1 ;;;; -*- coding: utf-8; -*-
2 changes in sbcl-0.9.16 relative to sbcl-0.9.15:
3 * feature: implemented the READER-METHOD-CLASS and
4 WRITER-METHOD-CLASS portion of the Class Initialization Protocol
6 * optimization: faster LOGCOUNT implementation on x86 and x86-64
8 * optimization: hashing of general arrays and vectors has been
9 improved. (reported by Any Fingerhut)
10 * enhancement: SB-INTROSPECT is now able to find definitions of
11 profiled functions. (thanks to Troels Henriksen)
12 * fixed bug #337: use of MAKE-METHOD in method combination now works
13 even in the presence of user-defined method classes. (reported by
14 Bruno Haible and Pascal Costanza)
15 * fixed bug #339(c): if there are applicable methods not part of any
16 long-form method-combination group, call INVALID-METHOD-ERROR.
17 (reported by Bruno Haible)
18 * bug fix: improved the handling of type declarations and the
19 detection of violations for keyword arguments with non-constant
21 * bug fix: potentially erronous calls to PATHNAME and
22 MERGE-PATHNAMES were being flushed in some cases.
23 (reported by Richard Kreuter)
24 * bug fix: compiled calls to TYPEP were mishandling obsolete
25 instances. (reported by James Bielman and Attila Lendvai)
26 * bug fix: format strings with ~> without matching ~< no longer
27 trigger an AVER, but signal an understandable error instead.
28 (reported by Antonio Martinez)
29 * bug fix: specifying an output-file without a pathname-type for
30 COMPILE-FILE or COMPILE-FILE-PATHNAME resulted in using the type
31 of input-file instead of "fasl". (reported by Robert Dodier)
33 changes in sbcl-0.9.15 relative to sbcl-0.9.14:
34 * added support for the ucs-2 external format. (contributed by Ivan
36 * minor incompatible change: pretty printing of objects of type
37 (cons symbol) is, in the default pprint-dispatch-table, now
38 sensitive to whether the symbol satisfies FBOUNDP. (thanks to
40 * minor incompatible change: SB-MOP:FINALIZE-INHERITANCE is now
41 called later in a class's lifetime, possibly as late as when the
42 first instance of the class is created. Previously,
43 SB-MOP:FINALIZE-INHERITANCE was called by the system as soon as a
44 class became finalizeable.
45 * fixed bug: FILE-POSITION sometimes returned inconsistent results
46 for multibyte external-format streams. (thanks to "vbzoli")
47 * fixed bug: CHANGE-CLASS would fail to preserve the values of slots
48 with :ALLOCATION :CLASS inherited from superclasses of the
50 * fixed bug: anonymous classes can now be created using the :NAME
51 initarg and MAKE-INSTANCE / REINITIALIZE-INSTANCE, as specified by
52 AMOP. (reported by Leonid Slobodov on comp.lang.lisp)
53 * fixed bug: core-files saved with :EXECUTABLE T can again be
54 executed when SBCL_HOME isn't set. (reported by James Knight)
55 * fixed bug: toplevel LOCALLY forms with declarations could
56 occasionally get miscompiled. (reported by Yaroslav Kavenchuk)
57 * fixed bug: printing from several different threads using different
58 values of *print-case* could cause invalid output, due to
59 some internal special variables of the printer not being bound
60 thread-locally (reported by Max Mikhanosha)
61 * fixed bug: SPECIALIZER metaobjects (including anonymous classes
62 and EQL-SPECIALIZERs) can be used as specializers to DEFMETHOD.
63 (reported by Pascal Costanza)
64 * fixed bug: FINALIZE-INHERITANCE is called from
65 REINITIALIZE-INSTANCE on classes when the class has previously
66 been finalized, as required by AMOP.
67 * minor code generation optimizations:
68 ** better register allocation in CLOS dispatching functions
69 ** overflow detection when coercing signed bytes to fixnums on x86-64
70 ** is now implemented with one IMUL instruction instead of three shifts
71 ** more efficient bit-vector access on x86 and x86-64
72 ** more efficient access to raw structure slots on x86 and x86-64
73 * fixed some bugs revealed by Paul Dietz' test suite:
74 ** ENSURE-DIRECTORIES-EXIST.8: ENSURE-DIRECTORIES-EXIST must
77 changes in sbcl-0.9.14 relative to sbcl-0.9.13:
78 * feature: thread support on Solaris/x86, and experimental thread support
80 * feature: SBCL now tries to signal a STORAGE-CONDITION when running out
82 * feature: SBCL now provides USE-VALUE and STORE-VALUE restarts in the
83 default method for SLOT-UNBOUND.
84 * minor incompatible change: prevent the user from specializing the
85 new-value argument to SB-MOP:SLOT-VALUE-USING-CLASS. It's
86 somewhat counter to the intent of the protocol, I (CSR) think, and
87 additionally it just doesn't work in SBCL as currently
88 implemented, thanks to optimizations (that are always valid for
89 the other three SLOT-VALUEish functions, but not for the setter).
90 * minor incompatible change: the :SB-LDB feature is now enabled by
91 default, and DISABLE-DEBUGGER and ENABLE-DEBUGGER also affect
92 the low-level debugger.
93 * enchancement: RUN-PROGRAM is now able to extract the file-descriptor
94 from SYNONYM-STREAM and TWO-WAY-STEAMS provided they can be decomposed
95 down to an SB-SYS:FD-STREAM, allowing direct communication in
96 more cases. Temporary files and pipes are still used as a fallback
98 * thread-safety: DECODE-UNIVERSAL-TIME & GET-DECODED-TIME depended
99 on the non-reentrant C functions localtime(3) and gmtime(3).
100 * bug fix: global optimization policy was not visible in LOCALLY and
102 * bug fix: class objects can be used as specializers in methods.
103 (reported by Pascal Costanza)
104 * bug fix: native unparsing of pathnames with :DIRECTORY NIL failed
105 with a type error. (reported by blitz_ on #lisp)
106 * bug fix: unparsing logical pathnames with :NAME :WILD :TYPE NIL
107 failed with a type error. (reported by Pascal Bourguignon)
108 * bug fix: merging pathnames against defaults with :DIRECTORY
109 starting with '(:RELATIVE :BACK) should preserve the :BACK.
110 (reported by James Y Knight)
111 * bug fix: saving large (>2GB) cores on x86-64 now works
112 * bug fix: a x86-64 backend bug when compiling (setf aref) with a
113 constant index and a (simple-array (signed-byte 32)) array
114 * bug fix: NAME-CHAR on an invalid symbol no longer signals an
115 error (patch by Robert J. Macomber)
116 * feature: TIME now displays the amount of run-time spent in GC
117 * bug fix: The debugger now does a better job of respecting
118 (PUSH '(*PRINT-CIRCLE* . T) SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*)
119 when printing SB-DEBUG:*DEBUG-CONDITION*. (This is a debugger-only
120 workaround for bug 403.)
121 * bug fix: floating point exception handling now works on FreeBSD
122 (thanks to NIIMI Satoshi)
123 * fixed some bugs revealed by Paul Dietz' test suite:
124 ** MISC.641: LET-conversion were not supposed to work in late
126 * improvements to the Win32/x86 port:
127 ** GET-INTERNAL-RUN-TIME implemented, thanks to Frank Buss.
128 ** improved SB-BSD-SOCKETS support, thanks to Timothy Ritchey.
130 changes in sbcl-0.9.13 relative to sbcl-0.9.12:
131 * new feature: source path information is generated for macro-expansion
132 errors for use in IDE's like Slime (thanks to Helmut Eller)
133 * bug fix: calls to the compiler no longer modify *RANDOM-STATE*
134 * bug fix: compiler does not loop forever on an invalid type in
136 * improvement: compilation of most CLOS applications is significantly
138 * optimization: added a limited bytecode compiler for simple toplevel
139 forms, speeding up compilation and FASL loading
141 changes in sbcl-0.9.12 relative to sbcl-0.9.11:
142 * minor incompatible change: in sbcl-0.9.11 (but not earlier
143 versions) SB-MOP:FINALIZE-INHERITANCE would recursively descend
144 into subclasses of the finalized class. Now user calls to
145 FINALIZE-INHERITANCE finalize just the one class, and calls by the
146 system return before any subclasses are finalized.
147 * minor incompatible change: The reader no longer ignores errors
148 regarding non-existent packages in #+ and #- feature tests.
149 * new feature: command line options --no-sysinit, --no-userinit to
150 inhibit loading the corresponding init files
151 * new feature: add a generic function SOCKET-SEND to SB-BSD-SOCKETS,
152 for sending data through UDP sockets (thanks to François-René Rideau)
153 * minor incompatible change: SIGPIPE is ignored and "Broken pipe"
154 error is signalled instead (thanks to François-René Rideau)
155 * minor incompatible change: Error signalling behaviour of lexical
156 operations violating package locks has changed slightly. Refer to
157 documentation on package locks for details.
158 * enhancement: EVAL can process IF-expressions without resorting to the
160 * bug fix: Some operations involving SIN, COS, or TAN weren't being
161 constant-folded properly due to missing out-of-line functions.
162 (reported by Mika Pihlajamäki)
163 * bug fix: LISTEN sometimes returned T even in cases where no data was
164 immediately available from the stream
165 * fixed bug: types of the last two arguments to SET-SYNTAX-FROM-CHAR
166 were reversed. (reported by Levente Mészáros)
167 * fixed bug: Tests for the (VECTOR T) type gave the wrong answer
168 when given a vector displaced to an adjustable array. (reported
170 * bug fix: derivation of float boundaries from numbers outside the
171 appropriate float range (reported by John Wiseman)
172 * bug fix: MAKE-LOAD-FORM-SAVING-SLOTS accepts en empty slot name
174 * bug fix: precomputing cache entries for generic functions with
175 some subclasses of specializers as yet invalid does not attempt to
176 fill a cache line with a negative offset. (reported by Levente
178 * improvements to DOCUMENTATION for TYPE and STRUCTURE doc-types:
179 allow condition class objects as arguments to DOCUMENTATION and
180 (SETF DOCUMENTATION); only find and set documentation for
181 structure names for the STRUCTURE doc-type. (suggested by Gary
183 * improvements to the Win32/x86 port:
184 ** (user-homedir-pathname) and default initialization file
185 locations now know about the user's "Documents and Settings"
186 directory (thanks to Yaroslav Kavenchuk)
187 ** run-program is implemented (thanks to Mike Thomas)
188 ** sockets support (thanks to Timothy Ritchey)
189 ** better backtrace support (thanks to Alastair Bridgewater)
190 ** sb-grovel supported
191 ** asdf-install and sb-posix work somewhat
192 ** capable of running Slime using SWANK:*COMMUNICATION-STYLE* NIL
193 * improvements to the Solaris/x86 port:
194 ** works on Solaris 11/Solaris Express
195 ** floating-point exception handling support
196 ** support for the breakpoint-based TRACE :ENCAPSULATE NIL facility
197 * fixed some bugs revealed by Paul Dietz' test suite:
198 ** REMOVE-METHOD returns its generic function argument even when
199 the method is not one of the generic functions' methods.
200 ** objects declared as MEMBER types can be admissible arguments to
202 ** printing characters should simply be printed by the FORMAT ~:C
204 ** compiler failure when compiling functions with hairy constant
205 defaults for optional parameters.
206 ** compiler produces wrong code when MAYBE-INLINE-expanding a
207 function, which is already optimized.
209 changes in sbcl-0.9.11 relative to sbcl-0.9.10:
210 * new platform: experimental support for SBCL x86/Darwin, including
211 MacOS X 10.4.5 on Intel.
212 * new feature: Unicode character names are now known to the system
213 (through CHAR-NAME and NAME-CHAR).
214 * new feature: the filesystem location of SBCL's core file is
215 exposed to lisp through the variable SB-EXT:*CORE-PATHNAME*.
216 * minor incompatible change: the contrib modules SB-POSIX and
217 SB-BSD-SOCKETS no longer depend on stub C libraries; the intent of
218 this change is to make it easier to distribute
219 location-independent binaries.
220 * bug fix: as implied by AMOP, standardized classes no longer have
221 slots named by external symbols of public packages. (reported by
223 * optimization: calling functions via constant symbols -- as in
224 (FUNCALL 'FOO) -- is now roughly as efficient as calling them
225 via the function object as in (FUNCALL #'FOO).
226 * optimization: on x86 and x86-64, the calling convention for the
227 case of dealing with an unknown number of values has been altered
228 to be friendlier to the prediction heuristics implemented,
229 particularly with reference to CALL and RET pairing. (thanks to
230 Alastair Bridgewater)
231 * optimization: on x86, the code for access to thread-local symbol
232 values and binding/unbinding of thread-local symbols is smaller.
233 (thanks to Alastair Bridgewater)
234 * enhancement: CONSTANTP is now able to determine constantness of
235 more complex forms, including calls to constant-foldable standardized
236 functions and some special forms beyond QUOTE.
237 * fixed bug: occasional GC crashes on Solaris/x86
238 * optimization: x86-64 supports stack allocation of results of simple
239 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT
240 * enchancement: the PROCESS-INPUT and -OUTPUT streams created by
241 SB-EXT:RUN-PROGRAM can be used for both character and byte IO
242 (thanks to James Knight)
243 * fixed bug: CL:LISTEN always returns NIL at end of file, as required
244 by the standard (thanks to Stephen Compall)
246 changes in sbcl-0.9.10 relative to sbcl-0.9.9:
247 * new feature: new SAVE-LISP-AND-DIE keyword argument :EXECUTABLE can
248 be used for bundling the runtime and the core file into one
249 executable binary. This feature is not currently supported on all SBCL
250 platforms. (thanks to James Bielman and NIIMI Satoshi)
251 * new feature: a generational or ephemeral garbage collector is now
252 the default on the PowerPC platform (both Linux and Darwin). The
253 old Cheney (stop and copy) collector is a build-time option.
254 (thanks to Cyrus Harmon, after Raymond Toy)
255 * minor incompatible change: the method by which SBCL finds its
256 contributed modules has changed; it no longer relies on symbolic
257 links from an $SBCL_HOME/systems directory, but searches directly
258 in the subdirectories of $SBCL_HOME.
259 * enhancement: the dynamic heap size on the Linux/PPC platform is
260 markedly larger, even using the older Cheney garbage collector.
261 * fixed bug #399: full call to DATA-VECTOR-REF in accesses to
262 certain complicated string types. (reported by Gary King)
263 * fixed bug: STRING-TO-OCTETS and OCTETS-TO-STRING did not convert
264 character codes. (thanks to Yaroslav Kavenchuk and Ivan Boldyrev)
265 * fixed bug: DOCUMENTATION on structures no longer signals an
266 error if no structure type of the right name exists. (reported by
268 * fixed bug: compiled calls to MAKE-INSTANCE where there is an
269 applicable non-standard primary or :AROUND method on
270 INITIALIZE-INSTANCE or SHARED-INITIALIZE and a non-keyword initarg
271 no longer cause unbound variable errors. (reported by Kevin Reid)
272 * fixed bug: metacircle resolution in cases where methods have slots
273 added before the slots from STANDARD-METHOD. (reported by Jean
275 * fixed bug: the Power PC assembly code for calling into lisp
276 sometimes computed its return address wrongly (depending on the
277 15th bit of the address, fixed at link time). This used to
278 manifest itself in a segmentation violation while building PCL.
279 (reported by Kevin Rosenberg, Eric Marsden, Lars Brinkhoff and
280 many others over the years)
281 * fixed bug: ROOM no longer reports silly numbers for stack usage if
282 the stack pointer (treated as a signed integer) is negative.
283 (thanks to Peter van Eynde)
285 changes in sbcl-0.9.9 relative to sbcl-0.9.8:
286 * new platform: experimental support for the Windows operating
287 system has been added. (thanks to Alastair Bridgewater)
288 * fixed several bugs in and robustified the PPC FFI (including
289 callbacks). (thanks to Cyrus Harmon and Heiner Schwarte)
290 * bug fix: allow non-simple string symbol names (reported by Paul
292 * bug fix: interrupt handling on NetBSD (thanks to Richard M
294 * bug fix: saving a core corrupted callbacks on x86/x86-64
295 * bug fix: closed a loophole in metacircularity detection and
296 grounding in the PCL implementation of CLOS.
297 * bug fix: TRUENAME on "/" no longer returns a relative pathname.
298 (reported by tomppa on #lisp)
299 * bug fix: clear the direction flag on Lisp -> C transitions, as
300 required by the x86-64 ABI. Fixes mysterious GC crashes on SuSE.
301 (reported by Andrej Grozin and Hendrik Maryns)
302 * optimization: major improvements to GC efficiency on GENCGC platforms
303 * optimization: faster implementation of EQUAL
304 * optimization: emit more efficient opcodes for some common
305 immediate->register MOV instructions on x86-64. (thanks to Lutz Euler)
306 * optimization: several other minor code-generation improvements on x86-64
307 * fixed segfaults on x86 FreeBSD 7-current. (thanks to NIIMI Satoshi)
309 changes in sbcl-0.9.8 relative to sbcl-0.9.7:
310 * minor incompatible change: (SETF CLASS-NAME) and (SETF
311 GENERIC-FUNCTION-NAME) are generic functions once more (reverting
312 the change in the 0.9.7 release). (SETF CLASS-NAME) is specified
313 by ANSI as a generic function, and for consistency (SETF
314 GENERIC-FUNCTION-NAME) is treated likewise.
315 * fixed bug #233.b: make constraint propagation notice when a variable
316 value is changed after it is referenced but before it is used
317 * fixed bug #296: no more arbitrary behaviour from filenames with
318 odd characters as --load arguments. (reported by Adam Warner)
319 * fixed bug #390: :CHARACTER-SET pathname components now work as
320 expected. (reported by Tim Daly Jr)
321 * fixed bug #391: complicated :TYPE intersections in slot
322 definitions no longer cause an error in PCL internals.
323 * fixed bug #392: plugged a hole in the obsolete-instance protocol.
324 * bug fix: FILE-STRING-LENGTH is now external-format sensitive,
325 returning the number of octets which would be written to the
326 file-stream. (thanks to Robert J. Macomber)
327 * bug fix: the SB-SIMPLE-STREAMS contrib passes external-format
328 arguments to the internal stream functions. (thanks to David
330 * bug fix: stack exhaustion is now handled on the x86/SunOS
331 platform. (thanks to Daisuke Homma)
332 * bug fix: more accurate ROOM results on GENCGC platforms
333 * optimization: improved type inference for arithmetic-for
334 index variables in LOOP
335 * optimization: faster floating-point SQRT on x86-64
336 * fixed some bugs revealed by Paul Dietz' test suite:
337 ** DOCUMENTATION returns NIL instead of "" for method combinations
338 that don't have a docstring
340 changes in sbcl-0.9.7 relative to sbcl-0.9.6:
341 * minor incompatible change: (SETF CLASS-NAME) and (SETF
342 GENERIC-FUNCTION-NAME) are no longer generic functions, and
343 therefore are not customizeable by user code (as seems to be at
344 least permitted and maybe required by AMOP). As a consolation,
345 however, the SBCL implementation of these functions now calls
346 REINITIALIZE-INSTANCE as specified by AMOP.
347 * bug fix: slot-definition documentation is propagated as per ANSI
348 7.6.2 to EFFECTIVE-SLOT-DEFINITIONS.
349 COMPUTE-EFFECTIVE-SLOT-DEFINITION now receives a :DOCUMENTATION
350 argument, as do eslotd initialization methods. (from Pascal
351 Costanza's "Closer" project)
352 * bug fix: REINITIALIZE-INSTANCE on generic functions calls
353 COMPUTE-DISCRIMINATING-FUNCTION (almost) unconditionally, as
355 * bug fix: it is now possible to have more than one subclass of
356 STANDARD-GENERIC-FUNCTION without causing stack overflow.
357 (reported by Bruno Haible, Pascal Costanza and others)
358 * bug fix: the dependent update protocol now works for generic
359 functions. (thanks to Gerd Moellmann; reported by Bruno Haible
361 * bug fix: condition-class instances corresponding to
362 DEFINE-CONDITION forms are now created eagerly. (reported by
363 Kalle Olavi Niemitalo on comp.lang.lisp)
364 * bug fix: floating point printing is more accurate in some
365 circumstances. (thanks to Simon Alexander)
366 * bug fix: *COMPILE-FILE-PATHNAME* now contains the user's pathname
367 merged with *DEFAULT-PATHNAME-DEFAULTS*.
368 * bug fix: callbacks on OS X now preserve stack-alignment.
369 (thanks to Cyrus Harmon)
370 * enhancement: the x86-64 disassembler is much better at
371 disassembling SSE instructions. (thanks to Lutz Euler)
372 * enhancement: CHECK-TYPE on a variable now allows type-inference
373 (thanks to Vincent Arkesteijn)
374 * optimization: improved performance of EUC-JP external format.
375 (thanks to NIIMI Satoshi)
376 * optimization: performance improvements to IO on file streams of
377 :ELEMENT-TYPE CHARACTER
378 * optimization: much faster memory allocation on x86-64
379 * optimization: faster garbage collections (latency and throughput)
381 * optimization: faster bignum subtraction and division on x86 and x86-64
382 (thanks to Lutz Euler)
383 * optimization: more accurate type-inference for loops that use a
384 floating point index variable or a negative step.
386 changes in sbcl-0.9.6 relative to sbcl-0.9.5:
387 * bug fix: add a workaround to SBCL looping infinitely at startup on
388 Linux kernels with apparently buggy implementations of personality().
389 (thanks to Svein Ove Aas)
390 * bug fix: Unicode symbols are correctly printed in LDB backtraces
391 (thanks to David Lichteblau)
392 * bug fix: local bindings shadow global symbol macros for
393 macroexpansion. (reported by Matthew D Swank on comp.lang.lisp)
394 * optimization: non-open coded uses of numeric comparison operators
395 (e.g. >) no longer cons when called with more than one parameter
396 on platforms supporting dynamic-extent allocation.
397 * enhancement: saving cores with foreign code loaded is now supported
398 on MIPS/Linux in addition to the previously supported platforms.
399 * bug fix: FIND-RESTART now tests for activity, not applicability when given
400 a restart object as identifier. (reported by Helmut Eller for CMUCL)
401 * bug fix: division by zero in sb-sprof when no samples were collected
402 * bug fix: a race when a slow to arrive sigprof signal killed sbcl
403 * bug fix: asdf-install uses CRLF as required by the HTTP spec.
404 (thanks to Alexander Kjeldaas)
405 * new feature: ignoring the return values of destructive functions
406 like NREVERSE signals a compile-time style-warning.
407 (thanks to Kevin Reid)
408 * new feature: the alignment of alien structure fields can be
409 explicitly specified. (thanks to Cyrus Harmon)
410 * bug fix: run cleanup forms (in all threads) when receiving a SIGTERM
411 and dump core on SIGQUIT
413 ** incompatible change: threads do not inherit values of specials
414 from their parents (see manual)
415 ** bug fix: threads stacks belonging to dead threads are freed by the
416 next exiting thread, no need to gc to collect thread stacks anymore
417 ** minor incompatible change: INTERRUPT-THREAD-ERROR-ERRNO removed
418 ** WITH-RECURSIVE-LOCK can be nested in a WITH-MUTEX for the same lock
419 ** bug fix: dynamic variable and thread start related gc lossage
420 ** bug fix: job control (SIGSTOP/SIGCONT) no longer confuses threads
422 * fixed some bugs revealed by Paul Dietz' test suite:
423 ** SUBTYPEP is slightly more accurate on heinously complicated
424 CONS types where some of the members have uncertain (in the
425 NIL, NIL sense) type relationships to each other.
427 ** Cores produced by SAVE-LISP-AND-DIE on GENCGC platforms are
428 no longer purified unless :PURIFY T is explicitly specified.
429 ** Non-purified cores are significantly smaller than before
431 changes in sbcl-0.9.5 relative to sbcl-0.9.4:
432 * new feature: timers based on Zach Beane's excellent timer package
433 * added support for the following external formats: koi8-u,
434 x-mac-cyrillic, cp437, cp850, cp852, cp855, cp857, cp860, cp861,
435 cp862, cp863, cp864, cp865, cp866, cp869, cp874, iso-8859-2,
436 iso-8859-3, iso-8859-4, iso-8859-5, iso-8859-6, iso-8859-7,
437 iso-8859-8, iso-8859-9, iso-8859-10, iso-8859-11, iso-8859-13,
438 iso-8859-14, cp1250, cp1251, cp1252, cp1253, cp1254,
439 cp1255,cp1256, cp1257, cp1258 (contributed by Ivan Boldyrev)
440 * incompatible change: a threaded SBCL will no longer revert to
441 non-threaded mode on non-NPTL systems, but refuse to start entirely.
442 * bug fix: interrupts are disabled until startup is complete; no
443 more sigsegvs when receiving a signal to soon
444 * optimization: faster 32-bit SB-ROTATE-BYTE:ROTATE-BYTE on non-x86/ppc
446 * bug fix: add a workaround for the memory randomization features in
447 Linux kernels >= 2.6.12 that interfere with SBCL's memory maps. This
448 workaround will only be in effect on systems with the proc filesystem
450 * bug fix: printing objects of type HASH-TABLE signals a
451 PRINT-NOT-READABLE error when *READ-EVAL* is NIL. (reported by
453 * bug fix: GET-INTERNAL-REAL-TIME now works even for processes that
454 have been running for over 50 days. (reported by Gilbert Baumann)
455 * bug fix: the logic for getting names of functions gets less
456 confused when confronded with alternate-metaclass
457 funcallable-instances. (reported by Cyrus Harmon)
458 * bug fix: FUNCTIONP and (LAMBDA (X) (TYPEP X 'FUNCTION)) are now
459 consistent, even on internal alternate-metaclass objects.
460 * bug fix: flush closure information collected by physical
461 environment analysis prepass before the main pass. (bug reported
463 * bug fix: compiler pack phase does not modify a hashtable, which is
464 iterated. (reported by Bryan O'Connor, fixed by Rob MacLachlan)
465 * bug fix: backquote does not optimize (LIST* x y) to CONS when x
466 may be expanded. (reported by Alexander <malishev> on c.l.l)
467 * bug fix: no more style-warnings when compiling code using LAST
468 under high-SPEED low-SPACE settings. (reported by David Wallin
470 * fixed some incompatibilities between SBCL's MOP and the MOP
472 ** the METAOBJECT class is now implemented;
473 ** FUNCALLABLE-STANDARD-OBJECT is now a subclass of
474 STANDARD-OBJECT, as required;
475 ** the classes STANDARD-CLASS and FUNCALLABLE-STANDARD-CLASS are
476 now compatible (as required by VALIDATE-SUPERCLASS); there
477 remains a consistency requirement over the presence or absence
478 of the FUNCTION class in the superclass list of finalized
479 classes; see the manual for more details;
480 ** the :AROUND method for COMPUTE-SLOTS on
481 FUNCALLABLE-STANDARD-CLASS now honours the primary method's
482 requested slot ordering.
484 ** bug fix: parent thread now can be gc'ed even with a live
486 ** bug fix: binding a special with PROGV to no value is not
487 the same as that symbol not having been bound (thanks to
489 ** bug fix: binding specials is thread safe (thanks to
491 ** bug fix: interrupt handlers are now per-process, RUN-PROGRAM
492 and SB-SPROF do not die with 'no handler for signal XX in
493 interrupt_handle_now(..)' anymore
494 ** bug fix: WITH-TIMEOUT works with multiple running threads
495 * fixed some bugs revealed by Paul Dietz' test suite:
496 ** ENSURE-GENERIC-FUNCTION should take a method class object for
497 the :method-class keyword argument.
