignum subtraction and division on x86 and x86-64
400 (thanks to Lutz Euler)
401 * optimization: more accurate type-inference for loops that use a
402 floating point index variable or a negative step.
404 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
405 * bug fix: add a workaround to SBCL looping infinitely at startup on
406 Linux kernels with apparently buggy implementations of personality().
407 (thanks to Svein Ove Aas)
408 * bug fix: Unicode symbols are correctly printed in LDB backtraces
409 (thanks to David Lichteblau)
410 * bug fix: local bindings shadow global symbol macros for
411 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
412 * optimization: non-open coded uses of numeric comparison operators
413 (e.g. >) no longer cons when called with more than one parameter
414 on platforms supporting dynamic-extent allocation.
415 * enhancement: saving cores with foreign code loaded is now supported
416 on MIPS/Linux in addition to the previously supported platforms.
417 * bug fix: FIND-RESTART now tests for activity, not applicability when given
418 a restart object as identifier. (reported by Helmut Eller for CMUCL)
419 * bug fix: division by zero in sb-sprof when no samples were collected
420 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
421 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
422 (thanks to Alexander Kjeldaas)
423 * new feature: ignoring the return values of destructive functions
424 like NREVERSE signals a compile-time style-warning.
425 (thanks to Kevin Reid)
426 * new feature: the alignment of alien structure fields can be
427 explicitly specified. (thanks to Cyrus Harmon)
428 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
429 and dump core on SIGQUIT
431 ** incompatible change: threads do not inherit values of specials
432 from their parents (see manual)
433 ** bug fix: threads stacks belonging to dead threads are freed by the
434 next exiting thread, no need to gc to collect thread stacks anymore
435 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
436 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
437 ** bug fix: dynamic variable and thread start related gc lossage
438 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
440 * fixed some bugs revealed by Paul Dietz' test suite:
441 ** SUBTYPEP is slightly more accurate on heinously complicated
442 CONS types where some of the members have uncertain (in the
443 NIL, NIL sense) type relationships to each other.
445 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
446 no longer purified unless :PURIFY T is explicitly specified.
447 ** Non-purified cores are significantly smaller than before
449 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
450 * new feature: timers based on Zach Beane's excellent timer package
451 * added support for the following external formats: koi8-u,
452 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
453 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
454 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
455 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
456 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
457 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
458 * incompatible change: a threaded SBCL will no longer revert to
459 non-threaded mode on non-NPTL systems, but refuse to start entirely.
460 * bug fix: interrupts are disabled until startup is complete; no
461 more sigsegvs when receiving a signal to soon
462 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
464 * bug fix: add a workaround for the memory randomization features in
465 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
466 workaround will only be in effect on systems with the proc filesystem
468 * bug fix: printing objects of type HASH-TABLE signals a
469 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
471 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
472 have been running for over 50 days. (reported by Gilbert Baumann)
473 * bug fix: the logic for getting names of functions gets less
474 confused when confronded with alternate-metaclass
475 funcallable-instances. (reported by Cyrus Harmon)
476 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
477 consistent, even on internal alternate-metaclass objects.
478 * bug fix: flush closure information collected by physical
479 environment analysis prepass before the main pass. (bug reported
481 * bug fix: compiler pack phase does not modify a hashtable, which is
482 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
483 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
484 may be expanded. (reported by Alexander <malishev> on c.l.l)
485 * bug fix: no more style-warnings when compiling code using LAST
486 under high-SPEED low-SPACE settings. (reported by David Wallin
488 * fixed some incompatibilities between SBCL's MOP and the MOP
490 ** the METAOBJECT class is now implemented;
491 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
492 STANDARD-OBJECT, as required;
493 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
494 now compatible (as required by VALIDATE-SUPERCLASS); there
495 remains a consistency requirement over the presence or absence
496 of the FUNCTION class in the superclass list of finalized
497 classes; see the manual for more details;
498 ** the :AROUND method for COMPUTE-SLOTS on
499 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
500 requested slot ordering.
502 ** bug fix: parent thread now can be gc'ed even with a live
504 ** bug fix: binding a special with PROGV to no value is not
505 the same as that symbol not having been bound (thanks to
507 ** bug fix: binding specials is thread safe (thanks to
509 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
510 and SB-SPROF do not die with 'no handler for signal XX in
511 interrupt_handle_now(..)' anymore
512 ** bug fix: WITH-TIMEOUT works with multiple running threads
513 * fixed some bugs revealed by Paul Dietz' test suite:
514 ** ENSURE-GENERIC-FUNCTION should take a method class object for
515 the :method-class keyword argument.
517 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
518 * new port: the Solaris operating system on x86 processors is now
519 mostly supported, though some rough edges in the environment
520 remain. (thanks to Daisuke Homma)
521 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
522 space for its heap. (thanks to Thiemo Seufer)
523 * optimization: SBCL on MIPS platforms now supports dynamic-extent
524 closures. (thanks to Thiemo Seufer)
525 * minor incompatible change: eof selects abort in the debugger.
526 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
527 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
529 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
530 deprecated, and will go away in a future revision of SBCL.
531 * minor incompatible change: GC-ON and GC-OFF are no longer
532 implemented with a counter, it does not matter how many times gc
533 is switched on or off
534 * bug fix: discriminating functions for generic function classes
535 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
536 make invalid assumptions about method precedence order. (reported
538 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
539 platforms. (thanks to Thiemo Seufer)
540 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
541 * bug fix: correct pathname printing with printer escaping is on.
542 (thanks to Kevin Reid)
543 * bug fix: complex VOP definitions in "user-space" no longer trigger
544 package locks. (reported by Zach Beane)
545 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
546 for funcallable-instance functions, and is no different from
548 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
549 characters in the current readtable. (reported by Nicholas Neuss)
550 * bug fix: SUBTYPEP on various CONS types returns more a more
551 accurate acknowledgment of its certainty.
552 * optimizations: REMOVE-DUPLICATES now runs in linear time on
553 lists in some cases. This partially fixes bug 384.
554 * flush all standard streams before prompting in the REPL and the
556 * bug fix: signal handling and triggering gc do not conflict
557 directly anymore, in particular a high frequency sb-sprof does
558 not prevent gc from running
559 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
560 approximation for timezone and DST information between the
561 universal time epoch and the smallest negative 32-bit time_t.
562 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
563 year 1899 so long as the encoded time is non-negative.
564 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
565 the correct answer for non-constant rotate amounts when performing
566 an inline 32-bit rotation.
568 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
569 there is only one thread in the session
570 ** bug fix: memory leak for streams created in one thread and
571 written to in another
572 ** bug fix: lockup when compiled with gcc4
573 ** bug fix: race that allows the gc to be triggered when gc is
575 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
576 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
578 ** bug fix: finished threads are gc'ed properly
579 * fixed some bugs revealed by Paul Dietz' test suite:
580 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
581 has arguments with a different set of applicable methods from
582 the orignal arguments.
583 ** The type error thrown by MAP now has a correct expected-type
585 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
586 name a compiled function.
587 ** Three MISC tests where a large bignum was improperly coerced to
588 a float that couldn't represent that bignum during type
589 derivation were fixed.
590 ** SUBTYPEP can now handle types involving the negation of a
591 list-form FUNCTION type.
592 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
593 as (COMPLEX (AND RATIO (NOT FIXNUM))).
594 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
596 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
597 * New feature: Experimental support for bivalent streams: streams
598 opened with :element-type :default now allow character and binary
599 (unsigned-byte 8) I/O
600 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
601 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
603 * bug fix: no more highly sporadic "couldn't check whether ~S is
604 readable" when reading a stream and an interrupt hits in the middle
605 of a select system call
606 * compiler better recognizes complex arrays (reported by Eduardo
608 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
609 for variables on SBCL built with linkage-tables. (reported by Luis
611 * various error reporting improvements.
612 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
613 (Thanks to Thiemo Seufer)
614 * bug fix: nested reader invokations work correctly
615 * bug fix: it is possible to have simultaneous references to foreign
616 code and foreign data with the same name.
618 ** added x86-64 support
619 ** incompatible change: the threading api now works with thread
620 objects instead of thread ids
621 ** bug fix: threads are protected from signals and interruption when
622 starting up or going down
623 ** bug fix: a race where an exiting thread could lose its stack to gc
624 ** bug fix: don't halt on infinite error in threads if possible
625 ** fixed numerous gc deadlocks introduced in the pthread merge
626 ** bug fix: fixed thread safety issues in read and print
627 ** bug fix: debugger doesn't hang on session lock if interrupted at
628 an inappropriate moment
629 ** bug fix: run-program is now thread safe(r)
630 ** bug fix: inner with-recursive-lock no longer releases the mutex
631 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
632 * fixed some bugs revealed by Paul Dietz' test suite:
633 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
634 EXPECTED-TYPE slots filled.
635 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
637 ** MAKE-CONDITION accepts classes as type-designators.
638 ** COMPILE may never return NIL.
639 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
640 range before calling Unix time functions
642 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
643 * numerous signal handling fixes to increase stability
644 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
645 * minor incompatible change: we now correctly canonize default
646 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
647 as per the MOP, rather than the historical (INITARG INITFUNCTION
648 INITFORM). (reported by Bruno Haible)
649 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
650 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
651 for more information.
652 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
653 pathname is a directory pathname.
654 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
655 * dynamic space size on PPC has been increased to 768Mb. (thanks to
657 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
658 :METACLASS argument in addition to a class name. (reported by
659 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
660 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
661 (reported by Bruno Haible; patch from Gerd Moellmann)
662 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
664 * bug fix: more cleanups to the floating point exception handling on
665 x86-64 (thanks to James Knight)
666 * bug fix: the compiler does not try to constant fold calls of
667 COERCE and BIT-* functions when they return freshly constructed
668 objects. (reported by Nikodemus Siivola and Paul Dietz)
669 * optimization: improved the MIPS versions of generic (in the
670 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
671 * optimization: direct conversion of (unsigned-byte 32) to floats on
672 the PowerPC platform.
673 * optimization: structure instances with raw slots now use less
674 memory, and probably show better memory locality. (thanks to
676 * optimization: DYNAMIC-EXTENT declarations for lists and closures
677 are treated as requests for stack allocation on the x86-64,
678 Alpha-32, PPC and SPARC platforms.
679 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
680 the wrong file descriptor; implementation of SOCKET-OPEN-P.
681 (thanks to Tony Martinez)
683 ** gcing a dead thread can no longer lead to lockups
684 ** threads block signals until they are set up properly
685 ** errno is no longer shared by threads
686 ** interrupt-thread restores the eflags register on x86
687 ** fixed some lockups due to gc/thread interaction
688 * fixed some bugs revealed by Paul Dietz' test suite:
689 ** invalid dotted lists no longer raise a read error when
691 ** signal an error if a symbol that names a declaration is used
692 as the name of a type, or vice versa
693 ** allow using the (declare (typespec var*)) abbreviation for
694 (declare (type typespec var*)) with all type specifiers
695 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
696 ** always return NIL from PPRINT-POP when OBJECT is NIL
697 ** don't signal errors when pretty-printing malformed LABELS,
698 FLET or MACROLET forms
699 ** declarations in a DOLIST body are also in scope for the
701 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
703 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
706 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
707 * fixed cross-compiler leakages that prevented building a 32-bit
708 target with a 64-bit host compiler.
709 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
710 opened with :IF-EXISTS :OVERWRITE.
711 * fixed bug 281: error for an invalid qualifier in a short-form method
712 combination method is not signalled until the faulty method is called.
713 * bug fix: iteration variable type inferrer failed to deal with open
714 intervals. (reported by Alan Shields)
715 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
716 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
717 or more runtime options were provided to the sbcl binary.
718 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
719 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
720 * optimization: there's now a fast-path for fixnum arguments in the
721 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
722 * optimization: the code generated on x86-64 is more compact thanks
723 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
724 * bug fix: floating-point exception handling now partly works on
725 x86-64. (Thanks to James Knight)
726 * improvement to the MIPS backend from Thiemo Seufer: C-style
727 64-bit long long arguments and return values to alien functions
729 * fixed some bugs revealed by Paul Dietz' test suite:
730 ** the type-error signalled from WARN has a filled-in DATUM slot.
731 ** the type-error required when a stream is not associated with
732 a file has the stream as its datum.
733 ** type-errors on single-floats on x86-64 no longer have
734 :INVALID-OBJECT as the datum
735 ** the type-errors signalled for invalid function names now have
736 a correct expected type
737 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
738 for structures defined with a :TYPE. Documentation strings for
739 typed structures are no longer immediately discarded
740 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
741 broadcast streams and synonym streams. FILE-LENGTH now also works
742 on broadcast streams.
744 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
745 * incompatible change: the --noprogrammer option, deprecated since
746 version 0.7.5, has been removed. Please use the equivalent
747 --disable-debugger option instead.
748 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
750 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
751 inoperational for a while now) has been completely removed.
752 * null lexical environments are now printed as #<NULL-LEXENV>,
753 significantly reducing the amount of clutter in typical backtraces.
754 * documentation on weak pointers, finalization, and after GC hooks
755 has been added to the manual.
756 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
757 as other specialized array types, is much faster. SUBSEQ and
758 COPY-SEQ on such arrays have also been sped up.
759 * optimization: EQL is now more efficient when at least other argument
760 is known to be of type (OR FIXNUM (NOT NUMBER)).
761 * fixed bug: the runtime is now less vulnerable to changes in the
762 size of the SBCL object on OS X, and virtual memory is reserved for
763 all spaces that need to be at a fixed address.
764 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
765 * fixed bug: finalizers and after GC hooks that cause consing are now
767 * fixed bug: compiler error messages and summaries are now printed to
768 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
769 * fixed inference of the upper bound of an iteration variable.
770 (reported by Rajat Datta).
771 * fixed bug 211e: calling local functions with duplicated constant
772 keyword argument no longer causes a bogus style warning about an
774 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
775 ftype declaration to be disregarded. (reported by Dave Roberts)
776 * fixed bug 373: caused by erronous compilation of references to alien
777 variables in the runtime on ppc/darwin.
778 * fixed bug 376: CONJUGATE type deriver.
779 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
780 (reported by Baughn on #lisp)
781 * a cross-compiler bug on non-x86 platforms has been identified and
782 fixed. (thanks to Bruno Haible)
783 * improvements to the MIPS runtime code for increased stability.
784 (thanks to Thiemo Seufer)
785 * increased the maximimum compact environment size to allow
786 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
787 * partial workaround for bug 135: don't name the function we're
788 calling for hairy cases of EVAL, so as not to accumulate one environment
789 entry per call to EVAL. (reported by Kevin Reid)
790 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
791 * optimization: type testing for non-vector arrays should be faster.
792 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
794 * bug fix: setting 31st element of a bit vector to zero did not work
796 * bug fix: redefining a class definition which failed due to a
797 previous accessor / function clash now works (but see BUGS entry
798 #380 for more problems in this area). (thanks to Zach Beane)
799 * the long form of DEFINE-METHOD-COMBINATION disables method group
800 checking when given a single method group with pattern *. (thanks
802 * on x86 compiler supports stack allocation of results of simple
803 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
804 * support for building and running on Mac OS X 10.4 "Tiger" has been added
805 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
806 * fixed some bugs related to Unicode integration:
807 ** the restarts for recovering from input and output encoding
808 errors only appear when there is in fact such an error to
810 * fixed some bugs revealed by Paul Dietz' test suite:
811 ** MISC.549 and similar: late transformation of unsafe type
812 assertions into derived types caused unexpected code
814 ** SCALE-FLOAT type deriver is less wrong.
815 ** type derivers for EXP, LOG and similar functions compute result
816 types for complex arguments better.
817 ** (MISC.563) CONJUGATE type deriver works for very restricted
819 ** out-of-line type testers for character strings are available.
820 ** EQUAL compiler transform understands specialness of objects
822 ** accessing double-floats stored on the stack now works on x86-64.
823 ** debugger internals could sometimes create invalid lispobjs,
824 resulting in GC crashes.
825 ** MISC.548: type check weakening can convert required type into
827 ** initialization forms for bindings are not in scope of free special
830 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
831 * incompatible change: thread support for non-NPTL systems has
832 been removed - locking is buggy and unreliable. A threaded
833 SBCL build will now warn at startup and refuse to create
834 new threads, unless futex support is detected in the kernel
835 * incompatible change: the top level REPL now has only an ABORT
836 restart associated with it, not TOPLEVEL and ABORT as it used to.
837 TOP and TOPLEVEL are now available as debugger commands for
838 returning to the top level.
839 * incompatible change: forms evaluated in the REPL now use the
840 global optimization policy.
841 * incompatible change: user- and system-initialization files are
842 no longer processed with LOAD, but by READ and EVAL; hence the
843 global optimization policy, startup package, readtable, etc,
845 * The .fasl file format number has been incremented because of
846 various incompatible changes.
847 * internal entry point details and argument counts no longer appear
848 in backtraces unless explicitly requested by setting
849 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
850 * built-in and standard functions no longer have names like "top
851 level local call to FOO".
852 * fixed bug 32: functions defined in non-null lexical environments
853 now have more legible printed representation
854 * fixed bug 33: functions defined in non-null lexical environemnts
855 are now more amenable to inspection by INSPECT.