499 changes in sbcl-0.9.4 relative to sbcl-0.9.3:
500 * new port: the Solaris operating system on x86 processors is now
501 mostly supported, though some rough edges in the environment
502 remain. (thanks to Daisuke Homma)
503 * enhancement: SBCL on MIPS platforms now has a much larger dynamic
504 space for its heap. (thanks to Thiemo Seufer)
505 * optimization: SBCL on MIPS platforms now supports dynamic-extent
506 closures. (thanks to Thiemo Seufer)
507 * minor incompatible change: eof selects abort in the debugger.
508 * minor incompatible change: *INVOKE-DEBUGGER-HOOK* is run before
509 *DEBUGGER-HOOK* => *DEBUGGER-HOOK* is not run when the debugger
511 * minor incompatible change: SB-KERNEL:INSTANCE-LAMBDA is
512 deprecated, and will go away in a future revision of SBCL.
513 * minor incompatible change: GC-ON and GC-OFF are no longer
514 implemented with a counter, it does not matter how many times gc
515 is switched on or off
516 * bug fix: discriminating functions for generic function classes
517 with non-standard methods for COMPUTE-APPLICABLE-METHODS no longer
518 make invalid assumptions about method precedence order. (reported
520 * bug fix: TRUNCATE now correctly signals division by zero on MIPS
521 platforms. (thanks to Thiemo Seufer)
522 * bug fix: degree sign (<U00B0>) could not be encoded in KOI8-R.
523 * bug fix: correct pathname printing with printer escaping is on.
524 (thanks to Kevin Reid)
525 * bug fix: complex VOP definitions in "user-space" no longer trigger
526 package locks. (reported by Zach Beane)
527 * fixed bug 343: SB-KERNEL:INSTANCE-LAMBDA is no longer necessary
528 for funcallable-instance functions, and is no different from
530 * bug fix: PARSE-INTEGER no longer depends on the whitespaceness of
531 characters in the current readtable. (reported by Nicholas Neuss)
532 * bug fix: SUBTYPEP on various CONS types returns more a more
533 accurate acknowledgment of its certainty.
534 * optimizations: REMOVE-DUPLICATES now runs in linear time on
535 lists in some cases. This partially fixes bug 384.
536 * flush all standard streams before prompting in the REPL and the
538 * bug fix: signal handling and triggering gc do not conflict
539 directly anymore, in particular a high frequency sb-sprof does
540 not prevent gc from running
541 * bug fix: DECODE-UNIVERSAL-TIME now uses a more reasonable
542 approximation for timezone and DST information between the
543 universal time epoch and the smallest negative 32-bit time_t.
544 * bug fix: ENCODE-UNIVERSAL-TIME no longer errors when passed the
545 year 1899 so long as the encoded time is non-negative.
546 * bug fix: on the x86 platform, SB-ROTATE-BYTE:ROTATE-BYTE returns
547 the correct answer for non-constant rotate amounts when performing
548 an inline 32-bit rotation.
550 ** bug fix: RELEASE-FOREGROUND doesn't choke on session lock if
551 there is only one thread in the session
552 ** bug fix: memory leak for streams created in one thread and
553 written to in another
554 ** bug fix: lockup when compiled with gcc4
555 ** bug fix: race that allows the gc to be triggered when gc is
557 ** bug fix: one less memory fault in INTERRUPT-THREAD, again
558 ** bug fix: gc and INTERRUPT-THREAD don't hang when the RT signal
560 ** bug fix: finished threads are gc'ed properly
561 * fixed some bugs revealed by Paul Dietz' test suite:
562 ** CALL-NEXT-METHOD signals an error (in safe code) when the call
563 has arguments with a different set of applicable methods from
564 the orignal arguments.
565 ** The type error thrown by MAP now has a correct expected-type
567 ** DISASSEMBLE now throws a TYPE-ERROR when its argument does not
568 name a compiled function.
569 ** Three MISC tests where a large bignum was improperly coerced to
570 a float that couldn't represent that bignum during type
571 derivation were fixed.
572 ** SUBTYPEP can now handle types involving the negation of a
573 list-form FUNCTION type.
574 ** SUBTYPEP also now handles certain COMPLEX type specifiers such
575 as (COMPLEX (AND RATIO (NOT FIXNUM))).
576 ** READ-BYTE and WRITE-BYTE no longer take stream designators.
578 changes in sbcl-0.9.3 relative to sbcl-0.9.2:
579 * New feature: Experimental support for bivalent streams: streams
580 opened with :element-type :default now allow character and binary
581 (unsigned-byte 8) I/O
582 * Support for the koi8-r external format. (thanks to Ivan Boldyrev)
583 * Bug fix: OPEN no longer fails when *PRINT-READABLY* is T. (thanks
585 * bug fix: no more highly sporadic "couldn't check whether ~S is
586 readable" when reading a stream and an interrupt hits in the middle
587 of a select system call
588 * compiler better recognizes complex arrays (reported by Eduardo
590 * bug fix: out-of-line SB-SYS:FOREIGN-SYMBOL-ADDRESS did not work
591 for variables on SBCL built with linkage-tables. (reported by Luis
593 * various error reporting improvements.
594 * optimizations: LOGNOR on fixnums is improved in the MIPS backend.
595 (Thanks to Thiemo Seufer)
596 * bug fix: nested reader invokations work correctly
597 * bug fix: it is possible to have simultaneous references to foreign
598 code and foreign data with the same name.
600 ** added x86-64 support
601 ** incompatible change: the threading api now works with thread
602 objects instead of thread ids
603 ** bug fix: threads are protected from signals and interruption when
604 starting up or going down
605 ** bug fix: a race where an exiting thread could lose its stack to gc
606 ** bug fix: don't halt on infinite error in threads if possible
607 ** fixed numerous gc deadlocks introduced in the pthread merge
608 ** bug fix: fixed thread safety issues in read and print
609 ** bug fix: debugger doesn't hang on session lock if interrupted at
610 an inappropriate moment
611 ** bug fix: run-program is now thread safe(r)
612 ** bug fix: inner with-recursive-lock no longer releases the mutex
613 * fixed a bug in (DECODE-UNIVERSAL-TIME 0) (reported by Paul Dietz)
614 * fixed some bugs revealed by Paul Dietz' test suite:
615 ** TYPE-ERRORs from signalled by COERCE now have DATUM and
616 EXPECTED-TYPE slots filled.
617 ** MULTIPLE-VALUE-SETQ always returns the primary value of the
619 ** MAKE-CONDITION accepts classes as type-designators.
620 ** COMPILE may never return NIL.
621 ** ENCODE-UNIVERSAL-TIME now guards against times out of Unix's
622 range before calling Unix time functions
624 changes in sbcl-0.9.2 relative to sbcl-0.9.1:
625 * numerous signal handling fixes to increase stability
626 * Support for EUC-JP external format. (thanks to NIIMI Satoshi)
627 * minor incompatible change: we now correctly canonize default
628 initargs, making them be a list of (INITARG INITFORM INITFUNCTION)
629 as per the MOP, rather than the historical (INITARG INITFUNCTION
630 INITFORM). (reported by Bruno Haible)
631 * new feature: WITH-COMPILATION-UNIT now accepts a non-standard
632 :SOURCE-PLIST option. See (DOCUMENTATION #'WITH-COMPILATION-UNIT T)
633 for more information.
634 * TRUENAME and PROBE-FILE now correctly resolve symlinks even if the
635 pathname is a directory pathname.
636 * SB-SPROF now works (more) reliably on non-GENCGC platforms.
637 * dynamic space size on PPC has been increased to 768Mb. (thanks to
639 * SB-MOP:ENSURE-CLASS-USING-CLASS now accepts a class as the
640 :METACLASS argument in addition to a class name. (reported by
641 Bruno Haible for CMUCL, patch for CMUCL by Gerd Moellmann)
642 * RESTART-CASE can now be wrapped around CALL-METHOD forms.
643 (reported by Bruno Haible; patch from Gerd Moellmann)
644 * bug fix: sbcl runtime can now be compiled with gcc4 (thanks to
646 * bug fix: more cleanups to the floating point exception handling on
647 x86-64 (thanks to James Knight)
648 * bug fix: the compiler does not try to constant fold calls of
649 COERCE and BIT-* functions when they return freshly constructed
650 objects. (reported by Nikodemus Siivola and Paul Dietz)
651 * optimization: improved the MIPS versions of generic (in the
652 generic sense) arithmetic routines. (thanks to Thiemo Seufer)
653 * optimization: direct conversion of (unsigned-byte 32) to floats on
654 the PowerPC platform.
655 * optimization: structure instances with raw slots now use less
656 memory, and probably show better memory locality. (thanks to
658 * optimization: DYNAMIC-EXTENT declarations for lists and closures
659 are treated as requests for stack allocation on the x86-64,
660 Alpha-32, PPC and SPARC platforms.
661 * contrib improvement: it's harder to cause SOCKET-CLOSE to close()
662 the wrong file descriptor; implementation of SOCKET-OPEN-P.
663 (thanks to Tony Martinez)
665 ** gcing a dead thread can no longer lead to lockups
666 ** threads block signals until they are set up properly
667 ** errno is no longer shared by threads
668 ** interrupt-thread restores the eflags register on x86
669 ** fixed some lockups due to gc/thread interaction
670 * fixed some bugs revealed by Paul Dietz' test suite:
671 ** invalid dotted lists no longer raise a read error when
673 ** signal an error if a symbol that names a declaration is used
674 as the name of a type, or vice versa
675 ** allow using the (declare (typespec var*)) abbreviation for
676 (declare (type typespec var*)) with all type specifiers
677 ** circularity detection works properly with PPRINT-LOGICAL-BLOCK
678 ** always return NIL from PPRINT-POP when OBJECT is NIL
679 ** don't signal errors when pretty-printing malformed LABELS,
680 FLET or MACROLET forms
681 ** declarations in a DOLIST body are also in scope for the
683 ** COMPILE-FILE accepts all pathname designators as INPUT-FILE
685 ** the ELEMENT-TYPE keyword argument to WITH-OUTPUT-STREAM is
688 changes in sbcl-0.9.1 relative to sbcl-0.9.0:
689 * fixed cross-compiler leakages that prevented building a 32-bit
690 target with a 64-bit host compiler.
691 * fixed a bug in CLOSE :ABORT T: no longer attempts to remove files
692 opened with :IF-EXISTS :OVERWRITE.
693 * fixed bug 281: error for an invalid qualifier in a short-form method
694 combination method is not signalled until the faulty method is called.
695 * bug fix: iteration variable type inferrer failed to deal with open
696 intervals. (reported by Alan Shields)
697 * bug fix: dynamically loading (via LOAD-SHARED-OBJECT or similar)
698 "frameworks" on Mac OS X no longer causes an EXC_BAD_ACCESS if two
699 or more runtime options were provided to the sbcl binary.
700 * compiled code is not steppable if COMPILATION-SPEED >= DEBUG.
701 * contrib improvement: implement SB-POSIX:MKSTEMP (Yannick Gingras)
702 * optimization: there's now a fast-path for fixnum arguments in the
703 generic subtraction routines on x86/x86-64. (Thanks to Lutz Euler)
704 * optimization: the code generated on x86-64 is more compact thanks
705 to not outputting unneccessary prefix bytes. (Thanks to Lutz Euler)
706 * bug fix: floating-point exception handling now partly works on
707 x86-64. (Thanks to James Knight)
708 * improvement to the MIPS backend from Thiemo Seufer: C-style
709 64-bit long long arguments and return values to alien functions
711 * fixed some bugs revealed by Paul Dietz' test suite:
712 ** the type-error signalled from WARN has a filled-in DATUM slot.
713 ** the type-error required when a stream is not associated with
714 a file has the stream as its datum.
715 ** type-errors on single-floats on x86-64 no longer have
716 :INVALID-OBJECT as the datum
717 ** the type-errors signalled for invalid function names now have
718 a correct expected type
719 ** (SETF (DOCUMENTATION ... 'STRUCTURE)) no longer signals an error
720 for structures defined with a :TYPE. Documentation strings for
721 typed structures are no longer immediately discarded
722 ** FILE-STRING-LENGTH and STREAM-EXTERNAL-FORMAT now work on
723 broadcast streams and synonym streams. FILE-LENGTH now also works
724 on broadcast streams.
726 changes in sbcl-0.9.0 relative to sbcl-0.8.21:
727 * incompatible change: the --noprogrammer option, deprecated since
728 version 0.7.5, has been removed. Please use the equivalent
729 --disable-debugger option instead.
730 * incompatible change: finalizers and *AFTER-GC-HOOKS* are now run with
732 * incompatible change: support for *BEFORE-GC-HOOKS* (that have been
733 inoperational for a while now) has been completely removed.
734 * null lexical environments are now printed as #<NULL-LEXENV>,
735 significantly reducing the amount of clutter in typical backtraces.
736 * documentation on weak pointers, finalization, and after GC hooks
737 has been added to the manual.
738 * optimization: REPLACE on declared (UNSIGNED-BYTE 8) vectors, as well
739 as other specialized array types, is much faster. SUBSEQ and
740 COPY-SEQ on such arrays have also been sped up.
741 * optimization: EQL is now more efficient when at least other argument
742 is known to be of type (OR FIXNUM (NOT NUMBER)).
743 * fixed bug: the runtime is now less vulnerable to changes in the
744 size of the SBCL object on OS X, and virtual memory is reserved for
745 all spaces that need to be at a fixed address.
746 * fixed bug: finalizers are now thread-safe. (thanks to Gabor Mellis)
747 * fixed bug: finalizers and after GC hooks that cause consing are now
749 * fixed bug: compiler error messages and summaries are now printed to
750 *ERROR-OUTPUT*, not *STANDARD-OUTPUT*.
751 * fixed inference of the upper bound of an iteration variable.
752 (reported by Rajat Datta).
753 * fixed bug 211e: calling local functions with duplicated constant
754 keyword argument no longer causes a bogus style warning about an
756 * fixed bug 305: INLINE/NOTINLINE declaration no longer causes local
757 ftype declaration to be disregarded. (reported by Dave Roberts)
758 * fixed bug 373: caused by erronous compilation of references to alien
759 variables in the runtime on ppc/darwin.
760 * fixed bug 376: CONJUGATE type deriver.
761 * fixed infinite looping of ALIEN-FUNCALL, compiled with high DEBUG.
762 (reported by Baughn on #lisp)
763 * a cross-compiler bug on non-x86 platforms has been identified and
764 fixed. (thanks to Bruno Haible)
765 * improvements to the MIPS runtime code for increased stability.
766 (thanks to Thiemo Seufer)
767 * increased the maximimum compact environment size to allow
768 purify on images with large amounts of functions. (thanks to Cyrus Harmon)
769 * partial workaround for bug 135: don't name the function we're
770 calling for hairy cases of EVAL, so as not to accumulate one environment
771 entry per call to EVAL. (reported by Kevin Reid)
772 * improvements to the x86-64 disassembler. (thanks to Lutz Euler)
773 * optimization: type testing for non-vector arrays should be faster.
774 * fixed TRACE :ENCAPSULATE NIL, added support for :ENCAPSULATE NIL
776 * bug fix: setting 31st element of a bit vector to zero did not work
778 * bug fix: redefining a class definition which failed due to a
779 previous accessor / function clash now works (but see BUGS entry
780 #380 for more problems in this area). (thanks to Zach Beane)
781 * the long form of DEFINE-METHOD-COMBINATION disables method group
782 checking when given a single method group with pattern *. (thanks
784 * on x86 compiler supports stack allocation of results of simple
785 calls of MAKE-ARRAY, bound to variables, declared DYNAMIC-EXTENT.
786 * support for building and running on Mac OS X 10.4 "Tiger" has been added
787 ** Binaries built on Tiger will not run on 10.2 "Jaguar" currently
788 * fixed some bugs related to Unicode integration:
789 ** the restarts for recovering from input and output encoding
790 errors only appear when there is in fact such an error to
792 * fixed some bugs revealed by Paul Dietz' test suite:
793 ** MISC.549 and similar: late transformation of unsafe type
794 assertions into derived types caused unexpected code
796 ** SCALE-FLOAT type deriver is less wrong.
797 ** type derivers for EXP, LOG and similar functions compute result
798 types for complex arguments better.
799 ** (MISC.563) CONJUGATE type deriver works for very restricted
801 ** out-of-line type testers for character strings are available.
802 ** EQUAL compiler transform understands specialness of objects
804 ** accessing double-floats stored on the stack now works on x86-64.
805 ** debugger internals could sometimes create invalid lispobjs,
806 resulting in GC crashes.
807 ** MISC.548: type check weakening can convert required type into
809 ** initialization forms for bindings are not in scope of free special
812 changes in sbcl-0.8.21 (0.9alpha.1?) relative to sbcl-0.8.20:
813 * incompatible change: thread support for non-NPTL systems has
814 been removed - locking is buggy and unreliable. A threaded
815 SBCL build will now warn at startup and refuse to create
816 new threads, unless futex support is detected in the kernel
817 * incompatible change: the top level REPL now has only an ABORT
818 restart associated with it, not TOPLEVEL and ABORT as it used to.
819 TOP and TOPLEVEL are now available as debugger commands for
820 returning to the top level.
821 * incompatible change: forms evaluated in the REPL now use the
822 global optimization policy.
823 * incompatible change: user- and system-initialization files are
824 no longer processed with LOAD, but by READ and EVAL; hence the
825 global optimization policy, startup package, readtable, etc,
827 * The .fasl file format number has been incremented because of
828 various incompatible changes.
829 * internal entry point details and argument counts no longer appear
830 in backtraces unless explicitly requested by setting
831 SB-DEBUG:*SHOW-ENTRY-POINT-DETAILS*.
832 * built-in and standard functions no longer have names like "top
833 level local call to FOO".
834 * fixed bug 32: functions defined in non-null lexical environments
835 now have more legible printed representation
836 * fixed bug 33: functions defined in non-null lexical environemnts
837 are now more amenable to inspection by INSPECT.
838 * workaround for bug 354: XEPs no longer appear in backtraces unless
839 explicitly requested.
840 * fixed bug: receiving the signal which results from attempting to
841 write to mprotect()ed memory (SIGSEGV on Linux and some *BSDs,
842 SIGBUS on other *BSDs) on architectures where the C stack is also
843 the Lisp stack (x86 and x86-64) from foreign code no longer leads
844 to debugger confusion or wild instability. (reported by Cheuksan
846 * fixed bug: COUNT and EQUAL no longer issue compiler efficiency
847 notes when operating on objects known to be SIMPLE-BIT-VECTORs.
848 (reported by Lutz Euler)
849 * fixed bug: (TYPEP X '(MEMBER ...)) no longer returns a list in
850 compiled code. (reported by Paul Dietz)
851 * fixed bug 276b: mutating with MULTIPLE-VALUE-SETQ a binding of a
852 specialized parameter to a method to something that is not TYPEP
853 the specializer is now possible.
854 * fixed bug: the MAKE-INSTANCE optimization is now correct in the
855 face of package deletion.
856 * fixed bug: LOAD should bind *LOAD-PATHNAME* to the merged
857 pathname. (reported by David Tolpin on c.l.l)
858 * contrib improvement: the SB-SIMPLE-STREAMS contrib now defines
859 STRING-SIMPLE-STREAM and FILE-SIMPLE-STREAM as subclasses of
860 STRING-STREAM and FILE-STREAM, respectively.
861 * contrib improvement: SB-INTROSPECT handles more of SLIME's needs
862 than previously; in addition, its test suite is now run on build.
863 (thanks to Luke Gorrie)
864 * a more robust x86-64 disassembler. (thanks to Lutz Euler)
865 * optimization: added a immediate representation for single-floats
867 * fixed some bugs revealed by Paul Dietz' test suite:
868 ** MISC.564: defined out-of-line version of %ATAN2 on x86.
869 ** attempting to create a package with a colliding nickname causes
870 correctable errors to be signalled.
871 ** MISC.572-574: :START1 and :START2 broken for simple-base-strings.
872 ** several x86-64 backend bugs related to sign-extension of immediate
875 changes in sbcl-0.8.20 (0.9alpha.0?) relative to sbcl-0.8.19:
876 * fixed inspection of specialized arrays. (thanks to Simon Alexander)
877 * fixed disassembly of SHLD and SHRD on x86. (thanks to David
879 * fixed loading of multiply forward-referenced layouts.
880 (thanks to Cheuksan Wang)
881 * fixed bug 7: less verbose COMPILE-FILE output. Additionally, the
882 output is now directed to *STANDARD-OUTPUT* as specified by ANSI.
883 (see COMPILE-FILE documentation for details of :PRINT option)
884 * fixed bugs 19 and 317: fixed-format floating point printing is
885 more accurate. This also fixes a bug reported by Adam Warner
886 related to the ~@F format directive.
887 * fixed bug 371: bignum print/read inconsistency. (thanks to Harald
889 * fixed bug: SET-SYNTAX-FROM-CHAR correctly shallow-copies a
890 dispatch table if the from-char is a dispatch macro character.
891 * fixed bug: COUNT and EQUAL on bit vectors with lengths divisible
892 by the wordsize no longer ignore the last word. (reported by Lutz
894 * fixed bug in type checking of dynamic-extent variables. (reported
896 * optimization: sequence traversal functions use their freedom to
897 coerce function designators to functions.
898 * optimization: code with many calls to CLOS methods specialized on
899 CLOS classes has had redundant type checks removed.
900 * optimization: added declarations to speed up operations that access
901 the internal character database (for example STRING-UPCASE)
902 * optimization: comparison operations between floats and sufficiently small
903 fixnums no longer create extra rationals
904 * fixed some bugs related to Unicode integration:
905 ** portions of multibyte characters at the end of buffers for
906 character-based file input are correctly transferred to the
907 start of the buffer at the next read.
908 ** COMPILE-FILE now respects any EXTERNAL-FORMAT argument given,
909 passing it through to OPEN.
910 ** LOAD on source files likewise passes any EXTERNAL-FORMAT
911 argument given to internal calls to OPEN.
912 ** the built-in comment readers (introduced by character sequences
913 ";" and "#|") are more forgiving to encoding errors; they will
914 STYLE-WARN and then attempt to resync the stream at a character
915 boundary. (thanks to Teemu Kalvas)
916 * fixed some bugs in the x86-64 port:
917 ** Negative short int return values from c-calls are sign-extended
919 ** The stack is aligned to 16-bytes for c-calls, as required by
920 the ABI. (thanks to Cheuksan Wang)
921 ** The disassembler understands more x86-64. (thanks to Cheuksan Wang)
922 ** The regression tests use SB-ALIEN:INT instead of SB-ALIEN:INTEGER
923 for enums. (thanks to Vincent Arkesteijn)
924 ** Multiple small optimizations and bugfixes for floating point
926 * fixed some bugs revealed by Paul Dietz' test suite:
927 ** Space, Tab, Linefeed, Return and Page have the invalid
928 secondary constituent character trait.
929 ** SET-SYNTAX-FROM-CHAR correctly copies multiple-escape character
931 ** WITH-INPUT-FROM-STRING should only update the index place on
933 ** Pretty-printing backquoted forms when *PRINT-CIRCLE* is true
935 ** Bit-array operations (BIT-AND and similar) worked incorrectly
936 with one-dimensional arrays with fill pointers.
937 ** TYPE-OF failed on a complex with an integer realpart and a
939 ** compiler failure during type inference for the code of form
940 (IF (EQL X (THE ...)) ...) (MISC.535).
942 changes in sbcl-0.8.19 relative to sbcl-0.8.18:
943 * new port: SBCL now works in native 64-bit mode on x86-64/Linux
944 platform. The port supports 61-bit fixnums, large memory spaces
945 and reloading shared object files.
946 * enhancement: saving cores with foreign code loaded is now
947 supported on ppc/Darwin in addition to the previously supported
949 * enhancement: the statistical profiler now walks deeper into the
950 call stack for more meaningful call-graphs and accrued time
951 reports (x86/x86-64 only). It also now reports time spent in
953 * enhancement: it is now possible to trace most individual methods
954 of a generic function in addition to tracing the generic function
956 * bug fix: invalid :DEFAULT-INITARGS are detected in compiled calls
958 * bug fix: defaulted initargs are passed to INITIALIZE-INSTANCE and
959 SHARED-INITIALIZE methods from compiled calls to MAKE-INSTANCE.
960 * bug fix: COERCE to (COMPLEX FLOAT) of a complex number no longer
961 produces an error. (thanks to Vincent Arkesteijn)
962 * bug fix: NAMESTRING on pathnames with :WILD components in their
963 directories works correctly. (thanks to Artem V. Andreev)
964 * fixed bug 125: compiler preserves identity of closures. (reported
966 * bug fixed: functions with &REST arguments sometimes failed with
967 "Undefined function" when compiled with (DEBUG 3). (reported by
969 * bug fix: overflow during compiling of setting element of a bit
970 vector with constant index and value. (reported by Timmy Douglas)
971 * build fix: fixed the dependence on *LOAD-PATHNAME* and
972 *COMPILE-FILE-PATHNAME* being absolute pathnames.
973 * on x86 compiler partially supports stack allocation of dynamic-extent
975 * GO and RETURN-FROM do not check the extent of their exit points
976 when compiled with SAFETY 0.
977 * fixed some bugs related to Unicode integration:
978 ** encoding and decoding errors are now much more robustly
979 handled; it should now be possible to recover even from invalid
980 input or output to the terminal. (thanks to Teemu Kalvas)
981 ** provided a first cut at implementing STRING-TO-OCTETS and
982 OCTETS-TO-STRING. (thanks to Robert J. Macomber)
983 ** altered the SB-MD5 contributed module slightly, changing the
984 interface just enough for it to be supportable for builds where
985 lisp characters are not eight bits.
986 * fixed some bugs revealed by Paul Dietz' test suite:
987 ** the FORMATTER-generated functions for ~V[ conditionals require
988 the correct number of arguments.
989 ** READ-FROM-STRING returns the mandated second value when applied
990 to displaced strings.
991 ** the #\Rubout and #\Backspace characters are treated as invalid
992 constituent characters by the tokenizer.
994 changes in sbcl-0.8.18 relative to sbcl-0.8.17:
995 * new feature: reloading changed shared object files with
996 LOAD-SHARED-OBJECT now causes the new definitions to take effect.
997 * new feature: references to foreign variables and functions
998 can now be compiled and loaded before the corresponding shared
999 object file is loaded, as long as the foreign definitions are
1000 available at runtime.
1001 * Solaris 10 (aka SunOS 5.10) on the SPARC platform is now
1002 supported. (thanks to Dan Debertin)
1003 * SB-ALIEN enums can now be represented in Lisp by any symbols, not
1004 just keywords. (thanks to Vincent Arkesteijn)
1005 * fixed bug #331: structure-class instances corresponding to
1006 DEFSTRUCT forms are now created eagerly.
1007 * fixed bug #345: backtraces from calls to undefined functions work
1008 on x86 as well. Related bug #61 is now also partially fixed on x86
1009 (backtraces from throws to unknown catch tags.)
1010 * bug fix: lambda-list parsing is now stricter vrt. order and number
1011 of lambda-list keywords.
1012 * bug fix: as specified by AMOP, an error is signalled if a
1013 class-option appears multiple times in a DEFCLASS form. (reported
1015 * bug fix: RANDOM can be compiled when the compiler derives the type
1016 of its numeric argument as a disjoint set of small integers.
1017 (reported by Paul Dietz)
1018 * bug fix: starting a core saved with shared objects loaded when
1019 those objects are not available no longer causes threaded SBCL to
1020 hang. (reported by Sean Ross)
1021 * bug fix: evaluated FUNCTION no longer bypasses encapsulation (eg.
1023 * bug fix: (SETF MACRO-FUNCTION) now accepts an optional environment
1024 argument, which must always be NIL. (reported by Kalle Niemitalo)
1025 * bug fix: printing 1.0d+23 no longer results in an error.
1026 (reported by Rolf Wester for CMUCL; bug fix from Raymond Toy)
1027 * bug fix: structure slot setters preserve evaluation order. (thanks
1029 * fixed some bugs related to Unicode integration:
1030 ** RUN-PROGRAM can allow its child to take input from a Lisp
1031 stream. (reported by Stefan Scholl)
1032 ** ASDF-INSTALL successfully downloads .tar.gz files in all
1033 locales. (reported by Ken Causey)
1034 * fixed some bugs revealed by Paul Dietz' test suite:
1035 ** INCF, DECF and REMF evaluate their place form as specified in
1037 ** FORMATTER expands ~{ ~} iteration directives with V or #
1038 parameters correctly.
1039 ** FORMATTER deals with the ~@[ ~] conditional directive where the
1040 consequent uses no arguments correctly.
1041 ** the system has a partial understanding of the (COMPLEX RATIO)
1043 ** UPGRADED-COMPLEX-PART-TYPE obeys the lattice upgrading rules
1046 changes in sbcl-0.8.17 relative to sbcl-0.8.16:
1047 * new feature: a build-time option (controlled by the :SB-UNICODE
1048 keyword feature, enabled by default) for building the system with
1049 support for the entire 21-bit character space defined by the
1051 * new feature: the system now has rudimentary external-format
1052 support; the primary user-visible change at this time is that
1053 characters with the high bit set (such as lower-case-e-acute) will
1054 print correctly to a terminal in a UTF-8 environment.
1055 * minor incompatible change: BASE-CHAR no longer names a class;
1056 however, CHARACTER continues to do so, as required by ANSI.
1057 * minor incompatible change: SB-DEBUG:*DEBUG-PRINT-FOO* variables
1058 are no longer supported, and SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST*
1059 has been moved to the SB-EXT package (temporarily re-exported from
1061 * minor incompatible change: SB-C::*COMPILER-ERROR-PRINT-FOO* variables
1062 are no longer supported: use SB-EXT:*COMPILER-PRINT-VARIABLE-ALIST*
1064 * fixed bug #351: better error-handlind and reporting for malformed
1066 * fixed bug #350: bignum-printing is now more memory-efficient,
1067 allowing printing of very large bignums, eg. (expt 2 10000000).
1068 (reported by Bruno Haible)
1069 * fixed bug #302: better primitive-type selection for intersection
1071 * fixed bug #308: non-graphic characters now all have names, as
1072 required. (reported by Bruno Haible)
1073 * bug fix: redefining a class with different superclasses now correctly
1074 removes it from the direct-subclasses of its previous superclasses.
1075 (reported by David Morse)
1076 * bug fix: (SETF FIND-CLASS) using a FORWARD-REFERENCED-CLASS as the
1077 new value now works. (reported by Bruno Haible)
1078 * bug fix: correct canonicalization of multiple non-standard slot
1079 options in DEFCLASS as per AMOP 5.4.2. (reported by Bruno Haible)
1080 * bug fix: SB-MOP:CLASS-PROTOTYPE now signals an error if the class
1081 is not yet finalized, as required by AMOP. (reported by Bruno Haible)
1082 * bug fix: SB-MOP:ALLOCATE-INSTANCE method for instances of BUILT-IN-CLASS
1083 now exists, an signals an error.
1084 * bug fix: duplicate LOOP variable bindings now signal PROGRAM-ERROR
1085 during macroexpansion for non-iteration variables as well. (reported
1086 by Bruno Haible for CMUCL)
1087 * bug fix: Cyclic structures and unprintable objects in compiler
1088 messages no longer cause errors. (reported by Bruno Haible)
1089 * bug fix: READ, READ-PRESERVING-WHITESPACE, READ-DELIMITED-LIST,
1090 and READ-FROM-STRING all now return a primary value of NIL if
1091 *READ-SUPPRESS* is true. (reported by Bruno Haible for CMUCL)
1092 * bug fix: Default value of EOF-ERROR-P in READ-FROM-STRING is true.
1093 (reported by Bruno Haible for CMUCL)
1094 * bug fix: ERROR now signals a TYPE-ERROR if the arguments to ERROR
1095 do not designate a condition. (reported by Bruno Haible for
1097 * bug fix: UNINTERN, USE-PACKAGE, IMPORT and EXPORT all signal an
1098 SB-EXT:NAME-CONFLICT condition (subtype of PACKAGE-ERROR) in the
1099 name conflict situations in CLHS 11.1.1.2.5, and provide a restart
1100 permitting resolution in favour of any of the conflicting symbols.
1101 (reported by Bruno Haible for CMUCL)
1102 * bug fix: EQUAL compiler optimizations is less aggressive on
1103 strings which can potentially compare true despite having distinct
1104 specialized array element types.
1105 * bug fix: unit enumerations can be defined without dividing by
1106 zero. (thanks to Vincent Arkesteijn)
1107 * FORMAT compile-time argument count checking has been enhanced.
1108 (report from Bruno Haible for CMUCL)
1109 * a partial workaround for the bug 262: the compiler does not try to
1110 inline-expand a local function doing RETURN-FROM from a deleted
1111 BLOCK. (thanks to Peter Denno for the bug report and to David
1112 Wragg for the simple test case)
1113 * fixed some bugs revealed by Paul Dietz' test suite:
1114 ** RENAME-PACKAGE allows all package designators as new package
1116 ** constraint propagation and lambda variable substitution are
1117 more cautious in dealing with partially deleted code.
1118 ** compiler sometimes lost reoptimization passes.
1119 ** CERROR, given a condition as condition designator, passes its
1120 remaining arguments to the continue format control without
1122 ** Case-altering FORMAT directives work correctly on non-ASCII
1124 ** The REAL type specifier handles bounds outside the floating
1125 point ranges without signalling FLOATING-POINT-OVERFLOW.
1126 ** Functions with IR1-transformations can create intercomponent
1127 references to global functions.
1128 ** NIL parameter to the FORMAT directive ~^ means `unsupplied
1130 ** FORMAT ~R treats a nil value for its first parameter correctly.
1132 changes in sbcl-0.8.16 relative to sbcl-0.8.15:
1133 * enhancement: saving cores with foreign code loaded is now
1134 supported on x86/NetBSD and sparc/Linux in addition to the previously
1135 supported platforms.
1136 * bug fix: on some platforms repeated installations caused multiple
1137 copies of HTML documentation to be installed -- should not happen
1138 any more. (reported by Stefan Scholl)
1139 * bug fix: parsing self-recursive alien record types multiple times
1140 no longer causes infinite recursion. (reported by Thomas F. Burdick,
1141 original patch by Helmut Eller for CMUCL)
1142 * bug fix: stack-exhaustion detection works now on NetBSD as well.
1143 (thanks to Richard Kreuter)
1144 * bug fix: defining classes whose accessors are methods on existing
1145 generic functions in other (locked) packages no longer signals
1146 bogus package lock violations. (reported by François-René Rideau)
1147 * bug fix: special variables as DEFMETHOD parameters no longer have
1148 associated bogus type declarations. (reported by David Wragg and
1150 * bug fix: read-write consistency on streams of element-type
1151 (SIGNED-BYTE N) for N > 32. (reported by Bruno Haible for CMUCL)
1152 * bug fix: redefiniton of the only method of a generic function with
1153 no DEFGENERIC no longer emits a full WARNING. In addition,
1154 redefinition of generic functions with no DEFGENERIC to an
1155 incompatible lambda list now signals an error. (thanks to Zach
1157 * bug fix: DEFGENERIC now works even when there's a function of the
1158 same name in an enclosing lexical environment. (thanks to Zach
1160 * fixed compiler failure, caused by instrumenting code during
1161 IR1-optimization. (Debian bug report #273606 by Gabor Melis)
1162 * optimization: added loop analysis and improved register allocation
1163 * fixed some bugs revealed by Paul Dietz' test suite:
1164 ** POSITION on displaced vectors with non-zero displacement
1165 returns the right answer.
1166 ** (SIMPLE-STRING) is a valid type specifier for sequence
1168 ** *PRINT-LEVEL* handling for slotless structures is pedantically
1170 ** PPRINT-INDENT accepts a request for an indentation of any REAL.
1171 ** PPRINT-TAB (and the FORMAT ~T directive) now indent by the
1173 ** The justification version of the FORMAT ~< directive treats
1174 non-zero minpad parameter correctly.
1176 changes in sbcl-0.8.15 relative to sbcl-0.8.14:
1177 * incompatible change: SB-INT:*BEFORE-SAVE-INITIALIZATIONS* and
1178 SB-INT:*AFTER-SAVE-INITIALIZATIONS* have been renamed
1179 SB-EXT:*SAVE-HOOKS* and SB-EXT:*INIT-HOOKS*, and are now part of
1180 the supported interface.
1181 * new feature: Single-stepping of code compiled with DEBUG 2 or
1182 higher and (> DEBUG (MAX SPACE SPEED)) is now possible.
1183 * new feature: saving cores with foreign code loaded is now
1184 supported on x86/FreeBSD, x86/Linux, and sparc/SunOS. (based on
1185 Timothy Moore's work for CMUCL)
1186 * bug fix: DEFTYPE lambda-list parsing now binds unsupplied keyword
1187 parameters to * instead of NIL if no initform is supplied.
1188 (reported by Johan Bockgaard)
1189 * bug fix: DEFINE-COMPILER-MACRO lambda-list parsing now binds
1190 correctly when FUNCALL appears as the car of the form. Note:
1191 despite this FUNCALL forms are not currently subject to
1192 compiler-macro expansion. (port of Raymond Toy's fix for the
1193 same from CMUCL, reported by Johan Bockgaard)
1194 * bug fix: FOR ... ON ... -clauses in LOOP now work on dotted lists
1195 (thanks for Teemu Kalvas)
1196 * bug fix: in FORMAT ~^ inside ~:{ now correctly steps to the next
1197 case instead of terminating the iteration (thanks for Julian
1198 Squires, Sean Champ and Raymond Toy)
1199 * bug fix: incorrect expansion of defgeneric that caused a style
1200 warning. (thanks for Zach Beane)
1201 * bug fix: slot accessor effective method computation works properly
1202 for classes with multiple non-standard applicable methods on
1203 SB-MOP:SLOT-VALUE-USING-CLASS. (reported by Ralf Mattes)
1204 * on x86 compiler supports stack allocation of results of LIST and
1205 LIST*, bound to variables, declared DYNAMIC-EXTENT. (based on
1206 CMUCL implementation by Gerd Moellmann)
1207 * fixed some bugs revealed by Paul Dietz' test suite:
1208 ** FORMAT strings with both the ~<~:;~> form of the justification
1209 directive and pretty-printing directives cause an error.
1211 changes in sbcl-0.8.14 relative to sbcl-0.8.13:
1212 * incompatible change: the internal functions
1213 SB-KERNEL:32BIT-LOGICAL-FOO, intended for providing efficient
1214 logical operations on (UNSIGNED-BYTE 32) values, have been renamed
1215 to SB-KERNEL:WORD-LOGICAL-FOO. Modular arithmetic should be used
1216 instead of the old functions.
1217 * new feature: on platforms where "dladdr" is available foreign
1218 function names now appear in backtraces. (based on Helmut Eller's
1220 * documentation: documentation for SB-BSD-SOCKETS, SB-GRAY,
1221 SB-SIMPLE-STREAMS, and SB-PROFILE have been integrated into the user
1223 * documentation: ASDF manual is now distributed with SBCL.
1224 * bug fix: SBCL can now load its contributed modules with REQUIRE
1225 even if the system-provided entries have been removed from
1226 ASDF:*CENTRAL-REGISTRY*. Infinite recursion in REQUIRE is also
1228 * bug fix: backtraces involving undefined functions or assembly
1229 routines are more informative. (thanks to Brian Downing)
1230 * bug fix: mutually referent alien structures now work correctly.
1231 (reported by Rick Taube)
1232 * bug fix: structures defined by WITH-ALIEN can be referred to
1233 within other definitions in the same WITH-ALIEN.
1234 * bug fix: division operators (MOD, TRUNCATE and the like) with
1235 constant zero divisors and integer dividends no longer generate
1237 * bug fix: provide default methods for INPUT-STREAM-P and
1238 OUTPUT-STREAM-P specialized on SB-GRAY:FUNDAMENTAL-STREAM.
1239 * bug fix: improve the detection and resolution of MOP metacycles
1240 (where effective method computation depends on the generic function
1241 having its effective method computed). (reported by Bruno Haible)
1242 * bug fix: pass the right initargs to :BEFORE or :AFTER methods on
1243 SHARED-INITIALIZE or INITIALIZE-INSTANCE in optimized
1244 MAKE-INSTANCE constructors. (reported by Axel Schairer for cmucl)
1245 * bug fix: pathnames with relative directory components can be
1246 represented relative to default pathnames.
1247 * optimization: in taking the GCD of bignums, reduce the two bignums
1248 to approximately the same size (using Euclid's algorithm) before
1249 applying the more sophisticated binary GCD. (thanks to Juho
1251 * optimization: COUNT on bitvectors now operates word-at-a-time.
1252 * optimization: ASH with a positive, but not necessarily constant,
1253 (leftwards) shift, when in a modular context, is compiled to a
1255 * fixed some bugs revealed by Paul Dietz' test suite:
1256 ** FORMAT variable parameters ("~V<char>") are defaulted properly
1257 if the corresponding argument is NIL.
1258 ** FORMAT directives accepting a minpad parameter treat negative
1260 ** literal commas as character parameters to FORMAT directives are
1262 ** literal spaces directly after ~<Newline> directives within a
1263 format-logical-block (~:< ~@:>) do not induce :FILL-style
1264 conditional newlines.
1265 ** PRINT-UNREADABLE-OBJECT inserts spaces as specified (and only
1266 as specified: it no longer includes conditional newlines).
1267 ** PRINC-TO-STRING binds *PRINT-READABLY* to NIL (as well as
1269 ** PPRINT-LOGICAL-BLOCK signals a TYPE-ERROR if its :PREFIX or
1270 :PER-LINE-PREFIX argument does not evaluate to a string.
1272 changes in sbcl-0.8.13 relative to sbcl-0.8.12:
1273 * new feature: SB-PACKAGE-LOCKS. See the "Package Locks" section of
1274 the manual for details; for now, package locks can be disabled by
1275 removing :SB-PACKAGE-LOCKS in customize-target-features.lisp, but
1276 if no major problems are found then it is likely that they will be
1277 compiled in unconditionally.
1278 * major incompatible change: LOAD-FOREIGN and LOAD-1-FOREIGN are now
1279 unsupported operators on all platforms. To load a shared library
1280 into SBCL, use SB-ALIEN:LOAD-SHARED-OBJECT. To load a non-shared
1281 object file, link it into a shared library outside of SBCL and
1282 load it using SB-ALIEN:LOAD-SHARED-OBJECT. (incidentally fixes
1284 * bug workaround: ROOM T and the GC were not working together
1285 reliably, because invariants expected by the SB!VM:INSTANCE-USAGE
1286 reporting facility aren't preserved. That reporting has been
1287 disabled, so now until and unless someone figures out how to make it
1288 work reliably with the current GC, (ROOM T) is equivalent to (ROOM).
1289 * minor incompatible change: as threatened around sbcl-0.8.0, the
1290 home package of MOP-related symbols is now SB-MOP, not SB-PCL.
1291 The symbols are also exported from SB-PCL for backwards
1292 compatibility, but more so than before SB-PCL should be treated as
1293 an implementation-internal package.
1294 * fasl format changed: SBCL 0.8.13 fasls are incompatible with those
1296 * the SB-SPROF contrib now works on (most) non-x86 architectures.
1297 It is known as of this release not to work on the Alpha, however.
1298 * fixed bug #167: errors signalled due to illegal syntax in method
1299 bodies are now more legible.
1300 * fixed bug #338: instances of EQL-SPECIFIER are now valid type
1301 designators and can hence be used with TYPEP.
1302 * fixed bug #333: CHECK-TYPE now ensures that the type error
1303 signalled, if any, has the right object to be accessed by
1304 TYPE-ERROR-DATUM. (reported by Tony Martinez)
1305 * fixed bug #340: SETF of VALUES obeys the specification in ANSI
1306 5.1.2.3 for multiple-value place subforms. (reported by Kalle
1308 * fixed bug #334: programmatic addition of slots using specialized
1309 methods on SB-MOP:COMPUTE-SLOTS works for :ALLOCATION :INSTANCE
1310 and :ALLOCATION :CLASS slots.
1311 * fixed bug #269: SCALE-FLOAT scales floats by any integer, not just
1312 float exponents. (rereported by Peter Seibel)
1313 * fixed a bug: #\Space (and other whitespace characters) are no
1314 longer considered to be macro characters in standard syntax by
1315 GET-MACRO-CHARACTER.
1316 * fixed bug: initialization of condition class metaobjects no longer
1317 causes an instance of the condition to be created. (reported by
1319 * fixed bug: it is now possible to have slots such that
1320 SB-MOP:SLOT-DEFINITION-ALLOCATION of the effective slot
1321 description is neither :INSTANCE nor :CLASS.
1322 * fixed bug: the ctor optimization of MAKE-INSTANCE now respects
1323 user-defined methods on SLOT-BOUNDP-USING-CLASS and (SETF
1324 SLOT-VALUE-USING-CLASS), and no longer causes errors with
1325 non-standard SLOT-DEFINITION-ALLOCATION values.
1326 * fixed bugs: various race conditions handling exiting threads.
1327 CL-PPCRE's thread torture test now passes.
1328 * fixed arguably-a-bug: GC time proportional to number of threads in
1329 system even when most of them are idle
1330 * optimization: improved performance of BIT and SBIT on bit-vectors.
1331 * .fasl file incompatibility: The fasl file version number has
1332 been incremented because of changes associated with package locks.
1334 changes in sbcl-0.8.12 relative to sbcl-0.8.11:
1335 * minor incompatible change: the system no longer provides
1336 optimization hints (conditions which are TYPEP
1337 SB-EXT:COMPILER-NOTE) for conforming code in default compilation
1338 mode; these hints are still emitted when higher SPEED optimization
1340 * new contrib module: a sampling profiler (profiling by statistical
1341 sampling, rather than by instrumenting functions) is available as
1342 the SB-SPROF contrib. (thanks to Gerd Moellmann and Juho Snellman)
1343 * the behaviour of the standard function ED is now customizeable by
1344 third parties through a hook variable: see ED's documentation
1345 string for information on the protocol.
1346 * the compiler no longer emits efficiency notes for (FUNCALL X)
1347 when the type of X is uncertain under default optimization
1349 * fixed bug 276: mutating a binding of a specialized parameter to a
1350 method to something that is not TYPEP the specializer is now
1352 * fixed bugs 45d and 118: DOUBLE-FLOAT[-NEGATIVE]-EPSILON now
1353 exhibit the required behaviour on the x86 platform. (thanks to
1354 Peter van Eynde, Eric Marsden and Bruno Haible)
1355 * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
1356 even for difficult arguments. (reported by Peter Graves)
1357 * fixed bug 141a: the backquote printer now descends quoted
1359 * fixed another bug in backquote printing: no more destructive
1360 modification of the form's list structure. (reported by Brian
1362 * fixed bug in INTERRUPT-THREAD: pin the function, so that it cannot
1363 move between its address being taken and the call to
1364 interrupt_thread, fixing a crashing race condition.
1365 * the SB-POSIX contrib implementation has been adjusted so that it
1366 no longer exhibits ridiculously poor performance when constructing
1367 instances corresponding to C structs.
1369 changes in sbcl-0.8.11 relative to sbcl-0.8.10:
1370 * minor incompatible change: the sb-grovel contrib now treats C
1371 structures as alien (in the sense of SB-ALIEN) objects rather than
1372 as undistinguished (simple-array (unsigned-byte 8) (*))s. This
1373 has implications for memory management of client code
1374 (sb-grovel-returned objects must now be manually managed) and for
1375 type safety (alien objects now have full types).
1376 * new feature: the SB-EXT:MUFFLE-CONDITIONS declaration should be
1377 used to control emission of compiler diagnostics, rather than the
1378 SB-EXT:INHIBIT-WARNINGS OPTIMIZE quality. See the manual for
1379 documentation on this feature. The SB-EXT:INHIBIT-WARNINGS
1380 quality should be considered deprecated.
1381 * install.sh now installs the user manual as well
1382 * (not quite a new documentable feature, but worth considering in
1383 the light of the new SB-EXT:MUFFLE-CONDITIONS declaration): the
1384 beginnings of a semantically meaningful condition hierarchy is
1385 under development, for use in SB-EXT:MUFFLE-CONDITIONS and by
1387 * fixed bug: PARSE-NAMESTRING now accepts any valid pathaname
1388 designator as the defaults argument.
1389 * fixed bug: Displaced arrays whose displaced-to array has become
1390 too small now cause ARRAY-DIMENSION to signal an error, providing
1391 for safer bounds-checking. (reported by Bruno Haible)
1392 * fixed bug: DEFCLASS slot definitions with identical :READER and
1393 :WRITER names now signal a reasonable error. (reported by Thomas
1395 * fixed bug: CLOSE :ABORT T on appending stream no longer causes
1397 * fixed bug: Invalid slot specification errors now print correctly.
1398 (thanks to Zach Beane)
1399 * fixed bug 320: Shared to local slot value transfers in class
1400 redefinitions now happen corrently from superclasses as
1401 well. (reported by Bruno Haible)
1402 * fixed bug 316: SHIFTF now accepts VALUES forms. (reported by Bruno
1404 * fixed bug 322: DEFSTRUCT :TYPE LIST type predicates now handle
1405 improper lists correctly. (reported by Bruno Haible)
1406 * fixed bug 313: source-transform for <fun-name> was erroneously
1407 applied to a call of a value of a variable with name <fun-name>.
1408 (reported by Antonio Menezes Leitao)
1409 * fixed bug 307: The obsolete instance protocol ensures that
1410 subclasses are properly obsoleted. (thanks to Nikodemus Siivola)
1411 * fixed bug 298, revealed by Paul F. Dietz' test suite: SBCL can
1412 remove dead unknown-values globs from the middle of the stack.
1413 * added a new restart to *BREAK-ON-SIGNALS* handling to make it
1414 easier to resume long computations after using *BREAK-ON-SIGNALS*
1415 to diagnose and fix failures. (thanks to Nikodemus Siivola)
1416 * fixed bug reported by PFD in lisppaste #747 (and Bruno Haible from
1417 CLISP test suite): WRITE-TO-STRING is not constant-foldable.
1418 * fixed bugs in COMPLEX type specifier: UPGRADED-COMPLEX-PART-TYPE
1419 is now consistent with (COMPLEX <x>); bugs in treatment of COMPLEX
1420 MEMBER and UNION types have likewise been fixed. (thanks to Bruno
1422 * fixed a (fairly theoretical) bug in string printing: if
1423 *PRINT-READABLY* is true, signal PRINT-NOT-READABLE if the string
1424 does not have array-element-type equal to the most general string
1426 * fixed bug: SET-PPRINT-DISPATCH does not immediately resolve
1427 function name. (thanks to Nikodemus Siivola)
1428 * fixed bug: compile-time format string checker failed on
1429 non-closed ~{. (reported by Thomas F Burdick)
1430 * fixed bug: as reported by Kalle Olavi Niemitalo on #lisp IRC,
1431 don't warn on legal loop code involving more than one aggregate
1432 boolean. (thanks to Nikodemus Siivola)
1433 * fixed bug: as reported by Peter Graves on #lisp IRC, passing a NIL
1434 in keyword position to MAKE-PACKAGE was not being reported as
1436 * fixed bug: as reported by Juan Ripoll on cmucl-imp,
1437 MULTIPLE-VALUE-BIND should be able to lexically bind lambda list
1439 * fixed bugs 280 and 312: the checking for multiple definitions in a
1440 file is less likely to become confused by uses of inline
1442 * fixed bug: the #S reader macro performs the keyword coercion
1443 specified for slot names. (reported by Kalle Niemitalo)
1444 * fixed bug: lambda lists may contain symbols whose names start with
1445 & but are not lambda-list-keywords; their occurrence triggers a
1447 * fixed bug 321: define-method-combination argument lambda lists do
1448 not cause invalid code to be generated when &optional or &aux
1449 variables have default values. (reported by Bruno Haible)
1450 * fixed bug 327: system subclasses of STRUCTURE-OBJECT or CONDITION
1451 have CLOS classes; this appears also to have fixed reported
1452 instability in stack exhaustion detection.
1453 * fixed bug: the CONTROL-ERROR from ABORT, CONTINUE and
1454 MUFFLE-WARNING when no associated restart is present is now
1456 * optimization: rearranged the expansion of various defining macros
1457 so that each expands into only one top-level form in a
1458 :LOAD-TOPLEVEL context; this appears to decrease fasl sizes by
1460 * optimization: used a previously little-used slot in symbols to
1461 cache SXHASH values, yielding a 5-10% compiler speedup. (thanks
1463 * fixed some bugs revealed by Paul Dietz' test suite:
1464 ** MAKE-INSTANCES-OBSOLETE returns the class name when called with
1466 ** Fixed an optimization bug related to inheritance of initforms
1467 from local to shared slots.
1468 ** FILE-POSITION works as specified on BROADCAST-STREAMs.
1469 ** CAST optimizer forgot to flush argument derived type.
1470 ** print/read consistency on floats is now orders of magnitude
1471 more likely. (thanks also to Bruno Haible for a similar report
1473 ** removed stack cleaning in the cleanup part of UNWIND-PROTECT.
1474 ** IMAGPART is specified (infelicitously) to return (* 0 <thing>)
1475 for objects of type REAL. Make it so.
1476 ** SXHASH is specified (infelicitously) to respect similarity,
1477 which means that (SXHASH 0.0) must equal (SXHASH -0.0). Make
1478 it so. (thanks to Markus Ziegler)
1479 ** on the Alpha, the compiler succeeds in compiling functions
1480 returning a known number of arguments greater than 63.
1481 ** fixed handling of invalid NIL arguments in keyword position
1483 ** fixed non-pretty printing of arrays with *PRINT-RADIX* being
1485 ** provided a readably-printable representation for RANDOM-STATE
1487 ** ensured that pathnames, where they have a namestring, always
1488 print using #P"..." syntax.
1490 changes in sbcl-0.8.10 relative to sbcl-0.8.9:
1491 * Support for the forthcoming 2.0 version of the NetBSD kernel
1492 running on x86 hardware has been added. (thanks to Perry
1493 E. Metzger most immediately, and others for their past work)
1494 * SBCL now runs on OpenBSD 3.4. (Thanks to Scott Parish; 3.4 is the
1495 current release version; SBCL's OpenBSD support had been broken
1496 since about the time of OpenBSD's switch to ELF binary format.)
1497 * [placeholder for DX summary]
1498 ** user code with &REST lists declared dynamic-extent, under high
1499 speed or space and low safety and debug optimization policy.
1500 * The manual has been converted to Texinfo format and the debugger
1501 chapter from the cmucl manual has been added.