856 * workaround for bug 354: XEPs no longer appear in backtraces unless
857 explicitly requested.
858 * fixed bug: receiving the signal which results from attempting to
859 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
860 SIGBUS on other *BSDs) on architectures where the C stack is also
861 the Lisp stack (x86 and x86-64) from foreign code no longer leads
862 to debugger confusion or wild instability. (reported by Cheuksan
864 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
865 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
866 (reported by Lutz Euler)
867 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
868 compiled code. (reported by Paul Dietz)
869 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
870 specialized parameter to a method to something that is not TYPEP
871 the specializer is now possible.
872 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
873 face of package deletion.
874 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
875 pathname. (reported by David Tolpin on c.l.l)
876 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
877 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
878 STRING-STREAM and FILE-STREAM, respectively.
879 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
880 than previously; in addition, its test suite is now run on build.
881 (thanks to Luke Gorrie)
882 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
883 * optimization: added a immediate representation for single-floats
885 * fixed some bugs revealed by Paul Dietz' test suite:
886 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
887 ** attempting to create a package with a colliding nickname causes
888 correctable errors to be signalled.
889 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
890 ** several x86-64 backend bugs related to sign-extension of immediate
893 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
894 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
895 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
897 * fixed loading of multiply forward-referenced layouts.
898 (thanks to Cheuksan Wang)
899 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
900 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
901 (see COMPILE-FILE documentation for details of :PRINT option)
902 * fixed bugs 19 and 317: fixed-format floating point printing is
903 more accurate. This also fixes a bug reported by Adam Warner
904 related to the ~@F format directive.
905 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
907 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
908 dispatch table if the from-char is a dispatch macro character.
909 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
910 by the wordsize no longer ignore the last word. (reported by Lutz
912 * fixed bug in type checking of dynamic-extent variables. (reported
914 * optimization: sequence traversal functions use their freedom to
915 coerce function designators to functions.
916 * optimization: code with many calls to CLOS methods specialized on
917 CLOS classes has had redundant type checks removed.
918 * optimization: added declarations to speed up operations that access
919 the internal character database (for example STRING-UPCASE)
920 * optimization: comparison operations between floats and sufficiently small
921 fixnums no longer create extra rationals
922 * fixed some bugs related to Unicode integration:
923 ** portions of multibyte characters at the end of buffers for
924 character-based file input are correctly transferred to the
925 start of the buffer at the next read.
926 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
927 passing it through to OPEN.
928 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
929 argument given to internal calls to OPEN.
930 ** the built-in comment readers (introduced by character sequences
931 ";" and "#|") are more forgiving to encoding errors; they will
932 STYLE-WARN and then attempt to resync the stream at a character
933 boundary. (thanks to Teemu Kalvas)
934 * fixed some bugs in the x86-64 port:
935 ** Negative short int return values from c-calls are sign-extended
937 ** The stack is aligned to 16-bytes for c-calls, as required by
938 the ABI. (thanks to Cheuksan Wang)
939 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
940 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
941 for enums. (thanks to Vincent Arkesteijn)
942 ** Multiple small optimizations and bugfixes for floating point
944 * fixed some bugs revealed by Paul Dietz' test suite:
945 ** Space, Tab, Linefeed, Return and Page have the invalid
946 secondary constituent character trait.
947 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
949 ** WITH-INPUT-FROM-STRING should only update the index place on
951 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
953 ** Bit-array operations (BIT-AND and similar) worked incorrectly
954 with one-dimensional arrays with fill pointers.
955 ** TYPE-OF failed on a complex with an integer realpart and a
957 ** compiler failure during type inference for the code of form
958 (IF (EQL X (THE ...)) ...) (MISC.535).
960 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
961 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
962 platform. The port supports 61-bit fixnums, large memory spaces
963 and reloading shared object files.
964 * enhancement: saving cores with foreign code loaded is now
965 supported on ppc/Darwin in addition to the previously supported
967 * enhancement: the statistical profiler now walks deeper into the
968 call stack for more meaningful call-graphs and accrued time
969 reports (x86/x86-64 only). It also now reports time spent in
971 * enhancement: it is now possible to trace most individual methods
972 of a generic function in addition to tracing the generic function
974 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
976 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
977 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
978 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
979 produces an error. (thanks to Vincent Arkesteijn)
980 * bug fix: NAMESTRING on pathnames with :WILD components in their
981 directories works correctly. (thanks to Artem V. Andreev)
982 * fixed bug 125: compiler preserves identity of closures. (reported
984 * bug fixed: functions with &REST arguments sometimes failed with
985 "Undefined function" when compiled with (DEBUG 3). (reported by
987 * bug fix: overflow during compiling of setting element of a bit
988 vector with constant index and value. (reported by Timmy Douglas)
989 * build fix: fixed the dependence on *LOAD-PATHNAME* and
990 *COMPILE-FILE-PATHNAME* being absolute pathnames.
991 * on x86 compiler partially supports stack allocation of dynamic-extent
993 * GO and RETURN-FROM do not check the extent of their exit points
994 when compiled with SAFETY 0.
995 * fixed some bugs related to Unicode integration:
996 ** encoding and decoding errors are now much more robustly
997 handled; it should now be possible to recover even from invalid
998 input or output to the terminal. (thanks to Teemu Kalvas)
999 ** provided a first cut at implementing STRING-TO-OCTETS and
1000 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
1001 ** altered the SB-MD5 contributed module slightly, changing the
1002 interface just enough for it to be supportable for builds where
1003 lisp characters are not eight bits.
1004 * fixed some bugs revealed by Paul Dietz' test suite:
1005 ** the FORMATTER-generated functions for ~V[ conditionals require
1006 the correct number of arguments.
1007 ** READ-FROM-STRING returns the mandated second value when applied
1008 to displaced strings.
1009 ** the #\Rubout and #\Backspace characters are treated as invalid
1010 constituent characters by the tokenizer.
1012 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
1013 * new feature: reloading changed shared object files with
1014 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
1015 * new feature: references to foreign variables and functions
1016 can now be compiled and loaded before the corresponding shared
1017 object file is loaded, as long as the foreign definitions are
1018 available at runtime.
1019 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1020 supported. (thanks to Dan Debertin)
1021 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1022 just keywords. (thanks to Vincent Arkesteijn)
1023 * fixed bug #331: structure-class instances corresponding to
1024 DEFSTRUCT forms are now created eagerly.
1025 * fixed bug #345: backtraces from calls to undefined functions work
1026 on x86 as well. Related bug #61 is now also partially fixed on x86
1027 (backtraces from throws to unknown catch tags.)
1028 * bug fix: lambda-list parsing is now stricter vrt. order and number
1029 of lambda-list keywords.
1030 * bug fix: as specified by AMOP, an error is signalled if a
1031 class-option appears multiple times in a DEFCLASS form. (reported
1033 * bug fix: RANDOM can be compiled when the compiler derives the type
1034 of its numeric argument as a disjoint set of small integers.
1035 (reported by Paul Dietz)
1036 * bug fix: starting a core saved with shared objects loaded when
1037 those objects are not available no longer causes threaded SBCL to
1038 hang. (reported by Sean Ross)
1039 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1041 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1042 argument, which must always be NIL. (reported by Kalle Niemitalo)
1043 * bug fix: printing 1.0d+23 no longer results in an error.
1044 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1045 * bug fix: structure slot setters preserve evaluation order. (thanks
1047 * fixed some bugs related to Unicode integration:
1048 ** RUN-PROGRAM can allow its child to take input from a Lisp
1049 stream. (reported by Stefan Scholl)
1050 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1051 locales. (reported by Ken Causey)
1052 * fixed some bugs revealed by Paul Dietz' test suite:
1053 ** INCF, DECF and REMF evaluate their place form as specified in
1055 ** FORMATTER expands ~{ ~} iteration directives with V or #
1056 parameters correctly.
1057 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1058 consequent uses no arguments correctly.
1059 ** the system has a partial understanding of the (COMPLEX RATIO)
1061 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1064 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1065 * new feature: a build-time option (controlled by the :SB-UNICODE
1066 keyword feature, enabled by default) for building the system with
1067 support for the entire 21-bit character space defined by the
1069 * new feature: the system now has rudimentary external-format
1070 support; the primary user-visible change at this time is that
1071 characters with the high bit set (such as lower-case-e-acute) will
1072 print correctly to a terminal in a UTF-8 environment.
1073 * minor incompatible change: BASE-CHAR no longer names a class;
1074 however, CHARACTER continues to do so, as required by ANSI.
1075 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1076 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1077 has been moved to the SB-EXT package (temporarily re-exported from
1079 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1080 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1082 * fixed bug #351: better error-handlind and reporting for malformed
1084 * fixed bug #350: bignum-printing is now more memory-efficient,
1085 allowing printing of very large bignums, eg. (expt 2 10000000).
1086 (reported by Bruno Haible)
1087 * fixed bug #302: better primitive-type selection for intersection
1089 * fixed bug #308: non-graphic characters now all have names, as
1090 required. (reported by Bruno Haible)
1091 * bug fix: redefining a class with different superclasses now correctly
1092 removes it from the direct-subclasses of its previous superclasses.
1093 (reported by David Morse)
1094 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1095 new value now works. (reported by Bruno Haible)
1096 * bug fix: correct canonicalization of multiple non-standard slot
1097 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1098 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1099 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1100 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1101 now exists, an signals an error.
1102 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1103 during macroexpansion for non-iteration variables as well. (reported
1104 by Bruno Haible for CMUCL)
1105 * bug fix: Cyclic structures and unprintable objects in compiler
1106 messages no longer cause errors. (reported by Bruno Haible)
1107 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1108 and READ-FROM-STRING all now return a primary value of NIL if
1109 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1110 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1111 (reported by Bruno Haible for CMUCL)
1112 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1113 do not designate a condition. (reported by Bruno Haible for
1115 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1116 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1117 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1118 permitting resolution in favour of any of the conflicting symbols.
1119 (reported by Bruno Haible for CMUCL)
1120 * bug fix: EQUAL compiler optimizations is less aggressive on
1121 strings which can potentially compare true despite having distinct
1122 specialized array element types.
1123 * bug fix: unit enumerations can be defined without dividing by
1124 zero. (thanks to Vincent Arkesteijn)
1125 * FORMAT compile-time argument count checking has been enhanced.
1126 (report from Bruno Haible for CMUCL)
1127 * a partial workaround for the bug 262: the compiler does not try to
1128 inline-expand a local function doing RETURN-FROM from a deleted
1129 BLOCK. (thanks to Peter Denno for the bug report and to David
1130 Wragg for the simple test case)
1131 * fixed some bugs revealed by Paul Dietz' test suite:
1132 ** RENAME-PACKAGE allows all package designators as new package
1134 ** constraint propagation and lambda variable substitution are
1135 more cautious in dealing with partially deleted code.
1136 ** compiler sometimes lost reoptimization passes.
1137 ** CERROR, given a condition as condition designator, passes its
1138 remaining arguments to the continue format control without
1140 ** Case-altering FORMAT directives work correctly on non-ASCII
1142 ** The REAL type specifier handles bounds outside the floating
1143 point ranges without signalling FLOATING-POINT-OVERFLOW.
1144 ** Functions with IR1-transformations can create intercomponent
1145 references to global functions.
1146 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1148 ** FORMAT ~R treats a nil value for its first parameter correctly.
1150 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1151 * enhancement: saving cores with foreign code loaded is now
1152 supported on x86/NetBSD and sparc/Linux in addition to the previously
1153 supported platforms.
1154 * bug fix: on some platforms repeated installations caused multiple
1155 copies of HTML documentation to be installed -- should not happen
1156 any more. (reported by Stefan Scholl)
1157 * bug fix: parsing self-recursive alien record types multiple times
1158 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1159 original patch by Helmut Eller for CMUCL)
1160 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1161 (thanks to Richard Kreuter)
1162 * bug fix: defining classes whose accessors are methods on existing
1163 generic functions in other (locked) packages no longer signals
1164 bogus package lock violations. (reported by François-René Rideau)
1165 * bug fix: special variables as DEFMETHOD parameters no longer have
1166 associated bogus type declarations. (reported by David Wragg and
1168 * bug fix: read-write consistency on streams of element-type
1169 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1170 * bug fix: redefiniton of the only method of a generic function with
1171 no DEFGENERIC no longer emits a full WARNING. In addition,
1172 redefinition of generic functions with no DEFGENERIC to an
1173 incompatible lambda list now signals an error. (thanks to Zach
1175 * bug fix: DEFGENERIC now works even when there's a function of the
1176 same name in an enclosing lexical environment. (thanks to Zach
1178 * fixed compiler failure, caused by instrumenting code during
1179 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1180 * optimization: added loop analysis and improved register allocation
1181 * fixed some bugs revealed by Paul Dietz' test suite:
1182 ** POSITION on displaced vectors with non-zero displacement
1183 returns the right answer.
1184 ** (SIMPLE-STRING) is a valid type specifier for sequence
1186 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1188 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1189 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1191 ** The justification version of the FORMAT ~< directive treats
1192 non-zero minpad parameter correctly.
1194 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1195 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1196 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1197 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1198 the supported interface.
1199 * new feature: Single-stepping of code compiled with DEBUG 2 or
1200 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1201 * new feature: saving cores with foreign code loaded is now
1202 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1203 Timothy Moore's work for CMUCL)
1204 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1205 parameters to * instead of NIL if no initform is supplied.
1206 (reported by Johan Bockgaard)
1207 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1208 correctly when FUNCALL appears as the car of the form. Note:
1209 despite this FUNCALL forms are not currently subject to
1210 compiler-macro expansion. (port of Raymond Toy's fix for the
1211 same from CMUCL, reported by Johan Bockgaard)
1212 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1213 (thanks for Teemu Kalvas)
1214 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1215 case instead of terminating the iteration (thanks for Julian
1216 Squires, Sean Champ and Raymond Toy)
1217 * bug fix: incorrect expansion of defgeneric that caused a style
1218 warning. (thanks for Zach Beane)
1219 * bug fix: slot accessor effective method computation works properly
1220 for classes with multiple non-standard applicable methods on
1221 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1222 * on x86 compiler supports stack allocation of results of LIST and
1223 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1224 CMUCL implementation by Gerd Moellmann)
1225 * fixed some bugs revealed by Paul Dietz' test suite:
1226 ** FORMAT strings with both the ~<~:;~> form of the justification
1227 directive and pretty-printing directives cause an error.
1229 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1230 * incompatible change: the internal functions
1231 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1232 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1233 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1234 instead of the old functions.
1235 * new feature: on platforms where "dladdr" is available foreign
1236 function names now appear in backtraces. (based on Helmut Eller's
1238 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1239 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1241 * documentation: ASDF manual is now distributed with SBCL.
1242 * bug fix: SBCL can now load its contributed modules with REQUIRE
1243 even if the system-provided entries have been removed from
1244 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1246 * bug fix: backtraces involving undefined functions or assembly
1247 routines are more informative. (thanks to Brian Downing)
1248 * bug fix: mutually referent alien structures now work correctly.
1249 (reported by Rick Taube)
1250 * bug fix: structures defined by WITH-ALIEN can be referred to
1251 within other definitions in the same WITH-ALIEN.
1252 * bug fix: division operators (MOD, TRUNCATE and the like) with
1253 constant zero divisors and integer dividends no longer generate
1255 * bug fix: provide default methods for INPUT-STREAM-P and
1256 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1257 * bug fix: improve the detection and resolution of MOP metacycles
1258 (where effective method computation depends on the generic function
1259 having its effective method computed). (reported by Bruno Haible)
1260 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1261 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1262 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1263 * bug fix: pathnames with relative directory components can be
1264 represented relative to default pathnames.
1265 * optimization: in taking the GCD of bignums, reduce the two bignums
1266 to approximately the same size (using Euclid's algorithm) before
1267 applying the more sophisticated binary GCD. (thanks to Juho
1269 * optimization: COUNT on bitvectors now operates word-at-a-time.
1270 * optimization: ASH with a positive, but not necessarily constant,
1271 (leftwards) shift, when in a modular context, is compiled to a
1273 * fixed some bugs revealed by Paul Dietz' test suite:
1274 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1275 if the corresponding argument is NIL.
1276 ** FORMAT directives accepting a minpad parameter treat negative
1278 ** literal commas as character parameters to FORMAT directives are
1280 ** literal spaces directly after ~<Newline> directives within a
1281 format-logical-block (~:< ~@:>) do not induce :FILL-style
1282 conditional newlines.
1283 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1284 as specified: it no longer includes conditional newlines).
1285 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1287 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1288 :PER-LINE-PREFIX argument does not evaluate to a string.
1290 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1291 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1292 the manual for details; for now, package locks can be disabled by
1293 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1294 if no major problems are found then it is likely that they will be
1295 compiled in unconditionally.
1296 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1297 unsupported operators on all platforms. To load a shared library
1298 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1299 object file, link it into a shared library outside of SBCL and
1300 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1302 * bug workaround: ROOM T and the GC were not working together
1303 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1304 reporting facility aren't preserved. That reporting has been
1305 disabled, so now until and unless someone figures out how to make it
1306 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1307 * minor incompatible change: as threatened around sbcl-0.8.0, the
1308 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1309 The symbols are also exported from SB-PCL for backwards
1310 compatibility, but more so than before SB-PCL should be treated as
1311 an implementation-internal package.
1312 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1314 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1315 It is known as of this release not to work on the Alpha, however.
1316 * fixed bug #167: errors signalled due to illegal syntax in method
1317 bodies are now more legible.
1318 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1319 designators and can hence be used with TYPEP.
1320 * fixed bug #333: CHECK-TYPE now ensures that the type error
1321 signalled, if any, has the right object to be accessed by
1322 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1323 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1324 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1326 * fixed bug #334: programmatic addition of slots using specialized
1327 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1328 and :ALLOCATION :CLASS slots.
1329 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1330 float exponents. (rereported by Peter Seibel)
1331 * fixed a bug: #\Space (and other whitespace characters) are no
1332 longer considered to be macro characters in standard syntax by
1333 GET-MACRO-CHARACTER.
1334 * fixed bug: initialization of condition class metaobjects no longer
1335 causes an instance of the condition to be created. (reported by
1337 * fixed bug: it is now possible to have slots such that
1338 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1339 description is neither :INSTANCE nor :CLASS.
1340 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1341 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1342 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1343 non-standard SLOT-DEFINITION-ALLOCATION values.
1344 * fixed bugs: various race conditions handling exiting threads.
1345 CL-PPCRE's thread torture test now passes.
1346 * fixed arguably-a-bug: GC time proportional to number of threads in
1347 system even when most of them are idle
1348 * optimization: improved performance of BIT and SBIT on bit-vectors.
1349 * .fasl file incompatibility: The fasl file version number has
1350 been incremented because of changes associated with package locks.
1352 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1353 * minor incompatible change: the system no longer provides
1354 optimization hints (conditions which are TYPEP
1355 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1356 mode; these hints are still emitted when higher SPEED optimization
1358 * new contrib module: a sampling profiler (profiling by statistical
1359 sampling, rather than by instrumenting functions) is available as
1360 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1361 * the behaviour of the standard function ED is now customizeable by
1362 third parties through a hook variable: see ED's documentation
1363 string for information on the protocol.
1364 * the compiler no longer emits efficiency notes for (FUNCALL X)
1365 when the type of X is uncertain under default optimization
1367 * fixed bug 276: mutating a binding of a specialized parameter to a
1368 method to something that is not TYPEP the specializer is now
1370 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1371 exhibit the required behaviour on the x86 platform. (thanks to
1372 Peter van Eynde, Eric Marsden and Bruno Haible)
1373 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1374 even for difficult arguments. (reported by Peter Graves)
1375 * fixed bug 141a: the backquote printer now descends quoted
1377 * fixed another bug in backquote printing: no more destructive
1378 modification of the form's list structure. (reported by Brian
1380 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1381 move between its address being taken and the call to
1382 interrupt_thread, fixing a crashing race condition.
1383 * the SB-POSIX contrib implementation has been adjusted so that it
1384 no longer exhibits ridiculously poor performance when constructing
1385 instances corresponding to C structs.
1387 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1388 * minor incompatible change: the sb-grovel contrib now treats C
1389 structures as alien (in the sense of SB-ALIEN) objects rather than
1390 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1391 has implications for memory management of client code
1392 (sb-grovel-returned objects must now be manually managed) and for
1393 type safety (alien objects now have full types).
1394 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1395 used to control emission of compiler diagnostics, rather than the
1396 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1397 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1398 quality should be considered deprecated.
1399 * install.sh now installs the user manual as well
1400 * (not quite a new documentable feature, but worth considering in
1401 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1402 beginnings of a semantically meaningful condition hierarchy is
1403 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1405 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1406 designator as the defaults argument.
1407 * fixed bug: Displaced arrays whose displaced-to array has become
1408 too small now cause ARRAY-DIMENSION to signal an error, providing
1409 for safer bounds-checking. (reported by Bruno Haible)
1410 * fixed bug: DEFCLASS slot definitions with identical :READER and
1411 :WRITER names now signal a reasonable error. (reported by Thomas
1413 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1415 * fixed bug: Invalid slot specification errors now print correctly.
1416 (thanks to Zach Beane)
1417 * fixed bug 320: Shared to local slot value transfers in class
1418 redefinitions now happen corrently from superclasses as
1419 well. (reported by Bruno Haible)
1420 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1422 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1423 improper lists correctly. (reported by Bruno Haible)
1424 * fixed bug 313: source-transform for <fun-name> was erroneously
1425 applied to a call of a value of a variable with name <fun-name>.
1426 (reported by Antonio Menezes Leitao)
1427 * fixed bug 307: The obsolete instance protocol ensures that
1428 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1429 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1430 remove dead unknown-values globs from the middle of the stack.
1431 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1432 easier to resume long computations after using *BREAK-ON-SIGNALS*
1433 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1434 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1435 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1436 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1437 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1438 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1440 * fixed a (fairly theoretical) bug in string printing: if
1441 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1442 does not have array-element-type equal to the most general string
1444 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1445 function name. (thanks to Nikodemus Siivola)
1446 * fixed bug: compile-time format string checker failed on
1447 non-closed ~{. (reported by Thomas F Burdick)
1448 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1449 don't warn on legal loop code involving more than one aggregate
1450 boolean. (thanks to Nikodemus Siivola)
1451 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1452 in keyword position to MAKE-PACKAGE was not being reported as
1454 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1455 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1457 * fixed bugs 280 and 312: the checking for multiple definitions in a
1458 file is less likely to become confused by uses of inline
1460 * fixed bug: the #S reader macro performs the keyword coercion
1461 specified for slot names. (reported by Kalle Niemitalo)
1462 * fixed bug: lambda lists may contain symbols whose names start with
1463 & but are not lambda-list-keywords; their occurrence triggers a
1465 * fixed bug 321: define-method-combination argument lambda lists do
1466 not cause invalid code to be generated when &optional or &aux
1467 variables have default values. (reported by Bruno Haible)
1468 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1469 have CLOS classes; this appears also to have fixed reported
1470 instability in stack exhaustion detection.
1471 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1472 MUFFLE-WARNING when no associated restart is present is now
1474 * optimization: rearranged the expansion of various defining macros
1475 so that each expands into only one top-level form in a
1476 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1478 * optimization: used a previously little-used slot in symbols to
1479 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1481 * fixed some bugs revealed by Paul Dietz' test suite:
1482 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1484 ** Fixed an optimization bug related to inheritance of initforms
1485 from local to shared slots.
1486 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1487 ** CAST optimizer forgot to flush argument derived type.
1488 ** print/read consistency on floats is now orders of magnitude
1489 more likely. (thanks also to Bruno Haible for a similar report
1491 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1492 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1493 for objects of type REAL. Make it so.
1494 ** SXHASH is specified (infelicitously) to respect similarity,
1495 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1496 it so. (thanks to Markus Ziegler)
1497 ** on the Alpha, the compiler succeeds in compiling functions
1498 returning a known number of arguments greater than 63.
1499 ** fixed handling of invalid NIL arguments in keyword position
1501 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1503 ** provided a readably-printable representation for RANDOM-STATE
1505 ** ensured that pathnames, where they have a namestring, always
1506 print using #P"..." syntax.
1508 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1509 * Support for the forthcoming 2.0 version of the NetBSD kernel
1510 running on x86 hardware has been added. (thanks to Perry
1511 E. Metzger most immediately, and others for their past work)
1512 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1513 current release version; SBCL's OpenBSD support had been broken
1514 since about the time of OpenBSD's switch to ELF binary format.)
1515 * [placeholder for DX summary]
1516 ** user code with &REST lists declared dynamic-extent, under high
1517 speed or space and low safety and debug optimization policy.
1518 * The manual has been converted to Texinfo format and the debugger
1519 chapter from the cmucl manual has been added.
1520 * A facility has been added to extract documentation strings from
1521 sbcl and store them as Texinfo-formatted snippets for inclusion in
1522 the manual (via Texinfo's @include directive)
1523 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1524 the test case to Dave Roberts)
1525 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1526 pointers. (reported by Sean Ross)
1527 * bug fix: PROFILE output is printed nicely even for large numerical
1528 values. (thanks to Zach Beane)
1529 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1530 greater than 32 handle EOF correctly.
1531 * bug fix: on X86 an immediate argument of the IMUL instruction is
1532 correctly printed in disassembly. (thanks to Lutz Euler)
1533 * bug fix: class slots in redefined classes preserve their old
1534 values. (thanks to Bruno Haible and Nikodemus Siivola)
1535 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1536 succeeds. (reported by Marco Baringer)
1537 * fixed some bugs revealed by Paul Dietz' test suite:
1538 ** READ-SEQUENCE now works on ECHO-STREAMs.
1539 ** RATIONALIZE works more according to its specification. (thanks
1542 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1543 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1544 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1545 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1546 should matter to you only if you rebind the printer control
1547 variables and then find you want different bindings in the
1548 debugger than in the ordinary execution of your program.)
1549 * The runtime build system has been tweaked to support building
1550 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1551 assembler and linker. (thanks to Nikodemus Siivola)
1552 * Unbound, undefined, undeclared variables now trigger full
1553 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1554 more often programmer error than deliberate exploitation of undefined
1556 * optimization: the hash algorithm for strings has changed to one
1557 that is less vulnerable to spurious collisions. (thanks to Juho
1559 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1560 needless bounds checking. (thanks to Juho Snellman)
1561 * optimization: implemented multiplication as a modular
1562 (UNSIGNED-BYTE 32) operation on the PPC backend.
1563 * fixed some bugs revealed by Paul Dietz' test suite:
1564 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1566 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1567 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1568 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1570 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1572 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1574 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1575 their output stream on EOF from read.
1576 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1577 have been read to end-of-file.
1578 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1580 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1581 description of determination of which consecutive characters
1583 ** Printing the "Space" character with escaping on now yields "#\\ ",
1584 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1585 ** Reading floating-point numbers with *READ-BASE* set to a number
1586 less than 10 works correctly.
1587 ** Reading floating-point numbers with *READ-BASE* set to a number
1588 more than 10 works correctly.
1589 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1590 the readtable currently in effect.
1592 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1593 * minor incompatible change: parsing of namestrings on a physical
1594 (Unix) host has changed; numbers after the final #\. in a
1595 namestring are no longer interpreted as a version field. This is
1596 intented to be largely invisible to the user, except that the
1597 meaning of the namestring "*.*.*" has changed: it now refers to a
1598 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1599 should usually be replaced by
1600 (make-pathname :name :wild :type :wild :version :wild)
1601 with the added benefit that this is more likely to be portable.
1602 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1603 option to OPEN now signals an error if the file being opened
1604 exists; this may have an impact on existing code.
1605 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1606 signals received, on the PowerPC platforms (both Linux and
1607 Darwin). (thanks to Pierre Mai for pointing out the location of
1609 * several fixes on OS X: The system now builds and runs cleanly on
1610 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1611 (thanks to Brian Mastenbrook)
1612 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1613 second-resolution: integer multiples of 1/3600 between -24 and 24.
1614 (thanks to Vincent Arkesteijn)
1615 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1616 type when called with 1 argument; PEEK-CHAR checked type of
1617 PEEK-TYPE only after having read first character from a
1618 stream. (reported by Peter Graves)
1619 * bug fix: the garbage collector now has much better locality
1620 behaviour, and in particular no longer treats all memory as being
1621 exhausted when many small objects point to each other in a deeply
1623 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1624 recognized as being TYPEP their class.
1625 * bug fix: the PUSHNEW documentation string has been corrected.
1626 (thanks to Vincent Arkesteijn)
1627 * bug fix: defaulting of the value for the last of an atypically
1628 large number of multiple values being bound was not being
1629 performed correctly on the Alpha or PPC platforms
1630 * optimization: implemented multiplication as a modular
1631 (UNSIGNED-BYTE 32) operation on the x86 backend.
1632 * optimization: SEARCH on simple-base-strings can now be open-coded.
1633 (see also contrib/compiler-extras.lisp for inspiration for
1634 teaching the compiler about the Boyer-Moore algorithm).
1635 * value, returned by MAX (and MIN) called with several EQUALP, but
1636 not EQL, arguments now does not depend on compiler settings.
1637 * fixed some bugs revealed by Paul Dietz' test suite:
1638 ** in stack analysis liveness information is propagated from
1639 non-local entry points.
1640 ** pathwise CAST removing failed when the CAST node did not start
1642 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1643 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1645 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1646 host is already defined.
1647 ** RENAME-FILE works on streams instead of signalling an internal
1649 ** PEEK-CHAR uses the current readtable when determining whether
1650 or not a character is whitespace.
1651 ** MERGE-PATHNAMES handles the case when the pathname does not
1652 specify a name while the default-pathname specifies a version.
1653 ** Pathnames now stand a better chance of respecting print/read
1655 ** Attempting to use standardized file system operators with a
1656 pathname with invalid :DIRECTORY components signals a
1658 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1660 ** DIRECTORY on logical pathnames is more correct.
1661 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1662 signal a TYPE-ERROR if their argument is not a stream.
1663 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1664 designator argument does not designate a stream.
1665 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1666 examining the synonym.
1667 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1669 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1670 element-type larger than ([UN]SIGNED-BYTE 32).
1672 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1673 * When built with the :SB-FUTEX feature, threaded builds now take
1674 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1675 for faster/more reliable mutex and condition variable support.
1676 * Incompatible change (but one you probably shouldn't have been using
1677 anyway): the interface and code for arbitrating between multiple
1678 threads in the same user session has been redesigned.
1679 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1680 error when called without an explicit environment argument.
1681 (thanks to Nikodemus Siivola)
1682 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1683 confused about their position. (thanks to Adam Warner and Gerd
1685 * bug fix: Pretty printing backquoted forms with unquotations in the
1686 argument list position of various code constructs such as LAMBDA
1687 now works correctly. (reported by Paul Dietz)
1688 * bug fix: Pretty printing unquotations no longer loses all
1689 stream position information.
1690 * optimization: performance of string output streams is now less
1691 poor for multiple small sequence writes.
1692 * optimization: performance of CSUBTYPEP in the presence of complex
1693 expressions involving CONS and NOT many times has been improved.
1694 (reported by Paul Dietz)
1695 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1697 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1698 streams for the REPL, for the debugger, and so on. (thanks to
1700 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1701 * fixed some bugs revealed by Paul Dietz' test suite:
1702 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1703 is ignored if it is too small, rather than propagating through
1704 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1705 ** extremely complex negations of CONS types were not being
1706 sufficiently canonicalized, leading to inconsistencies in
1708 ** VALUES tranformer lost derived type.
1710 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1711 * fixed a bootstrapping bug: the build process no longer assumes
1712 that the various BOOLE-related constants have the same value in
1713 host and target lisps. (noted by Paul Dietz' test suite on an
1714 SBCL binary built from CLISP)
1715 * The system can now be dynamically linked on the MIPS platform,
1716 which enables dynamic loading of foreign code from Lisp. (thanks
1717 to Ralf Baechle for discussions on the MIPS ABI)
1718 * The system now records debugging information for its own source
1719 files in a filesystem-position-independent manner, relative to
1720 the "SYS" logical host.
1721 * fixed a compiler bug: MV-LET convertion did not check references
1722 to the "max args" entry point. (reported by Brian Downing)
1723 * tweaked disassembly notes to be less confident about proclaiming
1724 some instruction as an LRA. (thanks to Brian Downing)
1725 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1726 now each have their own history, command character, and other
1727 characteristics. (thanks to David Lichteblau)
1728 * fixed some bugs revealed by Paul Dietz' test suite:
1729 ** compiler failure in compiling LOGAND expressions including a
1731 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1732 shift greater than 32.
1733 ** FLUSH-DEST did not mark blocks for type check regeneration.
1734 ** HANDLER-CASE failed to accept declarations in handler clauses
1735 in some circumstances.
1737 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1738 * New code in contrib/sb-introspect (still probably not entirely
1739 stable yet) provides some support for smart Lisp development
1740 environments like SLIME.
1741 * The conditions signalled for errors occurring when loading .fasl
1742 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1743 in a way which should help ASDF recover gracefully.
1744 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1745 stuff has been cleaned up. If you code contrib/ stuff, this might
1746 affect you, and you can look at contrib/README, contrib/STANDARDS,
1747 and/or the 0.8.4.27 diff to check.
1748 * In full calls the compiler now does not generate checks for declared
1749 argument types for all arguments.
1750 * various threading fixes
1751 ** and some experimental patches which didn't make it into
1752 the main tree for this release, but which are shipped in
1753 contrib/experimental-thread.patch as a possible fix for some
1754 failures (deadlock, spinning...) in GC-intensive multithreaded
1756 * fixed PPC build problem (source code incompatibility of different
1757 library versions): added offsetof() hackery which attempts to divine
1758 where glibc maintainers put uc_mcontext today
1759 * fixed bug 282: compiler does not trust type assertions while passing
1760 arguments to a full call.
1761 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1762 &OPTIONAL VALUES type parameter.