1502 * A facility has been added to extract documentation strings from
1503 sbcl and store them as Texinfo-formatted snippets for inclusion in
1504 the manual (via Texinfo's @include directive)
1505 * bug fix: compiler emitted division in optimized DEREF. (thanks for
1506 the test case to Dave Roberts)
1507 * bug fix: multidimensional simple arrays loaded from FASLs had fill
1508 pointers. (reported by Sean Ross)
1509 * bug fix: PROFILE output is printed nicely even for large numerical
1510 values. (thanks to Zach Beane)
1511 * bug fix: streams with element-type (SIGNED-BYTE <N>) for <N>
1512 greater than 32 handle EOF correctly.
1513 * bug fix: on X86 an immediate argument of the IMUL instruction is
1514 correctly printed in disassembly. (thanks to Lutz Euler)
1515 * bug fix: class slots in redefined classes preserve their old
1516 values. (thanks to Bruno Haible and Nikodemus Siivola)
1517 * bug fix: compilation of funcalls of CXX+R (e.g. CDDR) now
1518 succeeds. (reported by Marco Baringer)
1519 * fixed some bugs revealed by Paul Dietz' test suite:
1520 ** READ-SEQUENCE now works on ECHO-STREAMs.
1521 ** RATIONALIZE works more according to its specification. (thanks
1524 changes in sbcl-0.8.9 relative to sbcl-0.8.8:
1525 * deprecation of old extension: *DEBUG-PRINT-LEVEL* and
1526 *DEBUG-PRINT-LENGTH* are now deprecated in favor of the new, more
1527 general SB-DEBUG:*DEBUG-PRINT-VARIABLE-ALIST* mechanism. (This
1528 should matter to you only if you rebind the printer control
1529 variables and then find you want different bindings in the
1530 debugger than in the ordinary execution of your program.)
1531 * The runtime build system has been tweaked to support building
1532 (on SPARC/SunOS) using a C compiler which invokes Sun's own
1533 assembler and linker. (thanks to Nikodemus Siivola)
1534 * Unbound, undefined, undeclared variables now trigger full
1535 WARNINGs, not just STYLE-WARNINGs, on the assumption that this is
1536 more often programmer error than deliberate exploitation of undefined
1538 * optimization: the hash algorithm for strings has changed to one
1539 that is less vulnerable to spurious collisions. (thanks to Juho
1541 * optimization: VECTOR-POP, VECTOR-PUSH-EXTEND and REPLACE do less
1542 needless bounds checking. (thanks to Juho Snellman)
1543 * optimization: implemented multiplication as a modular
1544 (UNSIGNED-BYTE 32) operation on the PPC backend.
1545 * fixed some bugs revealed by Paul Dietz' test suite:
1546 ** ADJUST-ARRAY now copies the datum in a zero rank array if
1548 ** ADJUST-ARRAY no longer adjusts non-adjustable arrays.
1549 ** MAKE-STRING-INPUT-STREAM accepts NIL as a value for :END.
1550 ** MAKE-STRING-INPUT-STREAM functions correctly for strings with
1552 ** CLEAR-INPUT accepts NIL and T for its (optional) stream
1554 ** Ratios can now be printed correctly with *PRINT-BASE* bound to
1556 ** ECHO-STREAMs no longer attempt to echo the end of file value to
1557 their output stream on EOF from read.
1558 ** CONCATENATED-STREAM-STREAMS discards constituent streams which
1559 have been read to end-of-file.
1560 ** CLOSE works as expected on the null CONCATENATED-STREAM, and on
1562 ** Printing symbols with *PRINT-CASE* :CAPITALIZE respects the
1563 description of determination of which consecutive characters
1565 ** Printing the "Space" character with escaping on now yields "#\\ ",
1566 rather than "#\\Space", as mandated by ANSI 22.1.3.2.
1567 ** Reading floating-point numbers with *READ-BASE* set to a number
1568 less than 10 works correctly.
1569 ** Reading floating-point numbers with *READ-BASE* set to a number
1570 more than 10 works correctly.
1571 ** Printing with *PRINT-READABLY* targets the standard readtable, not
1572 the readtable currently in effect.
1574 changes in sbcl-0.8.8 relative to sbcl-0.8.7:
1575 * minor incompatible change: parsing of namestrings on a physical
1576 (Unix) host has changed; numbers after the final #\. in a
1577 namestring are no longer interpreted as a version field. This is
1578 intented to be largely invisible to the user, except that the
1579 meaning of the namestring "*.*.*" has changed: it now refers to a
1580 pathname with :TYPE :WILD :NAME #<pattern "*.*">. This namestring
1581 should usually be replaced by
1582 (make-pathname :name :wild :type :wild :version :wild)
1583 with the added benefit that this is more likely to be portable.
1584 As a consequence of this change, the :IF-EXISTS :NEW-VERSION
1585 option to OPEN now signals an error if the file being opened
1586 exists; this may have an impact on existing code.
1587 * fixed bug 190: RUN-PROGRAM should now work properly, respecting
1588 signals received, on the PowerPC platforms (both Linux and
1589 Darwin). (thanks to Pierre Mai for pointing out the location of
1591 * several fixes on OS X: The system now builds and runs cleanly on
1592 Panther (10.3), and works around sigreturn bug (no more SIGFPEs).
1593 (thanks to Brian Mastenbrook)
1594 * bug fix: DECODE-UNIVERSAL-TIME now accepts timezone arguments with
1595 second-resolution: integer multiples of 1/3600 between -24 and 24.
1596 (thanks to Vincent Arkesteijn)
1597 * bug fix: functions =, /=, <, <=, >, >= did not check the argument
1598 type when called with 1 argument; PEEK-CHAR checked type of
1599 PEEK-TYPE only after having read first character from a
1600 stream. (reported by Peter Graves)
1601 * bug fix: the garbage collector now has much better locality
1602 behaviour, and in particular no longer treats all memory as being
1603 exhausted when many small objects point to each other in a deeply
1605 * bug fix: arrays specialized on (UNSIGNED-BYTE 15) are now
1606 recognized as being TYPEP their class.
1607 * bug fix: the PUSHNEW documentation string has been corrected.
1608 (thanks to Vincent Arkesteijn)
1609 * bug fix: defaulting of the value for the last of an atypically
1610 large number of multiple values being bound was not being
1611 performed correctly on the Alpha or PPC platforms
1612 * optimization: implemented multiplication as a modular
1613 (UNSIGNED-BYTE 32) operation on the x86 backend.
1614 * optimization: SEARCH on simple-base-strings can now be open-coded.
1615 (see also contrib/compiler-extras.lisp for inspiration for
1616 teaching the compiler about the Boyer-Moore algorithm).
1617 * value, returned by MAX (and MIN) called with several EQUALP, but
1618 not EQL, arguments now does not depend on compiler settings.
1619 * fixed some bugs revealed by Paul Dietz' test suite:
1620 ** in stack analysis liveness information is propagated from
1621 non-local entry points.
1622 ** pathwise CAST removing failed when the CAST node did not start
1624 ** INPUT-STREAM-P, OUTPUT-STREAM-P, STREAM-ELEMENT-TYPE and
1625 OPEN-STREAM-P signal a TYPE-ERROR if their argument is not a
1627 ** LOAD-LOGICAL-PATHNAME-TRANSLATIONS returns NIL if the logical
1628 host is already defined.
1629 ** RENAME-FILE works on streams instead of signalling an internal
1631 ** PEEK-CHAR uses the current readtable when determining whether
1632 or not a character is whitespace.
1633 ** MERGE-PATHNAMES handles the case when the pathname does not
1634 specify a name while the default-pathname specifies a version.
1635 ** Pathnames now stand a better chance of respecting print/read
1637 ** Attempting to use standardized file system operators with a
1638 pathname with invalid :DIRECTORY components signals a
1640 ** OPEN :DIRECTION :IO no longer fails to work on non-existent
1642 ** DIRECTORY on logical pathnames is more correct.
1643 ** CLEAR-INPUT, CLEAR-OUTPUT, FINISH-OUTPUT and FORCE-OUTPUT
1644 signal a TYPE-ERROR if their argument is not a stream.
1645 ** READ-BYTE and WRITE-BYTE signal a TYPE-ERROR if their stream
1646 designator argument does not designate a stream.
1647 ** OPEN-STREAM-P and INPUT-STREAM-P on synonym streams work by
1648 examining the synonym.
1649 ** STREAM-ELEMENT-TYPE and FRESH-LINE on broadcast-streams now
1651 ** OPEN and WITH-OPEN-STREAM allow opening streams with
1652 element-type larger than ([UN]SIGNED-BYTE 32).
1654 changes in sbcl-0.8.7 relative to sbcl-0.8.6:
1655 * When built with the :SB-FUTEX feature, threaded builds now take
1656 advantage of the "fast userspace mutex" facility in Linux kernel 2.6
1657 for faster/more reliable mutex and condition variable support.
1658 * Incompatible change (but one you probably shouldn't have been using
1659 anyway): the interface and code for arbitrating between multiple
1660 threads in the same user session has been redesigned.
1661 * bug fix: GET-SETF-EXPANSION no longer throws an internal type
1662 error when called without an explicit environment argument.
1663 (thanks to Nikodemus Siivola)
1664 * bug fix: buffered :DIRECTION :IO streams are less likely to become
1665 confused about their position. (thanks to Adam Warner and Gerd
1667 * bug fix: Pretty printing backquoted forms with unquotations in the
1668 argument list position of various code constructs such as LAMBDA
1669 now works correctly. (reported by Paul Dietz)
1670 * bug fix: Pretty printing unquotations no longer loses all
1671 stream position information.
1672 * optimization: performance of string output streams is now less
1673 poor for multiple small sequence writes.
1674 * optimization: performance of CSUBTYPEP in the presence of complex
1675 expressions involving CONS and NOT many times has been improved.
1676 (reported by Paul Dietz)
1677 * ASDF-INSTALL bug fix: now parses *PROXY* properly. (thanks to
1679 * SB-SIMPLE-STREAMS enhancement: simple-streams can now be used as
1680 streams for the REPL, for the debugger, and so on. (thanks to
1682 * DEFINE-CODITION is more efficient. (thanks to Brian Mastenbrook)
1683 * fixed some bugs revealed by Paul Dietz' test suite:
1684 ** the value of the :REHASH-THRESHOLD argument to MAKE-HASH-TABLE
1685 is ignored if it is too small, rather than propagating through
1686 to cause DIVIDE-BY-ZERO or FLOATING-POINT-OVERFLOW errors.
1687 ** extremely complex negations of CONS types were not being
1688 sufficiently canonicalized, leading to inconsistencies in
1690 ** VALUES tranformer lost derived type.
1692 changes in sbcl-0.8.6 relative to sbcl-0.8.5:
1693 * fixed a bootstrapping bug: the build process no longer assumes
1694 that the various BOOLE-related constants have the same value in
1695 host and target lisps. (noted by Paul Dietz' test suite on an
1696 SBCL binary built from CLISP)
1697 * The system can now be dynamically linked on the MIPS platform,
1698 which enables dynamic loading of foreign code from Lisp. (thanks
1699 to Ralf Baechle for discussions on the MIPS ABI)
1700 * The system now records debugging information for its own source
1701 files in a filesystem-position-independent manner, relative to
1702 the "SYS" logical host.
1703 * fixed a compiler bug: MV-LET convertion did not check references
1704 to the "max args" entry point. (reported by Brian Downing)
1705 * tweaked disassembly notes to be less confident about proclaiming
1706 some instruction as an LRA. (thanks to Brian Downing)
1707 * contrib update: SB-ACLREPL is now threadsafe; multiple listeners
1708 now each have their own history, command character, and other
1709 characteristics. (thanks to David Lichteblau)
1710 * fixed some bugs revealed by Paul Dietz' test suite:
1711 ** compiler failure in compiling LOGAND expressions including a
1713 ** Implementation of ASH-MOD32 on X86 and PPC did not work for the
1714 shift greater than 32.
1715 ** FLUSH-DEST did not mark blocks for type check regeneration.
1716 ** HANDLER-CASE failed to accept declarations in handler clauses
1717 in some circumstances.
1719 changes in sbcl-0.8.5 relative to sbcl-0.8.4:
1720 * New code in contrib/sb-introspect (still probably not entirely
1721 stable yet) provides some support for smart Lisp development
1722 environments like SLIME.
1723 * The conditions signalled for errors occurring when loading .fasl
1724 files have been systematized (inheriting from SB-EXT:INVALID-FASL)
1725 in a way which should help ASDF recover gracefully.
1726 * The REQUIRE/PROVIDE behavior of *MODULE-PROVIDER-FUNCTIONS*
1727 stuff has been cleaned up. If you code contrib/ stuff, this might
1728 affect you, and you can look at contrib/README, contrib/STANDARDS,
1729 and/or the 0.8.4.27 diff to check.
1730 * In full calls the compiler now does not generate checks for declared
1731 argument types for all arguments.
1732 * various threading fixes
1733 ** and some experimental patches which didn't make it into
1734 the main tree for this release, but which are shipped in
1735 contrib/experimental-thread.patch as a possible fix for some
1736 failures (deadlock, spinning...) in GC-intensive multithreaded
1738 * fixed PPC build problem (source code incompatibility of different
1739 library versions): added offsetof() hackery which attempts to divine
1740 where glibc maintainers put uc_mcontext today
1741 * fixed bug 282: compiler does not trust type assertions while passing
1742 arguments to a full call.
1743 * fixed bug 261: compiler allows NIL or "no value" to be accepted for
1744 &OPTIONAL VALUES type parameter.
1745 * fix bug 214: algorithm for noting rejected templates is now more
1746 similar to that of template seletion. (also reported by rydis on
1748 * fixed bug 141b: printing backquoted information readably and prettily
1749 inserts a space where necessary.
1750 * bug fix: obviously wrong type specifiers such as (FIXNUM 1) or
1751 (CHARACTER 10) are now reported as errors, rather than propagated
1752 as unknown types. (reported by piso on #lisp)
1753 * bug fix: the :IF-EXISTS argument to OPEN now behaves correctly
1754 with values NIL and :ERROR. (thanks to Milan Zamazal)
1755 * fixed bug 191c: CLOS now does proper keyword argument checking as
1756 described in CLHS 7.6.5 and 7.6.5.1.
1757 * bug fix: LOOP forms using NIL as a for-as-arithmetic counter no
1758 longer raise an error; further, using a list as a for-as-arithmetic
1759 counter now raises a meaningful error.
1760 * fixed bug 213a: even fairly unreasonable CONS type specifiers are
1761 now understood by sequence creation functions such as MAKE-SEQUENCE
1763 * fixed bug 46k: READ-BYTE now signals an error when asked to read from
1764 a STRING-INPUT-STREAM.
1765 * compiler enhancement: SIGNUM is now better able to derive the type
1767 * type declarations inside WITH-SLOTS are checked. (reported by
1769 * fixed some bugs revealed by Paul Dietz' test suite:
1770 ** incorrect optimization of TRUNCATE for a positive first
1771 argument and negative second.
1772 ** compiler failure in let-convertion during flushing dead code.
1773 ** compiler failure while deriving type of TRUNCATE on an
1774 interval, containing 0.
1775 ** ASH of a negative bignum by a negative bignum count now returns
1777 ** intersection of CONS types now canonicalizes properly, fixing
1778 inconsistencies in SUBTYPEP.
1780 changes in sbcl-0.8.4 relative to sbcl-0.8.3:
1781 * incompatible change: The --disable-debugger command line
1782 option now clobbers the debugger at a more fundamental
1783 level, by redefining #'INVOKE-DEBUGGER instead of by
1784 rebinding *DEBUGGER-HOOK*. The main difference is that BREAK
1785 is specified by ANSI to ignore *DEBUGGER-HOOK* and
1786 INVOKE-DEBUGGER regardless. Under the old system, BREAK would
1787 enter the debugger REPL and then suffer recursive errors
1788 because *DEBUG-IO* is also messed up in --disable-debugger mode;
1789 while under the new system, BREAK in --disable-debugger mode
1790 terminates the system just as an unhandled error would.
1791 * fixed compiler performance when processing loops with a step >1;
1792 * bug fix: DOCUMENTATION now retrieves generic function
1793 documentation. Also, DOCUMENTATION and (SETF DOCUMENTATION)
1794 support has been systematized, and now supports the methods
1795 specified by ANSI, along with a default method and a method for
1796 slot documentation. (reported by Nathan Froyd)
1797 * bug fix: effective methods associated with a generic function are
1798 no longer cached over a change of that generic function's method
1799 combination. (reported by Andreas Fuchs)
1800 * bug fix: RUN-PROGRAM now does not fail if some element in $PATH
1801 names a non-existent directory. (thanks to Andreas Fuchs)
1802 * bug fix: ROUND and TRUNCATE could, under certain circumstances on
1803 the PPC platform, lead to stack corruption; this has been fixed.
1804 (reported by Rainer Joswig)
1805 * bug fix: ASH on an (UNSIGNED-BYTE 32) with a shift of -32 or lower
1806 no longer ever returns 1 instead of 0. (thanks to Lars Brinkhoff)
1807 * fixed bug 285: TRUNCATE on bignum arguments, and indeed bignum
1808 arithmetic in general, is now much more reliable on the PPC
1810 * bug fix: LOGCOUNT on (UNSIGNED-BYTE 64) objects on the Alpha
1811 platform now returns the right answer.
1812 * optimization: restored some effective method precomputation in
1813 CLOS (turned off by an ANSI fix in sbcl-0.8.3); the amount of
1814 precomputation is now tunable.
1815 * optimization: compiler-internal data structure use has been
1816 reviewed, and changes have been made that should improve the
1817 performance of the compiler by about 20%.
1818 * optimization: performance of FILL (and :INITIAL-ELEMENT) on
1819 simple-base-strings and simple-bit-vectors is improved.
1820 * optimization: the optimization of 32-bit logical and arithmetic
1821 functions introduced in version 0.8.3 on the x86 has been
1822 implemented on the mips, ppc and sparc platforms; an
1823 implementation of the same facility, but for 64-bit arithmetic,
1824 has been added for the alpha.
1825 * microoptimization: the compiler is better able to make use of the
1826 x86 LEA instruction for multiplication by constants.
1827 * bug fix: in some situations compiler did not report usage of
1828 generic arithmetic in (SPEED 3) policy.
1829 * bug 145b fix: compiler used wrong type specifier while converting
1830 MEMBER-types to numeric.
1831 * bug fix: COMPILE-FILE must bind *READTABLE*. (reported by Doug
1833 * bug fix: (SETF AREF) on byte-sized-element arrays with constant
1834 index argument now works properly on the Alpha platform.
1835 * bug fix: floating point exception treatment on the Alpha platform
1837 * bug fix: FILE-POSITION works much better on string input and
1838 output streams. (thanks to Nikodemus Siivola)
1839 * bug fix: many threading/garbage collection symptoms sorted.
1840 SB-THREAD:INTERRUPT-THREAD now safe to call on a thread that
1841 might be pseudo-atomic.
1842 * internal change: Stopping for GC is now done with signals not
1843 ptrace. GC is now done in whichever thread wanted it, instead of
1845 * bug fix: GC hooks (missing since 0.8) reinstated, so finalizers
1847 * bug fix: result form in DO is not contained in the implicit
1849 * incompatible change: ICR structure is changed; the value part of
1850 CONTINUATION is now called LVAR; corresponding functions are
1851 renamed (e.g. SB-C::CONTINUATION-TYPE has become SB-C::LVAR-TYPE).
1852 * added type deriver for ISQRT (thanks to Robert E. Brown).
1853 * bug fix: better support for loading from the command line when an
1854 initialization file sets (READTABLE-CASE *READTABLE*). (thanks
1856 * fixed some bugs revealed by Paul Dietz' test suite:
1857 ** the RETURN clause in LOOP is now equivalent to DO (RETURN ...).
1858 ** ROUND and FROUND now give the right answer when given very
1859 small float arguments.
1860 ** (FLOAT X) for X of type DOUBLE-FLOAT now returns X in all
1862 ** optimizer for (EXPT X 0) did not work for X not of type FLOAT.
1863 ** (GCD 0 <negative-integer>) returned <negative-integer>.
1864 ** LCM should return a non-negative integer.
1865 ** PARSE-INTEGER returned the index of a terminator instead of the
1866 upper bounding index of a substring in case :JUNK-ALLOWED NIL.
1867 ** PARSE-INTEGER returned an incorrect index being applied to a
1869 ** LCM with two arguments of 0 returns 0 rather than signalling
1871 ** unsigned addition of a 32-bit constant with the high bit set no
1872 longer causes an internal compiler error on the x86.
1873 ** LOGBITP accepts a non-negative bignum as its INDEX argument.
1874 ** compiler incorrectly derived types of DPB and DEPOSIT-FIELD
1875 with negative last argument.
1876 ** byte specifiers with zero size and position no longer cause
1877 an error during type derivation.
1878 ** bignum multiplication on the Alpha platform now returns the
1880 * porting: The system now builds on SuSE AMD64, although it still
1881 generates a 32-bit binary.
1882 * .fasl file incompatibility: The fasl file version number has
1883 been incremented (because of the changes to internal compiler
1884 data structures referred to above).
1886 changes in sbcl-0.8.3 relative to sbcl-0.8.2:
1887 * SBCL now builds and runs on MacOS X (version 10.2), or perhaps
1888 more accurately, on the Darwin kernel running on PowerPC hardware.
1889 (thanks to Brian Mastenbrook, Pierre Mai and Patrik Nordebo)
1890 * Compiler code deletion notes now signal a condition of type
1891 SB-EXT:CODE-DELETION-NOTE (a subtype of SB-EXT:COMPILER-NOTE) with
1892 an associated MUFFLE-WARNING restart.
1893 * The compiler now performs limited argument count validation of
1894 constant format strings in FORMAT, and where appropriate in ERROR,
1895 CERROR and WARN. (thanks to Gerd Moellmann)
1896 * New ASDF-INSTALL contrib can be used for automatic download and
1897 installation of third-party Lisp code from CCLAN or other sites
1899 * Threaded builds (:SB-THREAD) now support SB-THREAD:INTERRUPT-THREAD,
1900 which forces another thread to execute a function supplied by the
1902 * bug 75 fix: WITH-OUTPUT-TO-STRING (and MAKE-STRING-OUTPUT-STREAM)
1903 now accept and act upon their :ELEMENT-TYPE keyword argument.
1904 (reported by Martin Atzmueller, Edi Weitz)
1905 * bug fix: FILE-POSITION now accepts position designators up to
1906 ARRAY-DIMENSION-LIMIT or the extreme of the off_t range, whichever
1907 is the greater. (thanks to Patrik Nordebo)
1908 * bug fix: MAKE-ARRAY ignored :INITIAL-CONTENTS NIL. (reported by
1909 Kalle Olavi Niemitalo)
1910 * bug fix: the CLASS-PROTOTYPE of the GENERIC-FUNCTION class is now
1911 printable. (reported by Eric Marsden)
1912 * bug fix in sb-posix: mmap() now works on systems with a 64-bit
1913 off_t, including Darwin and FreeBSD. (thanks to Andreas Fuchs)
1914 * x86 bug fix in control stack exhaustion checking: now shows backtrace
1915 * bug fix in WITH-TIMEOUT: now the body can have more than one form.
1916 (thanks to Stig Sandoe)
1917 * bug fix in READ-SEQUENCE: READ-SEQUENCE following PEEK-CHAR or
1918 UNREAD-CHAR now correctly includes the unread character in the
1919 target sequence. (thanks to Gerd Moellmann)
1920 * bug fix in threaded builds: the system can now be suspended and
1921 resumed by shell job control with minimal disruption.
1922 * bug fixes in times and timezones >2038AD
1923 * better handling of "where is GNU make?" problem in build scripts
1924 (thanks to Nikodemus Siivola)
1925 * new optimization: inside a named function any reference to a
1926 function with the same name is considered to be a self-reference;
1927 this behaviour is controlled with SB-C::RECOGNIZE-SELF-CALLS
1928 optimization quality.
1929 * new optimization on x86: logical functions and + now have
1930 optimized (UNSIGNED-BYTE 32) versions, which are automatically
1931 used when the result is truncated to 32 bits.
1932 * VALUES declaration is partially enabled.
1933 * fixes in SB-GROVEL (thanks to Andreas Fuchs)
1934 * fixed some bugs revealed by Paul Dietz' test suite:
1935 ** The system now obeys the constraint imposed by
1936 UPGRADED-ARRAY-ELEMENT-TYPE that the upgraded array element
1937 types form a lattice under type intersection.
1938 ** FFLOOR, FTRUNCATE, FCEILING and FROUND work with integers.
1939 ** ASSOC now ignores NIL elements in an alist.
1940 ** CEILING now gives the right answer with MOST-NEGATIVE-FIXNUM
1941 and (1+ MOST-POSITIVE-FIXNUM) answers.
1942 ** The addition of a method with invalid qualifiers to a generic
1943 function does not cause an error to be signalled immediately;
1944 a warning is signalled, and the error is generated only on
1945 calling the generic function.
1946 * changed .fasl file version number, in part to add type codes for
1947 new array subtypes UNSIGNED-BYTE 7, 15, 29, and 31 mandated by
1948 obscure ANSI requirements
1950 changes in sbcl-0.8.2 relative to sbcl-0.8.1:
1951 * fixed bug 148: failure to inline-expand a local function left
1952 garbage, confusing the compiler.
1953 * fixed bugs 3cd: structure slot readers perform type check if the
1954 slot can have an invalid value (i.e. it is either not initialized
1955 or can be written with a less specific slot writer).
1956 * bug fix: the compiler now traps array references to elements off
1957 the end of an array; previously, the bounds checking in some
1958 circumstances could go off-by-one.
1959 * improved MACHINE-VERSION, especially on Linux (thanks to Lars
1961 * type declarations for array element types now obey the description
1962 on the CLHS page "Declaration TYPE", as per discussions on
1963 sbcl-help around 2003-05-08. This means that a declaration
1964 (TYPE (ARRAY FOO) BAR) means that, within the scope of the
1965 declaration, all references to BAR will be asserted or assumed
1966 (with THE, so dependent on compiler policy) to involve objects of
1967 type FOO. Note that no such declaration is implied in
1968 (MAKE-ARRAY .. :ELEMENT-TYPE 'FOO).
1969 * declared types of functions from the "Conditions"
1970 chapter. (reported by Paul Dietz)
1971 * bug fix: CERROR accepts a function as its first argument.
1972 * bug fix: NTH an NTHCDR accept a bignum as index
1973 arguments. (reported by Adam Warner)
1974 * optimization: character compare routines now optimize comparing
1975 against a constant character. (reported by Gilbert Baumann)
1976 * bug fix: (SETF AREF) on byte-sized-element arrays with constant index
1977 argument now works properly on the MIPS platform.
1978 * fixed compiler failure on (TYPEP x '(NOT (MEMBER 0d0))).
1979 * repeated evaluation of the same DEFSTRUCT, a slot of which is
1980 declared to have a functional type, does not cause an error
1982 * fixed bug: sometimes MAKE-INSTANCE did not work with classes with
1983 many :DEFAULT-INITARGS. (reported by Istvan Marko)
1984 * fixed bug: if last continuation of a deleted block has a
1985 destination, this destination should be deleted too. (reported by
1987 * fixed a bug in the bootstrap process: the host compiler's values
1988 of ARRAY-DIMENSION-LIMIT and ARRAY-TOTAL-SIZE-LIMIT no longer leak
1989 into the newly-built SBCL. (reported by Eric Marsden on #lisp,
1990 test case from Patrik Nordebo)
1991 * improved the ability of the disassembler on the PPC platform to
1992 provide helpful disassembly notes.
1993 * SB-MOP:CLASS-PROTOTYPE on built-in-classes returns an instance of
1994 the class in more cases than previously.
1995 * bug fix: FILE-POSITION now understands :START and :END for
1996 STRING-INPUT-STREAMs. (thanks to Nikodemus Siivola)
1997 * bug fix: (SIGNED-BYTE 8) streams no longer return (UNSIGNED-BYTE
1998 8) data. (thanks to David Lichteblau)
1999 * bug fix: it is possible to add a method to a generic function
2000 without lambda list.
2001 * bug fix: reader failed to signal END-OF-FILE inside an
2002 object representation. (reported by Nikodemus Siivola)
2003 * fixed some bugs revealed by Paul Dietz' test suite:
2004 ** LAST and [N]BUTLAST should accept a bignum.
2005 ** condition slot accessors are methods.
2006 ** (VECTOR NIL) is a subtype of STRING.
2008 changes in sbcl-0.8.1 relative to sbcl-0.8.0:
2009 * minor incompatible change: some nonsensical specialized lambda
2010 lists (used in DEFMETHOD) which were previously ignored now signal
2012 * minor incompatible change: the system is now aware of the types of
2013 variables in the COMMON-LISP package, and will signal errors for
2014 most violations of these type constraints (where previously they
2015 were silently accepted).
2016 * minor incompatible change: COMPILE-FILE now uses the freedom
2017 afforded (ANSI 3.2.2.3) to use derived function types for
2018 functions defined in the same file. This also permits the system
2019 to warn on static type mismatches and function
2020 redefinition. (Currently it does not work with high DEBUG level.)
2021 * minor incompatible change: VALUES declaration is disabled.
2022 * When issuing notes, the compiler now signals a condition of type
2023 SB-EXT:COMPILER-NOTE, and provides an associated MUFFLE-WARNING
2024 restart for use in user handlers. It is expected that the
2025 COMPILER-NOTE condition will eventually become a condition
2026 supertype to a hierarchy of note types, which will then be
2027 handleable in a similar fashion. However, at the moment, no such
2028 note subtypes yet exist. (SB-INT:SIMPLE-COMPILER-NOTE exists,
2029 but it's an implementation detail, not a classification for the
2031 * Changes in type checking closed the following bugs:
2032 ** type checking of unused values (192b, 194d, 203);
2033 ** template selection based on unsafe type assertions (192c, 236);
2034 ** type checking in branches (194bc).
2035 * A short form of VALUES type specifier has ANSI meaning (it has
2036 increased the number of situations when SBCL cannot perform type
2038 * fixed bug in DEFSTRUCT: once again, naming structure slots with
2039 keywords or constants is permissible.
2040 * STREAM-READ-SEQUENCE and STREAM-WRITE-SEQUENCE now have methods
2041 defined on the relevant FUNDAMENTAL-BINARY-{INPUT,OUTPUT}-STREAM
2042 classes. (thanks to Antonio Martinez)
2043 * improved ANSIness in DESCRIBE: The DESCRIBE function no longer
2044 outputs FRESH-LINE or TERPRI, and no longer converts its stream
2045 argument to a pretty-print stream. Instead, it leaves any such
2046 operations to DESCRIBE-OBJECT methods.
2047 * bug fix: APROPOS now respects the EXTERNAL-ONLY flag. (reported
2049 * bug fix: NIL is now a valid destructuring argument in DEFMACRO
2050 lambda lists. (thanks to David Lichteblau)
2051 * bug fix: Defining a generic function with a :METHOD-CLASS being a
2052 subclass of STANDARD-METHOD no longer causes stack exhaustion.
2053 (thanks to Gerd Moellmann)
2054 * fixed bug 246: increased compilation speed of long
2055 MULTIPLE-VALUE-BIND (and likewise of NTH-VALUE with a constant
2057 * a contributed module implementing COMPILER-LET and MACROEXPAND-ALL
2059 * DEFCONSTANT now throws a condition of type
2060 SB-EXT:DEFCONSTANT-UNEQL if it is being asked to redefine a
2061 constant to a non-EQL value; CONTINUE and ABORT restarts
2062 respectively change and preserve the value.
2063 * fixed bug 63: The code walker, part of the implementation of CLOS,
2064 is now better at handling symbol macros.
2065 * bug fix: There is no longer an internal implementation type named
2066 CL:LENGTH. (reported by Raymond Toy)
2067 * bug fix: In macro-like defining macros/special operators the
2068 implicit block does not enclose the lambda list.
2069 * fixed bugs 10 and 43: Bare VALUES, AND, OR and MEMBER symbols (not
2070 enclosed in parentheses) are not suitable as type specifiers, and
2071 their use properly signals an error now.
2072 * bug fix: An argument count mismatch for a type specifier in code
2073 being compiled no longer causes an unhandled error at compile
2074 time, but signals a compile-time warning.
2075 * fixed simple vector readable printing
2076 * bug fix: DESCRIBE takes more care over whether the class
2077 precedence list slot of a class is bound before accessing it.
2078 (reported by Markus Krummenacker)
2079 * bug fix: FORMATTER can successfully compile pretty-printer format
2080 strings which use variants of the ~* directive inside.
2081 * bug fix: SEARCH now applies its TEST predicate to the elements of
2082 the arguments in the correct order. (thanks to Wolfhard Buss)
2083 * fixed bug 235b: The compiler uses return types of MAPCAR and friends
2084 in type inference. (thanks to Robert E. Brown)
2085 * bug fix: Reading in symbols with an explicit package name of ""
2086 (e.g. '||::FOO) now works correctly. (reported by Henrik Motakef)
2087 * fixed some bugs revealed by Paul Dietz' test suite:
2088 ** NIL is now allowed as a structure slot name.
2089 ** Arbitrary numbers, not just REALs, are allowed in certain
2090 circumstances in LOOP for-as-arithmetic clauses.
2091 ** Multiple class redefinitions before slot access no longer
2092 causes a type error.
2093 ** (SETF FIND-CLASS) now accepts NIL as an argument to remove the
2094 association between the name and a class.
2095 ** Generic functions with non-standard method-combination and over
2096 five methods all of which return constants no longer return NIL
2097 after the first few invocations. (thanks to Gerd Moellmann)
2098 ** CALL-NEXT-METHOD with no arguments now passes the original
2099 values of the arguments, even in the presence of assignment.
2100 ** Functions [N]SUBST*, LAST, NRECONC, [N]SUBLIS may return any
2102 ** DISASSEMBLE works with closures and funcallable instances.
2103 ** ADD-METHOD now returns the generic function, not the new method.
2104 ** FIND-METHOD signals an error if the lengths of the specializers
2105 is incompatible with the generic function, even if the ERRORP
2107 ** TYPE-OF returns recognizeable subtypes of all built-in-types of
2108 which its argument is a member.
2109 ** DEFCLASS only redefines the class named by its class-name
2110 argument if that name is the proper name of the class;
2111 otherwise, it creates a new class.
2112 ** SLOT-UNBOUND now correctly initalizes the CELL-ERROR-NAME slot
2113 of the UNBOUND-SLOT condition to the name of the slot.
2114 ** (SETF (AREF bv 0) ...) did not work for bit vectors.
2115 ** SLOT-UNBOUND and SLOT-MISSING now have their return values
2116 treated by SLOT-BOUNDP, SLOT-VALUE, (SETF SLOT-VALUE) and
2117 SLOT-MAKUNBOUND in the specified fashion.
2119 changes in sbcl-0.8.0 relative to sbcl-0.8alpha.0
2120 * SBCL now builds using CLISP (version of late April 2003 from CVS) as
2121 cross-compilation host. As a consequence, we can now bootstrap our
2122 way up to SBCL starting with a bare gcc toolchain and human-readable
2123 source code (first the source to CLISP, then the source to SBCL).
2124 * A contributed module containing a partial implementation of the
2125 simple-streams interface has been included. (thanks to Rudi
2127 * A contributed module implementing the RFC1321 Message Digest
2128 Algorithm, known as MD5, has been included.
2129 * minor incompatible change: The :NEGATIVE-ZERO-IS-NOT-ZERO feature
2130 no longer has any effect, as the code controlled by this feature
2131 has been deleted. (As far as we know, no-one has ever built using
2132 this feature, and its semantics were confused in any case).
2133 * minor incompatible change: As a consequence of making SLOT-EXISTS-P
2134 work on conditions (as required by the ANSI specification),
2135 SLOT-VALUE, (SETF SLOT-VALUE) and SLOT-BOUNDP likewise have the
2136 expected behaviour on conditions. Users should note, however,
2137 that such behaviour is not required by the ANSI specification,
2138 and so use of this behaviour may render their code unportable.
2139 * fixed some bugs revealed by Paul Dietz' test suite:
2140 ** the GENERIC-FUNCTION type is no longer disjoint from FUNCTION
2142 ** &ENVIRONMENT parameter in macro lambda list is bound first.
2143 ** SXHASH on condition objects no longer returns NIL.
2144 ** :ALLOCATION :CLASS slots are better treated; their values are
2145 updated on class redefinition, and initforms inherited from
2146 superclasses are applied.
2147 ** REMOVE-METHOD returns its generic function argument even when
2148 no method was removed.
2149 ** SHARED-INITIALIZE now initializes the values of the requested
2150 slots, including those with :ALLOCATION :CLASS.
2151 ** ALLOCATE-INSTANCE now works on structure classes defined via
2152 DEFSTRUCT (and not just by those from DEFCLASS :METACLASS
2154 ** SLOT-EXISTS-P now works on conditions, as well as structures
2156 ** MAKE-LOAD-FORM now has the required methods on
2157 STRUCTURE-OBJECT, CONDITION and STANDARD-OBJECT.
2158 ** MAKE-LOAD-FORM-SAVING-SLOTS no longer returns a special
2159 keyword, and now implements the SLOT-NAMES argument.
2160 ** methods with &OPTIONAL arguments no longer allow too many
2161 arguments to be passed in the call without error.
2162 ** DEFGENERIC now checks that the :ARGUMENT-PRECEDENCE-ORDER
2163 option is consistent with the required arguments of the generic
2164 function lambda list.
2165 * bug fix: REQUIRE accepts a string designator. (Thanks to
2167 * bug fix: SB-MOP:DIRECT-SLOT-DEFINITION-CLASS and
2168 SB-MOP:EFFECTIVE-SLOT-DEFINITION-CLASS now have the
2169 specified-by-AMOP lambda list of (CLASS &REST INITARGS).
2170 * bug fix: The compiler now checks for duplicated variables in macro
2172 * bug fix: SETQ on globals returns the correct value.
2173 * fixed bug 47.d: (DEFGENERIC IF (X)) now signals a PROGRAM-ERROR,
2174 not a COMPILER-ERROR (followed by some other strange error on
2175 choosing the CONTINUE restart).
2176 * bug fix: make.sh and friends are now more consistent in the way that
2177 they look for GNU "make".
2179 changes in sbcl-0.8alpha.0 relative to sbcl-0.7.14
2180 * experimental native threads support (on x86 Linux >=2.4 only).
2181 This is not compiled in by default: you need to add :SB-THREAD to
2182 the target features. See the "Beyond ANSI" chapter of the manual
2184 * fix for longstanding nonANSIism: The old distinction between
2185 CL:CLASS objects and SB-PCL:CLASS objects has been eliminated.
2186 The return value from CL:FIND-CLASS is now a CLOS class, and
2187 likewise that of CL:CLASS-OF; CL:BUILT-IN-CLASS,
2188 CL:STRUCTURE-CLASS and CL:STANDARD-CLASS name CLOS classes.
2189 * An interface to the MetaObject Protocol, as described in Kiczales,
2190 des Rivieres and Bobrow's "The Art of the Metaobject Protocol",
2191 MIT Press, 1991, is available from the SB-MOP package.
2192 * incompatible change: the SB-PCL package should now be considered
2193 a private implementation detail, and no longer a semi-private MOP
2195 * minor incompatible change: due to rearrangement for threads, the
2196 control stack and binding stack are now allocated at arbitrary
2197 addresses instead of being hardcoded per-port. Users affected by
2198 this probably have to be doing advanced things with shared
2199 libraries, and will know who they are.
2200 * minor incompatible change: Previously, all --eval forms used were
2201 processed with READ before any of them were processed with EVAL.
2202 Now each --eval form is processed with both READ and EVAL before
2203 the next --eval form is processed. (Thus package operations like
2204 sbcl --eval "(defpackage :foo)" --eval "(print 'foo::bar)" now
2205 work as the user might reasonably expect.)
2206 * minor incompatible change: *STANDARD-INPUT* is now only an
2207 INPUT-STREAM, not a BIDIRECTIONAL-STREAM. (thanks to Antonio
2209 * minor incompatible change: Y-OR-N-P is now character-oriented,
2210 not line oriented. Also, YES-OR-NO-P now works without errors.
2211 (thanks to Antonio Martinez)
2212 * sb-aclrepl module improvements: an integrated inspector, added
2213 repl features, and a bug fix to :trace command.
2214 * Known functions, which cannot be open coded by the backend, are
2215 considered to be able to check types of their arguments. (fixing
2216 a bug report by Nathan J. Froyd)
2217 * fixed a bug in computing method discriminating functions: It is
2218 now possible to define methods specialized on classes which have
2219 forward-referenced superclasses. (thanks to Gerd Moellmann)
2220 * fixed evaluation order in optional entries (reported by Gilbert
2222 * SB-MOP:ENSURE-CLASS-USING-CLASS now takes its arguments in the
2223 specified-by-AMOP order of (CLASS NAME &REST ARGS &KEY).
2224 * SB-MOP:COMPUTE-EFFECTIVE-SLOT-DEFINITION now takes the
2225 required-by-AMOP NAME argument, as well as CLASS and
2226 DIRECT-SLOT-DEFINITIONS. (thanks to Kevin Rosenberg)
2227 * fixed bug 20: DEFMETHOD can define methods using names that are
2228 not the proper names of classes to designate class specializers.
2229 * bug fix: INTERACTIVE-STREAM-P now works on streams associated with
2230 Unix file descriptors, instead of blowing up. (thanks to Antonio
2232 * Garbage collection refactoring: user-visible change is that a
2233 call to the GC function during WITHOUT-GCING will not do garbage
2234 collection until the end of the WITHOUT-GCING. If you were doing
2235 this you were probably losing anyway.
2236 * fixed bug in MEMBER type: (MEMBER 0.0) is not the same as
2237 (SINGLE-FLOAT 0.0 0.0), because of the existence of -0.0 which is
2238 TYPEP the latter but not the former.
2239 * The compiler issues a full WARNING for calls to undefined functions
2240 with names from the CL package.
2241 * MAP-INTO for a vector destination is open coded. (reported by
2242 Brian Downing on c.l.l)
2243 * bug fix: the long form of DEFINE-METHOD-COMBINATION now accepts a
2244 documentation string.
2245 * fixed some bugs revealed by Paul Dietz' test suite:
2246 ** COPY-ALIST now signals an error if its argument is a dotted
2248 ** Condition slots are now accessed more correctly in the presence
2249 of multiple initargs for a given slot.
2250 ** The USE-VALUE, CONTINUE and STORE-VALUE functions now correctly
2251 exclude restarts of the same name associated with a different
2253 ** DEFCLASS of forward-referenced classes with another
2254 forward-referenced class in the superclasses list no longer
2256 ** Condition slots are now initialized once each, not multiple
2257 times. (thanks to Gerd Moellmann)
2258 ** CONVERT-MORE-CALL failed on a lambda list (&KEY). (thanks to
2260 ** &WHOLE and &REST arguments in macro lambda lists are patterns.
2261 ** NSET-EXCLUSIVE-OR does not return extra elements when its
2262 arguments contain duplicated elements.
2263 ** RESTART-CASE understands local macros.
2264 ** RESTART-CASE associates exactly its own restarts with a condition.
2265 ** ENDP in safe mode checks its argument to be of type LIST.
2266 ** COPY-SYMBOL in a threaded build no longer fails when the symbol
2267 in question is unbound.
2268 ** Optimized MAKE-INSTANCE functions no longer cause internal
2269 assertion failures in the presence of duplicate initargs.
2270 ** SLOT-MAKUNBOUND returns the instance acted upon, not NIL.
2271 ** Side-effectful :DEFAULT-INITARGS have their side-effects
2272 propagated even in the ctor optimized implementation of
2274 ** :ALLOW-OTHER-KEYS NIL is now accepted in an initarg list.
2276 changes in sbcl-0.7.14 relative to sbcl-0.7.13:
2277 * a better implementation of SXHASH on (simple) bit vectors,
2278 measured both in execution speed and in distribution of results
2279 over the positive fixnums, has been installed. Likewise, a better
2280 implementation of EQUAL for simple bit vectors is now available.
2281 * fixed CEILING optimization for a divisor of form 2^k.
2282 * fixed bug 240 (emitting extra style warnings "using the lexical
2283 binding of the symbol *XXX*" for &OPTIONAL arguments). (reported
2284 by Antonio Martinez)
2285 * fixed SXHASH, giving different results for NIL depending on type
2286 declarations (SYMBOL or LIST). (thanks to Gerd Moellmann)
2287 * fixed bug in DEFPARAMETER and DEFVAR: they could assign a lexical
2288 variable. (found by Rolf Wester)
2289 * SBCL does not ignore type declarations for special
2290 variables. (reported by rif on c.l.l 2003-03-05)
2291 * some bug fixes in contrib/sb-aclrepl/
2292 * fixed some bugs revealed by Paul Dietz' test suite:
2293 ** a bug in the CONS type specifier, whereby the CAR and CDR
2294 types got intertwined, has been fixed;
2295 ** the type system is now able to reason about the interaction
2296 between INTEGER and RATIO types more completely;
2297 ** APPEND, [N]REVERSE and NRECONC check that those their
2298 arguments, which must be proper lists, are really so;
2299 ** An array specialized to be unable to hold elements has been
2300 implemented, as required -- yes, really -- by ANSI;
2301 ** GETF and GET-PROPERTIES throw a TYPE-ERROR, not a SIMPLE-ERROR,
2302 on malformed property lists;
2304 changes in sbcl-0.7.13 relative to sbcl-0.7.12:
2305 * incompatible packaging change: in line with Unix convention,
2306 SBCL now looks for its core file in /usr/{local/,}lib/sbcl/sbcl.core
2307 if it's not in $SBCL_HOME. It also sets SBCL_HOME to match.
2308 * REQUIRE and PROVIDE are now optionally capable of doing something
2309 useful. See the documentation string for REQUIRE.
2310 * infrastructure for a managed SBCL contrib system: contributed
2311 modules in this release include:
2312 ** the ASDF system definition facility;
2313 ** an interface to the BSD Sockets API;
2314 ** an ACL-like convenience interface to the repl;
2315 (thanks to Kevin Rosenberg)
2316 ** an implementation of ROTATE-BYTE, with efficient implementation
2318 * fixed a bug in LOG, so that LOG of a rational argument near 1 now
2319 gives a closer approximation to the right answer than previously.
2320 (thanks to Raymond Toy)
2321 * fixed bug 157: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE and
2322 UPGRADED-COMPLEX-PART-TYPE now take (ignored, in all situations)
2323 optional environment arguments, as required by ANSI.
2324 * fixed bugs in other functions taking environment objects, allowing
2325 calls with an explicit NIL environment argument to be compiled
2327 * fixed bug 228: primary return values from
2328 FUNCTION-LAMBDA-EXPRESSION are either NIL or suitable for input to
2329 COMPILE or FUNCTION.
2330 * fixed a bug in DEFSTRUCT: predicates for :NAMED structures with
2331 :TYPE will no longer signal errors on innocuous objects.
2332 * fixed bug 231b: SETQ is better at respecting type declarations in
2333 the lexical environment.
2334 * fixed a bug in DEFCLASS: classes named by symbols with no or
2335 unprintable packages can now be defined.
2336 * fixed a bug in RESTART-BIND: The :TEST-FUNCTION option had been
2337 carelessly renamed to :TEST-FUN. (thanks to Robert E. Brown)
2338 * fixed compiler failure related to checking types of functions.
2339 (reported by Robert E. Brown)
2340 * the compiler is now much more consistent in its error-checking
2341 treatment of bounding index arguments to sequence functions: in
2342 (SAFETY 3) code, errors will be signalled in almost all cases if
2343 invalid sequence bounding indices are passed to functions defined
2344 by ANSI to operate on sequences.
2345 * fixed a bug in the build procedure: documentation of SBCL-specific
2346 packages is now preserved and available in the final Lisp image.
2347 * lifted FDEFINITION lookup out of loops in the implementation of
2348 many list operations. (thanks to Robert E. Brown)
2349 * fixed a bug in the reader: the #n# reader macro now works for
2350 objects of type STANDARD-OBJECT. (reported by Tony Martinez)
2351 * the compiler is now aware that SYMBOL-FUNCTION returns a FUNCTION
2352 and that READ-DELIMITED-LIST returns a LIST. (thanks to Robert
2353 E. Brown and Tony Martinez respectively)
2354 * PCL is now smarter about SLOT-VALUE, (SETF SLOT-VALUE) and
2355 SLOT-BOUNDP: in particular, it is now able to optimize them much
2356 better, and is now not vulnerable to having packages renamed.
2357 Furthermore, a compliance bug has been fixed: SLOT-MISSING is now
2358 always called when a slot is not present in an instance. (thanks
2360 * fixed a bug related to CONCATENATED-STREAMs: PEEK-CHAR will no
2361 longer signal an error on unreading a character following EOF on
2362 the previous constituent stream. (thanks to Tony Martinez)
2363 * fixed some bugs revealed by Paul Dietz' test suite:
2364 ** ARRAY-IN-BOUNDS-P now allows arbitrary integers as arguments,
2365 not just nonnegative fixnums;
2366 ** the logical bit-array operators such as BIT-AND now accept an
2367 explicit NIL for their "opt-arg" argument (to indicate a
2368 freshly-consed result bit-array);
2369 ** ELT now signals an error on an invalid sequence index in safe
2371 ** the type system is now cleverer about negations of numeric
2372 types, and consequently understands the BIGNUM and RATIO types
2374 ** the type system is now cleverer about the interaction between
2375 INTEGER and RATIO types: while bugs still remain, many more
2376 cases are accurately computed;
2377 ** in TYPECASE, OTHERWISE now only introduces an otherwise-clause
2378 if it is in the last clause;
2379 ** CONSTANTLY now correctly returns a side-effect-free function in
2381 ** DECLARE is no longer treated as a special-operator; in
2382 particular, SPECIAL-OPERATOR-P no longer returns T for DECLARE;
2383 * incremented fasl file version number due to the change in the
2384 DEFSTRUCT-SLOT-DESCRIPTION structure.
2386 changes in sbcl-0.7.12 relative to sbcl-0.7.11:
2387 * minor incompatible change: code processed by the "interpreter" or
2388 EVAL now has a compilation optimization policy of (DEBUG 2)
2389 (changed from (DEBUG 1)) to improve debuggability of interactive
2390 development, and to allow the use of the debug RETURN command in
2392 * an experimental implementation of the RETURN command for the
2393 debugger has been included. (thanks to Frederik Kuivinen)
2394 * fixed bug 62: constraints were not propagated into a loop.
2395 * fixed bug in embedded calls of SORT (reported and investigated by
2397 * fixed some bugs revealed by Paul Dietz' test suite:
2398 ** printing and reading of arrays with some dimensions having
2399 length 0 (thanks to Gerd Moellmann);
2400 ** BOA constructor with &AUX argument without a default value does
2401 not cause a type error;
2402 ** CONSTANTP now returns true for all self-evaluating objects.
2404 changes in sbcl-0.7.11 relative to sbcl-0.7.10:
2405 * fixed bug 127: DEFSTRUCT now does not clobber old structure
2406 accessors that are related by inheritance, as specified in the
2407 :CONC-NAME section of the specification of DEFSTRUCT. (thanks to
2408 Valtteri Vuorikoski)
2409 * The compiler is now able to inline functions that were defined in
2410 a complex lexical environment (e.g. inside a MACROLET).
2411 * fixed bug in DESCRIBE, which now works on rank-0 arrays. (thanks
2413 * Support for the upcoming FreeBSD-5.0 release has been included.
2414 (thanks to Dag-Erling Smorgrav)
2415 * fixed bug 219: DEFINE-COMPILER-MACRO no longer has compile-time
2416 effect when it is not in a toplevel context.
2417 * fixed bug 222: DEFMETHOD and SYMBOL-MACROLET interactions now
2418 stand a better chance of being correct. (thanks to Gerd
2420 * fixed bug in COERCE, which now signals an error on coercing a
2421 rational to a bounded real type which excludes the expected
2423 * The compiler is now able to derive types more accurately from the
2424 COERCE and COMPILE functions.
2425 * fixed bug 223: functional binding is considered to be constant
2426 only for symbols in the CL package.
2427 * fixed bug 231: SETQ did not check the type of a variable being set
2428 (reported by Robert E. Brown)
2429 * A new optimization for MAKE-INSTANCE has been included, fixing
2430 various bugs (including relating to :ALLOCATION :CLASS slots and
2431 :DEFAULT-INITARGS over-eager evalueation). (thanks to Gerd
2433 * fixed some LOOP bugs revealed by Paul Dietz' test suite:
2434 ** As required by ANSI, LOOP now disallows anonymous collection
2435 clauses such as COLLECT I in conjunction with aggregate boolean
2436 clauses such as THEREIS (= I 1);
2437 ** LOOP now signals an error when any variable is reused in the
2438 same loop (including the potentially useful construct analogous
2439 to WITH A = 1 WITH A = (1+ A);
2440 ** IT is only a special loop symbol within the first clause of a
2441 conditional loop clause;
2442 ** LOOP with a typed iteration variable over a hashtable now
2443 signals a type error iff it should.
2444 * fixed some other bugs revealed by Paul Dietz' test suite:
2445 ** FILE-STREAM now names the class previously known as FD-STREAM;
2446 ** in DEFSTRUCT, a bare :CONC-NAME (or a :CONC-NAME with no
2447 argument) no longer signals an error;
2448 ** likewise in DEFSTRUCT, :CONC-NAME NIL now respects the package
2449 of the slot symbol, rather than using the current package
2450 ((:CONC-NAME "") continues to intern the slot's name in the
2452 * incremented fasl file version number, because of the incompatible
2453 change to the DEFSTRUCT-DESCRIPTION structure, and again because
2454 of the new implementation of DEFINE-COMPILER-MACRO.
2456 changes in sbcl-0.7.10 relative to sbcl-0.7.9:
2457 * Support for building SBCL for MIPS platforms running in
2458 little-endian mode has now been checked in, and basic
2459 functionality on said platforms verified.
2460 * minor incompatible change: PCL now records the pathname of a file
2461 in which methods and the like are defined, rather than its
2463 * minor incompatible change: TRUENAME now considers the truename of
2464 a file naming a directory to be the pathname with :DIRECTORY
2465 component indicating that directory.
2466 * minor incompatible change: a NAMED clause in the extended form of
2467 LOOP no longer causes a BLOCK named NIL to surround the LOOP. The
2468 reason for the previous behaviour is unclear.