1763 * fix bug 214: algorithm for noting rejected templates is now more
1764 similar to that of template seletion. (also reported by rydis on
1766 * fixed bug 141b: printing backquoted information readably and prettily
1767 inserts a space where necessary.
1768 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1769 (CHARACTER 10) are now reported as errors, rather than propagated
1770 as unknown types. (reported by piso on #lisp)
1771 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1772 with values NIL and :ERROR. (thanks to Milan Zamazal)
1773 * fixed bug 191c: CLOS now does proper keyword argument checking as
1774 described in CLHS 7.6.5 and 7.6.5.1.
1775 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1776 longer raise an error; further, using a list as a for-as-arithmetic
1777 counter now raises a meaningful error.
1778 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1779 now understood by sequence creation functions such as MAKE-SEQUENCE
1781 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1782 a STRING-INPUT-STREAM.
1783 * compiler enhancement: SIGNUM is now better able to derive the type
1785 * type declarations inside WITH-SLOTS are checked. (reported by
1787 * fixed some bugs revealed by Paul Dietz' test suite:
1788 ** incorrect optimization of TRUNCATE for a positive first
1789 argument and negative second.
1790 ** compiler failure in let-convertion during flushing dead code.
1791 ** compiler failure while deriving type of TRUNCATE on an
1792 interval, containing 0.
1793 ** ASH of a negative bignum by a negative bignum count now returns
1795 ** intersection of CONS types now canonicalizes properly, fixing
1796 inconsistencies in SUBTYPEP.
1798 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1799 * incompatible change: The --disable-debugger command line
1800 option now clobbers the debugger at a more fundamental
1801 level, by redefining #'INVOKE-DEBUGGER instead of by
1802 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1803 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1804 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1805 enter the debugger REPL and then suffer recursive errors
1806 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1807 while under the new system, BREAK in --disable-debugger mode
1808 terminates the system just as an unhandled error would.
1809 * fixed compiler performance when processing loops with a step >1;
1810 * bug fix: DOCUMENTATION now retrieves generic function
1811 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1812 support has been systematized, and now supports the methods
1813 specified by ANSI, along with a default method and a method for
1814 slot documentation. (reported by Nathan Froyd)
1815 * bug fix: effective methods associated with a generic function are
1816 no longer cached over a change of that generic function's method
1817 combination. (reported by Andreas Fuchs)
1818 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1819 names a non-existent directory. (thanks to Andreas Fuchs)
1820 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1821 the PPC platform, lead to stack corruption; this has been fixed.
1822 (reported by Rainer Joswig)
1823 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1824 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1825 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1826 arithmetic in general, is now much more reliable on the PPC
1828 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1829 platform now returns the right answer.
1830 * optimization: restored some effective method precomputation in
1831 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1832 precomputation is now tunable.
1833 * optimization: compiler-internal data structure use has been
1834 reviewed, and changes have been made that should improve the
1835 performance of the compiler by about 20%.
1836 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1837 simple-base-strings and simple-bit-vectors is improved.
1838 * optimization: the optimization of 32-bit logical and arithmetic
1839 functions introduced in version 0.8.3 on the x86 has been
1840 implemented on the mips, ppc and sparc platforms; an
1841 implementation of the same facility, but for 64-bit arithmetic,
1842 has been added for the alpha.
1843 * microoptimization: the compiler is better able to make use of the
1844 x86 LEA instruction for multiplication by constants.
1845 * bug fix: in some situations compiler did not report usage of
1846 generic arithmetic in (SPEED 3) policy.
1847 * bug 145b fix: compiler used wrong type specifier while converting
1848 MEMBER-types to numeric.
1849 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1851 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1852 index argument now works properly on the Alpha platform.
1853 * bug fix: floating point exception treatment on the Alpha platform
1855 * bug fix: FILE-POSITION works much better on string input and
1856 output streams. (thanks to Nikodemus Siivola)
1857 * bug fix: many threading/garbage collection symptoms sorted.
1858 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1859 might be pseudo-atomic.
1860 * internal change: Stopping for GC is now done with signals not
1861 ptrace. GC is now done in whichever thread wanted it, instead of
1863 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1865 * bug fix: result form in DO is not contained in the implicit
1867 * incompatible change: ICR structure is changed; the value part of
1868 CONTINUATION is now called LVAR; corresponding functions are
1869 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1870 * added type deriver for ISQRT (thanks to Robert E. Brown).
1871 * bug fix: better support for loading from the command line when an
1872 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1874 * fixed some bugs revealed by Paul Dietz' test suite:
1875 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1876 ** ROUND and FROUND now give the right answer when given very
1877 small float arguments.
1878 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1880 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1881 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1882 ** LCM should return a non-negative integer.
1883 ** PARSE-INTEGER returned the index of a terminator instead of the
1884 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1885 ** PARSE-INTEGER returned an incorrect index being applied to a
1887 ** LCM with two arguments of 0 returns 0 rather than signalling
1889 ** unsigned addition of a 32-bit constant with the high bit set no
1890 longer causes an internal compiler error on the x86.
1891 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1892 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1893 with negative last argument.
1894 ** byte specifiers with zero size and position no longer cause
1895 an error during type derivation.
1896 ** bignum multiplication on the Alpha platform now returns the
1898 * porting: The system now builds on SuSE AMD64, although it still
1899 generates a 32-bit binary.
1900 * .fasl file incompatibility: The fasl file version number has
1901 been incremented (because of the changes to internal compiler
1902 data structures referred to above).
1904 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1905 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1906 more accurately, on the Darwin kernel running on PowerPC hardware.
1907 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1908 * Compiler code deletion notes now signal a condition of type
1909 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1910 an associated MUFFLE-WARNING restart.
1911 * The compiler now performs limited argument count validation of
1912 constant format strings in FORMAT, and where appropriate in ERROR,
1913 CERROR and WARN. (thanks to Gerd Moellmann)
1914 * New ASDF-INSTALL contrib can be used for automatic download and
1915 installation of third-party Lisp code from CCLAN or other sites
1917 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1918 which forces another thread to execute a function supplied by the
1920 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1921 now accept and act upon their :ELEMENT-TYPE keyword argument.
1922 (reported by Martin Atzmueller, Edi Weitz)
1923 * bug fix: FILE-POSITION now accepts position designators up to
1924 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1925 is the greater. (thanks to Patrik Nordebo)
1926 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1927 Kalle Olavi Niemitalo)
1928 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1929 printable. (reported by Eric Marsden)
1930 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1931 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1932 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1933 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1934 (thanks to Stig Sandoe)
1935 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1936 UNREAD-CHAR now correctly includes the unread character in the
1937 target sequence. (thanks to Gerd Moellmann)
1938 * bug fix in threaded builds: the system can now be suspended and
1939 resumed by shell job control with minimal disruption.
1940 * bug fixes in times and timezones >2038AD
1941 * better handling of "where is GNU make?" problem in build scripts
1942 (thanks to Nikodemus Siivola)
1943 * new optimization: inside a named function any reference to a
1944 function with the same name is considered to be a self-reference;
1945 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1946 optimization quality.
1947 * new optimization on x86: logical functions and + now have
1948 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1949 used when the result is truncated to 32 bits.
1950 * VALUES declaration is partially enabled.
1951 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1952 * fixed some bugs revealed by Paul Dietz' test suite:
1953 ** The system now obeys the constraint imposed by
1954 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1955 types form a lattice under type intersection.
1956 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1957 ** ASSOC now ignores NIL elements in an alist.
1958 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1959 and (1+ MOST-POSITIVE-FIXNUM) answers.
1960 ** The addition of a method with invalid qualifiers to a generic
1961 function does not cause an error to be signalled immediately;
1962 a warning is signalled, and the error is generated only on
1963 calling the generic function.
1964 * changed .fasl file version number, in part to add type codes for
1965 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1966 obscure ANSI requirements
1968 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1969 * fixed bug 148: failure to inline-expand a local function left
1970 garbage, confusing the compiler.
1971 * fixed bugs 3cd: structure slot readers perform type check if the
1972 slot can have an invalid value (i.e. it is either not initialized
1973 or can be written with a less specific slot writer).
1974 * bug fix: the compiler now traps array references to elements off
1975 the end of an array; previously, the bounds checking in some
1976 circumstances could go off-by-one.
1977 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1979 * type declarations for array element types now obey the description
1980 on the CLHS page "Declaration TYPE", as per discussions on
1981 sbcl-help around 2003-05-08. This means that a declaration
1982 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1983 declaration, all references to BAR will be asserted or assumed
1984 (with THE, so dependent on compiler policy) to involve objects of
1985 type FOO. Note that no such declaration is implied in
1986 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1987 * declared types of functions from the "Conditions"
1988 chapter. (reported by Paul Dietz)
1989 * bug fix: CERROR accepts a function as its first argument.
1990 * bug fix: NTH an NTHCDR accept a bignum as index
1991 arguments. (reported by Adam Warner)
1992 * optimization: character compare routines now optimize comparing
1993 against a constant character. (reported by Gilbert Baumann)
1994 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1995 argument now works properly on the MIPS platform.
1996 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1997 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1998 declared to have a functional type, does not cause an error
2000 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
2001 many :DEFAULT-INITARGS. (reported by Istvan Marko)
2002 * fixed bug: if last continuation of a deleted block has a
2003 destination, this destination should be deleted too. (reported by
2005 * fixed a bug in the bootstrap process: the host compiler's values
2006 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
2007 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
2008 test case from Patrik Nordebo)
2009 * improved the ability of the disassembler on the PPC platform to
2010 provide helpful disassembly notes.
2011 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
2012 the class in more cases than previously.
2013 * bug fix: FILE-POSITION now understands :START and :END for
2014 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
2015 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
2016 8) data. (thanks to David Lichteblau)
2017 * bug fix: it is possible to add a method to a generic function
2018 without lambda list.
2019 * bug fix: reader failed to signal END-OF-FILE inside an
2020 object representation. (reported by Nikodemus Siivola)
2021 * fixed some bugs revealed by Paul Dietz' test suite:
2022 ** LAST and [N]BUTLAST should accept a bignum.
2023 ** condition slot accessors are methods.
2024 ** (VECTOR NIL) is a subtype of STRING.
2026 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2027 * minor incompatible change: some nonsensical specialized lambda
2028 lists (used in DEFMETHOD) which were previously ignored now signal
2030 * minor incompatible change: the system is now aware of the types of
2031 variables in the COMMON-LISP package, and will signal errors for
2032 most violations of these type constraints (where previously they
2033 were silently accepted).
2034 * minor incompatible change: COMPILE-FILE now uses the freedom
2035 afforded (ANSI 3.2.2.3) to use derived function types for
2036 functions defined in the same file. This also permits the system
2037 to warn on static type mismatches and function
2038 redefinition. (Currently it does not work with high DEBUG level.)
2039 * minor incompatible change: VALUES declaration is disabled.
2040 * When issuing notes, the compiler now signals a condition of type
2041 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2042 restart for use in user handlers. It is expected that the
2043 COMPILER-NOTE condition will eventually become a condition
2044 supertype to a hierarchy of note types, which will then be
2045 handleable in a similar fashion. However, at the moment, no such
2046 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2047 but it's an implementation detail, not a classification for the
2049 * Changes in type checking closed the following bugs:
2050 ** type checking of unused values (192b, 194d, 203);
2051 ** template selection based on unsafe type assertions (192c, 236);
2052 ** type checking in branches (194bc).
2053 * A short form of VALUES type specifier has ANSI meaning (it has
2054 increased the number of situations when SBCL cannot perform type
2056 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2057 keywords or constants is permissible.
2058 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2059 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2060 classes. (thanks to Antonio Martinez)
2061 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2062 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2063 argument to a pretty-print stream. Instead, it leaves any such
2064 operations to DESCRIBE-OBJECT methods.
2065 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2067 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2068 lambda lists. (thanks to David Lichteblau)
2069 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2070 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2071 (thanks to Gerd Moellmann)
2072 * fixed bug 246: increased compilation speed of long
2073 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2075 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2077 * DEFCONSTANT now throws a condition of type
2078 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2079 constant to a non-EQL value; CONTINUE and ABORT restarts
2080 respectively change and preserve the value.
2081 * fixed bug 63: The code walker, part of the implementation of CLOS,
2082 is now better at handling symbol macros.
2083 * bug fix: There is no longer an internal implementation type named
2084 CL:LENGTH. (reported by Raymond Toy)
2085 * bug fix: In macro-like defining macros/special operators the
2086 implicit block does not enclose the lambda list.
2087 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2088 enclosed in parentheses) are not suitable as type specifiers, and
2089 their use properly signals an error now.
2090 * bug fix: An argument count mismatch for a type specifier in code
2091 being compiled no longer causes an unhandled error at compile
2092 time, but signals a compile-time warning.
2093 * fixed simple vector readable printing
2094 * bug fix: DESCRIBE takes more care over whether the class
2095 precedence list slot of a class is bound before accessing it.
2096 (reported by Markus Krummenacker)
2097 * bug fix: FORMATTER can successfully compile pretty-printer format
2098 strings which use variants of the ~* directive inside.
2099 * bug fix: SEARCH now applies its TEST predicate to the elements of
2100 the arguments in the correct order. (thanks to Wolfhard Buss)
2101 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2102 in type inference. (thanks to Robert E. Brown)
2103 * bug fix: Reading in symbols with an explicit package name of ""
2104 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2105 * fixed some bugs revealed by Paul Dietz' test suite:
2106 ** NIL is now allowed as a structure slot name.
2107 ** Arbitrary numbers, not just REALs, are allowed in certain
2108 circumstances in LOOP for-as-arithmetic clauses.
2109 ** Multiple class redefinitions before slot access no longer
2110 causes a type error.
2111 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2112 association between the name and a class.
2113 ** Generic functions with non-standard method-combination and over
2114 five methods all of which return constants no longer return NIL
2115 after the first few invocations. (thanks to Gerd Moellmann)
2116 ** CALL-NEXT-METHOD with no arguments now passes the original
2117 values of the arguments, even in the presence of assignment.
2118 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2120 ** DISASSEMBLE works with closures and funcallable instances.
2121 ** ADD-METHOD now returns the generic function, not the new method.
2122 ** FIND-METHOD signals an error if the lengths of the specializers
2123 is incompatible with the generic function, even if the ERRORP
2125 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2126 which its argument is a member.
2127 ** DEFCLASS only redefines the class named by its class-name
2128 argument if that name is the proper name of the class;
2129 otherwise, it creates a new class.
2130 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2131 of the UNBOUND-SLOT condition to the name of the slot.
2132 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2133 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2134 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2135 SLOT-MAKUNBOUND in the specified fashion.
2137 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2138 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2139 cross-compilation host. As a consequence, we can now bootstrap our
2140 way up to SBCL starting with a bare gcc toolchain and human-readable
2141 source code (first the source to CLISP, then the source to SBCL).
2142 * A contributed module containing a partial implementation of the
2143 simple-streams interface has been included. (thanks to Rudi
2145 * A contributed module implementing the RFC1321 Message Digest
2146 Algorithm, known as MD5, has been included.
2147 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2148 no longer has any effect, as the code controlled by this feature
2149 has been deleted. (As far as we know, no-one has ever built using
2150 this feature, and its semantics were confused in any case).
2151 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2152 work on conditions (as required by the ANSI specification),
2153 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2154 expected behaviour on conditions. Users should note, however,
2155 that such behaviour is not required by the ANSI specification,
2156 and so use of this behaviour may render their code unportable.
2157 * fixed some bugs revealed by Paul Dietz' test suite:
2158 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2160 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2161 ** SXHASH on condition objects no longer returns NIL.
2162 ** :ALLOCATION :CLASS slots are better treated; their values are
2163 updated on class redefinition, and initforms inherited from
2164 superclasses are applied.
2165 ** REMOVE-METHOD returns its generic function argument even when
2166 no method was removed.
2167 ** SHARED-INITIALIZE now initializes the values of the requested
2168 slots, including those with :ALLOCATION :CLASS.
2169 ** ALLOCATE-INSTANCE now works on structure classes defined via
2170 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2172 ** SLOT-EXISTS-P now works on conditions, as well as structures
2174 ** MAKE-LOAD-FORM now has the required methods on
2175 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2176 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2177 keyword, and now implements the SLOT-NAMES argument.
2178 ** methods with &OPTIONAL arguments no longer allow too many
2179 arguments to be passed in the call without error.
2180 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2181 option is consistent with the required arguments of the generic
2182 function lambda list.
2183 * bug fix: REQUIRE accepts a string designator. (Thanks to
2185 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2186 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2187 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2188 * bug fix: The compiler now checks for duplicated variables in macro
2190 * bug fix: SETQ on globals returns the correct value.
2191 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2192 not a COMPILER-ERROR (followed by some other strange error on
2193 choosing the CONTINUE restart).
2194 * bug fix: make.sh and friends are now more consistent in the way that
2195 they look for GNU "make".
2197 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2198 * experimental native threads support (on x86 Linux >=2.4 only).