2469 * more systematization and improvement of CLOS and MOP conformance
2470 in PCL (thanks to Gerd Moellman and Pierre Mai):
2471 ** the standard ANSI CL generic function NO-NEXT-METHOD is now
2473 ** DEFINE-METHOD-COMBINATION no longer signals an error for
2474 primary methods with no specializers;
2475 ** the MOP generic function GENERIC-FUNCTION-DECLARATIONS is now
2477 ** the Readers for Class Metaobjects methods CLASS-DIRECT-SLOTS
2478 and CLASS-DIRECT-DEFAULT-INITARGS have been implemented for
2479 FORWARD-REFERENCED-CLASSes; error reporting on
2480 CLASS-DEFAULT-INITARGS, CLASS-PRECEDENCE-LIST and CLASS-SLOTS
2482 ** SXHASH on CLOS instances now uses a slot internal to the
2483 instance to return different numbers on distinct instances,
2484 while preserving the same return value through invocations of
2486 ** DEFMETHOD signals errors when methods with longer incongruent
2487 lambda lists are added to generic functions;
2488 ** COMPUTE-CLASS-PRECEDENCE-LIST now has a method specialized on
2489 CLASS, as specified in AMOP;
2490 ** COMPUTE-SLOTS :AROUND now assigns locations sequentially based
2491 on the order returned by the primary method for classes of
2492 class STANDARD-CLASS;
2493 ** DEFINE-METHOD-COMBINATION now works with the :ARGUMENTS option.
2494 * fixed some bugs shown by Paul Dietz' test suite:
2495 ** DOLIST puts its body in TAGBODY;
2496 ** SET-EXCLUSIVE-OR sends arguments to :TEST function in the
2498 ** MULTIPLE-VALUE-SETQ evaluates side-effectful places before
2499 value producing form;
2500 ** if more variables are given to PROGV than values, extra
2501 variables are bound and made to have no value;
2502 ** NSUBSTITUTE on list arguments gets the right answer with
2504 ** ELT signals an error of type TYPE-ERROR when the index argument
2505 is not a valid sequence index;
2506 ** LOOP signals (at macroexpansion time) an error of type
2507 PROGRAM-ERROR when duplicate variable names are found;
2508 ** LOOP supports DOWNTO and ABOVE properly; (thanks to Matthew Danish)
2509 ** FUNCALL of special-operators now cause an error of type
2511 ** PSETQ now works as required in the presence of side-effecting
2512 symbol-macro places;
2513 ** NCONC accepts any object as its last argument;
2514 ** :COUNT argument to sequence functions may be BIGNUM; (thanks to
2516 ** loop-for-as-package does not require a package to be explicitely
2518 ** LOOP WITH now treats NIL in the d-var-spec correctly as an
2520 * fixed bug 166: compiler preserves "there is a way to go"
2521 invariant when deleting code.
2522 * fixed bug 172: macro lambda lists with required arguments after
2523 &REST arguments now cause an error to be signalled. (thanks to
2525 * fixed Entomotomy PEEK-CHAR-WRONGLY-ECHOS-TO-ECHO-STREAM
2526 bug. (thanks to Matthew Danish)
2527 * fixed bug 225: STRING-STREAM is now a class. (reported by Gilbert
2529 * fixed bug 136: CALL-NEXT-METHOD no longer gets confused when
2530 arguments are lexically rebound. (thanks to Gerd Moellmann and
2532 * fixed bug 194: error messages are now more informative when there
2533 is no primary method applicable in a call to a generic
2534 function. (thanks to Gerd Moellmann)
2535 * fixed bug in command line argument checking (thanks to Julian
2537 * fixed bug in COUNT-IF, making it handle :FROM-END correctly
2538 (thanks to Matthew Danish)
2539 * incremented fasl file version number, because of the
2540 SXHASH-related changes in the layout of CLOS data structures
2542 changes in sbcl-0.7.9 relative to sbcl-0.7.8:
2543 * minor incompatible change: The runtime (the Unix executable named
2544 "sbcl") is now much pickier about the .core files it will load.
2545 Essentially it now requires .core files to descend from the same
2546 build (not just the same sources or LISP-IMPLEMENTATION-VERSION)
2547 as the runtime does. (The intent is to prevent the crashes which
2548 can occur, and which can even be reported as mysterious failures,
2549 when people patch the sources or change the build parameters
2550 without changing LISP-IMPLEMENTATION-VERSION, then mix and match
2551 sbcl and .core files.)
2552 * fixed bug: VALUES-LIST is no longer optimized away.
2553 * fixed bug 142: The FFI conversion of C string values to Lisp
2554 string values no longer conses excessively. (thanks to Nathan
2555 Froyd porting Raymond Toy's fix to CMU CL)
2556 * began to systematize and improve MOP conformance in PCL (thanks to
2557 Nathan Froyd, Gerd Moellman and Pierre Mai):
2558 ** SLOT-DEFINITION-ALLOCATION now returns :CLASS, not the class
2560 ** GENERIC-FUNCTION-ARGUMENT-PRECEDENCE-ORDER is now implemented;
2561 ** FINALIZE-INHERITANCE is now called on class finalization;
2562 ** DOCUMENTATION and (SETF DOCUMENTATION) now have the correct
2563 argument precedence order.
2564 * fixed bug 202: The compiler no longer fails on functions whose
2565 derived types contradict their declared type.
2566 * DEFMACRO is implemented via EVAL-WHEN instead of IR1 translation,
2567 so it can be non-toplevel.
2568 * The fasl file version number has changed (because of the new
2569 implementation of DEFMACRO).
2570 * (mostly) fixed bugs 46b and 46c: sequence functions now check, in
2571 safe code, that any length requirement by their type-specifier
2572 argument is valid. The exceptions to this are described in bug
2574 * fixed bugs 46h and 46i: TWO-WAY- and CONCATENATED-STREAM creation
2575 functions now check the types of their inputs as required by ANSI.
2576 * fixed bug 48c: SYMBOL-MACROLET signals PROGRAM-ERROR when an
2577 introduced symbol is DECLAREd to be SPECIAL.
2578 * fixed reading of (COMPLEX DOUBLE-FLOAT) literals from fasl files
2579 * fixed bug: :COUNT argument to sequence functions may be negative
2580 * fixed bug: body of DO-SYMBOLS may contain declarations
2581 * fixed bug: PUSHNEW now evaluates its arguments from left to right
2582 (reported by Paul F. Dietz, fixed by Gerd Moellman)
2583 * fixed bug: PUSH, PUSHNEW and POP now evaluate a place given by a
2584 symbol macro only once
2585 * fixed printing of call frame when argument list is unavailable
2586 * fixed bug: :ALLOW-OTHER-KEYS is an allowed keyword name
2587 * compiler no longer signals WARNING on unknown keyword
2590 changes in sbcl-0.7.8 relative to sbcl-0.7.7:
2591 * A beta-quality port to the mips architecture running Linux,
2592 based on the old CMUCL backend, has been made. It has been tested
2593 on a big-endian kernel, and works sufficiently well to be able to
2594 rebuild itself; it has not been tested in little-endian mode.
2595 * fixed an inconsistency between gencgc.c and purify.c which made
2596 dumping/loading .core files unreliable
2597 * fixed bug 120a: The compiler now deals correctly with IFs where
2598 the consequent is the same as the alternative, instead of
2599 misderiving the return type. (thanks to Alexey Dejneka)
2600 * fixed bug 113: Logical pathnames are now dumpable (the logical
2601 host is resolved at load-time, throwing an error if it is not
2603 * fixed bug 174: FORMAT's error message is slightly clearer when a
2604 non-printing character is used in a format directive.
2605 * fixed several bugs in compiler checking of type declarations, i.e.
2606 violations of the Python "declarations are assertions" principle
2607 (thanks to Alexey Dejneka)
2608 * fixed several bugs in PCL's error checking (thanks to Gerd
2610 * fixed bug: printing of FILE-ERROR (thanks to Antonio
2612 * fixed bug in compilation of functions as first class values
2613 (thanks to Antonio Martinez-Shotton)
2614 * The compiler's handling TYPE-ERRORs which it can prove will
2615 inevitably happen at runtime has been cleaned up and corrected
2616 in several ways. (thanks to Alexey Dejneka)
2617 * improved argument type checking for various basic arithmetic
2618 operations (MAX, +, LOGXOR, etc.) which have had so much TLC
2619 lavished on them in the past that they can be compiled in many
2620 ways in different special cases
2621 * fixed bug 181: compiler checks validity of user supplied type
2623 * cleaned up code flushing in optimization: Function calls which
2624 should signal errors for safety purposes (e.g. which ANSI says
2625 should signal errors when their arguments are of incorrect type)
2626 are no longer optimized away.
2627 * added new extension: SB-DEBUG:BACKTRACE-AS-LIST
2628 * incremented fasl file version number, because changes in the
2629 implementation of sequence functions like COERCE caused
2630 internal utility functions like COERCE-TO-SIMPLE-VECTOR (used
2631 in old inline expansions) to become undefined. (Actually these
2632 changes were later undone, so we might very well be binary
2633 compatible with 0.7.7 after all, but leaving the version number
2634 incremented seemed like the simplest and most conservative
2637 changes in sbcl-0.7.7 relative to sbcl-0.7.6:
2638 * An alpha-quality port to the parisc architecture running Linux,
2639 based on the old CMUCL backend, has been made. This, even more so
2640 than the other backends, should be considered still a work in
2641 progress; known problems include that the Linux kernel in 64-bit
2642 mode does not propagate the correct sigcontext structure to
2643 userspace, and consequently SBCL on a parisc64 kernel will not
2645 * fixed bug 189: The compiler now respects NOTINLINE declarations for
2646 functions declared in FLET and LABELS. (I.e. "LET conversion" is
2647 suppressed.) Also now that the compiler is looking at declarations
2648 in the environment, it checks optimization declarations as well,
2649 and suppresses inlining when (> DEBUG SPEED).
2650 * More fixes have been made to treatment of floating point exception
2651 treatment and other Unix signals. In particular, floating point
2652 exceptions no longer cause Bus errors on the SPARC/Linux platform.
2653 * The detection and handling of control stack exhaustion (infinite
2654 or very deeply nested recursion) has changed. Stack exhaustion
2655 detection is now done by write-protecting pages at the OS level
2656 and applies at all optimization settings; when found, a
2657 SB-KERNEL:CONTROL-STACK-EXHAUSTED condition (subclass of
2658 STORAGE-CONDITION) is signalled, so stack exhaustion can no longer
2659 be caught using IGNORE-ERRORS.
2660 * Bugs 65, 70, and 109 fixed: The compiler now preserves invariants
2661 correctly when transforming recursive LABELS functions to LETs.
2662 (thanks to Alexey Dejneka)
2663 * Bug 48a./b. fixed: SYMBOL-MACROLET now refuses to bind symbols
2664 that are names of constants or global variables.
2665 * Bug fix: DEFINE-ALIEN-ROUTINE now declaims the correct FTYPE for
2666 alien routines with docstrings.
2667 * Bug 184 fixed: Division of ratios by the integer 0 now signals an
2668 error of type DIVISION-BY-ZERO. (thanks to Wolfhard Buss and
2670 * Bug fix: Errors in PARSE-INTEGER are now of type PARSE-ERROR.
2671 (thanks to Eric Marsden)
2672 * Bug fix: COERCE to (COMPLEX FLOAT) of rationals now returns an
2673 object of type (COMPLEX FLOAT). (thanks to Wolfhard Buss)
2674 * Bug fix: The SPARC backend can now compile functions involving
2675 LOGAND and stack-allocated arguments. (thanks to Raymond Toy)
2676 * Bug fix: We no longer segfault on passing a non-FILE-STREAM stream
2677 to a functions expecting a PATHNAME-DESIGNATOR.
2678 * Bug fix: DEFGENERIC now enforces the ANSI restrictions on its
2679 lambda lists. (thanks to Alexey Dejneka)
2680 * Bug fix: changed encoding of PCL's internal MAKE-INSTANCE
2681 functions so that EXPORTing the name of the class doesn't cause
2682 MAKE-INSTANCE functions from earlier DEFCLASSes to get lost (thanks
2683 to Antonio Martinez for reporting this)
2684 * Bug 192 fixed: The internal primitive DATA-VECTOR-REF can now be
2685 constant-folded without failing an assertion. (thanks to Einar
2686 Floystad Dorum for reporting this)
2687 * Bugs 123 and 165 fixed: array specializations on as-yet-undefined
2688 types are now dealt with more correctly by the compiler.
2689 * Minor incompatible change: COMPILE-FILE-PATHNAME now merges its
2690 OUTPUT-FILE argument with its INPUT-FILE argument, resulting in
2691 behaviour analogous to RENAME-FILE. This puts its behaviour more
2692 in line with ANSI's wording on COMPILE-FILE-PATHNAME. (thanks to
2694 * The fasl file version number has changed again. (because of the
2695 bug fix involving the names of PCL MAKE-INSTANCE functions)
2697 changes in sbcl-0.7.6 relative to sbcl-0.7.5:
2698 * bug fix: Floating point exceptions are treated much more
2699 consistently on the x86/Linux and PPC/Linux platforms.
2700 * Array initialization with :INITIAL-ELEMENT is now much faster for
2701 cases when the compiler cannot open code the array creation, but
2702 does know what the UPGRADED-ARRAY-ELEMENT-TYPE will be. General
2703 array accesses have also seen a speed increase.
2704 * bug fix: LOAD :IF-DOES-NOT-EXIST NIL now works when file type is
2705 specified. (This was at the root of some bad interactions between
2706 SBCL and ILISP: thanks to Gregory Wright for diagnosing this and
2708 * bug fix: Internal error arguments for undefined functions are now
2709 computed correctly on the PPC/Linux platform.
2710 * bug fix: Bad &REST syntax is now checked correctly. (thanks to
2711 Raymond Toy's patch for CMU CL)
2712 * Support for the Solaris 9 operating environment has been included
2713 (thanks to Daniel Merritt)
2714 * A very ugly but hopefully complete draft of the missing FFI chapter
2715 of the manual has been created by reformatting the corresponding
2716 CMU CL manual chapter into (currently very ugly and incoherent)
2717 DocBook and bringing it up to date for SBCL behavior. Thus, the
2718 manual is now essentially complete, at least by my extreme
2719 once-and-only-once standards, whereby it's acceptable to refer to
2720 the doc strings of SB-EXT functions as the primary documentation.
2721 * The fasl file version number has changed again, due to cleanup of
2722 (user-invisible) bitrotted stuff. (E.g. *!INITIAL-FDEFN-OBJECTS*
2723 is no longer a static symbol.)
2725 changes in sbcl-0.7.5 relative to sbcl-0.7.4:
2726 * SBCL now builds with OpenMCL (version 0.12) as the
2727 cross-compilation host; also, more progress has been made toward
2728 bootstrapping under CLISP.
2729 * SBCL now runs on the Tru64 (aka OSF/1) operating system on the
2731 * bug 158 fixed: The compiler can now deal with integer loop
2732 increments different from 1; fixing this turned out also to fix
2734 * bug 169 fixed: no more bogus warnings about using lexical bindings
2735 despite the presence of perfectly good SPECIAL declarations (thanks
2736 to David Lichteblau)
2737 * bug 175 fixed: CHANGE-CLASS is now more ANSI-conforming,
2738 accepting initargs. (thanks to Espen Johnsen and Pierre Mai)
2739 * bug 179 fixed: DIRECTORY can now deal with filenames with pattern
2741 * bug 180 fixed: Method combination specifications no longer ignore
2742 the :MOST-SPECIFIC-LAST option. (thanks to Pierre Mai)
2743 * bug fix: Structure type predicate functions now check their argument
2744 count as they should.
2745 * bug fix: Classes with :METACLASS STRUCTURE-CLASS now print
2746 correctly. (thanks to Pierre Mai)
2747 * minor incompatible change: The --noprogrammer option is deprecated
2748 in favor of the new --disable-debugger option, which is very similar.
2749 (The major difference is that it takes effect at a slightly different
2750 time at startup, causing handling of errors in --sysinit and
2751 --userinit files will be affected differently.) The
2752 SB-EXT:DISABLE-DEBUGGER and SB-EXT:ENABLE-DEBUGGER functions have
2753 been added to allow this functionality to be controlled from ordinary
2754 Lisp code. (ENABLE-DEBUGGER should help people like the Debian
2755 maintainers, who might want to run non-interactive scripts to
2756 build SBCL cores which will later be used interactively.)
2757 * minor incompatible change: The LOAD function no longer, when given
2758 a wild pathname to load, loads all files matching that pathname.
2759 Instead, an error of type FILE-ERROR is signalled.
2761 changes in sbcl-0.7.4 relative to sbcl-0.7.3:
2762 * bug 147 fixed: The compiler preserves its block link/count
2763 invariants more correctly now so that it doesn't crash. (thanks
2765 * Dynamic loading of object files in OpenBSD is now supported. (thanks
2767 * COMPILE now works correctly on macros. (thanks to Matthias Hoelzl)
2768 * GET-MACRO-CHARACTER and SET-MACRO-CHARACTER now represent
2769 no-value-for-this-character as NIL (as specified by ANSI).
2770 * HOST-NAMESTRING on physical pathnames now returns a string that is
2771 valid as a host argument to MERGE-PATHNAMES and to MAKE-PATHNAME.
2772 (thanks to Christophe Rhodes)
2773 * The Alpha port handles icache flushing more correctly. (thanks to
2775 * More progress has been made toward bootstrapping under CLISP. (thanks
2776 to Christophe Rhodes)
2777 * The fasl file format has changed again, because dynamic loading
2778 on OpenBSD (which has non-ELF object files) motivated some cleanups
2779 in the way that foreign symbols are transformed and passed around.
2780 * minor incompatible change: The ASCII RUBOUT character, (CHAR-CODE 127),
2781 is no longer treated as whitespace by the reader, but instead as
2782 an ordinary character. Thus e.g. (READ-FROM-STRING "A
\7fB") returns
2783 |A
\7fB|, instead of A as it used to.
2785 changes in sbcl-0.7.3 relative to sbcl-0.7.2:
2786 * ANSI's DEFINE-SYMBOL-MACRO is now supported. (thanks to Nathan
2787 Froyd porting CMU CL code originally by Douglas Thomas Crosher)
2788 * SBCL now runs on the PPC archtiecture under Linux. It actually did
2789 this as of 0.7.1.45, but was left out of the previous news section
2790 (thanks to Dan Barlow)
2791 * SBCL now runs on the Solaris operating system on SPARC architectures
2792 (thanks to Christophe Rhodes's port of the CMUCL runtime)
2793 * cleanups to the runtime on SPARC, both Linux and Solaris, and for
2794 gcc>=3 (thanks to Nathan Froyd and Ingvar Mattsson)
2795 * SPARC backend cleanups, allowing builds of cores optimized for V8
2796 and V9 SPARCS, and also emission of code targeted to a particular
2797 backend chosen at runtime (thanks to Christophe Rhodes and Raymond
2799 * SBCL is closer to bootstrapping under CLISP, thanks to various
2800 fixes by Christophe Rhodes.
2801 * The fasl file format has changed again, to allow the compiler's
2802 INFO database to support symbol macros.
2803 * The user manual (in doc/) is formatted into HTML more nicely.
2804 (thanks to coreythomas)
2805 * The system is smarter about SUBTYPEP relationships, especially
2806 those involving NOT types (including types such as ATOM which are
2807 represented internally using NOT types). Thus SUBTYPEP is less
2808 likely to return (VALUES NIL NIL) in general, and in particular
2809 bugs 58 and (the remaining bits of) bug 50 are fixed. (thanks to
2811 * The fasl file format has changed again, because the internal
2812 representation of types now includes a new slot to support the new
2813 SUBTYPEP-of-NOT-types logic.
2814 * (not a change in the main branch of SBCL, but a related prototype
2815 which can hopefully be merged into the main branch of SBCL in the
2816 future:) Brian Spilsbury has produced a Unicode-enabled variant of
2817 sbcl-0.7.0, available as a patch against sbcl-0.7.0 at
2818 <http://designix.com.au/brian/SBCL/sbcl-0.7.0-unicode.p0.gz>.
2819 * Bug 151 fixed: GET-DISPATCH-MACRO-CHAR now returns NIL for undefined
2820 dispatch macro character combinations. (thanks to Alexey Dejneka)
2821 * Bugfix in PARSE-NAMESTRING: we now correctly parse unix namestrings
2822 that superficially look like logical namestrings correctly.
2823 * USER-HOMEDIR-PATHNAME now returns a (physical) pathname that SBCL
2825 * Bugfix in DEFSTRUCT: BOA constructor lambda lists now accept (name
2826 default supplied-p) for &optional and &key arguments. (thanks to
2829 changes in sbcl-0.7.2 relative to sbcl-0.7.1:
2830 * incompatible change: The compiler is now less aggressive about
2831 tail call optimization, doing it only when (> SPACE DEBUG) or
2832 (> SPEED DEBUG). (This is an incompatible change because there are
2833 programs which relied on the old CMU-CL-style behavior to optimize
2834 away their unbounded recursion which will now die of stack overflow.)
2835 * minor incompatible change: The default BYTES-CONSED-BETWEEN-GCS
2836 for non-GENCGC systems has been increased to 20M (since that
2837 seems much closer to the likely performance optimum for modern
2838 systems than the old 4M value was)
2839 * minor incompatible change: new larger values for *DEBUG-PRINT-LENGTH*
2840 and *DEBUG-PRINT-LEVEL*
2841 * SBCL runs on SPARC systems now. (thanks to Christophe Rhodes' port
2842 of CMU CL's support for SPARC, and various endianness and other
2843 SBCL portability fixes due to Christophe Rhodes and Dan Barlow)
2844 * new syntactic sugar for the Unix command line: --load foo.bar is now
2845 an alternate notation for --eval '(load "foo.bar")'.
2847 ** The system now detects stack overflow and handles it gracefully,
2848 at least for (OR (> SAFETY (MAX SPEED SPACE)) (= SAFETY 3))
2849 optimization settings. (This is a good thing in general, and
2850 its introduction in this version should be particularly timely
2851 for anyone whose code fails because of suppression of tail
2853 ** The system now hunts for the C variable "environ" in a more
2854 devious way, to avoid segfaults when the C library version differs
2855 between compile time and run time. (thanks to Christophe Rhodes)
2856 ** INTEGER-valued CATCH tags now work. (thanks to Alexey Dejneka,
2857 and also to Christophe Rhodes for porting the fix to non-X86 CPUs)
2858 ** The compiler no longer issues bogus style warnings for undefined
2859 classes in the same source file as the DEFCLASSes which defined
2860 them. (thanks to Stig E Sandoe for reporting and Martin Atzmueller
2862 ** fixes in CONDITION class precedence list for undefined function
2863 errors (thanks to Alexei Dejneka)
2864 ** *DEFAULT-PATHNAME-DEFAULTS* is used more consistently and
2865 correctly. (thanks to Dan Barlow)
2866 ** portability fixes aiming at bootstrapping under CLISP (thanks
2867 to Dave McDonald and Christophe Rhodes)
2868 ** FORMAT fixes (thanks to Robert Strandh and Dan Barlow)
2869 ** fixes in type translation and and type inference (thanks to
2871 ** fixes to optimizer internal errors (thanks to Alexei Dejneka)
2872 ** various fixes in the new ports (thanks to Dan Barlow)
2873 * several changes related to debugging:
2874 ** suppression of tail recursion, as noted above
2875 ** stack overflow detection, as noted above
2876 ** The default implementation of TRACE has changed. :ENCAPSULATE T
2877 is now the default. (For some time encapsulation has been more
2878 reliable than the breakpoint-based :ENCAPSULATE NIL
2879 implementation, at least on X86 systems; and I just noticed that
2880 encapsulation also seems closer to the spirit of the ANSI
2883 changes in sbcl-0.7.1 relative to sbcl-0.7.0:
2885 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are set
2886 up properly again. (There was a packaging bug in 0.7.0 which
2887 left their definitions in SB-SYS::LOAD-FOREIGN and
2888 SB-SYS::LOAD-1-FOREIGN. LOAD-FOREIGN and LOAD-1-FOREIGN are
2889 vital for most things which interface to C-level interfaces,
2890 like extensions working with sockets or databases or
2891 Perl-compatible regexes or whatever, and the need to fix
2892 this bug is the main reason that 0.7.1 was released so
2894 ** DEFGENERIC is now choosier about the methods it redefines, so that
2895 reLOADing a previously-LOADed file containing DEFGENERICs does
2896 the right thing now. Thus, the Lispy edit/reLOAD-a-little/test
2897 cycle now works as it should. (thanks to Alexey Dejneka)
2898 ** Bug 106 (types (COMPLEX FOO) where FOO is an obscure type) was
2899 fixed by Christophe Rhodes. (He actually submitted this patch
2900 months ago, and I delayed until after 0.7.0.)
2901 ** Bug 111 (internal compiler confusion about runtime checks on
2902 FUNCTION types) was fixed by Alexey Dejneka.
2903 * Some internal cleanups (getting rid of variables which aren't
2904 needed now that the byte interpreter is gone) caused the fasl
2905 file format number to change again.
2907 changes in sbcl-0.7.0 relative to sbcl-0.6.13:
2908 * major incompatible change: The default fasl file extension, i.e. the
2909 default extension for files produced by COMPILE-FILE, has changed
2910 to ".fasl", for all architectures. (No longer ".x86f" and ".axpf".)
2912 ** There are many changes in the implementation of the compiler.
2913 SBCL is now essentially a compiler-only implementation of ANSI
2914 Common Lisp. EVAL still "interprets" a few special cases, but
2915 almost all the interesting cases are handled by creating
2916 a LAMBDA expression, calling COMPILE on it, then calling
2917 FUNCALL on the result.
2918 ** The EVAL-WHEN code has been rewritten to be ANSI-compliant, and
2919 various related bugs (IR1-1, IR1-2, IR1-3, IR1-3a) have gone away.
2920 Since the code is newer, there might still be some new bugs
2921 (though not as many as before Martin Atzmueller's fixes:-). But
2922 the new code is substantially simpler and clearer, and hopefully
2923 any remaining bugs will be simpler, less fundamental, and more
2924 fixable then the bugs in the old code.
2925 ** The revised compiler is still a little unsteady on its feet.
2927 *** The debugging information it produces (particularly the names
2928 of FUNCTION objects) is sometimes much less useful than what
2929 the old compiler produced.
2930 *** The support for inlining FOO when you (DECLAIM (INLINE FOO))
2931 then do (DEFUN FOO ..) in a non-null lexical environment (e.g.
2932 within a MACROLET) has been temporarily weakened.
2933 ** There are new compiler optimizations for various functions:
2934 *** the sequence functions FIND, POSITION, FIND-IF, POSITION-IF,
2935 FIND-IF-NOT, POSITION-IF-NOT, and FILL
2936 *** the math functions TRUNCATE, FLOOR, and CEILING
2937 *** the function-of-all-trades COERCE
2938 Mostly these should be transparent, but there's one
2939 potentially-annoying problem (bug 117): when the compiler
2940 inline-expands a function and does type analysis on the result,
2941 it can create control paths which have type mismatches, and
2942 when it can't prove that those control paths aren't taken,
2943 it will issue WARNINGs about the type mismatches. This is
2944 a particular problem in practice for the new sequence functions.
2945 It's not clear how this should be fixed, and for now, a
2946 workaround is given in the entry for 117 in the BUGS file.
2947 ** (Because of the interaction between the two previous items --
2948 occasional inlining problems and new inline expansions -- some
2949 of the new sequence function optimizations won't really kick in
2950 completely until debugging information, and then inlining, are
2951 straightened out in some future version.)
2952 * minor incompatible changes:
2953 ** As part of a bug fix by Christophe Rhodes to DIRECTORY behavior,
2954 DIRECTORY no longer implicitly promotes NIL slots of its
2955 pathname argument to :WILD. In particular, when you ask for the
2956 contents of a directory (which you used to be able to do without
2957 explicit wildcards, e.g. (DIRECTORY "/tmp/")) you now need to use
2958 explicit wildcards, e.g. (DIRECTORY "/tmp/*.*").