2199 This is not compiled in by default: you need to add :SB-THREAD to
2200 the target features. See the "Beyond ANSI" chapter of the manual
2202 * fix for longstanding nonANSIism: The old distinction between
2203 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2204 The return value from CL:FIND-CLASS is now a CLOS class, and
2205 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2206 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2207 * An interface to the MetaObject Protocol, as described in Kiczales,
2208 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2209 MIT Press, 1991, is available from the SB-MOP package.
2210 * incompatible change: the SB-PCL package should now be considered
2211 a private implementation detail, and no longer a semi-private MOP
2213 * minor incompatible change: due to rearrangement for threads, the
2214 control stack and binding stack are now allocated at arbitrary
2215 addresses instead of being hardcoded per-port. Users affected by
2216 this probably have to be doing advanced things with shared
2217 libraries, and will know who they are.
2218 * minor incompatible change: Previously, all --eval forms used were
2219 processed with READ before any of them were processed with EVAL.
2220 Now each --eval form is processed with both READ and EVAL before
2221 the next --eval form is processed. (Thus package operations like
2222 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2223 work as the user might reasonably expect.)
2224 * minor incompatible change: *STANDARD-INPUT* is now only an
2225 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2227 * minor incompatible change: Y-OR-N-P is now character-oriented,
2228 not line oriented. Also, YES-OR-NO-P now works without errors.
2229 (thanks to Antonio Martinez)
2230 * sb-aclrepl module improvements: an integrated inspector, added
2231 repl features, and a bug fix to :trace command.
2232 * Known functions, which cannot be open coded by the backend, are
2233 considered to be able to check types of their arguments. (fixing
2234 a bug report by Nathan J. Froyd)
2235 * fixed a bug in computing method discriminating functions: It is
2236 now possible to define methods specialized on classes which have
2237 forward-referenced superclasses. (thanks to Gerd Moellmann)
2238 * fixed evaluation order in optional entries (reported by Gilbert
2240 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2241 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2242 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2243 required-by-AMOP NAME argument, as well as CLASS and
2244 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2245 * fixed bug 20: DEFMETHOD can define methods using names that are
2246 not the proper names of classes to designate class specializers.
2247 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2248 Unix file descriptors, instead of blowing up. (thanks to Antonio
2250 * Garbage collection refactoring: user-visible change is that a
2251 call to the GC function during WITHOUT-GCING will not do garbage
2252 collection until the end of the WITHOUT-GCING. If you were doing
2253 this you were probably losing anyway.
2254 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2255 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2256 TYPEP the latter but not the former.
2257 * The compiler issues a full WARNING for calls to undefined functions
2258 with names from the CL package.
2259 * MAP-INTO for a vector destination is open coded. (reported by
2260 Brian Downing on c.l.l)
2261 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2262 documentation string.
2263 * fixed some bugs revealed by Paul Dietz' test suite:
2264 ** COPY-ALIST now signals an error if its argument is a dotted
2266 ** Condition slots are now accessed more correctly in the presence
2267 of multiple initargs for a given slot.
2268 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2269 exclude restarts of the same name associated with a different
2271 ** DEFCLASS of forward-referenced classes with another
2272 forward-referenced class in the superclasses list no longer
2274 ** Condition slots are now initialized once each, not multiple
2275 times. (thanks to Gerd Moellmann)
2276 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2278 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2279 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2280 arguments contain duplicated elements.
2281 ** RESTART-CASE understands local macros.
2282 ** RESTART-CASE associates exactly its own restarts with a condition.
2283 ** ENDP in safe mode checks its argument to be of type LIST.
2284 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2285 in question is unbound.
2286 ** Optimized MAKE-INSTANCE functions no longer cause internal
2287 assertion failures in the presence of duplicate initargs.
2288 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2289 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2290 propagated even in the ctor optimized implementation of
2292 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2294 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2295 * a better implementation of SXHASH on (simple) bit vectors,
2296 measured both in execution speed and in distribution of results
2297 over the positive fixnums, has been installed. Likewise, a better
2298 implementation of EQUAL for simple bit vectors is now available.
2299 * fixed CEILING optimization for a divisor of form 2^k.
2300 * fixed bug 240 (emitting extra style warnings "using the lexical
2301 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2302 by Antonio Martinez)
2303 * fixed SXHASH, giving different results for NIL depending on type
2304 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2305 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2306 variable. (found by Rolf Wester)
2307 * SBCL does not ignore type declarations for special
2308 variables. (reported by rif on c.l.l 2003-03-05)
2309 * some bug fixes in contrib/sb-aclrepl/
2310 * fixed some bugs revealed by Paul Dietz' test suite:
2311 ** a bug in the CONS type specifier, whereby the CAR and CDR
2312 types got intertwined, has been fixed;
2313 ** the type system is now able to reason about the interaction
2314 between INTEGER and RATIO types more completely;
2315 ** APPEND, [N]REVERSE and NRECONC check that those their
2316 arguments, which must be proper lists, are really so;
2317 ** An array specialized to be unable to hold elements has been
2318 implemented, as required -- yes, really -- by ANSI;
2319 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2320 on malformed property lists;
2322 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2323 * incompatible packaging change: in line with Unix convention,
2324 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2325 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2326 * REQUIRE and PROVIDE are now optionally capable of doing something
2327 useful. See the documentation string for REQUIRE.
2328 * infrastructure for a managed SBCL contrib system: contributed
2329 modules in this release include:
2330 ** the ASDF system definition facility;
2331 ** an interface to the BSD Sockets API;
2332 ** an ACL-like convenience interface to the repl;
2333 (thanks to Kevin Rosenberg)
2334 ** an implementation of ROTATE-BYTE, with efficient implementation
2336 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2337 gives a closer approximation to the right answer than previously.
2338 (thanks to Raymond Toy)
2339 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2340 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2341 optional environment arguments, as required by ANSI.
2342 * fixed bugs in other functions taking environment objects, allowing
2343 calls with an explicit NIL environment argument to be compiled
2345 * fixed bug 228: primary return values from
2346 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2347 COMPILE or FUNCTION.
2348 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2349 :TYPE will no longer signal errors on innocuous objects.
2350 * fixed bug 231b: SETQ is better at respecting type declarations in
2351 the lexical environment.
2352 * fixed a bug in DEFCLASS: classes named by symbols with no or
2353 unprintable packages can now be defined.
2354 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2355 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2356 * fixed compiler failure related to checking types of functions.
2357 (reported by Robert E. Brown)
2358 * the compiler is now much more consistent in its error-checking
2359 treatment of bounding index arguments to sequence functions: in
2360 (SAFETY 3) code, errors will be signalled in almost all cases if
2361 invalid sequence bounding indices are passed to functions defined
2362 by ANSI to operate on sequences.
2363 * fixed a bug in the build procedure: documentation of SBCL-specific
2364 packages is now preserved and available in the final Lisp image.
2365 * lifted FDEFINITION lookup out of loops in the implementation of
2366 many list operations. (thanks to Robert E. Brown)
2367 * fixed a bug in the reader: the #n# reader macro now works for
2368 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2369 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2370 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2371 E. Brown and Tony Martinez respectively)
2372 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2373 SLOT-BOUNDP: in particular, it is now able to optimize them much
2374 better, and is now not vulnerable to having packages renamed.
2375 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2376 always called when a slot is not present in an instance. (thanks
2378 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2379 longer signal an error on unreading a character following EOF on
2380 the previous constituent stream. (thanks to Tony Martinez)
2381 * fixed some bugs revealed by Paul Dietz' test suite:
2382 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2383 not just nonnegative fixnums;
2384 ** the logical bit-array operators such as BIT-AND now accept an
2385 explicit NIL for their "opt-arg" argument (to indicate a
2386 freshly-consed result bit-array);
2387 ** ELT now signals an error on an invalid sequence index in safe
2389 ** the type system is now cleverer about negations of numeric
2390 types, and consequently understands the BIGNUM and RATIO types
2392 ** the type system is now cleverer about the interaction between
2393 INTEGER and RATIO types: while bugs still remain, many more
2394 cases are accurately computed;
2395 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2396 if it is in the last clause;
2397 ** CONSTANTLY now correctly returns a side-effect-free function in
2399 ** DECLARE is no longer treated as a special-operator; in
2400 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2401 * incremented fasl file version number due to the change in the
2402 DEFSTRUCT-SLOT-DESCRIPTION structure.
2404 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2405 * minor incompatible change: code processed by the "interpreter" or
2406 EVAL now has a compilation optimization policy of (DEBUG 2)
2407 (changed from (DEBUG 1)) to improve debuggability of interactive
2408 development, and to allow the use of the debug RETURN command in
2410 * an experimental implementation of the RETURN command for the
2411 debugger has been included. (thanks to Frederik Kuivinen)
2412 * fixed bug 62: constraints were not propagated into a loop.
2413 * fixed bug in embedded calls of SORT (reported and investigated by
2415 * fixed some bugs revealed by Paul Dietz' test suite:
2416 ** printing and reading of arrays with some dimensions having
2417 length 0 (thanks to Gerd Moellmann);
2418 ** BOA constructor with &AUX argument without a default value does
2419 not cause a type error;
2420 ** CONSTANTP now returns true for all self-evaluating objects.
2422 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2423 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2424 accessors that are related by inheritance, as specified in the
2425 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2426 Valtteri Vuorikoski)
2427 * The compiler is now able to inline functions that were defined in
2428 a complex lexical environment (e.g. inside a MACROLET).
2429 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2431 * Support for the upcoming FreeBSD-5.0 release has been included.
2432 (thanks to Dag-Erling Smorgrav)
2433 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2434 effect when it is not in a toplevel context.
2435 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2436 stand a better chance of being correct. (thanks to Gerd
2438 * fixed bug in COERCE, which now signals an error on coercing a
2439 rational to a bounded real type which excludes the expected
2441 * The compiler is now able to derive types more accurately from the
2442 COERCE and COMPILE functions.
2443 * fixed bug 223: functional binding is considered to be constant
2444 only for symbols in the CL package.
2445 * fixed bug 231: SETQ did not check the type of a variable being set
2446 (reported by Robert E. Brown)
2447 * A new optimization for MAKE-INSTANCE has been included, fixing
2448 various bugs (including relating to :ALLOCATION :CLASS slots and
2449 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2451 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2452 ** As required by ANSI, LOOP now disallows anonymous collection
2453 clauses such as COLLECT I in conjunction with aggregate boolean
2454 clauses such as THEREIS (= I 1);
2455 ** LOOP now signals an error when any variable is reused in the
2456 same loop (including the potentially useful construct analogous
2457 to WITH A = 1 WITH A = (1+ A);
2458 ** IT is only a special loop symbol within the first clause of a
2459 conditional loop clause;
2460 ** LOOP with a typed iteration variable over a hashtable now
2461 signals a type error iff it should.
2462 * fixed some other bugs revealed by Paul Dietz' test suite:
2463 ** FILE-STREAM now names the class previously known as FD-STREAM;
2464 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2465 argument) no longer signals an error;
2466 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2467 of the slot symbol, rather than using the current package
2468 ((:CONC-NAME "") continues to intern the slot's name in the
2470 * incremented fasl file version number, because of the incompatible
2471 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2472 of the new implementation of DEFINE-COMPILER-MACRO.
2474 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2475 * Support for building SBCL for MIPS platforms running in
2476 little-endian mode has now been checked in, and basic
2477 functionality on said platforms verified.
2478 * minor incompatible change: PCL now records the pathname of a file
2479 in which methods and the like are defined, rather than its
2481 * minor incompatible change: TRUENAME now considers the truename of
2482 a file naming a directory to be the pathname with :DIRECTORY
2483 component indicating that directory.
2484 * minor incompatible change: a NAMED clause in the extended form of
2485 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2486 reason for the previous behaviour is unclear.
2487 * more systematization and improvement of CLOS and MOP conformance
2488 in PCL (thanks to Gerd Moellman and Pierre Mai):
2489 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2491 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2492 primary methods with no specializers;
2493 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2495 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2496 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2497 FORWARD-REFERENCED-CLASSes; error reporting on
2498 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2500 ** SXHASH on CLOS instances now uses a slot internal to the
2501 instance to return different numbers on distinct instances,
2502 while preserving the same return value through invocations of
2504 ** DEFMETHOD signals errors when methods with longer incongruent
2505 lambda lists are added to generic functions;
2506 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2507 CLASS, as specified in AMOP;
2508 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2509 on the order returned by the primary method for classes of
2510 class STANDARD-CLASS;
2511 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2512 * fixed some bugs shown by Paul Dietz' test suite:
2513 ** DOLIST puts its body in TAGBODY;
2514 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2516 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2517 value producing form;
2518 ** if more variables are given to PROGV than values, extra
2519 variables are bound and made to have no value;
2520 ** NSUBSTITUTE on list arguments gets the right answer with
2522 ** ELT signals an error of type TYPE-ERROR when the index argument
2523 is not a valid sequence index;
2524 ** LOOP signals (at macroexpansion time) an error of type
2525 PROGRAM-ERROR when duplicate variable names are found;
2526 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2527 ** FUNCALL of special-operators now cause an error of type
2529 ** PSETQ now works as required in the presence of side-effecting
2530 symbol-macro places;
2531 ** NCONC accepts any object as its last argument;
2532 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2534 ** loop-for-as-package does not require a package to be explicitely
2536 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2538 * fixed bug 166: compiler preserves "there is a way to go"
2539 invariant when deleting code.
2540 * fixed bug 172: macro lambda lists with required arguments after
2541 &REST arguments now cause an error to be signalled. (thanks to
2543 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2544 bug. (thanks to Matthew Danish)
2545 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2547 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2548 arguments are lexically rebound. (thanks to Gerd Moellmann and
2550 * fixed bug 194: error messages are now more informative when there
2551 is no primary method applicable in a call to a generic
2552 function. (thanks to Gerd Moellmann)
2553 * fixed bug in command line argument checking (thanks to Julian
2555 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2556 (thanks to Matthew Danish)
2557 * incremented fasl file version number, because of the
2558 SXHASH-related changes in the layout of CLOS data structures
2560 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2561 * minor incompatible change: The runtime (the Unix executable named
2562 "sbcl") is now much pickier about the .core files it will load.
2563 Essentially it now requires .core files to descend from the same
2564 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2565 as the runtime does. (The intent is to prevent the crashes which
2566 can occur, and which can even be reported as mysterious failures,
2567 when people patch the sources or change the build parameters
2568 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2569 sbcl and .core files.)
2570 * fixed bug: VALUES-LIST is no longer optimized away.
2571 * fixed bug 142: The FFI conversion of C string values to Lisp
2572 string values no longer conses excessively. (thanks to Nathan
2573 Froyd porting Raymond Toy's fix to CMU CL)
2574 * began to systematize and improve MOP conformance in PCL (thanks to
2575 Nathan Froyd, Gerd Moellman and Pierre Mai):
2576 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2578 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2579 ** FINALIZE-INHERITANCE is now called on class finalization;
2580 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2581 argument precedence order.
2582 * fixed bug 202: The compiler no longer fails on functions whose
2583 derived types contradict their declared type.
2584 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2585 so it can be non-toplevel.
2586 * The fasl file version number has changed (because of the new
2587 implementation of DEFMACRO).
2588 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2589 safe code, that any length requirement by their type-specifier
2590 argument is valid. The exceptions to this are described in bug
2592 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2593 functions now check the types of their inputs as required by ANSI.
2594 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2595 introduced symbol is DECLAREd to be SPECIAL.
2596 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2597 * fixed bug: :COUNT argument to sequence functions may be negative
2598 * fixed bug: body of DO-SYMBOLS may contain declarations
2599 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2600 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2601 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2602 symbol macro only once
2603 * fixed printing of call frame when argument list is unavailable
2604 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2605 * compiler no longer signals WARNING on unknown keyword
2608 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2609 * A beta-quality port to the mips architecture running Linux,
2610 based on the old CMUCL backend, has been made. It has been tested
2611 on a big-endian kernel, and works sufficiently well to be able to
2612 rebuild itself; it has not been tested in little-endian mode.
2613 * fixed an inconsistency between gencgc.c and purify.c which made
2614 dumping/loading .core files unreliable
2615 * fixed bug 120a: The compiler now deals correctly with IFs where
2616 the consequent is the same as the alternative, instead of
2617 misderiving the return type. (thanks to Alexey Dejneka)
2618 * fixed bug 113: Logical pathnames are now dumpable (the logical
2619 host is resolved at load-time, throwing an error if it is not
2621 * fixed bug 174: FORMAT's error message is slightly clearer when a
2622 non-printing character is used in a format directive.
2623 * fixed several bugs in compiler checking of type declarations, i.e.
2624 violations of the Python "declarations are assertions" principle
2625 (thanks to Alexey Dejneka)
2626 * fixed several bugs in PCL's error checking (thanks to Gerd
2628 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2630 * fixed bug in compilation of functions as first class values
2631 (thanks to Antonio Martinez-Shotton)
2632 * The compiler's handling TYPE-ERRORs which it can prove will
2633 inevitably happen at runtime has been cleaned up and corrected
2634 in several ways. (thanks to Alexey Dejneka)
2635 * improved argument type checking for various basic arithmetic
2636 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2637 lavished on them in the past that they can be compiled in many
2638 ways in different special cases
2639 * fixed bug 181: compiler checks validity of user supplied type
2641 * cleaned up code flushing in optimization: Function calls which
2642 should signal errors for safety purposes (e.g. which ANSI says
2643 should signal errors when their arguments are of incorrect type)
2644 are no longer optimized away.
2645 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2646 * incremented fasl file version number, because changes in the
2647 implementation of sequence functions like COERCE caused
2648 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2649 in old inline expansions) to become undefined. (Actually these
2650 changes were later undone, so we might very well be binary
2651 compatible with 0.7.7 after all, but leaving the version number
2652 incremented seemed like the simplest and most conservative
2655 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2656 * An alpha-quality port to the parisc architecture running Linux,
2657 based on the old CMUCL backend, has been made. This, even more so
2658 than the other backends, should be considered still a work in
2659 progress; known problems include that the Linux kernel in 64-bit
2660 mode does not propagate the correct sigcontext structure to
2661 userspace, and consequently SBCL on a parisc64 kernel will not
2663 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2664 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2665 suppressed.) Also now that the compiler is looking at declarations
2666 in the environment, it checks optimization declarations as well,
2667 and suppresses inlining when (> DEBUG SPEED).
2668 * More fixes have been made to treatment of floating point exception
2669 treatment and other Unix signals. In particular, floating point
2670 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2671 * The detection and handling of control stack exhaustion (infinite
2672 or very deeply nested recursion) has changed. Stack exhaustion
2673 detection is now done by write-protecting pages at the OS level
2674 and applies at all optimization settings; when found, a
2675 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2676 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2677 be caught using IGNORE-ERRORS.
2678 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2679 correctly when transforming recursive LABELS functions to LETs.
2680 (thanks to Alexey Dejneka)
2681 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2682 that are names of constants or global variables.
2683 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2684 alien routines with docstrings.
2685 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2686 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2688 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2689 (thanks to Eric Marsden)
2690 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2691 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2692 * Bug fix: The SPARC backend can now compile functions involving
2693 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2694 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2695 to a functions expecting a PATHNAME-DESIGNATOR.
2696 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2697 lambda lists. (thanks to Alexey Dejneka)
2698 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2699 functions so that EXPORTing the name of the class doesn't cause
2700 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2701 to Antonio Martinez for reporting this)
2702 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2703 constant-folded without failing an assertion. (thanks to Einar
2704 Floystad Dorum for reporting this)
2705 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2706 types are now dealt with more correctly by the compiler.
2707 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2708 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2709 behaviour analogous to RENAME-FILE. This puts its behaviour more
2710 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2712 * The fasl file version number has changed again. (because of the
2713 bug fix involving the names of PCL MAKE-INSTANCE functions)
2715 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2716 * bug fix: Floating point exceptions are treated much more
2717 consistently on the x86/Linux and PPC/Linux platforms.
2718 * Array initialization with :INITIAL-ELEMENT is now much faster for
2719 cases when the compiler cannot open code the array creation, but
2720 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2721 array accesses have also seen a speed increase.
2722 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2723 specified. (This was at the root of some bad interactions between
2724 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2726 * bug fix: Internal error arguments for undefined functions are now
2727 computed correctly on the PPC/Linux platform.
2728 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2729 Raymond Toy's patch for CMU CL)
2730 * Support for the Solaris 9 operating environment has been included
2731 (thanks to Daniel Merritt)
2732 * A very ugly but hopefully complete draft of the missing FFI chapter
2733 of the manual has been created by reformatting the corresponding
2734 CMU CL manual chapter into (currently very ugly and incoherent)
2735 DocBook and bringing it up to date for SBCL behavior. Thus, the
2736 manual is now essentially complete, at least by my extreme
2737 once-and-only-once standards, whereby it's acceptable to refer to
2738 the doc strings of SB-EXT functions as the primary documentation.
2739 * The fasl file version number has changed again, due to cleanup of
2740 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2741 is no longer a static symbol.)
2743 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2744 * SBCL now builds with OpenMCL (version 0.12) as the
2745 cross-compilation host; also, more progress has been made toward
2746 bootstrapping under CLISP.
2747 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2749 * bug 158 fixed: The compiler can now deal with integer loop
2750 increments different from 1; fixing this turned out also to fix
2752 * bug 169 fixed: no more bogus warnings about using lexical bindings
2753 despite the presence of perfectly good SPECIAL declarations (thanks
2754 to David Lichteblau)
2755 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2756 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2757 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2759 * bug 180 fixed: Method combination specifications no longer ignore
2760 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2761 * bug fix: Structure type predicate functions now check their argument
2762 count as they should.
2763 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2764 correctly. (thanks to Pierre Mai)
2765 * minor incompatible change: The --noprogrammer option is deprecated
2766 in favor of the new --disable-debugger option, which is very similar.
2767 (The major difference is that it takes effect at a slightly different
2768 time at startup, causing handling of errors in --sysinit and
2769 --userinit files will be affected differently.) The
2770 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2771 been added to allow this functionality to be controlled from ordinary
2772 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2773 maintainers, who might want to run non-interactive scripts to
2774 build SBCL cores which will later be used interactively.)
2775 * minor incompatible change: The LOAD function no longer, when given
2776 a wild pathname to load, loads all files matching that pathname.
2777 Instead, an error of type FILE-ERROR is signalled.
2779 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2780 * bug 147 fixed: The compiler preserves its block link/count
2781 invariants more correctly now so that it doesn't crash. (thanks
2783 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2785 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2786 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2787 no-value-for-this-character as NIL (as specified by ANSI).
2788 * HOST-NAMESTRING on physical pathnames now returns a string that is
2789 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2790 (thanks to Christophe Rhodes)
2791 * The Alpha port handles icache flushing more correctly. (thanks to
2793 * More progress has been made toward bootstrapping under CLISP. (thanks
2794 to Christophe Rhodes)
2795 * The fasl file format has changed again, because dynamic loading
2796 on OpenBSD (which has non-ELF object files) motivated some cleanups
2797 in the way that foreign symbols are transformed and passed around.
2798 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2799 is no longer treated as whitespace by the reader, but instead as
2800 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2801 |A
\7fB|, instead of A as it used to.
2803 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2804 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2805 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2806 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2807 this as of 0.7.1.45, but was left out of the previous news section
2808 (thanks to Dan Barlow)
2809 * SBCL now runs on the Solaris operating system on SPARC architectures
2810 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2811 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2812 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2813 * SPARC backend cleanups, allowing builds of cores optimized for V8
2814 and V9 SPARCS, and also emission of code targeted to a particular
2815 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2817 * SBCL is closer to bootstrapping under CLISP, thanks to various
2818 fixes by Christophe Rhodes.
2819 * The fasl file format has changed again, to allow the compiler's
2820 INFO database to support symbol macros.
2821 * The user manual (in doc/) is formatted into HTML more nicely.
2822 (thanks to coreythomas)
2823 * The system is smarter about SUBTYPEP relationships, especially
2824 those involving NOT types (including types such as ATOM which are
2825 represented internally using NOT types). Thus SUBTYPEP is less
2826 likely to return (VALUES NIL NIL) in general, and in particular
2827 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2829 * The fasl file format has changed again, because the internal
2830 representation of types now includes a new slot to support the new
2831 SUBTYPEP-of-NOT-types logic.
2832 * (not a change in the main branch of SBCL, but a related prototype
2833 which can hopefully be merged into the main branch of SBCL in the
2834 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2835 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2836 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2837 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2838 dispatch macro character combinations. (thanks to Alexey Dejneka)
2839 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2840 that superficially look like logical namestrings correctly.
2841 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2843 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2844 default supplied-p) for &optional and &key arguments. (thanks to
2847 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2848 * incompatible change: The compiler is now less aggressive about
2849 tail call optimization, doing it only when (> SPACE DEBUG) or
2850 (> SPEED DEBUG). (This is an incompatible change because there are
2851 programs which relied on the old CMU-CL-style behavior to optimize
2852 away their unbounded recursion which will now die of stack overflow.)
2853 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2854 for non-GENCGC systems has been increased to 20M (since that
2855 seems much closer to the likely performance optimum for modern
2856 systems than the old 4M value was)
2857 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2858 and *DEBUG-PRINT-LEVEL*
2859 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2860 of CMU CL's support for SPARC, and various endianness and other
2861 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2862 * new syntactic sugar for the Unix command line: --load foo.bar is now
2863 an alternate notation for --eval '(load "foo.bar")'.
2865 ** The system now detects stack overflow and handles it gracefully,
2866 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2867 optimization settings. (This is a good thing in general, and
2868 its introduction in this version should be particularly timely
2869 for anyone whose code fails because of suppression of tail
2871 ** The system now hunts for the C variable "environ" in a more
2872 devious way, to avoid segfaults when the C library version differs
2873 between compile time and run time. (thanks to Christophe Rhodes)
2874 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2875 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2876 ** The compiler no longer issues bogus style warnings for undefined
2877 classes in the same source file as the DEFCLASSes which defined
2878 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2880 ** fixes in CONDITION class precedence list for undefined function
2881 errors (thanks to Alexei Dejneka)
2882 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2883 correctly. (thanks to Dan Barlow)
2884 ** portability fixes aiming at bootstrapping under CLISP (thanks
2885 to Dave McDonald and Christophe Rhodes)
2886 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2887 ** fixes in type translation and and type inference (thanks to
2889 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2890 ** various fixes in the new ports (thanks to Dan Barlow)
2891 * several changes related to debugging:
2892 ** suppression of tail recursion, as noted above
2893 ** stack overflow detection, as noted above
2894 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2895 is now the default. (For some time encapsulation has been more
2896 reliable than the breakpoint-based :ENCAPSULATE NIL
2897 implementation, at least on X86 systems; and I just noticed that
2898 encapsulation also seems closer to the spirit of the ANSI
2901 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2903 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2904 up properly again. (There was a packaging bug in 0.7.0 which
2905 left their definitions in SB-SYS::LOAD-FOREIGN and
2906 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2907 vital for most things which interface to C-level interfaces,
2908 like extensions working with sockets or databases or
2909 Perl-compatible regexes or whatever, and the need to fix
2910 this bug is the main reason that 0.7.1 was released so
2912 ** DEFGENERIC is now choosier about the methods it redefines, so that
2913 reLOADing a previously-LOADed file containing DEFGENERICs does
2914 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2915 cycle now works as it should. (thanks to Alexey Dejneka)
2916 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2917 fixed by Christophe Rhodes. (He actually submitted this patch
2918 months ago, and I delayed until after 0.7.0.)
2919 ** Bug 111 (internal compiler confusion about runtime checks on
2920 FUNCTION types) was fixed by Alexey Dejneka.
2921 * Some internal cleanups (getting rid of variables which aren't
2922 needed now that the byte interpreter is gone) caused the fasl
2923 file format number to change again.
2925 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2926 * major incompatible change: The default fasl file extension, i.e. the
2927 default extension for files produced by COMPILE-FILE, has changed
2928 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2930 ** There are many changes in the implementation of the compiler.
2931 SBCL is now essentially a compiler-only implementation of ANSI
2932 Common Lisp. EVAL still "interprets" a few special cases, but
2933 almost all the interesting cases are handled by creating
2934 a LAMBDA expression, calling COMPILE on it, then calling
2935 FUNCALL on the result.
2936 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2937 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2938 Since the code is newer, there might still be some new bugs
2939 (though not as many as before Martin Atzmueller's fixes:-). But
2940 the new code is substantially simpler and clearer, and hopefully
2941 any remaining bugs will be simpler, less fundamental, and more
2942 fixable then the bugs in the old code.
2943 ** The revised compiler is still a little unsteady on its feet.
2945 *** The debugging information it produces (particularly the names
2946 of FUNCTION objects) is sometimes much less useful than what
2947 the old compiler produced.
2948 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2949 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2950 within a MACROLET) has been temporarily weakened.
2951 ** There are new compiler optimizations for various functions:
2952 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2953 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2954 *** the math functions TRUNCATE, FLOOR, and CEILING
2955 *** the function-of-all-trades COERCE
2956 Mostly these should be transparent, but there's one
2957 potentially-annoying problem (bug 117): when the compiler
2958 inline-expands a function and does type analysis on the result,
2959 it can create control paths which have type mismatches, and
2960 when it can't prove that those control paths aren't taken,
2961 it will issue WARNINGs about the type mismatches. This is
2962 a particular problem in practice for the new sequence functions.
2963 It's not clear how this should be fixed, and for now, a
2964 workaround is given in the entry for 117 in the BUGS file.
2965 ** (Because of the interaction between the two previous items --
2966 occasional inlining problems and new inline expansions -- some
2967 of the new sequence function optimizations won't really kick in
2968 completely until debugging information, and then inlining, are
2969 straightened out in some future version.)
2970 * minor incompatible changes:
2971 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2972 DIRECTORY no longer implicitly promotes NIL slots of its
2973 pathname argument to :WILD. In particular, when you ask for the
2974 contents of a directory (which you used to be able to do without
2975 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2976 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2977 ** changes in behavior that ANSI explicitly defines to be
2978 implementation dependent:
2979 *** The new compiler-only implementation still conforms with ANSI,
2980 but acts a little different than before. Besides the obvious
2981 changes in performance tradeoffs (that the cost per form passed
2982 to EVAL has gone up, and the cost per form executed by EVAL
2983 has gone down), the behavior of the system changes a little
2984 because there are no longer any interpreted function objects.
2985 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2986 e.g. doing COMPILE on the output of interactive DEFUN is
2988 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2990 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2991 has changed from (:CL) to NIL.
2992 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2993 e.g. #\Newline and #\Tab, don't have names specified in the
2994 ANSI Common Lisp standard, is now based on their ASCII symbolic
2995 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2996 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2997 are no longer used for output.
2998 ** changes in internal implementation constants:
2999 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
3000 4 million. (If your application spends a lot of time GCing and
3001 you have a lot of RAM, you might want to experiment with
3002 increasing it even more.)
3003 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
3004 However, almost all old code should still continue to work without
3005 immediate update, as SB-C-CALL is now a (deprecated) nickname
3007 ** Old operator names in the style DEF-FOO are now deprecated in
3008 favor of new corresponding names DEFINE-FOO, for consistency with
3009 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
3010 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
3011 internal symbols, but a few supported extensions like
3012 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
3013 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
3014 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
3015 etc. as you get deeper into recursive calls to the debugger
3016 command loop, instead of the old "5]", "5]]", "5]]]"
3017 sequence. (I was motivated to do this when squabbles between
3018 ILISP and SBCL left me very deeply nested in the debugger. In the
3019 short term, this change will probably provoke more ILISP/SBCL
3020 squabbles, but hopefully it will be an improvement in the long run.)
3021 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3022 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3023 function, since declaiming return types involving aliens is
3024 (1) annoyingly messy to do by hand and (2) vital to efficient
3025 compilation of code which calls such functions.
3026 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3027 longer reexported by the SB-EXT package. They're solely useful
3028 for alien code, so it seems more logical that you should get
3029 them from the SB-ALIEN package, not in SB-EXT.
3030 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3031 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3032 features. Instead, the code that they used to control is always
3033 built into the system.
3034 * many other bug fixes
3035 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3036 advantage of the new EVAL-WHEN stuff and to clean them up in
3037 general, and they are now more ANSI-compliant in a number of
3038 ways. Martin Atzmueller is responsible for a lot of this.
3039 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3041 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3042 *** correct ERROR type for various file operations
3043 *** some fixes for Lisp streams
3044 *** DEFMETHOD syntax checking
3045 *** changing old weird representation of debug information as
3046 strings (which, among their other deficiencies, don't transform
3047 correctly when you rename packages, and don't change their
3048 print representation when you change things like *PACKAGE*
3049 and *PRINT-LENGTH*) to symbols and lists of symbols
3050 He also made several improvements and fixed several bugs in DESCRIBE.
3051 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3053 *** misbehavior of WRITE-STRING/WRITE-LINE
3054 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3055 and several other LOOP problems as well
3056 *** DIRECTORY when similar filenames are present
3057 *** DEFGENERIC with :METHOD options
3058 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3059 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3060 *** argument ordering in FIND with :TEST option
3061 *** mishandled package designator argument in APROPOS-LIST
3062 *** various problems in the backquote readmacro
3063 *** a bug in APROPOS
3064 *** probably some others that I'm not describing very well here,
3065 since the CVS log documents them by reference to sbcl-devel
3066 messages, and the SourceForge archives aren't working well.:-(
3067 ** Dan Barlow improved the Alpha port (and is making progress on the
3068 PPC port, for those of you who think different).
3069 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3070 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3071 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3072 continues to work on the SPARC port (for those of you in a position
3073 to look down upon our little PC-compatible boxes from a great height).
3074 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3075 and uses that copy, rather than the current dynamic value, when
3076 it's trying to decide whether to truncate output. Thus e.g.
3077 (let ((*print-lines* 50))
3078 (pprint-logical-block (stream nil)
3080 (let ((*print-lines* 8))
3081 (print (aref possiblybigthings i) stream)))))
3082 should now truncate the logical block only at 50 lines, instead of
3083 often truncating it at 8 lines, as it did before.