2959 ** changes in behavior that ANSI explicitly defines to be
2960 implementation dependent:
2961 *** The new compiler-only implementation still conforms with ANSI,
2962 but acts a little different than before. Besides the obvious
2963 changes in performance tradeoffs (that the cost per form passed
2964 to EVAL has gone up, and the cost per form executed by EVAL
2965 has gone down), the behavior of the system changes a little
2966 because there are no longer any interpreted function objects.
2967 COMPILED-FUNCTION-P is now synonymous with FUNCTIONP, and
2968 e.g. doing COMPILE on the output of interactive DEFUN is
2970 *** The value of INTERNAL-TIME-UNITS-PER-SECOND has been increased
2972 *** The default for the USE list in MAKE-PACKAGE and DEFPACKAGE
2973 has changed from (:CL) to NIL.
2974 *** The CHAR-NAME of unprintable ASCII characters which, unlike
2975 e.g. #\Newline and #\Tab, don't have names specified in the
2976 ANSI Common Lisp standard, is now based on their ASCII symbolic
2977 names (#\Nul, #\Soh, #\Stx, etc.) The old CMU-CL-style names
2978 (#\Null, #\^a, #\^b, etc.) are still accepted by NAME-CHAR, but
2979 are no longer used for output.
2980 ** changes in internal implementation constants:
2981 *** The default value of *BYTES-CONSED-BETWEEN-GCS* has doubled, to
2982 4 million. (If your application spends a lot of time GCing and
2983 you have a lot of RAM, you might want to experiment with
2984 increasing it even more.)
2985 ** The SB-C-CALL package has been merged into the SB-ALIEN package.
2986 However, almost all old code should still continue to work without
2987 immediate update, as SB-C-CALL is now a (deprecated) nickname
2989 ** Old operator names in the style DEF-FOO are now deprecated in
2990 favor of new corresponding names DEFINE-FOO, for consistency with
2991 the naming convention used in the ANSI standard (DEFSTRUCT, DEFVAR,
2992 DEFINE-CONDITION, DEFINE-MODIFY-MACRO..). This mostly affects
2993 internal symbols, but a few supported extensions like
2994 SB-ALIEN:DEF-ALIEN-FUNCTION are also affected. (So e.g.
2995 DEF-ALIEN-FUNCTION becomes DEFINE-ALIEN-FUNCTION.)
2996 ** The debugger prompt sequence now goes "5]", "5[2]", "5[3]",
2997 etc. as you get deeper into recursive calls to the debugger
2998 command loop, instead of the old "5]", "5]]", "5]]]"
2999 sequence. (I was motivated to do this when squabbles between
3000 ILISP and SBCL left me very deeply nested in the debugger. In the
3001 short term, this change will probably provoke more ILISP/SBCL
3002 squabbles, but hopefully it will be an improvement in the long run.)
3003 ** SB-ALIEN:DEFINE-ALIEN-FUNCTION (also known by the old deprecated
3004 name DEF-ALIEN-FUNCTION) now does DECLAIM FTYPE for the defined
3005 function, since declaiming return types involving aliens is
3006 (1) annoyingly messy to do by hand and (2) vital to efficient
3007 compilation of code which calls such functions.
3008 ** SB-ALIEN:LOAD-FOREIGN and SB-ALIEN:LOAD-1-FOREIGN are no
3009 longer reexported by the SB-EXT package. They're solely useful
3010 for alien code, so it seems more logical that you should get
3011 them from the SB-ALIEN package, not in SB-EXT.
3012 ** :SB-CONSTRAIN-FLOAT-TYPE, :SB-PROPAGATE-FLOAT-TYPE, and
3013 :SB-PROPAGATE-FUN-TYPE are no longer considered to be optional
3014 features. Instead, the code that they used to control is always
3015 built into the system.
3016 * many other bug fixes
3017 ** DEFSTRUCT and DEFCLASS have been substantially updated to take
3018 advantage of the new EVAL-WHEN stuff and to clean them up in
3019 general, and they are now more ANSI-compliant in a number of
3020 ways. Martin Atzmueller is responsible for a lot of this.
3021 ** Besides the cleanups discussed above, Martin Atzmueller fixed
3023 *** fixes in READ-SEQUENCE and WRITE-SEQUENCE
3024 *** correct ERROR type for various file operations
3025 *** some fixes for Lisp streams
3026 *** DEFMETHOD syntax checking
3027 *** changing old weird representation of debug information as
3028 strings (which, among their other deficiencies, don't transform
3029 correctly when you rename packages, and don't change their
3030 print representation when you change things like *PACKAGE*
3031 and *PRINT-LENGTH*) to symbols and lists of symbols
3032 He also made several improvements and fixed several bugs in DESCRIBE.
3033 ** Alexey Dejneka fixed many bugs, including classic bugs and bugs he
3035 *** misbehavior of WRITE-STRING/WRITE-LINE
3036 *** LOOP over keys of a hash table, LOOP bugs 49b and 81 and 103,
3037 and several other LOOP problems as well
3038 *** DIRECTORY when similar filenames are present
3039 *** DEFGENERIC with :METHOD options
3040 *** bug 126, in (MAKE-STRING N :INITIAL-ELEMENT #\SPACE))
3041 *** bug in the optimization of ARRAY-ELEMENT-TYPE
3042 *** argument ordering in FIND with :TEST option
3043 *** mishandled package designator argument in APROPOS-LIST
3044 *** various problems in the backquote readmacro
3045 *** a bug in APROPOS
3046 *** probably some others that I'm not describing very well here,
3047 since the CVS log documents them by reference to sbcl-devel
3048 messages, and the SourceForge archives aren't working well.:-(
3049 ** Dan Barlow improved the Alpha port (and is making progress on the
3050 PPC port, for those of you who think different).
3051 ** Besides the DIRECTORY fixes and changes mentioned elsewhere,
3052 Christophe Rhodes cleaned up the system self-test scripts (in tests/*),
3053 contributed the optimization of FIND-IF-NOT and POSITION-IF-NOT, and
3054 continues to work on the SPARC port (for those of you in a position
3055 to look down upon our little PC-compatible boxes from a great height).
3056 ** PPRINT-LOGICAL-BLOCK now copies the *PRINT-LINES* value on entry
3057 and uses that copy, rather than the current dynamic value, when
3058 it's trying to decide whether to truncate output. Thus e.g.
3059 (let ((*print-lines* 50))
3060 (pprint-logical-block (stream nil)
3062 (let ((*print-lines* 8))
3063 (print (aref possiblybigthings i) stream)))))
3064 should now truncate the logical block only at 50 lines, instead of
3065 often truncating it at 8 lines, as it did before.
3066 * The doc/cmucl/ directory, containing old CMU CL documentation
3067 from the time of the fork, is no longer part of the base system.
3068 SourceForge has shut down its anonymous FTP service, and with it
3069 my original plan for distributing the old CMU CL documentation
3070 there. For now, if you need these files you can download an old
3071 SBCL source release and extract them from it.
3072 * The fasl file version number changed again, for dozens of reasons,
3073 some of which are apparent above.
3075 changes in sbcl-0.6.13 relative to sbcl-0.6.12:
3076 * a port to the Compaq/DEC Alpha CPU, thanks to Dan Barlow
3077 * Martin Atzmueller ported Tim Moore's marvellous CMU CL DISASSEMBLE
3078 patch, so that DISASSEMBLE output is much nicer.
3079 * The code in the SB-PROFILE package now seems reasonably stable.
3080 I still haven't decided what the final interface should look like
3081 (I'd like PROFILE to interact cleanly with TRACE, since both
3082 facilities use function encapsulation) but if you have a need
3083 for profiling now, you can probably use it successfully with
3084 the current CMU-CL-style interface.
3085 * Pathnames and *DEFAULT-DIRECTORY-DEFAULTS* are much more
3086 ANSI-compliant, thanks to various fixes and tests from Dan Barlow.
3087 Also, at Dan Barlow's suggestion, TRUENAME on a dangling symbolic
3088 link now returns the dangling link itself, and for similar
3089 reasons, TRUENAME on a cyclic symbolic link returns the cyclic
3090 link itself. (In these cases the old code signalled an error and
3091 looped endlessly, respectively.) Thus, DIRECTORY now works even
3092 in the presence of dangling and cyclic symbolic links.
3093 * Compiler trace output (the :TRACE-FILE option to COMPILE-FILE)
3094 is now a supported extension again, since the consensus on
3095 sbcl-devel was that it can be useful for ordinary development
3096 work, not just for debugging SBCL itself.
3097 * The default for SB-EXT:*DERIVE-FUNCTION-TYPES* has changed to
3098 NIL, i.e. ANSI behavior, i.e. the compiler now recognizes
3099 that currently-defined functions might be redefined later with
3100 different return types.
3101 * Hash tables can be printed readably, as inspired by CMU CL code
3102 of Eric Marsden and SBCL code of Martin Atzmueller.
3103 * better error handling in CLOS method combination, thanks to
3104 Martin Atzmueller porting Pierre Mai's CMU CL patches
3105 * more overflow fixes for >16Mbyte I/O buffers
3106 * A bug in READ has been fixed, so that now a single Ctrl-D
3107 character suffices to cause end-of-file on character streams.
3108 In particular, now you only need one Ctrl-D at the command
3109 line (not two) to exit SBCL.
3110 * fixed bug 26: ARRAY-DISPLACEMENT now returns (VALUES NIL 0) for
3112 * fixed bug 107 (reported as a CMU CL bug by Erik Naggum on
3113 comp.lang.lisp 2001-06-11): (WRITE #*101 :RADIX T :BASE 36) now
3114 does the right thing.
3115 * The implementation of some type tests, especially for CONDITION
3116 types, is now tidier and maybe faster, due to CMU CL code
3117 originally by Douglas Crosher, ported by Martin Atzmueller.
3118 * Some math functions have been fixed, and there are new
3119 optimizers for deriving the types of COERCE and ARRAY-ELEMENT-TYPE,
3120 thanks to Raymond Toy's work on CMU CL, ported by Martin Atzmueller.
3121 * (There are also some new optimizers in contrib/*-extras.lisp. Those
3122 aren't built into sbcl-0.6.13, but are a sneak preview of what's
3123 likely to be built into sbcl-0.7.0.)
3124 * A bug in COPY-READTABLE was fixed. (Joao Cachopo's patch to CMU
3125 CL, ported to SBCL by Martin Atzmueller)
3126 * DESCRIBE now gives more information in some cases. (Pierre Mai's
3127 patch to CMU CL, ported to SBCL by Martin Atzmueller)
3128 * Martin Atzmueller and Bill Newman fixed some bugs in INSPECT.
3129 * There's a new slam.sh hack to shorten the edit/compile/debug
3130 cycle for low-level changes to SBCL itself, and a new
3131 :SB-AFTER-XC-CORE target feature to control the generation of
3132 the after-xc.core file needed by slam.sh.
3133 * minor incompatible change: The ENTRY-POINTS &KEY argument to
3134 COMPILE-FILE is no longer supported, so that now every function
3135 gets an entry point, so that block compilation looks a little
3136 more like the plain vanilla ANSI section 3.2.2.3 scheme.
3137 * minor incompatible change: SB-EXT:GET-BYTES-CONSED now
3138 returns the number of bytes consed since the system started,
3139 rather than the number consed since the first time the function
3140 was called. (The new definition parallels ANSI functions like
3141 CL:GET-INTERNAL-RUN-TIME.)
3142 * minor incompatible change: The old CMU-CL-style DIRECTORY options,
3143 i.e. :ALL, :FOLLOW-LINKS, and :CHECK-FOR-SUBDIRS, are no longer
3144 supported. Now DIRECTORY always does the abstract Common-Lisp-y
3145 thing, i.e. :ALL T :FOLLOW-LINKS T :CHECK-FOR-SUBDIRS T.
3146 * Fasl file version numbers are now independent of the target CPU,
3147 since historically most system changes which required version
3148 number changes have affected all CPUs equally. Similarly,
3149 the byte fasl file version is now equal to the ordinary
3152 changes in sbcl-0.6.12 relative to sbcl-0.6.11:
3153 * incompatible change: The old SB-EXT:OPTIMIZE-INTERFACE declaration
3154 is no longer recognized. I apologize for this, because it was
3155 listed in SB-EXT as a supported extension, but I found that
3156 its existing behavior was poorly specified, as well as incorrectly
3157 specified, and it looked like too much of a mess to straighten it
3158 out. I have enough on my hands trying to get ANSI stuff to work..
3159 * many patches ported from CMU CL by Martin Atzmueller, with
3160 half a dozen bug fixes in pretty-printing and the debugger, and
3161 half a dozen others elsewhere
3162 * fixed bug 13: Floating point infinities are now supported again.
3163 They might still be a little bit flaky, but thanks to bug reports
3164 from Nathan Froyd and CMU CL patches from Raymond Toy they're not
3165 as flaky as they were.
3166 * The --noprogrammer command line option is now supported. (Its
3167 behavior is slightly different in detail from what the old man
3168 page claimed it would do, but it's still appropriate under the
3169 same circumstances that the man page talks about.)
3170 * The :SB-PROPAGATE-FLOAT-TYPE and :SB-PROPAGATE-FUN-TYPE features
3171 are now supported, and enabled by default. Thus, the compiler can
3172 handle many floating point and complex operations much less
3173 inefficiently. (Thus e.g. you can implement a complex FFT
3175 * The compiler now detects type mismatches between DECLAIM FTYPE
3176 and DEFUN better, and implements CHECK-TYPE more correctly, and
3177 SBCL builds under CMU CL again despite its non-ANSI EVAL-WHEN,
3178 thanks to patches from Martin Atzmueller.
3179 * various fixes to make the cross-compiler more portable to
3180 ANSI-conforming-but-different cross-compilation hosts (notably
3181 Lispworks for Windows, following bug reports from Arthur Lemmens)
3182 * A bug in READ-SEQUENCE for CONCATENATED-STREAM, and a gross
3183 ANSI noncompliance in DEFMACRO &KEY argument parsing, have been
3184 fixed thanks to Pierre Mai's CMU CL patches.
3185 * fixes to keep the system from overflowing internal counters when
3186 it tries to use i/o buffers larger than 16M bytes
3187 * fixed bug 45a: Various internal functions required to support
3188 complex special functions have been merged from CMU CL sources.
3189 (When I was first setting up SBCL, I misunderstood a compile-time
3190 conditional #-OLD-SPECFUN, and so accidentally deleted them.)
3191 * improved support for type intersection and union, fixing bug 12
3192 (e.g., now (SUBTYPEP 'KEYWORD 'SYMBOL)=>T,T) and some other
3193 more obscure bugs as well
3194 * some steps toward byte-compiling non-performance-critical
3195 parts of the system, courtesy of patches from Martin Atzmueller
3196 * Christophe Rhodes has made some debian packages of sbcl at
3197 <http://www-jcsu.jesus.cam.ac.uk/ftp/pub/debian/lisp>.
3198 From his sbcl-devel e-mail of 2001-04-08 they're not completely
3199 stable, but are nonetheless usable. When he's ready, I'd be happy
3200 to add them to the SourceForge "File Releases" section. (And if
3201 anyone wants to do RPMs or *BSD packages, they'd be welcome too.)
3202 * new fasl file format version number (because of changes in
3203 internal representation of (OR ..) types to accommodate the new
3204 support for (AND ..) types, among other things)
3206 changes in sbcl-0.6.11 relative to sbcl-0.6.10:
3207 * Martin Atzmueller pointed out that bugs #9 and #25 are gone in
3209 * bug 34 fixed by Martin Atzmueller: dumping/loading instances works
3211 * fixed bug 40: TYPEP, SUBTYPEP, UPGRADED-ARRAY-ELEMENT-TYPE,
3212 and UPGRADED-COMPLEX-PART-TYPE now work better with of compound
3213 types built from undefined types, e.g. '(VECTOR SOME-UNDEF-TYPE).
3214 * DESCRIBE now works on structure objects again.
3215 * Most function call argument type mismatches are now handled as
3216 STYLE-WARNINGs instead of full WARNINGs, since the compiler doesn't
3217 know whether the function will be redefined before the call is
3218 executed. (The compiler could flag local calls with full WARNINGs,
3219 as per the ANSI spec "3.2.2.3 Semantic Constraints", but right now
3220 it doesn't keep track of enough information to know whether calls
3221 are local in this sense.)
3222 * Compiler output is now more verbose, with messages truncated
3223 later than before. (There should be some supported way for users
3224 to override the default verbosity, but I haven't decided how to
3225 provide it yet, so this behavior is still controlled by the internal
3226 SB-C::*COMPILER-ERROR-PRINT-FOO* variables in
3227 src/compiler/ir1util.lisp.)
3228 * Fasl file format version numbers have increased again, because
3229 support for the Gray streams extension changes the layout of the
3230 system's STREAM objects.
3231 * The Gray subclassable streams extension now works, thanks to a
3232 patch from Martin Atzmueller.
3233 * The full LOAD-FOREIGN extension (not just the primitive
3234 LOAD-FOREIGN-1) now works, thanks to a patch from Martin Atzmueller.
3235 * The default behavior of RUN-PROGRAM has changed. Now, unlike CMU CL
3236 but like most other programs, it defaults to copying the Unix
3237 environment from the original process instead of starting the
3238 new process in an empty environment.
3239 * Extensions which manipulate the Unix environment now support
3240 an :ENVIRONMENT keyword option which doesn't smash case or
3241 do other bad things. The CMU-CL-style :ENV option is retained
3242 for porting convenience.
3243 * LOAD-FOREIGN (and LOAD-1-FOREIGN) now support logical pathnames,
3244 as per Daniel Barlow's suggestion and Martin Atzmueller's patch
3246 changes in sbcl-0.6.10 relative to sbcl-0.6.9:
3248 * A patch from Martin Atzmueller seems to have solved the SIGINT
3249 problem, and as far as we know, signal-handling now works cleanly.
3250 (If you find any new bugs, please report them!)
3251 * The system no longer defaults Lisp source file names to types
3252 ".l", ".cl", or ".lsp", but only to ".lisp".
3253 * The compiler no longer uses special default file extensions for
3254 byte-compiled code. (The ANSI definition of COMPILE-FILE-PATHNAME
3255 seems to expect a single default extension for all compiled code,
3256 and there's no compelling reason to try to stretch the standard
3257 to allow two different extensions.) Instead, byte-compiled files
3258 default to the same extension as native-compiled files.
3259 * Fasl file format version numbers have increased again, because
3260 a rearrangement of internal implementation packages made some
3261 dumped symbols in old fasl files unreadable in new cores.
3262 * DECLARE/DECLAIM/PROCLAIM logic is more nearly ANSI in general, with
3263 many fewer weird special cases.
3264 * Bug #17 (differing COMPILE-FILE behavior between logical and
3265 physical pathnames) has been fixed, and some related misbehavior too,
3266 thanks to a patch from Martin Atzmueller.
3267 * Bug #30 (reader problems) is gone, thanks to a CMU CL patch
3268 by Tim Moore, ported to SBCL by Martin Atzmueller.
3269 * Martin Atzmueller fixed several filesystem-related problems,
3270 including bug #36, in part by porting CMU CL patches, which were
3271 written in part by Paul Werkowski.
3272 * More compiler warnings in src/runtime/ are gone, thanks to
3273 more patches from Martin Atzmueller.
3274 * Martin Atzmueller pointed out that bug 37 was fixed by his patches
3277 changes in sbcl-0.6.9 relative to sbcl-0.6.8:
3279 * DESCRIBE now works on CONDITION objects.
3280 * The debugger now handles errors which arise when trying to print
3281 *DEBUG-CONDITION*, so that it's less likely to fall into infinite
3283 * The build system now uses an additional file, customize-target-features.lisp,
3284 to allow local modifications to the target *FEATURES* list. (The point of
3285 this is that now I can set up a custom configuration, e.g. with :SB-SHOW
3286 debugging features enabled, without having to worry about propagating it
3287 into everyone's system when I do a "cvs update".) When no
3288 customize-target-features.lisp file exists, the target *FEATURES* list
3289 should be constructed the same way as before.
3290 * fixed bugs in DEFCONSTANT ANSI-compatibility:
3291 ** DEFCONSTANT now tests reassignments using EQL, not EQUAL, in order to
3292 warn about behavior which is undefined under the ANSI spec. Note: This
3293 is specified by ANSI, but it's not very popular with programmers.
3294 If it causes you problems, take a look at the new SB-INT:DEFCONSTANT-EQX
3295 macro in the SBCL sources for an example of a workaround which you
3296 might use to make portable ANSI-standard code which does what you want.
3297 ** DEFCONSTANT's implementation is now based on EVAL-WHEN instead of on
3298 pre-ANSI IR1 translation magic, so it does the ANSI-specified thing
3299 when it's used as a non-toplevel form. (This is required in order
3300 to implement the DEFCONSTANT-EQX macro.)
3301 ** (DEFCONSTANT X 1) (DEFVAR X) (SETF X 2) no longer "works".
3302 ** Unfortunately, non-toplevel DEFCONSTANT forms can still do some
3303 funny things, due to bugs in the implementation of EVAL-WHEN
3304 (bug #IR1-3). This probably won't be fixed until 0.7.x. (Fortunately,
3305 non-toplevel DEFCONSTANTs are uncommon.)
3306 * The core file version number and fasl file version number have been
3307 incremented, because the old noncompliant DEFCONSTANT behavior involved
3308 calling functions which no longer exist, and because I also took the
3309 opportunity to chop an unsupported slot out of the DEBUG-SOURCE structure.
3310 * fixed bug 1 (error handling before read-eval-print loop starts), and
3311 redid debugger restarts and related debugger commands somewhat while
3313 ** The QUIT debugger command is gone, since it did something
3314 rather different than the SB-EXT:QUIT command, and since it never
3315 worked properly outside the main toplevel read/eval/print loop.
3316 Invoking the new TOPLEVEL restart provides the same functionality.
3317 ** The GO debugger command is also gone, since you can just invoke
3318 the CONTINUE restart directly instead.
3319 ** The TOP debugger command is also gone, since it's redundant with the
3320 FRAME 0 command, and since it interfered with abbreviations for the
3322 * The system now recovers better from non-PACKAGE values of the *PACKAGE*
3324 * The system now understands compound CONS types (e.g. (CONS FIXNUM T))
3325 as required by ANSI. (thanks to Douglas Crosher's CMU CL patches, with
3326 some porting work by Martin Atzmueller)
3327 * Martin Atzmueller reviewed the CMU CL mailing lists and came back
3328 with a boatload of patches which he ported to SBCL. Now that those
3330 ** The system tries to make sure that its low-priority messages
3331 are prefixed by semicolons, to help people who like to use
3332 syntax highlighting in their ILISP buffer. (This patch
3333 was originally due to Raymond Toy.)
3334 ** The system now optimizes INTEGER-LENGTH better, thanks to more
3335 patches originally written by Raymond Toy.
3336 ** The compiler understands coercion between single-value and
3337 multiple-VALUES type expressions better, getting rid of some very
3338 weird behavior, thanks to patches originally by Robert MacLachlan
3339 and Douglas Crosher.
3340 ** The system understands ANSI-style non-KEYWORD &KEY arguments in
3341 lambda lists, thanks to a patch originally by Pierre Mai.
3342 ** The system no longer bogusly warns about "abbreviated type
3344 ** The compiler gets less confused by inlining and RETURN-FROM,
3345 thanks to some patches originally by Tim Moore.
3346 ** The system no longer hangs when dumping circular lists to fasl
3347 files, thanks to a patch originally from Douglas Crosher.
3348 * Martin Atzmueller also fixed ROOM, so that it no longer fails with an
3349 undefined function error.
3350 * gave up on fixing bug 3 (forbidden-by-ANSI warning for type mismatch
3351 in structure slot initforms) for now, documented workaround instead:-|
3352 * fixed bug 4 (no WARNING for DECLAIM FTYPE of slot accessor function)
3353 * fixed bug 5: added stubs for various Gray stream functions called
3354 in the not-a-CL:STREAM case, so that even when Gray streams aren't
3355 installed, at least appropriate type errors are generated
3356 * fixed bug 8: better reporting of various PROGRAM-ERRORs
3357 * fixed bug 9: IGNORE and IGNORABLE now work reasonably and more
3358 consistently in DEFMETHOD forms.
3359 * removed bug 21 from BUGS, since Martin Atzmueller points out that
3360 it doesn't seem to affect SBCL after all
3361 * The C runtime system now builds with better optimization and many
3362 fewer warnings, thanks to lots of cleanups by Martin Atzmueller.
3364 changes in sbcl-0.6.8 relative to sbcl-0.6.7:
3366 * The system is now under CVS at SourceForge (instead of the
3367 CVS repository on my home machine).
3368 * The new signal handling code has been tweaked to treat register
3369 contents as (UNSIGNED-BYTE 32), as the old CMU CL code did,
3370 instead of (SIGNED-BYTE 32), as the C header files have it. (Code
3371 downstream, e.g. in debug-int.lisp, has implicit dependencies
3372 on the unsignedness of integer representation of machine words,
3373 and that caused the system to bomb out with infinite regress
3374 when trying to recover from type errors involving signed values,
3375 e.g. (BUTLAST '(1 2 3) -1).)
3376 * (BUTLAST NIL) and (NBUTLAST NIL) now return NIL as they should.
3377 (This was one of the bugs Peter Van Eynde reported back in July.)
3378 * The system now uses code inspired by Colin Walters' O(N)
3379 implementation of MAP (from the cmucl-imp@cons.org mailing
3380 list, 2 September 2000) when it can't use a DEFTRANSFORM to
3381 inline the MAP operation, and there is more than one
3382 sequence argument to the MAP call (so that it can't just
3383 do ETYPECASE once and for all based on the type of the
3384 single sequence argument). (The old non-inline implementation
3385 of the general M-argument sequence-of-length-N case required
3386 O(M*N*N) time when any of the sequence arguments were LISTs.)
3387 * The QUIT :UNIX-CODE keyword argument has been renamed to
3388 QUIT :UNIX-STATUS. (The old name still works, but is deprecated.)
3389 * Raymond Wiker's patches to port RUN-PROGRAM from CMU CL to SBCL
3391 * Raymond Wiker's patches to port dynamic loading from Linux to
3392 FreeBSD have been added.
3393 * The BUGS file is now more nearly up to date, thanks in large part
3394 to Martin Atzmueller's review of it.
3395 * The debugger now flushes standard output streams before it begins
3396 its output ("debugger invoked" and so forth).
3397 * The core version number and fasl file version number have both
3398 been incremented, because of incompatible changes in the layout
3400 * FINISH-OUTPUT is now called more consistently on QUIT. (It
3401 used to not be called for a saved Lisp image.)
3402 * Martin Atzmueller's version of a patch to fix a compiler crash,
3403 as posted on sbcl-devel 13 September 2000, has been installed.
3404 * Instead of installing Martin Atzmueller's patch for the
3405 compiler transform for SUBSEQ, I deleted the compiler transform,
3406 and transforms for some similar consing operations.
3407 * A bug in signal handling which kept TRACE from working on OpenBSD
3409 * added enough DEFTRANSFORMs to allow (SXHASH 'FOO) to be optimized
3410 away by constant folding
3411 * The system now defines its address space constants in one place
3412 (in the Lisp sources), and propagates them automatically elsewhere
3413 (through GENESIS and the sbcl.h file). Therefore, patching the
3414 address map is less unnecessarily tedious and error-prone. The
3415 Lisp names of address space constants have also been systematized.
3416 * CVS tags like dollar-Header-dollar have been removed from
3417 the sources, because they have never saved me trouble and
3418 they've been source of trouble working with patches and other
3419 diff-related operations.