3084 * The doc/cmucl/ directory, containing old CMU CL documentation
3085 from the time of the fork, is no longer part of the base system.
3086 SourceForge has shut down its anonymous FTP service, and with it
3087 my original plan for distributing the old CMU CL documentation
3088 there. For now, if you need these files you can download an old
3089 SBCL source release and extract them from it.
3090 * The fasl file version number changed again, for dozens of reasons,
3091 some of which are apparent above.
3093 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3094 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3095 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3096 patch, so that DISASSEMBLE output is much nicer.
3097 * The code in the SB-PROFILE package now seems reasonably stable.
3098 I still haven't decided what the final interface should look like
3099 (I'd like PROFILE to interact cleanly with TRACE, since both
3100 facilities use function encapsulation) but if you have a need
3101 for profiling now, you can probably use it successfully with
3102 the current CMU-CL-style interface.
3103 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3104 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3105 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3106 link now returns the dangling link itself, and for similar
3107 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3108 link itself. (In these cases the old code signalled an error and
3109 looped endlessly, respectively.) Thus, DIRECTORY now works even
3110 in the presence of dangling and cyclic symbolic links.
3111 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3112 is now a supported extension again, since the consensus on
3113 sbcl-devel was that it can be useful for ordinary development
3114 work, not just for debugging SBCL itself.
3115 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3116 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3117 that currently-defined functions might be redefined later with
3118 different return types.
3119 * Hash tables can be printed readably, as inspired by CMU CL code
3120 of Eric Marsden and SBCL code of Martin Atzmueller.
3121 * better error handling in CLOS method combination, thanks to
3122 Martin Atzmueller porting Pierre Mai's CMU CL patches
3123 * more overflow fixes for >16Mbyte I/O buffers
3124 * A bug in READ has been fixed, so that now a single Ctrl-D
3125 character suffices to cause end-of-file on character streams.
3126 In particular, now you only need one Ctrl-D at the command
3127 line (not two) to exit SBCL.
3128 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3130 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3131 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3132 does the right thing.
3133 * The implementation of some type tests, especially for CONDITION
3134 types, is now tidier and maybe faster, due to CMU CL code
3135 originally by Douglas Crosher, ported by Martin Atzmueller.
3136 * Some math functions have been fixed, and there are new
3137 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3138 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3139 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3140 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3141 likely to be built into sbcl-0.7.0.)
3142 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3143 CL, ported to SBCL by Martin Atzmueller)
3144 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3145 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3146 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3147 * There's a new slam.sh hack to shorten the edit/compile/debug
3148 cycle for low-level changes to SBCL itself, and a new
3149 :SB-AFTER-XC-CORE target feature to control the generation of
3150 the after-xc.core file needed by slam.sh.
3151 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3152 COMPILE-FILE is no longer supported, so that now every function
3153 gets an entry point, so that block compilation looks a little
3154 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3155 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3156 returns the number of bytes consed since the system started,
3157 rather than the number consed since the first time the function
3158 was called. (The new definition parallels ANSI functions like
3159 CL:GET-INTERNAL-RUN-TIME.)
3160 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3161 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3162 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3163 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3164 * Fasl file version numbers are now independent of the target CPU,
3165 since historically most system changes which required version
3166 number changes have affected all CPUs equally. Similarly,
3167 the byte fasl file version is now equal to the ordinary
3170 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3171 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3172 is no longer recognized. I apologize for this, because it was
3173 listed in SB-EXT as a supported extension, but I found that
3174 its existing behavior was poorly specified, as well as incorrectly
3175 specified, and it looked like too much of a mess to straighten it
3176 out. I have enough on my hands trying to get ANSI stuff to work..
3177 * many patches ported from CMU CL by Martin Atzmueller, with
3178 half a dozen bug fixes in pretty-printing and the debugger, and
3179 half a dozen others elsewhere
3180 * fixed bug 13: Floating point infinities are now supported again.
3181 They might still be a little bit flaky, but thanks to bug reports
3182 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3183 as flaky as they were.
3184 * The --noprogrammer command line option is now supported. (Its
3185 behavior is slightly different in detail from what the old man
3186 page claimed it would do, but it's still appropriate under the
3187 same circumstances that the man page talks about.)
3188 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3189 are now supported, and enabled by default. Thus, the compiler can
3190 handle many floating point and complex operations much less
3191 inefficiently. (Thus e.g. you can implement a complex FFT
3193 * The compiler now detects type mismatches between DECLAIM FTYPE
3194 and DEFUN better, and implements CHECK-TYPE more correctly, and
3195 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3196 thanks to patches from Martin Atzmueller.
3197 * various fixes to make the cross-compiler more portable to
3198 ANSI-conforming-but-different cross-compilation hosts (notably
3199 Lispworks for Windows, following bug reports from Arthur Lemmens)
3200 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3201 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3202 fixed thanks to Pierre Mai's CMU CL patches.
3203 * fixes to keep the system from overflowing internal counters when
3204 it tries to use i/o buffers larger than 16M bytes
3205 * fixed bug 45a: Various internal functions required to support
3206 complex special functions have been merged from CMU CL sources.
3207 (When I was first setting up SBCL, I misunderstood a compile-time
3208 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3209 * improved support for type intersection and union, fixing bug 12
3210 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3211 more obscure bugs as well
3212 * some steps toward byte-compiling non-performance-critical
3213 parts of the system, courtesy of patches from Martin Atzmueller
3214 * Christophe Rhodes has made some debian packages of sbcl at
3215 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3216 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3217 stable, but are nonetheless usable. When he's ready, I'd be happy
3218 to add them to the SourceForge "File Releases" section. (And if
3219 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3220 * new fasl file format version number (because of changes in
3221 internal representation of (OR ..) types to accommodate the new
3222 support for (AND ..) types, among other things)
3224 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3225 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3227 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3229 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3230 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3231 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3232 * DESCRIBE now works on structure objects again.
3233 * Most function call argument type mismatches are now handled as
3234 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3235 know whether the function will be redefined before the call is
3236 executed. (The compiler could flag local calls with full WARNINGs,
3237 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3238 it doesn't keep track of enough information to know whether calls
3239 are local in this sense.)
3240 * Compiler output is now more verbose, with messages truncated
3241 later than before. (There should be some supported way for users
3242 to override the default verbosity, but I haven't decided how to
3243 provide it yet, so this behavior is still controlled by the internal
3244 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3245 src/compiler/ir1util.lisp.)
3246 * Fasl file format version numbers have increased again, because
3247 support for the Gray streams extension changes the layout of the
3248 system's STREAM objects.
3249 * The Gray subclassable streams extension now works, thanks to a
3250 patch from Martin Atzmueller.
3251 * The full LOAD-FOREIGN extension (not just the primitive
3252 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3253 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3254 but like most other programs, it defaults to copying the Unix
3255 environment from the original process instead of starting the
3256 new process in an empty environment.
3257 * Extensions which manipulate the Unix environment now support
3258 an :ENVIRONMENT keyword option which doesn't smash case or
3259 do other bad things. The CMU-CL-style :ENV option is retained
3260 for porting convenience.
3261 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3262 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3264 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3266 * A patch from Martin Atzmueller seems to have solved the SIGINT
3267 problem, and as far as we know, signal-handling now works cleanly.
3268 (If you find any new bugs, please report them!)
3269 * The system no longer defaults Lisp source file names to types
3270 ".l", ".cl", or ".lsp", but only to ".lisp".
3271 * The compiler no longer uses special default file extensions for
3272 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3273 seems to expect a single default extension for all compiled code,
3274 and there's no compelling reason to try to stretch the standard
3275 to allow two different extensions.) Instead, byte-compiled files
3276 default to the same extension as native-compiled files.
3277 * Fasl file format version numbers have increased again, because
3278 a rearrangement of internal implementation packages made some
3279 dumped symbols in old fasl files unreadable in new cores.
3280 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3281 many fewer weird special cases.
3282 * Bug #17 (differing COMPILE-FILE behavior between logical and
3283 physical pathnames) has been fixed, and some related misbehavior too,
3284 thanks to a patch from Martin Atzmueller.
3285 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3286 by Tim Moore, ported to SBCL by Martin Atzmueller.
3287 * Martin Atzmueller fixed several filesystem-related problems,
3288 including bug #36, in part by porting CMU CL patches, which were
3289 written in part by Paul Werkowski.
3290 * More compiler warnings in src/runtime/ are gone, thanks to
3291 more patches from Martin Atzmueller.
3292 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3295 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3297 * DESCRIBE now works on CONDITION objects.
3298 * The debugger now handles errors which arise when trying to print
3299 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3301 * The build system now uses an additional file, customize-target-features.lisp,
3302 to allow local modifications to the target *FEATURES* list. (The point of
3303 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3304 debugging features enabled, without having to worry about propagating it
3305 into everyone's system when I do a "cvs update".) When no
3306 customize-target-features.lisp file exists, the target *FEATURES* list
3307 should be constructed the same way as before.
3308 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3309 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3310 warn about behavior which is undefined under the ANSI spec. Note: This
3311 is specified by ANSI, but it's not very popular with programmers.
3312 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3313 macro in the SBCL sources for an example of a workaround which you
3314 might use to make portable ANSI-standard code which does what you want.
3315 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3316 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3317 when it's used as a non-toplevel form. (This is required in order
3318 to implement the DEFCONSTANT-EQX macro.)
3319 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3320 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3321 funny things, due to bugs in the implementation of EVAL-WHEN
3322 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3323 non-toplevel DEFCONSTANTs are uncommon.)
3324 * The core file version number and fasl file version number have been
3325 incremented, because the old noncompliant DEFCONSTANT behavior involved
3326 calling functions which no longer exist, and because I also took the
3327 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3328 * fixed bug 1 (error handling before read-eval-print loop starts), and
3329 redid debugger restarts and related debugger commands somewhat while
3331 ** The QUIT debugger command is gone, since it did something
3332 rather different than the SB-EXT:QUIT command, and since it never
3333 worked properly outside the main toplevel read/eval/print loop.
3334 Invoking the new TOPLEVEL restart provides the same functionality.
3335 ** The GO debugger command is also gone, since you can just invoke
3336 the CONTINUE restart directly instead.
3337 ** The TOP debugger command is also gone, since it's redundant with the
3338 FRAME 0 command, and since it interfered with abbreviations for the
3340 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3342 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3343 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3344 some porting work by Martin Atzmueller)
3345 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3346 with a boatload of patches which he ported to SBCL. Now that those
3348 ** The system tries to make sure that its low-priority messages
3349 are prefixed by semicolons, to help people who like to use
3350 syntax highlighting in their ILISP buffer. (This patch
3351 was originally due to Raymond Toy.)
3352 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3353 patches originally written by Raymond Toy.
3354 ** The compiler understands coercion between single-value and
3355 multiple-VALUES type expressions better, getting rid of some very
3356 weird behavior, thanks to patches originally by Robert MacLachlan
3357 and Douglas Crosher.
3358 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3359 lambda lists, thanks to a patch originally by Pierre Mai.
3360 ** The system no longer bogusly warns about "abbreviated type
3362 ** The compiler gets less confused by inlining and RETURN-FROM,
3363 thanks to some patches originally by Tim Moore.
3364 ** The system no longer hangs when dumping circular lists to fasl
3365 files, thanks to a patch originally from Douglas Crosher.
3366 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3367 undefined function error.
3368 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3369 in structure slot initforms) for now, documented workaround instead:-|
3370 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3371 * fixed bug 5: added stubs for various Gray stream functions called
3372 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3373 installed, at least appropriate type errors are generated
3374 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3375 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3376 consistently in DEFMETHOD forms.
3377 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3378 it doesn't seem to affect SBCL after all
3379 * The C runtime system now builds with better optimization and many
3380 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3382 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3384 * The system is now under CVS at SourceForge (instead of the
3385 CVS repository on my home machine).
3386 * The new signal handling code has been tweaked to treat register
3387 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3388 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3389 downstream, e.g. in debug-int.lisp, has implicit dependencies
3390 on the unsignedness of integer representation of machine words,
3391 and that caused the system to bomb out with infinite regress
3392 when trying to recover from type errors involving signed values,
3393 e.g. (BUTLAST '(1 2 3) -1).)
3394 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3395 (This was one of the bugs Peter Van Eynde reported back in July.)
3396 * The system now uses code inspired by Colin Walters' O(N)
3397 implementation of MAP (from the cmucl-imp@cons.org mailing
3398 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3399 inline the MAP operation, and there is more than one
3400 sequence argument to the MAP call (so that it can't just
3401 do ETYPECASE once and for all based on the type of the
3402 single sequence argument). (The old non-inline implementation
3403 of the general M-argument sequence-of-length-N case required
3404 O(M*N*N) time when any of the sequence arguments were LISTs.)
3405 * The QUIT :UNIX-CODE keyword argument has been renamed to
3406 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3407 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3409 * Raymond Wiker's patches to port dynamic loading from Linux to
3410 FreeBSD have been added.
3411 * The BUGS file is now more nearly up to date, thanks in large part
3412 to Martin Atzmueller's review of it.
3413 * The debugger now flushes standard output streams before it begins
3414 its output ("debugger invoked" and so forth).
3415 * The core version number and fasl file version number have both
3416 been incremented, because of incompatible changes in the layout
3418 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3419 used to not be called for a saved Lisp image.)
3420 * Martin Atzmueller's version of a patch to fix a compiler crash,
3421 as posted on sbcl-devel 13 September 2000, has been installed.
3422 * Instead of installing Martin Atzmueller's patch for the
3423 compiler transform for SUBSEQ, I deleted the compiler transform,
3424 and transforms for some similar consing operations.
3425 * A bug in signal handling which kept TRACE from working on OpenBSD
3427 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3428 away by constant folding
3429 * The system now defines its address space constants in one place
3430 (in the Lisp sources), and propagates them automatically elsewhere
3431 (through GENESIS and the sbcl.h file). Therefore, patching the
3432 address map is less unnecessarily tedious and error-prone. The
3433 Lisp names of address space constants have also been systematized.
3434 * CVS tags like dollar-Header-dollar have been removed from
3435 the sources, because they have never saved me trouble and
3436 they've been source of trouble working with patches and other
3437 diff-related operations.
3438 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3439 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3441 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3443 * The system has been ported to OpenBSD.
3444 * The system now compiles with a simple "sh make.sh" on the systems
3445 that it's supported on. I.e., now you no longer need to tweak
3446 text in the target-features.lisp-expr and symlinks in src/runtime/
3447 by hand, the make.sh takes care of it for you.
3448 * The system is no longer so grossly inefficient when compiling code
3449 involving vectors implemented as general (not simple) vectors (VECTOR T),
3450 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3451 which dares to use the various sequence functions on non-simple
3452 vectors, takes less of a performance hit.
3453 * There is now a primitive type predicate VECTOR-T-P
3454 to test for the (VECTOR T) type, so that e.g.
3455 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3456 can now be compiled with some semblance of efficiency. (The old code
3457 turned the type declaration into a full call to %TYPEP at runtime!)
3458 * AREF on (VECTOR T) is still not fast, since it's still compiled
3459 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3460 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3461 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3462 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3463 without making full calls to SUBTYPEP at runtime.
3464 (Some analogous efficiency issues for non-simple vectors specialized to
3465 element types other than T, or for non-simple multidimensional arrays,
3466 have not been addressed. They could almost certainly be handled the
3467 same way if anyone is motivated to do so.)
3468 * The changes in array handling break binary compatibility, so
3469 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3470 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3471 instead of (VALUES T T).
3472 * By following the instructions that Dan Barlow posted to sbcl-devel
3473 on 2 July 2000, I was able to enable primitive dynamic object
3474 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3475 functionality is not implemented (since it calls ld to resolve
3476 library references automatically, requiring RUN-PROGRAM for its
3477 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3478 try to resolve library references) is now supported.
3479 * The system now flushes the standard output streams when it terminates,
3480 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3481 them at several other probably-convenient times, e.g. in each pass of
3482 the toplevel read-eval-print loop, and after evaluating a form given
3483 as an "--eval" command-line option. (These changes were motivated by a
3484 discussion of stream flushing issues on cmucl-imp in August 2000.)
3485 * The source transform for TYPEP of array types no longer assumes
3486 that an array whose element type is a not-yet-defined type
3487 is implemented as an array of T, but instead punts, so that the
3488 type will be interpreted at runtime.
3489 * There is now some support for cross-compiling in make.sh: each of
3490 the phases of make.sh has its own script. (This should be transparent
3491 to people doing ordinary, non-cross-compile builds.)
3492 * Since my laptop doesn't have hundreds of megabytes of memory like
3493 my desktop machine, I became more motivated to do some items on
3494 my to-do list in order to reduce the size of the system a little:
3495 ** Arrange for various needed-only-at-cold-init things to be
3496 uninterned after cold init. To support this, those things have
3497 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3498 symbols with such names are now uninterned after cold init).
3499 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3500 fasl files for cold load.
3501 ** Remove the old compiler structure pooling code (which used to
3502 be conditional on the target feature :SB-ALLOC) completely.
3503 ** Redo the representation of some data in cold init to be more compact.
3504 (I also looked into supporting byte compiled code at bootstrap time,
3505 which would probably reduce the size of the system a lot, but that
3506 looked too complicated, so I punted for now.)