3420 * fixed the PROG1-vs.-PROGN bug in HANDLER-BIND (reported by
3421 ole.rohne@cern.ch on cmucl-help@cons.org 2000-10-25)
3423 changes in sbcl-0.6.7 relative to sbcl-0.6.6:
3425 * The system has been ported to OpenBSD.
3426 * The system now compiles with a simple "sh make.sh" on the systems
3427 that it's supported on. I.e., now you no longer need to tweak
3428 text in the target-features.lisp-expr and symlinks in src/runtime/
3429 by hand, the make.sh takes care of it for you.
3430 * The system is no longer so grossly inefficient when compiling code
3431 involving vectors implemented as general (not simple) vectors (VECTOR T),
3432 so code which dares to use VECTOR-PUSH-EXTEND and FILL-POINTER, or
3433 which dares to use the various sequence functions on non-simple
3434 vectors, takes less of a performance hit.
3435 * There is now a primitive type predicate VECTOR-T-P
3436 to test for the (VECTOR T) type, so that e.g.
3437 (DEFUN FOO (V) (DECLARE (TYPE (VECTOR T) V)) (AREF V 3))
3438 can now be compiled with some semblance of efficiency. (The old code
3439 turned the type declaration into a full call to %TYPEP at runtime!)
3440 * AREF on (VECTOR T) is still not fast, since it's still compiled
3441 as a full call to SB-KERNEL:DATA-VECTOR-REF, but at least the
3442 ETYPECASE used in DATA-VECTOR-REF is now compiled reasonably
3443 efficiently. (The old version made full calls to SUBTYPEP at runtime!)
3444 * (MAKE-ARRAY 12 :FILL-POINTER T) is now executed less inefficiently,
3445 without making full calls to SUBTYPEP at runtime.
3446 (Some analogous efficiency issues for non-simple vectors specialized to
3447 element types other than T, or for non-simple multidimensional arrays,
3448 have not been addressed. They could almost certainly be handled the
3449 same way if anyone is motivated to do so.)
3450 * The changes in array handling break binary compatibility, so
3451 *BACKEND-FASL-FILE-VERSION* has been bumped to 4.
3452 * (TYPEP (MAKE-ARRAY 12 :FILL-POINTER 4) 'VECTOR) now returns (VALUES T)
3453 instead of (VALUES T T).
3454 * By following the instructions that Dan Barlow posted to sbcl-devel
3455 on 2 July 2000, I was able to enable primitive dynamic object
3456 file loading code for Linux. The full-blown CMU CL LOAD-FOREIGN
3457 functionality is not implemented (since it calls ld to resolve
3458 library references automatically, requiring RUN-PROGRAM for its
3459 implementation), but a simpler SB-EXT:LOAD-1-FOREIGN (which doesn't
3460 try to resolve library references) is now supported.
3461 * The system now flushes the standard output streams when it terminates,
3462 unless QUIT is used with the RECKLESSLY-P option set. It also flushes
3463 them at several other probably-convenient times, e.g. in each pass of
3464 the toplevel read-eval-print loop, and after evaluating a form given
3465 as an "--eval" command-line option. (These changes were motivated by a
3466 discussion of stream flushing issues on cmucl-imp in August 2000.)
3467 * The source transform for TYPEP of array types no longer assumes
3468 that an array whose element type is a not-yet-defined type
3469 is implemented as an array of T, but instead punts, so that the
3470 type will be interpreted at runtime.
3471 * There is now some support for cross-compiling in make.sh: each of
3472 the phases of make.sh has its own script. (This should be transparent
3473 to people doing ordinary, non-cross-compile builds.)
3474 * Since my laptop doesn't have hundreds of megabytes of memory like
3475 my desktop machine, I became more motivated to do some items on
3476 my to-do list in order to reduce the size of the system a little:
3477 ** Arrange for various needed-only-at-cold-init things to be
3478 uninterned after cold init. To support this, those things have
3479 been renamed from FOO and *FOO* to !FOO and *!FOO* (i.e., all
3480 symbols with such names are now uninterned after cold init).
3481 ** Bind SB!C::*TOP-LEVEL-LAMBDA-MAX* to a nonzero value when building
3482 fasl files for cold load.
3483 ** Remove the old compiler structure pooling code (which used to
3484 be conditional on the target feature :SB-ALLOC) completely.
3485 ** Redo the representation of some data in cold init to be more compact.
3486 (I also looked into supporting byte compiled code at bootstrap time,
3487 which would probably reduce the size of the system a lot, but that
3488 looked too complicated, so I punted for now.)
3489 * The maximum signal nesting depth in the src/runtime/ support code has
3490 been reduced from 4096 to 256. (I don't know any reason for the very
3491 large old value. If the new smaller value turns out to break something,
3492 I'll probably just bump it back up.)
3493 * PPRINT-LOGICAL-BLOCK is now pickier about the types of its arguments,
3495 * Many, many bugs reported by Peter Van Eynde have been added to
3496 the BUGS list; some have even been fixed.
3497 * While enabling dynamic object file loading, I tried to make the
3498 code easier to understand, renaming various functions and variables
3499 with less ambiguous names, and changing some function calling
3500 conventions to be Lispier (e.g. returning NIL instead of 0 for failure).
3501 * While trying to figure out how to do the OpenBSD port, I tried to
3502 clean up some of the code in src/runtime/. In particular, I dropped
3503 support for non-POSIX signal handling, added various comments,
3504 tweaked the code to reduce the number of compilation warnings, and
3505 renamed some files to increase consistency.
3506 * To support the new automatic configuration functionality in make.sh,
3507 the source file target-features.lisp-expr has been replaced with the
3508 source file base-target-features.lisp-expr and the machine-generated
3509 file local-target-features.lisp-expr.
3510 * fixed a stupid quoting error in make.sh so that using CMU CL
3511 "lisp -batch" as cross-compilation host works again
3513 changes in sbcl-0.6.6 relative to sbcl-0.6.5:
3515 * DESCRIBE no longer tries to call itself recursively to describe
3516 bound/fbound values, so that it no longer fails on symbols which are
3517 bound to themselves (like keywords, T, and NIL).
3518 * DESCRIBE now works on generic functions.
3519 * The printer now prints less-screwed-up representations of closures
3520 (not naively trying to bogusly use the %FUNCTION-NAME accessor on them).
3521 * A private symbol is used instead of the :EMPTY keyword previously
3522 used to mark empty slots in hash tables. Thus
3523 (DEFVAR *HT* (MAKE-HASH-TABLE))
3524 (SETF (GETHASH :EMPTY *HT*) :EMPTY)
3525 (MAPHASH (LAMBDA (K V) (FORMAT T "~&~S ~S~%" K V)))
3526 now does what ANSI says that it should. (You can still get
3527 similar noncompliant behavior if bang on the hash table
3528 implementation with all the symbols you get back from
3529 DO-ALL-SYMBOLS, but at least that's a little harder to do.)
3530 This breaks binary compatibility, since tests for equality to
3531 :EMPTY are wired into things like the macroexpansion of
3532 WITH-HASH-TABLE-ITERATOR in FASL files produced by earlier
3534 * There's now a minimal placeholder implementation for CL:STEP,
3535 as required by ANSI.
3536 * An obscure bug in the interaction of the normal compiler, the byte
3537 compiler, inlining, and structure predicates has been patched
3538 by setting the flags for the DEFTRANSFORM of %INSTANCE-TYPEP as
3539 :WHEN :BOTH (as per Raymond Toy's suggestion on the cmucl-imp@cons.org
3541 * Missing ordinary arguments in a macro call are now detected even
3542 when the macro lambda list contains &KEY or &REST.
3543 * The debugger no longer complains about encountering the top of the
3544 stack when you type "FRAME 0" to explicitly instruct it to go to
3545 the top of the stack. And it now prints the frame you request even
3546 if it's the current frame (instead of saying "You are here.").
3547 * As specified by ANSI, the system now always prints keywords
3548 as #\: followed by SYMBOL-NAME, even when *PACKAGE* is the
3550 * The default initial SIZE of HASH-TABLEs is now smaller.
3551 * Type information from CLOS class dispatch is now propagated
3552 into DEFMETHOD bodies, so that e.g.
3553 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3555 is now basically equivalent to
3556 (DEFMETHOD FOO ((X SINGLE-FLOAT))
3557 (DECLARE (TYPE SINGLE-FLOAT X))
3559 and the compiler can compile (+ X 123.0) as a SINGLE-FLOAT-only
3560 operation, without having to do run-time type dispatch.
3561 * The macroexpansion of DEFMETHOD has been tweaked so that it has
3562 reasonable behavior when arguments are declared IGNORE or IGNORABLE.
3563 * Since I don't seem to be making big file reorganizations very often
3564 any more (and since my archive of sbcl-x.y.zv.tar.bz2 snapshots
3565 is overflowing my ability to conveniently back them up), I've finally
3566 checked the system into CVS. (The CVS repository is on my home system,
3567 not at SourceForge -- putting it on SourceForge might come later.)
3568 * SB-EXT:*GC-NOTIFY-STREAM* has been added, to control where the
3569 high-level GC-NOTIFY-FOO functions send their output. (There's
3570 still very little control of where low-level verbose GC functions
3571 send their output.) The SB-EXT:*GC-VERBOSE* variable now controls
3572 less than it used to -- the GC-NOTIFY-FOO functions are now under
3573 the control of *GC-NOTIFY-STREAM*, not *GC-VERBOSE*.
3574 * The system now stores the version string (LISP-IMPLEMENTATION-VERSION)
3575 in only one place in the source code, and propagates it automatically
3576 everywhere that it's needed. Thus e.g. when I bump the version from
3577 0.6.6 to 0.6.7, I'll only need to modify the sources in one place.
3578 * The C source files now include boilerplate legalese and documentation
3579 at the head of each file (just as the Lisp source files already did).
3580 * At Dan Barlow's suggestion, the hyperlink from the SBCL website
3581 to his page will be replaced with a link to his new CLiki service.
3583 changes in sbcl-0.6.5 relative to sbcl-0.6.4:
3585 * Raymond Wiker's patches to port the system to FreeBSD have been merged.
3586 * The build process now looks for GNU make under the default name "gmake",
3587 instead of "make" as it used to. If GNU make is not available as "gmake"
3588 on your system, you can change this default behavior by setting the
3589 GNUMAKE environment variable.
3590 * Replace #+SB-DOC with #!+SB-DOC in seq.lisp so that the system
3591 can build without error under CMU CL.
3593 changes in sbcl-0.6.4 relative to sbcl-0.6.3:
3595 * There is now a partial SBCL user manual (with some new text and some
3596 text cribbed from the CMU CL manual).
3597 * The beginnings of a profiler have been added (starting with the
3598 CMU CL profiler and simplifying and cleaning up). Eventually the
3599 main interface should be through the TRACE macro, but for now,
3600 it's still accessed through vaguely CMU-CL-style functions and macros
3601 exported from the package SB-PROFILE.
3602 * Some problems left over from porting CMU CL to the new
3603 cross-compilation bootstrap process have been cleaned up:
3604 ** DISASSEMBLE now works. (There was a problem in using DEFMACRO
3605 instead of SB!XC:DEFMACRO, compounded by an oversight on my
3606 part when getting rid of the compiler *BACKEND* stuff.)
3607 ** The value of *NULL-TYPE* was screwed up, because it was
3608 being initialized before the type system knew the final
3609 definition of the 'NULL type. This screwed up several key
3610 optimizations in the compiler, causing inefficiency in all sorts
3611 of places. (I found it because I wanted to understand why
3612 GET-INTERNAL-RUN-TIME was consing.)
3613 * fixed a bug in DEFGENERIC which was causing it to overwrite preexisting
3614 PROCLAIM FTYPE information. Unfortunately this broke binary
3615 compatibility again, since now the forms output by DEFGENERIC
3616 to refer to functions which didn't exist in 0.6.3.
3617 * added declarations so that SB-PCL::USE-CACHING-DFUN-P
3618 can use the new (as of 0.6.3) transform for SOME into MAP into
3620 * changed (MOD 1000000) type declarations for Linux timeval.tv_usec slot
3621 values to (INTEGER 0 1000000), so that the time code will no longer
3622 occasionally get blown up by Linux returning 1000000 microseconds
3623 * PRINT-UNREADABLE-OBJECT has been tweaked to make the spacing of
3624 its output conform to the ANSI spec. (Alas, this makes its output
3625 uglier in the :TYPE T :IDENTITY NIL case, but them's the breaks.)
3626 * A full call to MAP NIL with a single sequence argument no longer conses.
3627 * fixes to problems pointed out by Martin Atzmueller:
3628 * The manual page no longer talks about multiprocessing as though
3629 it were currently supported.
3630 * The ILISP support patches have been removed from the distribution,
3631 because as of version 5.10.1, ILISP now supports SBCL without us
3632 having to maintain patches.
3633 * added a modified version of Raymond Toy's recent CMU CL patch for
3634 EQUALP comparison of HASH-TABLE
3636 changes in sbcl-0.6.3 relative to sbcl-0.6.2:
3638 * The system still can't cross-compile itself with
3639 *TYPE-SYSTEM-INITIALIZED* (and all the consistency checks that
3640 entails), but at least it can compile more of itself that way
3641 than it used to be able to, and various buglets which were uncovered
3642 by trying to cross-compile itself that way have now been fixed.
3643 * This release breaks binary compatibility again. This time
3644 at least I've incremented the FASL file format version to 2, so that the
3645 problem can be detected reliably instead of just causing weird errors.
3646 * various new style warnings:
3647 ** using DEFUN, DEFMETHOD, or DEFGENERIC to overwrite an old definition
3648 ** using the deprecated EVAL/LOAD/COMPILE situation names in EVAL-WHEN
3649 ** using the lexical binding of a variable named in the *FOO* style
3650 * DESCRIBE has been substantially rewritten. It now calls DESCRIBE-OBJECT
3651 as specified by ANSI.
3652 * *RANDOM-STATE* is no longer automatically initialized from
3653 (GET-UNIVERSAL-TIME), but instead from a constant seed. Thus, the
3654 default behavior of the system is to repeat its behavior every time
3655 it's run. If you'd like to change this behavior, you can always
3656 explicitly set the seed from (GET-UNIVERSAL-TIME); whereas under the
3657 old convention there was no comparably easy way to get the system to
3658 repeat its behavior every time it was run.
3659 * Support for the pre-CLTL2 interpretation of FUNCTION declarations as
3660 FTYPE declarations has been removed, in favor of their ANSI
3661 interpretation as TYPE FUNCTION declarations. (See p. 228 of CLTL2.)
3662 * The quantifiers SOME, EVERY, NOTANY, and NOTEVERY no longer cons when
3663 the types of their sequence arguments can be determined at compile time.
3664 This is done through a new open code expansion for MAP which eliminates
3665 consing for (MAP NIL ..), and reduces consing otherwise, when sequence
3666 argument types can be determined at compile time.
3667 * The optimizer now transforms COERCE into an identity operation when it
3668 can prove that the coerced object is already of the correct type. (This
3669 can be a win for machine generated code, including the output of other
3670 optimization transforms, such as the MAP transform above.)
3671 * Credit information has been moved from source file headers into CREDITS.
3672 * Source file headers have been made more standard.
3673 * The CASE macro now compiles without complaining even when it has
3676 changes in sbcl-0.6.2 relative to sbcl-0.6.1:
3678 * (Note that the way that the PCL macroexpansions were rewritten
3679 to accommodate the change in DEFGENERIC below breaks binary
3680 compatibility. That is, fasl files compiled under sbcl-0.6.1 may
3681 not run under sbcl-0.6.2. Once we get out of alpha releases,
3682 i.e. hit release 1.0.0, we'll probably try to maintain binary
3683 compatibility between maintenance releases, e.g. between sbcl-1.4.3
3684 and sbcl-1.4.4. Until then, however, it might be fairly common
3685 for maintenance releases to break binary compatibility.)
3686 * A bug in the calculation of WARNINGS-P and FAILURE-P in COMPILE-FILE
3688 * The reporting of unhandled signals has been changed to print some
3689 explanatory text as well as the report form. (Previously only
3690 the report form was printed.)
3691 * The macroexpansion for DEFGENERIC now DECLAIMs the function that
3692 it defines, so that the compiler no longer issues undefined function
3693 warnings for compiled-but-not-yet-loaded generic functions.
3694 * The CLTL-style "LISP" and "USER" nicknames for the "COMMON-LISP"
3695 and "COMMON-LISP-USER" packages have been removed. Now only the "CL"
3696 and "CL-USER" standard nicknames from the "11.1.2 Standardized Packages"
3697 section of the ANSI spec are supported.
3698 * The "" nickname for the "KEYWORD" package has been removed.
3699 The reader still handles symbol tokens which begin with a package marker
3700 as keywords, but it doesn't expose its mechanism for doing so in the
3701 (PACKAGE-NICKNAMES (FIND-PACKAGE "KEYWORD")) list.
3702 * The system now issues STYLE-WARNINGs for contradictory TYPE
3703 proclamations. (Warnings for contradictory FTYPE proclamations would
3704 be nice too, but those can't be done usefully unless the type system
3705 is made smarter about FUNCTION types.)
3706 * The names of source files "*host-*.lisp" and "*target-*.lisp" have been
3707 systematized, so that "*target-*.lisp is supposed to exist only on the
3708 target and imply that there's a related file which exists on the
3709 host, and *host-*.lisp is supposed to exist only on the host and imply
3710 that there's a related file which exists on the target. This involves a
3711 lot of renaming. Hopefully the acute confusion caused by the renaming
3712 will be justified by the reduction in chronic confusion..
3713 ** runtime-type.lisp -> early-target-type.lisp
3714 ** target-type.lisp -> late-target-type.lisp
3715 ** early-host-format.lisp -> early-format.lisp
3716 ** late-host-format.lisp -> late-format.lisp
3717 ** host-error.lisp -> misc-error.lisp
3718 ** early-error.lisp -> early-target-error.lisp
3719 ** late-error.lisp -> late-target-error.lisp
3720 ** host-defboot.lisp -> early-defboot.lisp
3721 ** code/misc.lisp -> code/target-misc.lisp
3722 ** code/host-misc.lisp -> code/misc.lisp
3723 ** code/numbers.lisp -> code/target-numbers.lisp
3724 ** code/early-numbers.lisp -> numbers.lisp
3725 ** early-host-type.lisp -> early-type.lisp
3726 ** late-host-type.lisp -> late-type.lisp
3727 ** host-typep.lisp -> typep.lisp
3728 ** load.lisp -> target-load.lisp
3729 ** host-load.lisp -> load.lisp
3730 ** host-disassem.lisp -> disassem.lisp
3731 ** host-insts.lisp -> insts.lisp
3732 ** byte-comp.lisp -> target-byte-comp.lisp
3733 ** host-byte-comp.lisp -> byte-comp.lisp
3734 ** host-signal.lisp -> signal.lisp
3735 ** host-defstruct.lisp -> defstruct.lisp
3736 ** late-target-type.lisp -> deftypes-for-target.lisp
3737 Furthermore, several other previously target-only files foo.lisp (e.g.
3738 hash-table.lisp and random.lisp) have been split into a target-and-host
3739 foo.lisp file and a target-only target-foo.lisp file, with their key type
3740 definitions in the target-and-host part, so that the cross-compiler will
3741 know more about target types.
3742 * DEFSTRUCT BACKEND, and the BACKEND-valued *BACKEND* variable, have
3743 gone away. In their place are various *BACKEND-FOO* variables
3744 corresponding to the slots of the old structure.
3745 * A bug which caused the SB-COLD bootstrap-time package to be propagated
3746 into the target SBCL has been fixed.
3747 * The chill.lisp system for loading cold code into a running SBCL
3749 * Support for the CMU CL "scavenger hook" extension has been removed.
3750 (It was undocumented and unused in the CMU CL sources that SBCL was
3751 derived from, and stale in sbcl-0.6.1.)
3752 * Various errors in the cross-compiler type system were detected
3753 by running the cross-compiler with *TYPE-SYSTEM-INITIALIZED*
3754 (enabling various consistency checks). Many of them were fixed,
3755 but some hard problems remain, so the compiler is back to
3756 running without *TYPE-SYSTEM-INITIALIZED* for now.
3757 * As part of the cross-compiler type system cleanup, I implemented
3758 DEF!TYPE and got rid of early-ugly-duplicates.lisp.
3759 * I have started adding UNCROSS calls throughout the type system
3760 and the INFO database. (Thus perhaps eventually the blanket UNCROSS
3761 on cross-compiler input files will be able to go away, and various
3763 * CONSTANTP now returns true for quoted forms (as explicitly required
3766 changes in sbcl-0.6.1 relative to sbcl-0.6.0:
3768 * changed build optimization from (SAFETY 1) to (SAFETY 3) as a short-term
3769 fix for various type-unsafety bugs, e.g. failures with (LENGTH 123) and
3770 (MAKE-LIST -1). In the longer term, it ought to become true
3771 that declarations are assertions even at SAFETY 1. For now, it's not
3772 quite true even at SAFETY 3, but it's at least more nearly true..
3773 (Note that this change seems to increases the size of the system by
3774 O(5%) and to decrease the speed of the compiler by 20% or more.)
3775 * changed ALIEN printing to be much more abbreviated, as a short-term fix
3776 for the problem of printing dozens of lines of distracting information
3777 about low-level system machinery as part of the top stack frame
3778 on entry to the debugger when an undefined function was called.
3779 * tweaked the debugger's use of WITH-STANDARD-IO-SYNTAX so that *PACKAGE*
3780 is not reset to COMMON-LISP-USER.
3781 * Compilation of stuff related to dyncount.lisp has been made conditional
3782 on the :SB-DYNCOUNT target feature, so that the ordinary core system is
3783 smaller. The various dyncount-related symbols have been moved into
3784 a new "SB-DYNCOUNT" package.
3785 * tty-inspect.lisp has been renamed to inspect.lisp.
3786 * unix-glibc2.lisp has been renamed to unix.lisp, and the :GLIBC2
3787 feature has gone away. (When we eventually port to other flavors of
3788 libc and/or Unix, we'll try to make the differences between flavors
3789 invisible at the user level.)
3790 * Various other *FEATURES* tags, and/or their associated conditionals,
3791 have been removed if obsolescent, or given better documentation, or
3792 sometimes given more-mnemonic names.
3794 changes in sbcl-0.6.0 relative to sbcl-0.5.0:
3796 * tidied up "make.sh" script
3797 * tidied up system directory structure
3798 * better "clean.sh" behavior
3799 * added doc/FOR-CMUCL-DEVELOPERS
3800 * many many small tweaks to output format, e.g. removing possibly-confusing
3801 trailing #\. character in DESCRIBE-INSTANCE
3802 * (EQUALP #\A 'A) no longer signals an error.
3803 * new hashing code, including EQUALP hashing
3804 * tidied up Lisp initialization and toplevel
3805 * initialization files (e.g. /etc/sbclrc and $HOME/.sbclrc)
3806 * command line argument processing
3807 * added POSIX-GETENV function to deal with Unix-ish environment variables
3808 * more-Unixy handling of *STANDARD-INPUT* and other Lisp streams, e.g.
3809 terminating SBCL on EOF
3810 * non-verbose GC by default
3811 * There is no more "sbcl" shell script; the sbcl file is now the C
3812 runtime executable (just like CMU CL).
3813 * removed some unused fops, e.g. FOP-UNIFORM-VECTOR, FOP-CHARACTER, and
3815 * tweaked debug-info.lisp and debug-int.lisp to make the debugger store
3816 symbol and package information as Lisp native symbol and package objects
3817 instead of strings naming symbols and strings naming packages. This way,
3818 whenever packages are renamed (as in warm init), debug information is
3819 transformed along with everything else.
3820 * tweaked the optimization policy declarations which control the building
3821 of SBCL itself. Now, among other things, the system no longer saves
3822 source location debugging information. (This helps two problems at once
3823 by reducing SBCL size and by keeping SBCL from trying to look for its
3824 sources -- which may not exist -- when reporting errors.)
3825 * added src/cold/chill.lisp, to let SBCL read its own cold sources for
3826 debugging and testing purposes
3827 * cleaned up printing, making the printer call PRINT-OBJECT for
3828 instances, and using PRINT-UNREADABLE-OBJECT for most PRINT-OBJECT
3829 methods, giving nearly-ANSI behavior
3830 * converted almost all special variables to use *FOO* naming convention
3831 * deleted PARSE-TIME functionality, since it can be done portably
3832 * moved some files out of cold init into warm init
3833 * deleted DEFUN UNDEFINED-VALUE, replaced (UNDEFINED-VALUE) forms
3835 * regularized formatting of source files
3836 * added an install.sh script
3837 * fixed ridiculous memory usage of cross-compiler by making
3838 compiler/alloc.lisp not try to do pooling unless it can hook
3839 itself into the GC of the cross-compilation host. Now the system
3840 builds nicely on my old laptop.
3841 * added :SB-ALLOC in target-features.lisp-expr
3842 * deleted mention of :ANSI-DOC from target-features.lisp-expr (since it
3843 was not implemented)
3844 * re-did condition handling and note reporting in the compiler. Notes
3845 are no longer handled by signalling conditions. Style warnings
3846 and warnings are handled more correctly and reported in such a way
3847 that it's easy to find one or the other in your output (so that you
3848 can e.g. figure out which of many problems caused COMPILE-FILE to
3850 * changed the severity of several compiler warnings from full WARNING
3851 to STYLE-WARNING in order to conform with the ANSI spec; also changed
3852 compiler note reporting so that it doesn't use the condition system
3853 at all (and hence affects neither FAILURE-P nor WARNINGS-P in the
3854 COMPILE-FILE command)
3855 * made PROCLAIM and DECLAIM conform to ANSI. PROCLAIM is now an ordinary
3856 function. As a consequence, START-BLOCK and END-BLOCK declarations are
3857 no longer supported, since their implementation was deeply intertwingled
3858 with the magical, non-ANSI treatment that PROCLAIM received in CMU CL.
3859 * removed bogus "support" for compiler macros named (SETF FOO), and
3860 removed the compiler macro for SETF INFO (but only after making a fool
3861 of myself on the cmucl-imp mailing list by posting a bogus patch for
3862 DEFINE-COMPILER-MACRO..)
3863 * Compiled files containing forms which have side effects on the Lisp
3864 reader (such as DEFPACKAGE forms) are now handled more correctly.
3865 (Compiler queuing of top level lambdas has been suppressed by setting
3866 *TOP-LEVEL-LAMBDA-MAX* to 0. )
3867 * deleted various currently-unused source files, e.g. gengc.lisp. They
3868 may be added back at some point e.g. when porting to other architectures,
3869 but until they are it's distracting to distribute them and to try to
3871 * deleted "UNCROSS couldn't recurse through.." style warnings, since
3872 there were so many of them they're just distractions, and UNCROSS is
3873 known to be able to handle the current sources
3874 * moved PROFILE functionality into TRACE, so that it will be clear
3875 how the wrapping and unwrapping of functions when you profile them
3876 interacts with the wrapping and unwrapping of functions when you
3877 trace them. (Actually, the functionality isn't there yet, but at least
3878 the interface specification is there. Hopefully, the functionality will
3879 arrive with some future maintenance release.)
3880 * removed host-oops.lisp
3881 * changed signature of QUIT function to allow UNIX-CODE argument
3882 * fixed READ-SEQUENCE bug
3883 * tweaked verbose GC output so that it looks more like the progress
3884 output that ANSI specifies for functions like LOAD
3885 * set up the system on sourceforge.com, with home pages, mailing lists, etc.
3886 * added <http://sbcl.sourceforge.com> to the banner information printed by