3507 * The maximum signal nesting depth in the src/runtime/ support code has
3508 been reduced from 4096 to 256. (I don't know any reason for the very
3509 large old value. If the new smaller value turns out to break something,
3510 I'll probably just bump it back up.)
3511 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3513 * Many, many bugs reported by Peter Van Eynde have been added to
3514 the BUGS list; some have even been fixed.
3515 * While enabling dynamic object file loading, I tried to make the
3516 code easier to understand, renaming various functions and variables
3517 with less ambiguous names, and changing some function calling
3518 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3519 * While trying to figure out how to do the OpenBSD port, I tried to
3520 clean up some of the code in src/runtime/. In particular, I dropped
3521 support for non-POSIX signal handling, added various comments,
3522 tweaked the code to reduce the number of compilation warnings, and
3523 renamed some files to increase consistency.
3524 * To support the new automatic configuration functionality in make.sh,
3525 the source file target-features.lisp-expr has been replaced with the
3526 source file base-target-features.lisp-expr and the machine-generated
3527 file local-target-features.lisp-expr.
3528 * fixed a stupid quoting error in make.sh so that using CMU CL
3529 "lisp -batch" as cross-compilation host works again
3531 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3533 * DESCRIBE no longer tries to call itself recursively to describe
3534 bound/fbound values, so that it no longer fails on symbols which are
3535 bound to themselves (like keywords, T, and NIL).
3536 * DESCRIBE now works on generic functions.
3537 * The printer now prints less-screwed-up representations of closures
3538 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3539 * A private symbol is used instead of the :EMPTY keyword previously
3540 used to mark empty slots in hash tables. Thus
3541 (DEFVAR *HT* (MAKE-HASH-TABLE))
3542 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3543 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3544 now does what ANSI says that it should. (You can still get
3545 similar noncompliant behavior if bang on the hash table
3546 implementation with all the symbols you get back from
3547 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3548 This breaks binary compatibility, since tests for equality to
3549 :EMPTY are wired into things like the macroexpansion of
3550 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3552 * There's now a minimal placeholder implementation for CL:STEP,
3553 as required by ANSI.
3554 * An obscure bug in the interaction of the normal compiler, the byte
3555 compiler, inlining, and structure predicates has been patched
3556 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3557 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3559 * Missing ordinary arguments in a macro call are now detected even
3560 when the macro lambda list contains &KEY or &REST.
3561 * The debugger no longer complains about encountering the top of the
3562 stack when you type "FRAME 0" to explicitly instruct it to go to
3563 the top of the stack. And it now prints the frame you request even
3564 if it's the current frame (instead of saying "You are here.").
3565 * As specified by ANSI, the system now always prints keywords
3566 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3568 * The default initial SIZE of HASH-TABLEs is now smaller.
3569 * Type information from CLOS class dispatch is now propagated
3570 into DEFMETHOD bodies, so that e.g.
3571 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3573 is now basically equivalent to
3574 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3575 (DECLARE (TYPE SINGLE-FLOAT X))
3577 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3578 operation, without having to do run-time type dispatch.
3579 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3580 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3581 * Since I don't seem to be making big file reorganizations very often
3582 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3583 is overflowing my ability to conveniently back them up), I've finally
3584 checked the system into CVS. (The CVS repository is on my home system,
3585 not at SourceForge -- putting it on SourceForge might come later.)
3586 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3587 high-level GC-NOTIFY-FOO functions send their output. (There's
3588 still very little control of where low-level verbose GC functions
3589 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3590 less than it used to -- the GC-NOTIFY-FOO functions are now under
3591 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3592 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3593 in only one place in the source code, and propagates it automatically
3594 everywhere that it's needed. Thus e.g. when I bump the version from
3595 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3596 * The C source files now include boilerplate legalese and documentation
3597 at the head of each file (just as the Lisp source files already did).
3598 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3599 to his page will be replaced with a link to his new CLiki service.
3601 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3603 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3604 * The build process now looks for GNU make under the default name "gmake",
3605 instead of "make" as it used to. If GNU make is not available as "gmake"
3606 on your system, you can change this default behavior by setting the
3607 GNUMAKE environment variable.
3608 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3609 can build without error under CMU CL.
3611 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3613 * There is now a partial SBCL user manual (with some new text and some
3614 text cribbed from the CMU CL manual).
3615 * The beginnings of a profiler have been added (starting with the
3616 CMU CL profiler and simplifying and cleaning up). Eventually the
3617 main interface should be through the TRACE macro, but for now,
3618 it's still accessed through vaguely CMU-CL-style functions and macros
3619 exported from the package SB-PROFILE.
3620 * Some problems left over from porting CMU CL to the new
3621 cross-compilation bootstrap process have been cleaned up:
3622 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3623 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3624 part when getting rid of the compiler *BACKEND* stuff.)
3625 ** The value of *NULL-TYPE* was screwed up, because it was
3626 being initialized before the type system knew the final
3627 definition of the 'NULL type. This screwed up several key
3628 optimizations in the compiler, causing inefficiency in all sorts
3629 of places. (I found it because I wanted to understand why
3630 GET-INTERNAL-RUN-TIME was consing.)
3631 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3632 PROCLAIM FTYPE information. Unfortunately this broke binary
3633 compatibility again, since now the forms output by DEFGENERIC
3634 to refer to functions which didn't exist in 0.6.3.
3635 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3636 can use the new (as of 0.6.3) transform for SOME into MAP into
3638 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3639 values to (INTEGER 0 1000000), so that the time code will no longer
3640 occasionally get blown up by Linux returning 1000000 microseconds
3641 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3642 its output conform to the ANSI spec. (Alas, this makes its output
3643 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3644 * A full call to MAP NIL with a single sequence argument no longer conses.
3645 * fixes to problems pointed out by Martin Atzmueller:
3646 * The manual page no longer talks about multiprocessing as though
3647 it were currently supported.
3648 * The ILISP support patches have been removed from the distribution,
3649 because as of version 5.10.1, ILISP now supports SBCL without us
3650 having to maintain patches.
3651 * added a modified version of Raymond Toy's recent CMU CL patch for
3652 EQUALP comparison of HASH-TABLE
3654 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3656 * The system still can't cross-compile itself with
3657 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3658 entails), but at least it can compile more of itself that way
3659 than it used to be able to, and various buglets which were uncovered
3660 by trying to cross-compile itself that way have now been fixed.
3661 * This release breaks binary compatibility again. This time
3662 at least I've incremented the FASL file format version to 2, so that the
3663 problem can be detected reliably instead of just causing weird errors.
3664 * various new style warnings:
3665 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3666 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3667 ** using the lexical binding of a variable named in the *FOO* style
3668 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3669 as specified by ANSI.
3670 * *RANDOM-STATE* is no longer automatically initialized from
3671 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3672 default behavior of the system is to repeat its behavior every time
3673 it's run. If you'd like to change this behavior, you can always
3674 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3675 old convention there was no comparably easy way to get the system to
3676 repeat its behavior every time it was run.
3677 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3678 FTYPE declarations has been removed, in favor of their ANSI
3679 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3680 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3681 the types of their sequence arguments can be determined at compile time.
3682 This is done through a new open code expansion for MAP which eliminates
3683 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3684 argument types can be determined at compile time.
3685 * The optimizer now transforms COERCE into an identity operation when it
3686 can prove that the coerced object is already of the correct type. (This
3687 can be a win for machine generated code, including the output of other
3688 optimization transforms, such as the MAP transform above.)
3689 * Credit information has been moved from source file headers into CREDITS.
3690 * Source file headers have been made more standard.
3691 * The CASE macro now compiles without complaining even when it has
3694 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3696 * (Note that the way that the PCL macroexpansions were rewritten
3697 to accommodate the change in DEFGENERIC below breaks binary
3698 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3699 not run under sbcl-0.6.2. Once we get out of alpha releases,
3700 i.e. hit release 1.0.0, we'll probably try to maintain binary
3701 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3702 and sbcl-1.4.4. Until then, however, it might be fairly common
3703 for maintenance releases to break binary compatibility.)
3704 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3706 * The reporting of unhandled signals has been changed to print some
3707 explanatory text as well as the report form. (Previously only
3708 the report form was printed.)
3709 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3710 it defines, so that the compiler no longer issues undefined function
3711 warnings for compiled-but-not-yet-loaded generic functions.
3712 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3713 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3714 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3715 section of the ANSI spec are supported.
3716 * The "" nickname for the "KEYWORD" package has been removed.
3717 The reader still handles symbol tokens which begin with a package marker
3718 as keywords, but it doesn't expose its mechanism for doing so in the
3719 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3720 * The system now issues STYLE-WARNINGs for contradictory TYPE
3721 proclamations. (Warnings for contradictory FTYPE proclamations would
3722 be nice too, but those can't be done usefully unless the type system
3723 is made smarter about FUNCTION types.)
3724 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3725 systematized, so that "*target-*.lisp is supposed to exist only on the
3726 target and imply that there's a related file which exists on the
3727 host, and *host-*.lisp is supposed to exist only on the host and imply
3728 that there's a related file which exists on the target. This involves a
3729 lot of renaming. Hopefully the acute confusion caused by the renaming
3730 will be justified by the reduction in chronic confusion..
3731 ** runtime-type.lisp -> early-target-type.lisp
3732 ** target-type.lisp -> late-target-type.lisp
3733 ** early-host-format.lisp -> early-format.lisp
3734 ** late-host-format.lisp -> late-format.lisp
3735 ** host-error.lisp -> misc-error.lisp
3736 ** early-error.lisp -> early-target-error.lisp
3737 ** late-error.lisp -> late-target-error.lisp
3738 ** host-defboot.lisp -> early-defboot.lisp
3739 ** code/misc.lisp -> code/target-misc.lisp
3740 ** code/host-misc.lisp -> code/misc.lisp
3741 ** code/numbers.lisp -> code/target-numbers.lisp
3742 ** code/early-numbers.lisp -> numbers.lisp
3743 ** early-host-type.lisp -> early-type.lisp
3744 ** late-host-type.lisp -> late-type.lisp
3745 ** host-typep.lisp -> typep.lisp
3746 ** load.lisp -> target-load.lisp
3747 ** host-load.lisp -> load.lisp
3748 ** host-disassem.lisp -> disassem.lisp
3749 ** host-insts.lisp -> insts.lisp
3750 ** byte-comp.lisp -> target-byte-comp.lisp
3751 ** host-byte-comp.lisp -> byte-comp.lisp
3752 ** host-signal.lisp -> signal.lisp
3753 ** host-defstruct.lisp -> defstruct.lisp
3754 ** late-target-type.lisp -> deftypes-for-target.lisp
3755 Furthermore, several other previously target-only files foo.lisp (e.g.
3756 hash-table.lisp and random.lisp) have been split into a target-and-host
3757 foo.lisp file and a target-only target-foo.lisp file, with their key type
3758 definitions in the target-and-host part, so that the cross-compiler will
3759 know more about target types.
3760 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3761 gone away. In their place are various *BACKEND-FOO* variables
3762 corresponding to the slots of the old structure.
3763 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3764 into the target SBCL has been fixed.
3765 * The chill.lisp system for loading cold code into a running SBCL
3767 * Support for the CMU CL "scavenger hook" extension has been removed.
3768 (It was undocumented and unused in the CMU CL sources that SBCL was
3769 derived from, and stale in sbcl-0.6.1.)
3770 * Various errors in the cross-compiler type system were detected
3771 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3772 (enabling various consistency checks). Many of them were fixed,
3773 but some hard problems remain, so the compiler is back to
3774 running without *TYPE-SYSTEM-INITIALIZED* for now.
3775 * As part of the cross-compiler type system cleanup, I implemented
3776 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3777 * I have started adding UNCROSS calls throughout the type system
3778 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3779 on cross-compiler input files will be able to go away, and various
3781 * CONSTANTP now returns true for quoted forms (as explicitly required
3784 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3786 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3787 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3788 (MAKE-LIST -1). In the longer term, it ought to become true
3789 that declarations are assertions even at SAFETY 1. For now, it's not
3790 quite true even at SAFETY 3, but it's at least more nearly true..
3791 (Note that this change seems to increases the size of the system by
3792 O(5%) and to decrease the speed of the compiler by 20% or more.)
3793 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3794 for the problem of printing dozens of lines of distracting information
3795 about low-level system machinery as part of the top stack frame
3796 on entry to the debugger when an undefined function was called.
3797 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3798 is not reset to COMMON-LISP-USER.
3799 * Compilation of stuff related to dyncount.lisp has been made conditional
3800 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3801 smaller. The various dyncount-related symbols have been moved into
3802 a new "SB-DYNCOUNT" package.
3803 * tty-inspect.lisp has been renamed to inspect.lisp.
3804 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3805 feature has gone away. (When we eventually port to other flavors of
3806 libc and/or Unix, we'll try to make the differences between flavors
3807 invisible at the user level.)
3808 * Various other *FEATURES* tags, and/or their associated conditionals,
3809 have been removed if obsolescent, or given better documentation, or
3810 sometimes given more-mnemonic names.
3812 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3814 * tidied up "make.sh" script
3815 * tidied up system directory structure
3816 * better "clean.sh" behavior
3817 * added doc/FOR-CMUCL-DEVELOPERS
3818 * many many small tweaks to output format, e.g. removing possibly-confusing
3819 trailing #\. character in DESCRIBE-INSTANCE
3820 * (EQUALP #\A 'A) no longer signals an error.
3821 * new hashing code, including EQUALP hashing
3822 * tidied up Lisp initialization and toplevel
3823 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3824 * command line argument processing
3825 * added POSIX-GETENV function to deal with Unix-ish environment variables
3826 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3827 terminating SBCL on EOF
3828 * non-verbose GC by default
3829 * There is no more "sbcl" shell script; the sbcl file is now the C
3830 runtime executable (just like CMU CL).
3831 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3833 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3834 symbol and package information as Lisp native symbol and package objects
3835 instead of strings naming symbols and strings naming packages. This way,
3836 whenever packages are renamed (as in warm init), debug information is
3837 transformed along with everything else.
3838 * tweaked the optimization policy declarations which control the building
3839 of SBCL itself. Now, among other things, the system no longer saves
3840 source location debugging information. (This helps two problems at once
3841 by reducing SBCL size and by keeping SBCL from trying to look for its
3842 sources -- which may not exist -- when reporting errors.)
3843 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3844 debugging and testing purposes
3845 * cleaned up printing, making the printer call PRINT-OBJECT for
3846 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3847 methods, giving nearly-ANSI behavior
3848 * converted almost all special variables to use *FOO* naming convention
3849 * deleted PARSE-TIME functionality, since it can be done portably
3850 * moved some files out of cold init into warm init
3851 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3853 * regularized formatting of source files
3854 * added an install.sh script
3855 * fixed ridiculous memory usage of cross-compiler by making
3856 compiler/alloc.lisp not try to do pooling unless it can hook
3857 itself into the GC of the cross-compilation host. Now the system
3858 builds nicely on my old laptop.
3859 * added :SB-ALLOC in target-features.lisp-expr
3860 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3861 was not implemented)
3862 * re-did condition handling and note reporting in the compiler. Notes
3863 are no longer handled by signalling conditions. Style warnings
3864 and warnings are handled more correctly and reported in such a way
3865 that it's easy to find one or the other in your output (so that you
3866 can e.g. figure out which of many problems caused COMPILE-FILE to
3868 * changed the severity of several compiler warnings from full WARNING
3869 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3870 compiler note reporting so that it doesn't use the condition system
3871 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3872 COMPILE-FILE command)
3873 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3874 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3875 no longer supported, since their implementation was deeply intertwingled
3876 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3877 * removed bogus "support" for compiler macros named (SETF FOO), and
3878 removed the compiler macro for SETF INFO (but only after making a fool
3879 of myself on the cmucl-imp mailing list by posting a bogus patch for
3880 DEFINE-COMPILER-MACRO..)
3881 * Compiled files containing forms which have side effects on the Lisp
3882 reader (such as DEFPACKAGE forms) are now handled more correctly.
3883 (Compiler queuing of top level lambdas has been suppressed by setting
3884 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3885 * deleted various currently-unused source files, e.g. gengc.lisp. They
3886 may be added back at some point e.g. when porting to other architectures,
3887 but until they are it's distracting to distribute them and to try to
3889 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3890 there were so many of them they're just distractions, and UNCROSS is
3891 known to be able to handle the current sources
3892 * moved PROFILE functionality into TRACE, so that it will be clear
3893 how the wrapping and unwrapping of functions when you profile them
3894 interacts with the wrapping and unwrapping of functions when you
3895 trace them. (Actually, the functionality isn't there yet, but at least
3896 the interface specification is there. Hopefully, the functionality will
3897 arrive with some future maintenance release.)
3898 * removed host-oops.lisp
3899 * changed signature of QUIT function to allow UNIX-CODE argument
3900 * fixed READ-SEQUENCE bug
3901 * tweaked verbose GC output so that it looks more like the progress
3902 output that ANSI specifies for functions like LOAD
3903 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3904 * added <http://sbcl.sourceforge.com> to the banner information printed by